Current Protocols in Protein Science: 001

Current Protocols in Protein Science Edited by: John E. Coligan (Bethesda, Maryland); Ben M. Dunn (University of Florida...

Author: COLIGAN

63 downloads 3253 Views 78MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Current Protocols in Protein Science Edited by: John E. Coligan (Bethesda, Maryland); Ben M. Dunn (University of Florida); David W. Speicher (The Wistar Institute); Paul T. Wingfield (Bethesda, Maryland); Past Editor: Hidde L. Ploegh Series Editor: Gwen Taylor

Copyright © 2007 by John Wiley and Sons, Inc. ISBN: 978-0-471-11184-9 0-471-11184-8 - Looseleaf 0-471-14098-8 - CD-ROM Last updated: 26 Apr 2007

PREFACE roteins are like people: they are highly diverse, with very few traits shared by the entire population. Even close relatives often have very different structural (physical) characteristics and behave quite differently. This heterogeneity presents challenges for scientists interested in studying protein structure and function. The need for a readily accessible, updatable reference for protein methods has grown in tandem with the ongoing protein science renaissance driven by advances in molecular biology, genetics, structural biology, and related disciplines. Using recombinant methods, it is now practical to overexpress both normal and reengineered forms of a protein and to create chimeric proteins with unique new properties. The ability to express proteins usually found in low abundance at high levels and to manipulate their amino acid sequences provides unprecedented opportunities to study a vast array of proteins and their associated biological processes (both in vitro and in vivo). The combination of biochemical, biophysical, and high-resolution structural analyses of proteins with manipulation of the associated genes (in cultured cells or an entire mammalian organism) has already led to spectacular advances in many areas of biotechnology and molecular medicine.

P

Because proteins are so diverse, a wide range of techniques must often be considered and tested empirically for a particular protein, and each researcher’s arsenal must include both recent innovations and techniques that have been used for many years. Moreover, methods appropriate for purifying and initially characterizing the protein(s) associated with a biological activity of interest are often not suitable for analyzing the corresponding overexpressed recombinant protein. This manual includes both specific detailed protocols as well as strategies for adapting methods to particular projects at hand. It is designed to meet the needs of scientists with little prior experience of protein isolation and characterization, including graduate students and scientists trained in other biological disciplines. At the same time, the broad range of techniques presented here should ensure that even seasoned experts will find new and useful approaches and will benefit from the manual’s convenient compilation of standard information and methods that typically must be culled from many sources. The quarterly updating service will allow ongoing expansion of the core volume to include chapters covering critical topics that were deferred due to initial space limitations and to keep pace with the “cutting edge” of developments in this fast-moving field.

HOW TO USE THIS MANUAL Format and Organization This publication is available in looseleaf, CD-ROM and online formats. For looseleaf subscribers, binders are provided to accommodate the growth of the manual via the quarterly updates. The looseleaf binder format allows easy insertion of new pages, units, and chapters that are added. The index and table of contents are updated with each supplement. CD-ROM subscribers receive a completely new disc every quarter and should dispose of their outdated discs. The online version is available through institutional license and offers many useful features such as downloadable PDF files of units, enhanced text searching for contents from all Current Protocols titles, real-time updating, and superior color graphics. The material covered in all three versions is identical. Subjects in this manual are organized by chapters and sections, and protocols are contained in units. Units generally describe a method and include one or more protocols with

Current Protocols in Protein Science

iii Current Protocols in Protein Science

Supplement 41

listings of materials, steps and annotations, recipes for unique reagents and solutions, and commentaries on the “hows” and “whys” of the method; there are also “overview” units containing theoretical discussions that lay the foundation for subsequent protocols. Page numbering in the looseleaf version reflects the modular arrangement by unit; for example, page 2.3.5 refers to Chapter 2 (Computational Analysis), UNIT 2.3 (Protein Secondary Structure Prediction), page 5 of that particular unit. Many reagents and procedures are employed repeatedly throughout the manual. Instead of duplicating this information, cross-references among units are used extensively. Crossreferencing helps to ensure that lengthy and complex protocols are not overburdened with steps describing auxiliary procedures needed to prepare raw materials and analyze results. Certain units that describe commonly used techniques and recipes (e.g., gel electrophoresis, ion-exchange chromatography) are cross-referenced in other units that describe their application. Thus, whenever it is necessary to isolate or identify a protein band in a protocol, the appropriate unit in Chapter 10—describing various procedures for gel electrophoresis—is cross-referenced (i.e., UNIT 10.1). For some widely used techniques (such as dialysis), readers are cross-referenced to APPENDIX 3.

Introductory and Explanatory Information Because this publication is first and foremost a compilation of laboratory techniques in protein science, we have not provided extensive instructive material. We have, however, included explanatory information where required to help readers gain an intuitive grasp of the procedures. Chapter 1 provides an overview of strategies for protein purification and characterization, to help the reader plan for all aspects of an experimental study. Some subsequent chapters begin with special overview units that describe the state of the art of the topic matter and provide a context for the procedures that follow. Chapter and unit introductions describe how the protocols that follow connect to one another, and annotations to the actual protocol steps describe what is happening as a procedure is carried out. Finally, the Commentary that closes each protocol unit describes background information regarding the historical and theoretical development of the method, as well as alternative approaches, critical parameters, troubleshooting guidelines, anticipated results, and time considerations. All units contain cited references and many indicate key references to inform users of particularly useful background reading, original descriptions, or applications of a technique. Protocols Many units in the manual contain groups of protocols, each presented with a series of steps. The basic protocol is presented first in each unit and is generally the recommended or most universally applicable approach. Alternate protocols are provided where different equipment or reagents can be employed to achieve similar ends, where the starting material requires a variation in approach, or where requirements for the end product differ from those in the basic protocol. Support protocols describe additional steps that are required to perform the basic or alternate protocols; these steps are separated from the core protocol because they might be applicable to other uses in the manual, or because they are performed in a time frame separate from the basic protocol steps.

Preface

Reagents and Solutions Reagents required for a protocol are itemized in the materials list before the procedure begins. Many are common stock solutions, others are commonly used buffers or media, while others are solutions unique to a particular protocol. Recipes for the latter solutions are provided in each unit, following the protocols (and before the commentary) under

iv Supplement 41


the heading Reagents and Solutions. It is important to note that the names of some of these special solutions might be similar from unit to unit (e.g., SDS sample buffer) while the recipes differ; thus, make certain that reagents are prepared from the proper recipes. On the other hand, recipes for commonly used stock solutions and buffers are provided once in APPENDIX 2E. These universal recipes are cross-referenced parenthetically in the materials lists rather than repeated with every usage.

Commercial Suppliers Throughout the manual, we have recommended commercial suppliers of chemicals, biological materials, and equipment. In some cases, the noted brand has been found to be of superior quality or it is the only suitable product available in the marketplace. In other cases, the experience of the author of that protocol is limited to that brand. In the latter situation, recommendations are offered as an aid to the novice protein science experimenter in obtaining the tools of the trade. Experienced investigators are therefore encouraged to experiment with substituting their own favorite brands. Addresses, phone numbers, and facsimile numbers of all suppliers mentioned in this manual are provided in the SUPPLIERS APPENDIX.

Safety Considerations Anyone carrying out these protocols may encounter the following hazardous or potentially hazardous materials: (1) radioactive substances, (2) toxic chemicals and carcinogenic or teratogenic reagents, and (3) pathogenic and infectious biological agents. Readers are referred to APPENDIX 2A and APPENDIX 2B for guidelines regarding use, storage, and disposal of these materials. Although cautionary statements are included in the appropriate units with cross-references to this appendix, we emphasize that users must proceed with the prudence and precaution associated with good laboratory practice, and that all materials must be used in strict accordance with local and national regulations. Reader Response Most of the protocols included in this manual are used routinely in our own laboratories. These protocols work for us; to make them work for you we have annotated critical steps and included critical parameters and troubleshooting guides in the commentaries to most units. However, the successful evolution of this manual depends upon readers’ observations and suggestions. Consequently, a self-mailing reader-response survey can be found at the back of the manual (and is included with each supplement); we encourage readers to send in their comments. Feedback can also be submitted online from the Current Protocols homepage at http://www.currentprotocols.com using the “Contact Us” link. ACKNOWLEDGMENTS This manual is the product of dedicated efforts by many of our scientific colleagues who are acknowledged in each unit and by the hard work by the Current Protocols editorial staff at John Wiley and Sons. The publisher’s commitment and continuing support for a protein science manual, particularly through the efforts of Brian Crawford and Paul Majchrzyk, was essential for realizing this ambitious project. We are extremely grateful for the critical contributions by Virginia Chanda (Series Editor) who kept the editors and the contributors on track and played a key role in bringing the entire project to completion. Other skilled members of the Current Protocols staff who contributed to the project include Janet Blair, Hazel Chan, Elizabeth Konkle, Alice Ro, and Kathy Wisch. The extensive copyediting required to produce an accurate protocols manual was ably handled by Rebecca Barr, Karen Hopkin, Arline Keithe, Kathy Morgan, Connie Parks, and Allen Ranz, and electronic illustrations were prepared by Gae Xavier. Finally, we


v Current Protocols in Protein Science

Supplement 41

would like to thank two people who played important roles in initiating this project. Sarah Greene initially conceived this project and had a critical role in developing it. John A. Smith made important contributions during the early planning phases of the project. Helpful advice and criticisms of the project during early planning stages were provided by Fred Ausubel, Charles S. Craik, Bruce A. Cummingham, Mark Hermodson, Tony E. Hugli, Robert Kingston, Albert Light, Paul Matsudaira, David Moore, George Rose, and Kevin Struhl. Mark Hermodson read and critiqued the entire manual and individual chapters were thoughtfully reviewed by John Freed and John Kay. Finally, we are especially grateful to our many colleagues, who contributed material to this manual, for sharing their procedures and experiences.

RECOMMENDED BACKGROUND READING Branden, C. and Tooze, J. 1999. Introduction to Protein Structure, 2nd edition. Garland Publishing, New York. Easy-to-read overview of basic structural principles of proteins with extensive illustrations. Fersht, A. 1998. Structure and Mechanism in Protein Science: A Guide to Enzyme Catalysis and Protein Folding. 3rd edition. W. H. Freeman. Well established and clear review of protein structure and function, with an emphasis on enzymes. Hunte, C., von Jagow, G. and Schagger, H., editors. 2003. Membrane Protein Purification and Crystallization: A Practical Guide (2nd edition). Academic Press. Deals with specific problems encountered when purifying membrane-associated proteins. Link, A.J. (editor). 1998. 2-D Proteome Analysis Protocols. Humana Press. Extensive treatment of 2-D gel analysis and its role in proteomics, including preparations of samples from diverse sources, alternative 2-D gel analysis methods and data analysis/interpretation. Petsko, G.A. and Ringe, D. 2004. Protein Structure and Function. New Science Press Ltd., London. An updated look at protein structure in the genomics age. Includes discussion of sequence-to-structure and structure-to-function correlations. Speicher, D.W. (editor). 2004. Proteome analysis: Interpreting the Genome. Elsevier. Overview of most current proteomics methods and supporting technologies. Voet, D. and Voet, J.G. 2005. Biochemistry, 3rd edition. John Wiley & Sons. An updated biochemistry textbook whose protein-related chapters provide a comprehensive background in biochemistry.

John E. Coligan, Ben M. Dunn, Hidde L. Ploegh, David W. Speicher, and Paul T. Wingfield

Preface

vi Supplement 41


FOREWORD

O

ver the past five to ten years we have witnessed a revolution in biological research which is fueled by our increasing capacity to decipher biological information of three types: the digital information embedded in DNA with its four-letter alphabet; the three-dimensional information represented by proteins, the major executors of biological function; and the four-dimensional information of complex biological systems and networks representing the temporal and spatial interaction of multiple components. The analysis of complex systems and networks in their entirety is essential for scientists to understand the molecular basis of fascinating processes such as growth, development, and differentiation, and to extract the biological meaning of emergent properties such as consciousness, memory, and the ability to learn which evolved over billions of years. Not surprisingly, the linear nucleotide sequence in DNA has been the first type of information to be deciphered. Large-scale genetic mapping and DNA sequencing rapidly generate enormous amounts of biological information. In many ways, programs such as the human genome project represent some of the earliest attempts to understand biological complexity. Successful completion of the human genome and similar projects will pose the challenge of interpreting the information contained in billions of nucleotides and of explaining how the interplay of the products of perhaps 100,000 genes results in the myriad of biological phenotypes. Meeting these challenges will require new interdisciplinary strategies which draw from the expertise of scientists from disciplines as different as molecular biology, biochemistry, engineering, chemistry, applied mathematics, and computer science. Protein science is central to such integrated strategies. Separation of proteins from complex mixtures, followed by structural and functional analyses, control of protein function, post-translational processing, and modification and formation of macromolecular complexes of proteins and other biomolecules are but a few examples of topics which are essential for all those scientists who attempt to interpret the linear DNA sequence in terms of the three- and four-dimensional information of complex biological systems and networks. The following chapters detail the necessary protocols for this endeavor, provided by experts in experimental protein science. Two attractive features set this manual apart from other collections of research protocols. First, Current Protocols in Protein Science will be continuously updated and expanded by quarterly additions to the core edition. Secondly, this volume as well as the updates are also available on CD-ROM. Numerous cross-references within the manual by hypertext links, context based searching, and provisions for making individualized notebooks containing frequently used protocols are attractions of the CD-ROM version which are in tune with the increasing dependence on computers in biological research laboratories.

Current Protocols in Protein Science vii


Protein science is a diverse, experimentally challenging, and rapidly evolving discipline. This book provides expert guidance in experiment design and execution and promises to remain up-to-date, in contents and in format, for many years to come. Leroy E. Hood1 and Ruedi Aebersold1 1 Seattle, Washington

Foreword viii Current Protocols in Protein Science

Table of Contents

Preface Foreword

Chapter 1 Strategies of Protein Purification and Characterization Unit 1.1 Overview of Protein Purification and Characterization Unit 1.2 Strategies for Protein Purification Unit 1.3 Protein Purification Flow Charts Unit 1.4 Purification of Glutamate Dehydrogenase from Liver and Brain Unit 1.5 Overview of the Physical State of Proteins within Cells

Chapter 2 Computational Analysis Introduction Unit 2.1 Computational Methods for Protein Sequence Analysis Unit 2.2 Hydrophobicity Profiles for Protein Sequence Analysis Unit 2.3 Protein Secondary Structure Prediction Unit 2.4 Internet Basics Unit 2.5 Sequence Similarity Searching Using the BLAST Family of Programs Unit 2.6 Protein Databases on the Internet Unit 2.7 Protein Tertiary Structure Prediction Unit 2.8 Protein Tertiary Structure Modeling Unit 2.9 Comparative Protein Structure Prediction Unit 2.10 Protein Charge Determination Unit 2.11 Basic Protein Sequence Analysis

ix Current Protocols in Protein Science

Unit 2.12 Minimotif Miner: A Computational Tool to Investigate Protein Function, Disease, and Genetic Diversity

Chapter 3 Detection and Assay Methods Introduction Unit 3.1 Spectrophotometric Determination of Protein Concentration Unit 3.2 Quantitative Amino Acid Analysis Unit 3.3 In Vitro Radiolabeling of Peptides and Proteins Unit 3.4 Assays for Determination of Protein Concentration Unit 3.5 Kinetic Assay Methods Unit 3.6 Biotinylation of Proteins in Solution and on Cell Surfaces Unit 3.7 Metabolic Labeling with Amino Acids Unit 3.8 Analysis of Selenocysteine-Containing Proteins Unit 3.9 Solid-Phase Profiling of Proteins

Chapter 4 Extraction, Stabilization, and Concentration Introduction Unit 4.1 Overview of Cell Fractionation Unit 4.2 Purification of Organelles from Mammalian Cells Unit 4.3 Subcellular Fractionation of Tissue Culture Cells Unit 4.4 Desalting, Concentration, and Buffer Exchange by Dialysis and Ultrafiltration Unit 4.5 Selective Precipitation of Proteins Unit 4.6 Long-Term Storage of Proteins Unit 4.7 Extraction of Proteins from Plant Tissues

x Current Protocols in Protein Science

Chapter 5 Production of Recombinant Proteins Introduction Unit 5.1 Production of Recombinant Proteins in Escherichia coli Unit 5.2 Selection of Escherichia coli Expression Systems Unit 5.3 Fermentation and Growth of Escherichia coli for Optimal Protein Production Unit 5.4 Overview of the Baculovirus Expression System Unit 5.5 Protein Expression in the Baculovirus System Unit 5.6 Overview of Protein Expression in Saccharomyces cerevisiae Unit 5.7 Overview of Protein Expression in Pichia pastoris Unit 5.8 Culture of Yeast for the Production of Heterologous Proteins Unit 5.9 Overview of Protein Expression by Mammalian Cells Unit 5.10 Production of Recombinant Proteins in Mammalian Cells Unit 5.11 Overview of the Vaccinia Virus Expression System Unit 5.12 Preparation of Cell Cultures and Vaccinia Virus Stocks Unit 5.13 Generation of Recombinant Vaccinia Viruses Unit 5.14 Characterization of Recombinant Vaccinia Viruses and Their Products Unit 5.15 Gene Expression Using the Vaccinia Virus/ T7 RNA Polymerase Hybrid System Unit 5.16 Choice of Cellular Protein Expression System Unit 5.17 Use of the Gateway System for Protein Expression in Multiple Hosts Unit 5.18 Wheat Germ Cell-Free Expression System for Protein Production

Chapter 6 Purification of Recombinant Proteins Introduction Unit 6.1 Overview of the Purification of Recombinant Proteins Produced in Escherichia coli Unit 6.2 Preparation of Soluble Proteins from Escherichia coli

xi Current Protocols in Protein Science

Unit 6.3 Preparation and Extraction of Insoluble (Inclusion-Body) Proteins from Escherichia coli Unit 6.4 Overview of Protein Folding Unit 6.5 Folding and Purification of Insoluble (Inclusion Body) Proteins from Escherichia coli Unit 6.6 Expression and Purification of GST Fusion Proteins Unit 6.7 Expression and Purification of Thioredoxin Fusion Proteins Unit 6.8 Automated Large-Scale Purification of a Recombinant G-Protein-Coupled Neurotensin Receptor

Chapter 7 Characterization of Recombinant Proteins Introduction Unit 7.1 Overview of the Characterization of Recombinant Proteins Unit 7.2 Determining the Identity and Purity of Recombinant Proteins by UV Absorption Spectroscopy Unit 7.3 Determining the Identity and Structure of Recombinant Proteins Unit 7.4 Transverse Urea-Gradient Gel Electrophoresis Unit 7.5 Analytical Ultracentrifugation Unit 7.6 Determining the CD Spectrum of a Protein Unit 7.7 Determining the Fluorescence Spectrum of a Protein Unit 7.8 Light Scattering Unit 7.9 Measuring Protein Thermostability by Differential Scanning Calorimetry Unit 7.10 Characterizing Recombinant Proteins Using HPLC Gel Filtration and Mass Spectrometry Unit 7.11 Rapid Screening of E. coli Extracts by Heteronuclear NMR

Chapter 8 Conventional Chromatographic Separations Introduction Unit 8.1 Overview of Conventional Chromatography Unit 8.2 Ion-Exchange Chromatography

xii Current Protocols in Protein Science

Unit 8.3 Gel-Filtration Chromatography Unit 8.4 Hydrophobic-Interaction Chromatography Unit 8.5 Chromatofocusing Unit 8.6 Hydroxylapatite Chromatography Unit 8.7 HPLC of Peptides and Proteins Unit 8.8 Expanded-Bed Adsorption Chromatography

Chapter 9 Affinity Purification Introduction Unit 9.1 Lectin Affinity Chromatography Unit 9.2 Dye Affinity Chromatography Unit 9.3 Affinity Purification of Natural Ligands Unit 9.4 Metal-Chelate Affinity Chromatography Unit 9.5 Immunoaffinity Chromatography Unit 9.6 Purification of Sequence-Specific DNA-Binding Proteins by Affinity Chromatography Unit 9.7 Purification of DNA-Binding Proteins Using Biotin/Streptavidin Affinity Systems Unit 9.8 Immunoprecipitation Unit 9.9 Overview of Affinity Tags for Protein Purification

Chapter 10 Electrophoresis Introduction Unit 10.1 One-Dimensional SDS Gel Electrophoresis of Proteins Unit 10.2 One-Dimensional Isoelectric Focusing of Proteins in Slab Gels Unit 10.3 One-Dimensional Electrophoresis Using Nondenaturing Conditions Unit 10.4 Two-Dimensional Gel Electrophoresis

xiii Current Protocols in Protein Science

Unit 10.5 Protein Detection in Gels Using Fixation Unit 10.6 Protein Detection in Gels Without Fixation Unit 10.7 Electroblotting from Polyacrylamide Gels Unit 10.8 Detection of Proteins on Blot Membranes Unit 10.9 Capillary Electrophoresis of Proteins and Peptides Unit 10.10 Immunoblot Detection Unit 10.11 Autoradiography Unit 10.12 Overview of Digital Electrophoresis Analysis Unit 10.13 Capillary Electrophoresis of Peptides and Proteins Using Isoelectric Buffers

Chapter 11 Chemical Analysis Introduction Unit 11.1 Enzymatic Digestion of Proteins in Solution Unit 11.2 Enzymatic Digestion of Proteins on PVDF Membranes Unit 11.3 Enzymatic Digestion of Proteins in Gels for Mass Spectrometric Identification and Structural Analysis Unit 11.4 Chemical Cleavage of Proteins in Solution Unit 11.5 Chemical Cleavage of Proteins on Membranes Unit 11.6 Reversed-Phase Isolation of Peptides Unit 11.7 Removal of N-Terminal Blocking Groups from Proteins Unit 11.8 C-Terminal Sequence Analysis Unit 11.9 Amino Acid Analysis Unit 11.10 N-Terminal Sequence Analysis of Proteins and Peptides Unit 11.11 Determination of Disulfide-Bond Linkages in Proteins

Chapter 12 Post-Translational Modification: Glycosylation Introduction

xiv Current Protocols in Protein Science

Unit 12.1 Overview of Glycoconjugate Analysis Unit 12.2 Metabolic Radiolabeling of Animal Cell Glycoconjugates Unit 12.3 Inhibition of N-Linked Glycosylation Unit 12.4 Endoglycosidase and Glycoamidase Release of N-Linked Oligosaccharides Unit 12.5 Detection of Glycophospholipid Anchors on Proteins Unit 12.6 Determining the Structure of Oligosaccharides N- and O-Linked to Glycoproteins Unit 12.7 Determining the Structure of Glycan Moieties by Mass Spectrometry Unit 12.8 Detection and Analysis of Proteins Modified by O-Linked N-Acetylglucosamine

Chapter 13 Post-Translational Modification: Phosphorylation and Phosphatases Introduction Unit 13.1 Overview of Protein Phosphorylation Unit 13.2 Labeling Cultured Cells with 32Pi and Preparing Cell Lysates for Immunoprecipitation Unit 13.3 Phosphoamino Acid Analysis Unit 13.4 Detection of Phosphorylation by Immunological Techniques Unit 13.5 Detection of Phosphorylation by Enzymatic Techniques Unit 13.6 Preparation and Application of Polyclonal and Monoclonal Sequence-Specific AntiPhosphoamino Acid Antibodies Unit 13.7 Assays of Protein Kinases Using Exogenous Substrates Unit 13.8 Permeabilization Strategies to Study Protein Phosphorylation Unit 13.9 Phosphopeptide Mapping and Identification of Phosphorylation Sites Unit 13.10 Use of Protein Phosphatase Inhibitors

Chapter 14 Post-Translational Modification: Specialized Applications Introduction Unit 14.1 Analysis of Disulfide Bond Formation

xv Current Protocols in Protein Science

Unit 14.2 Analysis of Protein Acylation Unit 14.3 Analysis of Protein Prenylation and Carboxyl-Methylation Unit 14.4 Analysis of Oxidative Modification of Proteins Unit 14.5 Analysis of Protein Ubiquitination Unit 14.6 Analysis of Protein S-Nitrosylation Unit 14.7 Tyrosine O-Sulfation Unit 14.8 Analysis of Protein Sumoylation Unit 14.9 Identifying and Quantifying Sites of Protein Methylation by Heavy Methyl SILAC

Chapter 15 Chemical Modification of Proteins Introduction Unit 15.1 Modification of Cysteine Unit 15.2 Modification of Amino Groups

Chapter 16 Mass Spectrometry Introduction Unit 16.1 Overview of Peptide and Protein Analysis by Mass Spectrometry Unit 16.2 Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Analysis of Peptides Unit 16.3 Sample Preparation for MALDI Mass Analysis of Peptides and Proteins Unit 16.4 In-Gel Digestion of Proteins for MALDI-MS Fingerprint Mapping Unit 16.5 Searching Sequence Databases Over the Internet: Protein Identification Using MS-Fit Unit 16.6 Searching Sequence Databases Over the Internet: Protein Identification Using MS-Tag Unit 16.7 Enzymatic Approaches for Obtaining Amino Acid Sequence: On-Target Ladder Sequencing Unit 16.8 Introducing Samples Directly into Electrospray Ionization Mass Spectrometers Using a Nanospray Interface Unit 16.9 Introducing Samples Directly into Electrospray Ionization Mass Spectrometers Using Microscale Capillary Liquid Chromatography

xvi Current Protocols in Protein Science

Unit 16.10 Protein Identification Using a Quadrupole Ion Trap Mass Spectrometer and SEQUEST Database Matching Unit 16.11 De Novo Peptide Sequencing via Manual Interpretation of MS/MS Spectra

Chapter 17 Structural Biology Introduction Unit 17.1 Overview of Protein Structural and Functional Folds Unit 17.2 Electron Microscopy and Image Processing: An Essential Tool for Structural Analysis of Macromolecules Unit 17.3 Principles of Macromolecular X-Ray Crystallography Unit 17.4 Crystallization of Macromolecules Unit 17.5 Introduction to NMR of Proteins Unit 17.6 Probing Protein Structure and Dynamics by Hydrogen Exchange–Mass Spectrometry Unit 17.7 Introduction to Atomic Force Microscopy (AFM) in Biology Unit 17.8 Raman Spectroscopy of Proteins Unit 17.9 Crystallization of Integral Membrane Proteins

Chapter 18 Preparation and Handling of Peptides Introduction Unit 18.1 Introduction to Peptide Synthesis Unit 18.2 Synthesis of Multiple Peptides on Plastic Pins Unit 18.3 Synthetic Peptides for Production of Antibodies that Recognize Intact Proteins Unit 18.4 Native Chemical Ligation of Polypeptides Unit 18.5 Synthesis and Application of Peptide Dendrimers As Protein Mimetics Unit 18.6 Disulfide Bond Formation in Peptides Unit 18.7 Guide for Resin and Linker Selection in Solid-Phase Peptide Synthesis

xvii Current Protocols in Protein Science

Chapter 19 Identification of Protein Interactions Introduction Unit 19.1 Analysis of Protein-Protein Interactions Unit 19.2 Interaction Trap/Two-Hybrid System to Identify Interacting Proteins Unit 19.3 Phage-Based Expression Cloning to Identify Interacting Proteins Unit 19.4 Detection of Protein-Protein Interactions by Coprecipitation Unit 19.5 Imaging Protein-Protein Interactions by Fluorescence Resonance Energy Transfer (FRET) Microscopy Unit 19.6 High-Throughput Screening for Protein-Protein Interactions Using Yeast Two-Hybrid Arrays Unit 19.7 Identification of Protein Interactions by Far Western Analysis Unit 19.8 Scintillation Proximity Assay (SPA) Technology to Study Biomolecular Interactions Unit 19.9 Identifying Protein Interactions by Hydroxyl-Radical Protein Footprinting Unit 19.10 Visualization of Protein Interactions in Living Cells Using Bimolecular Fluorescence Complementation (BiFC) Analysis Unit 19.11 Production and Use of Trimeric Isoleucine Zipper Fusion Proteins to Study Surface Receptor Ligand Interactions Unit 19.12 Fluorescence Quenching Methods to Study Lipid-Protein Interactions Unit 19.13 Overview of Biacore Systems and Their Applications Unit 19.14 Using Biacore to Measure the Binding Kinetics of an Antibody-Antigen Interaction Unit 19.15 Identifying Small-Molecule Modulators of Protein-Protein Interactions Unit 19.16 Determination of Protein Contacts by Chemical Cross-Linking With EDC and Mass Spectrometry

Chapter 20 Quantitation of Protein Interactions Introduction Unit 20.1 Overview of the Quantitation of Protein Interactions Unit 20.2 Measuring Protein Interactions by Optical Biosensors Unit 20.3 Analytical Centrifugation: Equilibrium Approach

xviii Current Protocols in Protein Science

Unit 20.4 Titration Microcalorimetry Unit 20.5 Reduced-Scale Large-Zone Analytical Gel-Filtration Chromatography for Measurement of Protein Association Equilibria Unit 20.6 Size-Exclusion Chromatography with On-Line Light Scattering Unit 20.7 Analytical Ultracentrifugation: Sedimentation Velocity Analysis Unit 20.8 Spectroscopic Methods for the Determination of Protein Interactions Unit 20.9 Application of Amide Proton Exchange Mass Spectrometry for the Study of Protein-Protein Interactions Unit 20.10 Circular Dichroism to Study Protein Interactions Unit 20.11 Quantitative Determination of Protein Stability and Ligand Binding by Pulse Proteolysis

Chapter 21 Peptidases Introduction Unit 21.1 Proteases Unit 21.2 Papain-like Cysteine Proteases Unit 21.3 Overview of Pepsin-like Aspartic Peptidases Unit 21.4 Metalloproteases Unit 21.5 Purification and Characterization of Proteasomes from Saccharomyces cerevisiae Unit 21.6 Purification of the Eukaryotic 20S Proteasome Unit 21.7 Serpins (Serine Protease Inhibitors) Unit 21.8 Caspases Unit 21.9 Use of GFP as a Reporter for the Analysis of Sequence-Specific Proteases Unit 21.10 An Overview of Serine Proteases Unit 21.11 Over-Expression and Purification of Active Serine Proteases and Their Variants from Escherichia coli Inclusion Bodies Unit 21.12 Assaying Proteases in Cellular Environments Unit 21.13 Expression, Purification, and Characterization of Caspases

xix Current Protocols in Protein Science

Unit 21.14 Expression, Purification, and Characterization of Aspartic Endopeptidases: Plasmodium Plasmepsins and “Short” Recombinant Human Pseudocathepsin Unit 21.15 Zymography of Metalloproteinases Unit 21.16 Monitoring Metalloproteinase Activity Using Synthetic Fluorogenic Substrates Unit 21.17 Applications for Chemical Probes of Proteolytic Activity

Chapter 22 Gel-Based Proteome Analysis Introduction Unit 22.1 Overview of Proteome Analysis Unit 22.2 Protein Profiling Using Two-Dimensional Difference Gel Electrophoresis (2-D DIGE) Unit 22.3 Laser Capture Microdissection for Proteome Analysis Unit 22.4 Preparing Protein Extracts for Quantitative Two-Dimensional Gel Comparison Unit 22.5 Isolation of Organelles and Prefractionation of Protein Extracts Using Free-Flow Electrophoresis Unit 22.6 Protein Profiling by Microscale Solution Isoelectrofocusing (MicroSol-IEF)

Chapter 23 Non-Gel-Based Proteome Analysis Introduction Unit 23.1 Analysis of Protein Composition Using Multidimensional Chromatography and Mass Spectrometry Unit 23.2 Quantitative Protein Profile Comparisons Using the Isotope-Coded Affinity Tag Method Unit 23.3 Proteomic Analysis Using 2-D Liquid Separations of Intact Proteins From Whole-Cell Lysates Unit 23.4 Quantitative Protein Analysis Using Proteolytic [18O]Water Labeling Unit 23.5 Protein Microarrays

Chapter 24 Targeted Proteomics Introduction Unit 24.1 Human Serum and Plasma Proteomics

xx Current Protocols in Protein Science

Unit 24.2 Organelle Proteomics Unit 24.3 Glycoproteomics Using Chemical Immobilization

Chapter 25 Proteome Bioinformatics Introduction Unit 25.1 Bioinformatics Analysis for Interactive Proteomics

Chapter 26 Protein Engineering Introduction Unit 26.1 Misincorporation Proton-Alkyl Exchange (MPAX): Engineering Cysteine Probes into Proteins Unit 26.2 Combinatorial Recombination of Gene Fragments to Construct a Library of Chimeras Unit 26.3 Incorporation of Isotopically Enriched Amino Acids

Appendix 1 Useful Data 1A Characteristics of Amino Acids 1B Commonly Used Detergents 1C Conversion Factors and Half-Life Information for Radioactivity 1D Common Conversion Factors

Appendix 2 Laboratory Guidelines, Equipment, and Stock Solutions 2A Laboratory Safety 2B Safe Use of Radioisotopes 2C Centrifuges and Rotors 2D Standard Laboratory Equipment 2E Commonly Used Reagents

xxi Current Protocols in Protein Science

Appendix 3 Commonly Used Techniques 3A Use of Protein Folding Reagents 3B Dialysis 3C Techniques for Mammalian Cell Tissue Culture 3D Importing Biological Materials 3E Silanizing Glassware 3F Protein Precipitation Using Ammonium Sulfate 3G Statistics: Detecting Differences Among Groups 3H Analyzing Radioligand Binding Data

Appendix 4 Molecular Biology Techniques 4 Molecular Biology Techniques 4A Media Preparation and Bacteriological Tools 4B Growth in Liquid or Solid Media 4C Preparation of Plasmid DNA 4D Introduction of Plasmid DNA into Cells 4E Purification and Concentration of DNA from Aqueous Solutions 4F Agarose Gel Electrophoresis 4G Southern Blotting 4H Hybridization Analysis of DNA Blots 4I Digestion of DNA with Restriction Endonucleases 4J The Polymerase Chain Reaction 4K Quantitation of Nucleic Acids with Absorption Spectroscopy 4L Growth and Manipulation of Yeast

xxii Current Protocols in Protein Science

Appendix 5 Biophysical Methods: Data Analysis 5A Theoretical Aspects of the Quantitative Characterization of Ligand Binding Suppliers Selected Suppliers of Reagents and Equipment

xxiii Current Protocols in Protein Science

Overview of Protein Purification and Characterization AIMS AND OBJECTIVES Protein purification has an over 200-year history: the first attempts at isolating substances from plants having similar properties to “egg albumen,” or egg white, were reported in 1789 by Fourcroy. Many proteins from plants were purified in the nineteenth century, though most would not be considered pure by modern standards. A century later, ovalbumin was the first crystalline protein obtained (by Hofmeister in 1889). The year 1989 may not go down in history as a milestone in protein chemistry, but since then there has been a resurgence of interest in proteins after more than a decade of gene excitement. The aims of protein purification, up until the 1940s, were simply academic. To then, even the basic facts of protein structure were not fully appreciated, and pure proteins were needed just to study structure and test the rival theories of the pre-DNA days. During the Second World War, an acute need for blood proteins led to development of the Cohn fractionation procedure for purification of albumin and other proteins from serum (Cohn et al., 1946). This was the inception of large-scale protein purifications for commercial purposes; Cohn fractionation continues to be used to this day. As more proteins, and particularly enzymes, were purified and crystallized, they started to be used increasingly in diagnostic assays and enzymatic analyses, as well as in the largescale food, tanning, and detergent industries. Many enzymes used in industry are not in fact very pure, but as long as they do the job, that is sufficient. “Process” enzymes such as α-amylase, proteases, and lipases are pro-duced in ton quantities, mainly as secretion products in bacterial cultures, and may undergo only limited purification processes to mini-mize costs. At the other extreme, enzyme products for research and analysis require a high degree of purification to ensure that contaminating activities do not interfere with the intended use. Anyone familiar with molecular biology enzymes will appreciate how minute levels of contamination of DNase or RNase can completely destroy carefully planned experiments. The 1960s and 1970s could be described as the peak years for protein and enzyme research, and most of the methods used in protein puri-

Contributed by R.K. Scopes Current Protocols in Protein Science (1995) 1.1.1-1.1.6 Copyright © 2000 by John Wiley & Sons, Inc.

fication were established by then, at least in their principles. More recent developments have been mainly in instrumentation designed to optimize the application of each methodology. Developments in instrumentation have been stimulated by the rapid progress in molecular biology, because gene isolation has often been preceded by isolation of the gene product. Because such products can now be characterized sufficiently (i.e., partially sequenced) using minute amounts of protein, the need for large-scale or even moderate-scale procedures has decreased. Hence there has been an explosive development of modern equipment designed specifically for dealing with amounts of protein in the milligram to microgram range. On the other hand, structural studies using X-ray crystallography and nuclear magnetic resonance (NMR) require hundreds of milligrams of pure protein, so larger-scale equipment and procedures are still needed in the research laboratory. The nature of the proteins studied has also changed substantially. Whereas enzymes were once the most favored subjects, they have now been superceded by nonenzymatic proteins such as growth factors, hormone receptors, viral antigens, and membrane transporters. Many of these occur in minute amounts in the natural source, and their purification can be a major task. Heroic efforts in the past have used kilogram quantities of rather unpleasant starting materials, such as human organs, and ended up with a few micrograms of pure product. It is now more usual, however, to take the genetic approach: clone the gene before the protein has been isolated or even properly identified, then express it in a suitable host cell culture or organism. The expression level may be orders of magnitude higher than in the original source, which will make purification a relatively simple task. It can be useful to know beforehand some physical properties of the protein, to facilitate the development of a suitable purification protocol from the recombinant source. On the other hand, there are now several ways of preparing fusion proteins, which can be purified by affinity techniques without any knowledge of the properties of the target protein. Moreover, there are ways of modifying the expressed product to simplify purification further.

UNIT 1.1

Strategies of Protein Purification and Characterization

1.1.1 CPPS

Overview of Protein Purification and Characterization

Thus the approach to protein purification must first take into account the reason it is being done, as the methods will vary greatly with different requirements. At one extreme is the one-of-a-kind purification, in a well-financed and equipped laboratory, that is carried out to obtain a small amount of product for sequencing so that gene isolation can proceed. In this case, expense of equipment and reagents may be no problem, and a very low overall recovery of product can be acceptable, provided it is pure enough. At the other extreme are the requirements of commercial production of a protein in large amounts on a continuing basis, where high recovery and economy of processing are the chief parameters to be considered. There are many intermediate situations as well. Many publications in the area of protein research are entitled “Purification and characterization of…,” and describe a purification procedure in sufficient detail that it can be reproduced in another laboratory. The characterization section may include structural, functional, and genetic information, and carrying out such studies is likely to require at least milligram quantities of pure protein. Ideally the purification should involve a small number of steps, with good recovery at each step. If the recovery is poor (pir|S340.. SERUM ALBUMIN PRECURSOR (FRAGMENT).. albumin [Sus scrofa] Sheep mRNA for serum albumin. [Ovi.. SERUM ALBUMIN PRECURSOR. >pir|S069.. SERUM ALBUMIN PRECURSOR. >pir|A388.. serum albumin prepropeptide - rhes.. albumin [Bos taurus] Messenger RNA for rat preproalbumi.. SERUM ALBUMIN PRECURSOR. >pir|A938.. albumin - human serum albumin - dog Mature HSA [Homo sapiens] Human serum albumin [Unknown] synthetic mature HSA [Unknown] serum albumin [Homo sapiens] Human serum albumin (ALB) gene, co.. human serum albumin [Unknown] SERUM ALBUMIN PRECURSOR. >gp|M1252.. serum albumin precursor - human >g..

High Score 101 98 98 96 96 95 94 94 92 92 79 79 85 85 85 85 85 85 85 85

Smallest Sum Probability P(N) N 6.8e-08 1.8e-07 1.8e-07 3.4e-07 3.5e-07 4.8e-07 6.6e-07 6.6e-07 1.3e-06 1.3e-06 1.8e-06 4.0e-06 1.2e-05 1.2e-05 1.2e-05 1.2e-05 1.2e-05 1.2e-05 1.2e-05 1.2e-05

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

Alignment >sp|P35747|ALBU_HORSE SERUM ALBUMIN PRECURSOR. >pir|S34053|ABHOS serum albumin precursor - horse >gp|X74045|ECSERALB_1 preproalbumin [Equus caballus] Length = 607 Score = 101 (47.0 bits), Expect = 6.8e-08, P = 6.8e-08 Identities = 22/26 (84%), Positives = 22/26 (84%) Query: Sbjct:

1 TIKSEIAHVFNDLGEPVFKGLVLVAF 26 T KSEIAH FNDLGE FKGLVLVAF 26 THKSEIAHRFNDLGEKHFKGLVLVAF 51

Figure 2.1.2 BLAST output table and alignment for an unknown mammalian blood protein. The Query is the sequence of interest and the sbjct is the sequence region found in the database that matches the query. Note that in searches of the nonredundant databases at the NCBI, matches are found between the test sequence and the translation of mRNAs in GenBank. Important similarities have small probability scores, i.e., P(N) < 10−3. This analysis suggests that the unknown is an albumin, but it is not identical to any known sequence.

protein sequence database maintained at the NCBI. This database contains a compilation of protein sequences found in all of the molecular sequence databases with duplicate sequences removed. Figure 2.1.2 presents a table of the top 20 highest similarity scoring sequences and the best alignment.

FASTA Database Searching FASTA conducts a Lipman and Pearson (1985) search for similarity between a query sequence and any FASTA-formatted database or group of sequences collected in a database by the user. FASTA is a program developed for fast searching using an algorithm that allows for the insertion of gaps during the alignment

phase. This approximates evolutionary insertions and deletions during divergence. However, the gaps are generated to maximize the number of aligned residues. FASTA does a good job of generating a global alignment for a sequence. Figure 2.1.3 and Figure 2.1.4 present results from a FASTA search of the SWISS-PROT database with a short sequence fragment from an unknown mammalian blood protein. The FASTA output contains (1) a histogram that shows the distribution of relevant word matches found between the query sequence and the database searched; (2) the match statistics; and (3) the alignments generated for the query sequence and the matched sequence region. There

Computational Analysis

2.1.5 Current Protocols in Protein Science

Score Init1 Initn (-) (+) < 2 533 533:================================================== 4 4 4:== 6 134 134:================================================== 8 222 222:================================================== 10 1964 1964:================================================== 12 5289 5289:================================================== 14 7609 7609:================================================== 16 9434 9434:================================================== 18 2834 2834:================================================== 20 4914 4914:================================================== 22 2686 2686:================================================== 24 1808 1808:================================================== 26 1241 1241:================================================== 28 721 721:================================================== 30 375 374:================================================== 32 261 259:================================================== 34 138 137:================================================== 36 57 57:============================= 38 27 27:============== 40 18 18:========= 42 4 5:==+ 44 6 7:===+ 46 0 0: 48 0 1:+ 50 3 4:== 52 1 1:= 54 1 1:= 56 2 2:= 58 0 0: 60 0 0: 62 0 0: 64 0 0: 66 0 0: 68 0 0: 70 0 0: 72 0 0: 74 0 0: 76 0 0: 78 0 0: 80 0 0: > 80 6 6:=== Mean score calculations exclude scores greater than 72 mean initn score: 16.3 (s.d. 5.20) mean init1 score: 16.3 (s.d. 5.19) 1838 scores better than 26 saved, joining threshold: 27

Figure 2.1.3 FASTA histogram from a search of the SWISS-PROT database for sequences similar to a fragment from an unknown mammalian blood protein. Numbers of windows at each init1 score are plotted. Note that there are six highly significant alignments, as well as several sequences (those with init1 scores in the 48 to 56 range) that represent protein sequences from more distantly related species.

Computational Methods for Protein Sequence Analysis

are three similarity scores presented in the output from FASTA: init1, the highest similarity score allowing conservative substitutions; initn, the highest score generated by linking nonoverlapping regions; and opt, the score for the best alignment presented. In practice the histogram would be used to determine how many sequences have highly significant levels of similarity. The table of scores allows one to decide quickly how many and which sequence alignments should be viewed.

ALIGNMENT METHOD CONSIDERATIONS Dynamic programming is a different matrix method that is used to find the optimal global alignment of two sequences (Needleman and Wünsch, 1970; Sellers, 1974; Pearson and Miller, 1992; Panjukov, 1993). Let a be a sequence of length m and let b be a sequence of length n. Let D be the working matrix. Then, given a predefined scoring function, the basic steps of the alignment algorithm (Doolittle, 1986) are as follows. Assign the scores for all exact matches to the corresponding matrix el-


Alignment Score Table The best scores are: init1 initn opt Sw:Albu_Horse P35747 equus caballus (horse). serum album... 99 99 100 Sw:Albu_Rat P02770 rattus norvegicus (rat). serum albumi... 99 99 99 Sw:Albu_Pig P08835 sus scrofa (pig). serum albumin precu... 99 99 101 Sw:Albu_Bovin P02769 bos taurus (bovine). serum albumin ... 98 98 99 Sw:Albu_Sheep P14639 ovis aries (sheep). serum albumin p... 95 95 96 Sw:Albu_Human P02768 homo sapiens (human). serum albumin... 93 93 93 Sw:Albu_Chick P19121 gallus gallus (chicken). serum albu... 56 56 77 Sw:Albu_Mouse P07724 mus musculus (mouse). serum albumin... 55 55 55

Best Scoring Alignment SCORES

Init1: 99 Initn: 99 Opt: 84.6% identity in 26 aa overlap

100

10 20 TIKSEIAHVFNDLGEPVFKGLVLVAF | |||||| |||||| ||||||||| Albu_H MKWVTFVSLLFLFSSAYSRGVLRRDTHKSEIAHRFNDLGEKHFKGLVLVAFSQYLQQCPF 10 20 30 40 50 60 Pc12_P

Figure 2.1.4 Partial FASTA alignment table and best scoring alignment for the same search illustrated in Figure 2.1.3. The table shows the best alignment scores sorted by highest init1 score.

ements. Transform the matrix by beginning with element Dmn and adding its value (less any gap penalty) to Dm−1, n−1 and all the elements directly above it or to its left. Next, select the matrix element in column m − 2 and above row m − 1 that contains the largest score and repeat the summing procedure. Repeat the previous step until the entire matrix is transformed. Next, determine the path from the largest score in the matrix back toward the bottom right-hand corner. The path, with any gaps as indicated by off-diagonal moves, is then used to align a and b. This matching algorithm is used in most global alignment programs, which vary primarily in the sophistication of the scoring functions (similarity or distance). Note that this method assumes (incorrectly) that each residue has equal importance, and it ignores sequential constraints. Local alignments are usually found using some version of the Smith-Waterman method (1981), which applies when common regions are being sought in two sequences whose similarity is unknown or limited to a few regions, as is the case in database searches. The algorithm usually produces a single highest-scoring alignment with respect to the operations in E, although it has been successfully extended to produce the second-best and other alignments (Barton, 1990, 1993; Waterman and Eggert, 1991). In its simplest form (Barton, 1993), the local alignment method is as follows. Given two

sequences a and b, and a working (comparison) matrix H, the matrix elements of H are filled in by applying a scoring rule: the score assigned to a matrix element Hij is either zero or the maximum value of any of the three scores of its neighbors and predecessors that lie directly above, diagonally to the left, or directly to the left of Hij, adjusted for gaps or substitution of residues. The maximum element in H is found and traced back through the matrix to find the alignment.

SCORING MATRICES Scoring matrices (George et al., 1990; States and Boguski, 1990), such as the PAM, BLOSUM (Henikoff and Henikoff, 1993), and related MDM matrices (Dayhoff, 1978), are based on the compilation of frequencies of amino acid substitution in a set of homologous sequences. The data are used as an amino acid transition probability matrix. Increasing powers of the matrix correspond to increasing divergence. Lower powers are used for sequences that seem closely related and higher powers for those that are not so obviously related. Thus, PAM250 represents a matrix with 250 accepted protein mutations per 100 amino acids. The scoring matrix method has greatly improved the quality and sensitivity of alignments; however, it does have the deficiency that it ignores structure and the corresponding sequence constraints which are significant for predicting structure. Not all sequence comparison methods are




matrix methods. Programs such as FASTP and FASTN are representative of hash-coding methods (Dumas and Nunio, 1982; Wilbur and Lipman, 1983; Lipman and Pearson, 1985; Pearson, 1990, 1994). The methods are based on fast table lookups of subsequences of a target sequence using hash functions. Hash functions transform strings in numbers that serve as pointers into a table containing further information. Thus, a fixed-length subsequence (or k-tuple) of one sequence can quickly be checked for occurrence in another sequence after a hash table of k-tuples and their positions in the sequence is constructed for each sequence. FASTP focuses on regions having the highest density of identities and uses the PAM250 matrix. Readers are referred to Pearson (1994) for an introduction to using FASTA for comparing both DNA and protein sequences. The RDF2 and RSS programs for testing the statistical significance of FASTA alignments are also discussed. Other methods for sequence comparison include finite state machines that are capable of recognizing regular expressions, and various statistical approaches (States and Boguski, 1990; Tyler et al., 1991; Allison et al., 1992; Brendel et al., 1992). Regular expressions are formulas such as (a + b*)c, which represents the set of all strings that begin with the letter a or zero or more b’s and followed by a single c. Complicated patterns can be constructed and detected in sequences using the finite state method. Statistical approaches include the database-searching program BLAST (see above section on BLAST database searching), which approximates gapless alignments that optimize local similarity on the basis of maximal segment pairs. The BLAST strategy is to search for only those segment pairs that contain a word pair whose comparison score exceeds some threshold. It uses the PAM120 scoring matrix to allow for conservative substitutions of residues in a comparison. Statistical methods for sequence alignment (Karlin et al., 1988, 1989, 1991; Brendel, 1992; Pevzner, 1992; Mrazek and Kypr, 1993; Staden, 1994a-d) consider the statistical significance of a variety of protein sequence properties. Such properties include the relative composition of a sequence with respect to residue types; anomalous clusters, runs, and periodic variations of charge; internal repeats; local periodicities that might indicate regularity of structure; and the relative spacings of certain residue types. For a description of routines for

plotting hydrophobicity, charge, and hydrophobic moments using the program PIP, see Staden (1994a-d). In addition to the methods described above, interest in the development of theory and tools for linguistic (formal grammar) analysis of biological sequences is growing (Collado-Vides, 1991; Michaels et al., 1993; Searls, 1993). The next level of sequence analysis is the simultaneous comparison of more than two sequences, as described in the following section on Multiple Alignments.

MULTIPLE ALIGNMENTS Once a group of potentially homologous sequences has been identified, the next step could be to do a multiple alignment that compares the new sequence with the group of sequences determined to have highly significant similarities. There are several programs that will do multiple alignments. PILEUP from the GCG package does a good job with closely related sequences in an automated fashion. MACAW from NCBI, LINEUP from GCG, and DSCE (De Rijk and De Wachter, 1993) allow the investigator to generate an alignment manually. (The CD-ROM version of Current Protocols in Protein Science includes MACAW and Entrez from NCBI.) Figure 2.1.5 demonstrates a multiple alignment generated using PILEUP from five highscoring albumin sequences in the SWISSPROT database identified from the FASTA analysis (Fig. 2.1.4). PILEUP performs a cluster analysis based on pairwise comparisons of each sequence to be aligned. As well as the multiple alignment (Fig. 2.1.5), PILEUP produces a dendrogram (not shown) that displays the clustered relationships between the sequences. The dendrogram is not a phylogenetic tree, but a graphical representation of the clustered similarity statistics. After the multiple alignment is generated, one can begin to infer relationships between conserved regions and function or structure. The program Pro_Anal (Eroshkin et al., 1993) and the program for fast protein block searches (Fuchs, 1994) can help in this process. Programs like OBSTRUCT (Heringa et al., 1992) are useful in identifying clusters of protein sequences on the basis of structural and sequence similarity criteria. Kanaoka et al. (1989) have described a process of aligning protein sequences on the basis of hydrophobicity. Livingstone and Barton (1993) describe a strategy to generate multiple alignments for hierarchical analysis of residue conservation.


Albu_Bovin.Sw Albu_Chick.Sw Albu_Horse.Sw Albu_Human.Sw Albu_Mouse.Sw

1 MKWVTFISLL MKWVTLISFI MKWVTFVSLL MKWVTFISLL EADKSEIADR

LLFSSAYSRG FLFSSATSRN FLFSSAYSRG FLFSSAYSRG YNDLGEIHFK

VFRRDTHKSE LQRFARDAEH VLRRDTHKSE VFRRDAHKSE CAIPNLRENY

IAHRFKDLGE KSEIAHRYND IAHRFNDLGE VAHRFKDLGE GELADCCTKQ

50 EQFKGLVLIA LKEETFKAVA KHFKGLVLVA ENFKALVLIA EPERNECFLQ


51 FSQYLQQCPF MITFAQYLQR FSQYLQQCPF FAQYLQQCPF HKDDNPSLPP

DEHVKLVNEL CSYEGLSKLV EDHVKLVNEV EDHVKLVNEV FERPEAEAMC

TEFAKTCVAD KDVVDLAQKC TEFAKKCAAD TEFAKTCVAD TSFKENPTTF

ESHAGCEKSL VANEDAPECS ESAENCDKSL ESAENCDKSL MGHYLHEVAR

100 HTLFGDELCK KPLPSIILDE HTLFGDKLCT HTLFGDKLCT RHPYFYAPEL


101 VASLRETYGD ICQVEKLRDS VATLRATYGE VATLRETYGE LYYAEQYNEI

MADCCEKQEP YGAMADCCSK LADCCEKQEP MADCCAKQEP LTQCCAEADK

ERNECFLSHK ADPERNECFL ERNECFLTHK ERNECFLQHK ESCLTPKLDG

DDSPDLPKLK SFKVSQPDFV DDHPNLPKLK DDNPNLPRLV VKEKALVSSV

150 PDPNTLCDEF QPYQRPASDV PEPDAQCAAF RPEVDVMCTA RQRMKCSSMQ


151 KADEKKFWGK ICQEYQDNRV QEDPDKFLGK FHDNEETFLK KFGERAFKAW

YLYEIARRHP SFLGHFIYSV YLYEVARRHP KYLYEIARRH AVARLSQTFP

YFYAPELLYY ARRHPFLYAP YFYGPELLFH PYFYAPELLF NADFAEITKL

ANKYNGVFQD AILSFAVDFE AEEYKADFTE FAKRYKAAFT ATDLTKVNKE

200 CCQAEDKGAC HALQSCCKES CCPADDKLAC ECCQAADKAA CCHGDLLECA


201 LLPKIETMRE DVGACLDTKE LIPKLDALKE CLLPKLDELR DDRAELAKYM

KVLASSARQR IVMREKAKGV RILLSSAKER DEGKASSAKQ CENQATISSK

LRCASIQKFG SVKQQYFCGI LKCSSFQNFG RLKCASLQKF LQTCCDKPLL

ERALKAWSVA LKQFGDRVFQ ERAVKAWSVA GERAFKAWAV KKAHCLSEVE

250 RLSQKFPKAE ARQLIYLSQK RLSQKFPKAD ARLSQRFPKA HDTMPADLPA


251 FVEVTKLVTD YPKAPFSEVS FAEVSKIVTD EFAEVSKLVT IAADFVEDQE

LTKVHKECCH KFVHDSIGVH LTKVHKECCH DLTKVHTECC VCKNYAEAKD

GDLLECADDR KECCEGDMVE GDLLECADDR HGDLLECADD VFLGTFLYEY

ADLAKYICDN CMDDMARMMS ADLAKYICEH RADLAKYICE SRRHPDYSVS

300 QDTISSKLKE NLCSQQDVFS QDSISGKLKA NQDSISSKLK LLLRLAKKYE


301 CCDKPLLEKS GKIKDCCEKP CCDKPLLQKS ECCEKPLLEK ATLEKCCAEA

HCIAEVEKDA IVERSQCIME HCIAEVKEDD SHCIAEVEND NPPACYGTVL

IPENLPPLTA AEFDEKPADL LPSDLPALAA EMPADLPSLA AEFQPLVEEP

DFAEDKDVCK PSLVEKYIED DFAEDKEICK ADFVESKDVC KNLVKTNCDL

350 NYQEAKDAFL KEVCKSFEAG HYKDAKDVFL KNYAEAKDVF YEKLGEYGFQ


351 GSFLYEYSRR HDAFMAEFVY GTFLYEYSRR LGMFLYEYAR NAILVRYTQK

HPEYAVSVLL EYSRRHPEFS HPDYSVSLLL RHPDYSVVLL APQVSTPTLV

RLAKEYEATL IQLIMRIAKG RIAKTYEATL LRLAKTYETT EAARNLGRVG

EECCAKDDPH YESLLEKCCK EKCCAEADPP LEKCCAAADP TKCCTLPEDQ

400 ACYSTVFDKL TDNPAECYAN ACYRTVFDQF HECYAKVFDE RLPCVEDYLS

Figure 2.1.5 Multiple alignment of mammalian blood proteins using PILEUP. The first 400 residues of five high-scoring sequences identified from FASTA analysis (Fig. 2.1.4) are aligned.

Similarly, the program PIMA (Smith and Smith, 1992) uses an algorithm that takes advantage of secondary structure features to determine gap penalties during the multiple alignment process. Depiereux and Feytmans (1991) have used simultaneous multivariate multiple alignment methods to reveal a correspondence between physiochemical profiles and structurally conserved regions. The multiple alignment programs described above take advantage of pairwise comparison

algorithms and are subject to their limitations. Alignment algorithms that use a rigorous consideration of more than three sequences at once are computationally infeasible for most labs. Although some speed improvements have been obtained for the optimal alignment of three sequences simultaneously (Murata et al., 1985; Gotoh, 1986), new methods have to be developed for multiple sequence alignment. The earliest methods were sequence editors for interactive alignments. See Pearson and Miller




(1992) for a discussion of considerably improved dynamic programming algorithms for sequence analysis. Multiple sequence alignment tends to be order-sensitive. A sequence of pairwise comparisons may not lead to the best overall alignment for a given set of sequences because the sequences are not independent (States and Boguski, 1990)—that is, they are not necessarily equidistant from a common ancestor. The consensus sequence approach to multiple sequence alignment (Waterman, 1989, 1990; Waterman and Jones, 1990; Day and McMorris, 1993) is based on the construction of a single sequence that can be used to align all other given sequences; the sequence is determined by an iterative process of alignment followed by consensus sequence modification. A consensus residue can be stringently defined from a multiple alignment when the content of a column is represented by a single amino acid at that position. There are many multiple alignment algorithms—too many to describe here. A list of available sequence similarity and alignment software can be accessed via the Internet as described in the section below on Internet Resources. Once sequences have been aligned, the quality of the alignment should be assessed. It is clearly important to distinguish between chance similarities and true sequence similarities. One method for estimating the quality or statistical significance of an alignment (Doolittle, 1986) is to compare the alignment score with the mean score of a set of randomized sequences that are permutations of the original sequences and which preserve their length and relative composition. The further the alignment score is from the mean of the randomized set, the more confident we can be in the alignment. A possible objection to this data shuffling or permutation method is that the permuted sequences are not like real protein sequences (i.e., permuted sequences are unlikely to fold like real proteins). Other methods (States and Boguski, 1990) include the following: McLachlan’s (1972) “double matching probability” function, which is used to compute the frequency scores in infinite random sequences of the same composition as the given sequences; Reich and Meiske’s stochastic random sequence model for determining the significance of window scores in dot-matrix analysis (Reich and Meiske, 1987); and Argos’ significance tests (Argos and Vingron, 1990), based on actual protein sequences

instead of the randomized sequences used by McLachlan (1972). Despite efforts to determine the quality of alignments, it remains true that low statistical significance does not always imply a lack of biological significance (Collins and Coulson, 1990). Another important aspect of alignments is the accuracy of the original molecular sequence (States, 1992). Errors can be introduced in molecular sequence data in random or biased ways. The translation of nucleic acid sequences into amino acid sequences is very sensitive to indel errors: an error rate of 1% can lead to a frameshifting error in reading frames longer than 24 amino acids. Empirical estimates of sequence accuracy in current databases have revealed a rate of 3 to 5 errors per 1000 bases.

CLUSTER METHODS AND TREES Once a multiple alignment has been performed and regions of common similarity have been identified among the proteins, the question of evolutionary distance will arise. There are several programs for determining the level of divergence between a group of sequences. Some are available over the Internet via the World Wide Web (WWW; see also Internet Resources section). The programs DISTANCES and GROWTREE from the GCG package (Genetics Computer Group, 1994) calculate pairwise distance measures and plot a dendrogram of those measures. The programs depend on an accurate multiple alignment of the sequences involved. This means that the multiple alignments should be examined carefully and then edited as necessary to incorporate additional knowledge that might be available about the structure or function of the proteins. Figure 2.1.6 shows the table of pairwise phylogenetic distances calculated by DISTANCES for the top nine albumins identified from the FASTA analysis (Fig. 2.1.4), and the phylogram produced by GROWTREE is shown in Figure 2.1.7. The purpose of protein sequence comparison is to learn more about the structure, function, and evolution of proteins. It is known that different proteins and different parts of proteins change at characteristic but unequal rates in the course of evolution (Doolittle, 1986). Binding sites and catalytic units are conserved independently of their position, whereas surface regions change faster than buried regions. In addition, entire domains of a genome are subject to a number of editing operations such as domain joining, swapping, and reordering (States and Boguski, 1990). Thus, homologies


Matrix 1, dimension: 9 Key for column and row indices: 1 2 3 4 5 6 7 8 9

Albu_Bovin.Sw Albu_Chick.Sw Albu_Horse.Sw Albu_Human.Sw Albu_Mouse.Sw Albu_Pig.Sw Albu_Ranca.Sw Albu_Rat.Sw Albu_Sheep.Sw

1 2 3 4 5 6 7 8 9 __________________________________________________________________________ .. | 1 | 0.00 999.99 31.75 192.01 999.99 999.99 999.99 237.70 8.19 | 2 | 0.00 999.99 999.99 999.99 999.99 999.99 999.99 999.99 | 3 | 0.00 190.56 999.99 999.99 999.99 222.61 29.05 | 4 | 0.00 999.99 999.99 999.99 32.75 193.48 | 5 | 0.00 999.99 999.99 999.99 999.99 | 6 | 0.00 999.99 999.99 999.99 | 7 | 0.00 999.99 999.99 | 8 | 0.00 237.70 | 9 | 0.00

Figure 2.1.6 Evolutionary distance table for mammalian blood proteins calculated using the GCG DISTANCES program. Corrections to the calculations were done using the Kimura protein method as described in the GCG manual (Genetics Computer Group, 1994). The matrix values represent the number of matches between each pair of sequences divided by its length (999.9 = identity). These data were used to generate the GROWTREE phylogram presented in Figure 2.1.7.

Growtree Phylogram of: Albu_Bovin.Distances, Tree Tree_ February 3, 1995 15:02 Albu_Ranca.Sw Albu_Pig.Sw Albu_Mouse.Sw Albu_Chick.Sw Albu_Rat.Sw Albu_Human.Sw Albu_Horse.Sw Albu_Sheep.Sw Albu_Bovin.Sw

Figure 2.1.7 GROWTREE phylogram of Albu_Bovin.Distances, Tree Tree_1. Results of graphical cluster analysis of the distance data presented in Figure 2.1.6 are shown. In this representation sequences with the least divergence have the shortest branches.



might be found between entire proteins or just between certain parts of proteins.

IDENTIFICATION OF FUNCTIONAL SITES


Sequence comparisons are used to test for homology and to find functional sites and motifs. The goal is to establish structure and function for parts of an unknown sequence using known homologous or analogous (i.e., similar, but not homologous) sequences as sources of additional information. The goal is to obtain answers to questions of what functions the protein has and what its structure is. There are several paths the analysis can take. One approach that will reveal much about potential functional sites in the protein is to look for motifs. Motifs are defined in three ways: as consensus patterns of amino acids that correspond to regions of proteins containing the common residues; as the excised alignment of several sequences that contain the motif for use as a database discriminator; or as a single representative sequence (States and Boguski, 1990; Parry-Smith and Atwood, 1992). Motif searches are usually done using a program that takes advantage of a dictionary of sequence patterns which have been identified with specific protein activities. Many programs will do this analysis, and most use the PROSITE database of motifs found through analysis of the SWISS-PROT protein database. Figure 2.1.8 shows an example using the GCG MOTIFS program to analyze an albumin sequence. Note that the output file contains an alignment of the motif found and an extensive annotation describing the motif and listing the associated bibliography. The PIPL program developed by Staden (1994a-d) searches libraries of sequences to find patterns, where a pattern is a set of motifs with variable spacing. Patterns are constructed using logical operators to form a search term that is then used to scan a database such as PROSITE (Bairoch, 1992). The MOTIFS program of the Genetics Computer Group (1994) uses PROSITE as a dictionary of motifs. Weight matrices are used in scanning for motifs and developing a motif pattern. The profile analysis method (Gribskov et al., 1988, 1990; Gribskov, 1994) also involves weight matrices and is related to the consensus method. It is used for sensitive similarity matching, for homology seeking, and for scanning databases for motifs. It measures the similarity between a given sequence, called the target, and a group of aligned sequences called

the probe. The alignment of the probe uses a modified MDM78 mutational matrix. The matrix values of the profile are computed columnwise as the weighted average of the score for each residue at that position in the aligned sequences and the residue represented by the column of the profile; a row of the profile is the weighted average of the rows in the MDM78 table corresponding to the aligned residues at that position. The result is the set of properties that tend to be conserved at particular positions along the sequence. For detailed information on using profile analysis software to scan for homologies and motifs, see Gribskov (1994). A relatively new homologous sequence retrieval program (using the BLOSUM250 matrix) is dFLASH, which is available by E-mail server (Rigoutsos and Califano, 1994). There are many ways to proceed in computational analysis of new sequences. Each will provide a wealth of information to guide further laboratory experimentation. The following procedures illustrate two approaches to protein sequence analysis and can be adapted as required. Hodgman (1992) suggests the following protocol for protein function determination: 1. Split sequences into overlapping 200- to 300-residue segments. 2. Use a program such as FASTA to search a protein database for sequences similar to those in the segments. 3. If a match having >25% identity over 80 residues is found, then the search has been successful. 4a. For weaker matches, use dot-matrix plots to verify similarity. 4b. Alternatively, search the database again with the matching sequence to find related sequences for alignment to detect conserved positions. 5. Conserved regions can be used to search further with PIPL (Staden, 1994a-d), PROFILESEARCH (GCG), or SCRUTINEER (EMBL). 6. Use any biological information about the conserved regions to determine an analogous function for the test protein. 7. If the procedure has not been successful, use other resources such as PROSITE. 8. Check that all results are biologically credible. 9. If all else fails, (a) look for repeats within the sequence itself: if there are only two or three, it usually indicates that the protein has a repeated functional domain; (b) examine the distribution of particular amino acids or classes


Motifs Analysis Atp_Gtp_A

(A,G)x4GK(S,T) (G)x{4}GK(S) 425: HIVLS GESYSGKS TNARL

the motif from the Prosite database the pattern matched in this sequence the acutal region of sequence containing the motif

Attached annotation ***************************************** * ATP/GTP-binding site motif A (P-loop) * ***************************************** From sequence comparisons and crystallographic data analysis it has been shown [1,2,3,4,5] that an appreciable proportion of proteins that bind ATP or GTP share a number of more or less conserved sequence motifs. The best conserved of these motifs is a glycine-rich region, which probably forms a flexible loop between a betastrand and an alpha-helix. This loop interacts with one of the phosphate groups of the nucleotide. This sequence motif is generally referred to as the 'A' consensus sequence [1] or the 'P-loop' [5]. There are numerous ATP- or GTP-binding proteins in which the P-loop is found.We list below a number of protein families for which the relevance of the presence of such motif has been noted:

A list of proteins containing the motif has been omitted for brevity. Not all ATP- or GTP-binding proteins are picked-up by this motif. A number of proteins escape detection because the structure of their ATP-binding site is completely different from that of the P-loop. Examples of such proteins are the E1-E2 ATPases or the glycolytic kinases. In other ATP- or GTP-binding proteins the flexible loop exists in a slightly different form; this is the case for tubulins or protein kinases. A special mention must be reserved for adenylate kinase, in which there is a single deviation from the P-loop pattern: in the last position Gly is found instead of Ser or Thr. -Consensus pattern: [AG]-x(4)-G-K-[ST] -Sequences known to belong to this class detected by the pattern: a majority. -Other sequence(s) detected in SWISS-PROT: in addition to the proteins listed above, the 'A' motif is also found in a number of other proteins. Most of these proteins probably bind a nucleotide, but others are definitively not ATP- or GTP-binding (as for example chymotrypsin, or human ferritin light chain). -Expert(s) to contact by email: Koonin E.V. [email protected] -Last update: June 1994 / Text revised. [ 1] Walker J.E., Saraste M., Runswick M.J., Gay N.J. EMBO J. 1:945-951(1982). [ 2] Moller W., Amons R. FEBS Lett. 186:1-7(1985). [ 3] Fry D.C., Kuby S.A., Mildvan A.S. Proc. Natl. Acad. Sci. U.S.A. 83:907-911(1986). [ 4] Dever T.E., Glynias M.J., Merrick W.C.

Figure 2.1.8 MOTIFS analysis of an albumin sequence. This example shows results for a single motif found in the test protein. If the query sequence contains many motifs, the output can be quite extensive. Note that a long table of sequences which contain the motif and some of the references have been deleted due to space constraints.

for suggested functional sites; or (c) apply biological and biochemical knowledge to evaluate the composition of the sequence (e.g., regions rich in basic residues suggest the binding of nucleic acids, and acidic regions suggest generalized protein-protein interactions). Structure determination can begin with the ChouFasman (Chou and Fasman, 1974) or GarnierOsguthorpe-Robson algorithm (Garnier et al., 1978); both are discussed in UNIT 2.3. States and Boguski (1990) offer another scheme for the analysis of protein sequences:

1. Find all intrasequence repeats using dotmatrix methods, and, if repeats are found, split the sequence into pieces accordingly. 2. Use FASTA to find similar sequences or use PROFILESEARCH for a more sensitive search. 3. Compare the sequence to known motifs, for example, in the PROSITE database. 4. Search for hydrophobic segments using the Kyte-Doolittle (1982) method and for α helices and β strands using the hydrophobic moment (see UNITS 2.2 & 2.3).



5. If step 1 or 2 was successful, then search for helices using the variation moment. 6. If step 1 or 2 was successful, calculate the hydrophobic correlation to estimate similarity. 7. Keeping in mind that the results will be only 55% to 65% accurate, compare the outputs of several secondary structure prediction methods with information from all available homologous sequences. 8. For antibodies, identify possible antigenic regions using hydrophilicity plots and antigenicity index plots. 9. Look for unusual features such as leucine zippers; this can be done with any program, such as PROSITE, that takes advantage of a dictionary of motifs. 10. Determine potential sites of cleavage by various chemicals; cleavage sites can be predicted by computer methods and verified experimentally.

entific and medical researchers. A few are presented here. For a listing of molecular sequence and related software, get the Catalogue of Molecular Biology from Genethon by connecting via ftp ftp.genethon.fr and entering cd pub/resig/catalogu then mget bio.catal*. Amos Bairoch’s bibliographic database on sequence analysis (seqanalref), the Listing of Molecular Biology Databases (LiMB), a directory of researchers who apply artificial intelligence (AI) techniques to problems in molecular biology (aimbdb), and many other valuable files can be found using anonymous FTP to ncbi.nlm.nih.gov. Bairoch’s lists of molecular biology e-mail, FTP, and BBO servers are also available using anonymous FTP: type ftp expasy.hcuge.ch, then cd databases/info, then mget serv_ema.txt serv_ftp.txt serv_bbo.txt. For a copy of the BIOSCI FAQ, a collection of Frequently Asked Questions, use ftp net.bio.net, then cd pub/BIOSCI/doc and get biosci.FAQ.

INTERNET RESOURCES


The Internet has become an important resource in bioinformatics. There are new tools that make it easy to locate the latest software and databases. Since 1994 a new Internet-based network of hypertext tools that takes advantage of the World Wide Web (WWW) has emerged. To enter the WWW, you will need a network hypertext browser such as MOSAIC. Versions of MOSAIC, which was originally developed by the National Center for Supercomputer Applications (NCSA), are now freely available for most commonly used laboratory computers. If your local computer is connected to the Internet and your system has file transfer protocol (FTP) capability, connect to ftp.ncsa.uiuc.edu and log in under the username anonymous to retrieve a copy of MOSAIC for the computer you use most often. Once you have MOSAIC or another WWW browser running, we recommend connecting to the Australian National University’s Bioinformatics server (http://life.anu.edu.au) and selecting “Molecular biology” from the menu. Table 2.1.1 reproduces the list of resources available. Another current resource of hyperlinks is available through the author using MOSAIC or Netscape at the George Mason Comp uta tion al B iolo gy Resource (http://www.science.gmu.edu/~michaels/ comp_bio). As indicated by that list, there is a tremendous number of resources available on the Internet. In addition, there are many useful text and program files available by FTP that are of interest to molecular biologists and other sci-

LITERATURE CITED Allison, L., Wallace, C.S., and Yee, C.N. 1992. Finite-state models in the alignment of macromolecules. J. Mol. Evol. 35:77-89. Altschul, S.F., Gish, W., Miller, W., Myers, E.W., and Lipman, D.J. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-410. Argos, P. 1987. A sensitive procedure to compare amino acid sequences. J. Mol. Biol. 193:385396. Argos, P. and Vingron, M. 1990. Sensitivity comparison of protein amino acid sequences. Methods Enzymol. 183:352-365. Bairoch, A. 1992. PROSITE: A dictionary of protein sites and patterns. Nucl. Acids Res. 19:22412245. Barton, G.J. 1990. Protein multiple sequence alignment and flexible pattern matching. Methods Enzymol. 183:403-428. Barton, G.J. 1993. An efficient algorithm to locate all locally optimal alignments between two sequences allowing for gaps. CABIOS 9:729-734. Brendel, V. 1992. PROSET—A fast procedure to create nonredundant sets of protein sequences. Math. Comput. Modeling 16(6/7):37-43. Brendel, V., Bucher, P., Nourbaksh, I.R., Blaisdell, B.E., and Karlin, S. 1992. Methods and algorithms for statistical analysis of protein sequences. Proc. Natl. Acad. Sci. U.S.A. 89:20022006. Burks, C. 1990. The flow of nucleotide sequence data into data banks: Role and impact of largescale sequencing projects. In Computers and DNA, Santa Fe Institute (G. Bell and T. Marr, eds.) pp. 35-45. Addison-Wesley, Reading, Mass.


Table 2.1.1

Australian National University (ANU) Molecular Biology Resources

Databases French WWW genomic service, including Caenorhabditis elegans Database BLAST Database Searches at NCBI CompoundKB database—981 metabolic intermediate compounds Codon usage tables (major species) & EMBL mirror DNA Data Bank of Japan EC enzyme database EMBL data (Heidelberg) ESTDB—Expressed Sequence Tag Database (TIGR) FASTA Database search program (Virginia) Genbank searches (Indiana) Genbank/Swiss-Prot/Protein/PIR (via NIH) GENETHON Human Genome Centre Mendelian Inheritance in Man (index) Metabolic intermediate compounds Microbial germplasm Miscellaneous, e.g. codon usage, profiles (Weizmann) PIR (Houston) Prosite (via NIH) Protein databank (Brookhaven) REF52 2D Gel Protein Database REBASE—Restrict. Enzyme Data Base (NEB) REBASE restriction enzymes Swiss-Prot: EMBNet (Heidelberg) and ExPASy (Geneva) SWISS-2DPAGE—Two-dimensional Polyacrylamide Gel Electrophoresis Database (Geneva) Various (ICGEB, Italy) Biologists’s Control Panel Bibliographies/Tutorials Biosequence Comparison Sequence analysis (search index) Biocomputing bibliography Biological Journals—current titles Molecular biology journals References to molecular biology algorithms Periodical references to journals in molecular biology Software ANU software (gopher link) SIMPLE34—Detection of sequence repetition in nucleic acid MELANIE software packages for 2D PAGE computer analysis (Geneva) NCBI repository

Sequence Analysis Tools (Trieste) IUBio Biology Software and Database Archive (Indiana) Quest II software for 2D gel proteins CODA—Conservation Options and Decision Analysis RAPDistance (gopher link) RAPDistance (WWW link) News groups BioNet—various biological topics The Scientist (biweekly news with a biotech slant) BIOSCI mailing lists and newsgroup archives Special topics The Australian National Genomic Information Service (ANGIS) Human Genome Mapping Project (UK) Artificial Intelligence in molecular biology Database of molecular biologists working in AI Agricultural Genome World Wide Web server A Caenorhabditis elegans DataBase (ACeDB) Other sites Australian National Genomic Information Service (ANGIS) CAMIS Centre for Advanced Medical Informatics (Stanford) Caenorhabditis Genetics Centre (CGC) Chlamydomonas Genetics (Duke) EMBNet ExPASy (Geneva) Harvard Biological Laboratories Human Genome Center at Lawrence Berkeley Laboratory Johns Hopkins University, BioInformatics National Center for Biotechnology Information (NCBI) QUEST Protein Database Center Ribosomal Database Project (Argon Natl. Laboratory) Molecular biology archives (list) NIH GenoBase server Othersites (gopher links) Yeast Genome Information Server (Stanford) WAIS—EC enzymes WAIS—Molecular Biology WAIS—indexes sorted alphabetically (ADFA)



Chou, P.Y. and Fasman, G.D. 1974. Prediction of protein conformation. Biochemistry 13:222-244. Collado-Vides, J. 1991. The search for a grammatical theory of gene regulation is formally justified by showing the inadequacy of context-free grammars. CABIOS 7:321-326. Collins, J.F. and Coulson, A.F.W. 1990. Significance of protein sequence similarities. Methods Enzymol. 183:474-487. Day, W.H.E. and McMorris, F.R. 1993. A consensus program for molecular sequences. CABIOS 9:653-656. Dayhoff, M.O. 1978. Atlas of Protein Sequence and Structure. National Biomedical Research Foundation, Washington, D.C. Depiereux, E. and Feytmans, E. 1991. Simultaneous and multivariate alignment of protein sequences: Correspondence between physicochemical profiles and structurally conserved regions (SCR). Protein Eng. 4:603-613. De Rijk, P. and De Wachter, R. 1993. DCSE, an interactive tool for sequence alignment and secondary structure search. CABIOS 9:735-740. Doolittle, R.F. 1981. Similar amino acid sequences: Chance or common ancestry? Science 214:167339. Doolittle, R.F. 1986. Of URFs and ORFs: A Primer on How to Analyze Derived Amino Acid Sequences. University Science Books, Ann Arbor, Mich. Doolittle, R.F. 1989. Redundancies in protein sequences. In Prediction of Protein Structure and the Principles of Protein Conformation (G.D. Fasman, ed.) pp. 599-623. Plenum, New York.


Doolittle, R.F. 1990. What we have learned and will learn from sequence databases. In Computers and DNA, Santa Fe Institute (G. Bell and T. Marr, eds.) pp. 21-31. Addison-Wesley, Reading, Mass. Dumas, J.P. and Nunio, J. 1982. Efficient algorithm for folding and comparing nucleic acid sequences. Nucl. Acids Res. 10:197-206. Eroshkin, A.M., Zhilkin, P.A., and Fomin, V.I. 1993. Algorithm and computer program: Pro_Anal for analysis of relationship between structure and activity in a family of proteins or peptides. CABIOS 9:491-497. Fitch, W.M. 1966. An improved method of testing for evolutionary homology. J. Mol. Biol. 16:916. Fitch, W.M. 1969. Locating gaps in amino acid sequences to optimize the homology between two proteins. Biochem. Genet. 3:99-108. Fuchs, R. 1994. Fast protein block searches. CABIOS 10:79-80. Garnier, J., Osguthorpe, D.J., and Robson, B. 1978. Analysis of the accuracy and implications of simple methods for predicting the secondary structure of globular proteins. J. Mol. Biol. 120:97-120.

Genetics Computer Group. 1994. GCG Program Manual for the Wisconsin Package, Version 8, September 1994. Genetics Computer Group Inc., Madison, Wis. George, D., Hunt, L.T., and Barker, W.C. 1990. Mutation data matrix and its uses. Methods Enzymol. 183:333-351. Gibbs, A.J. and McIntyre, G.A. 1970. The diagram, a method for comparing sequences. J. Biochem. 16:1-11. Gotoh, O. 1986. Alignment of three biological sequences with an efficient traceback procedure. J. Theor. Biol. 121:327-337. Gribskov, M., Homyak, M., Edenfield, J., and Eisenberg, D. 1988. Profile scanning for three-dimensional structural patterns in protein sequences. CABIOS 4:61-66. Gribskov, M., Luethy, R., and Eisenberg, D. 1990. Profile analysis. Methods Enzymol. 183:146159. Gribskov, M. 1994. Profile analysis. Computer analysis of sequence data. Methods Mol. Biol. 24:247-266. Henikoff, S. and Henikoff, J.G. 1993. Proteins Struct. Funct. Genet. 17:49-61. Heringa, J., Sommerfeldt, H., Higgins, D.G., and Argos, P. 1992. OBSTRUCT: A program to obtain the largest cliques from a protein sequence set according to structural resolution and sequence similarity. CABIOS 8:599-600. Hodgman, T.C. 1992. Nucleic acid and protein sequence management. In Microcomputers in Biochemistry: A Practical Approach (C.F.A. Bryce, ed.) pp. 131-158. IRL Press, Oxford. Kanaoka, M., Kishimoto, F., Ueki, Y., and Umeyama, H. 1989. Alignment of protein sequences using the hydrophobic core scores. Protein Eng. 2:347-351. Karlin, S.P., Morris, M., Ghandour, G., and Leung M.-Y. 1988. Algorithms for identifying local molecular sequence features. CABIOS 4:41-51. Karlin, S.P., Ost, F., and Blaisdell, B.E. 1989. Patterns in DNA and amino acid sequences and their statistical significance. In Mathematical Methods for DNA Sequences (M.S. Waterman, ed.) pp. 133-157. CRC Press, Boca Raton, Fla. Karlin, S., Bucher, P., and Brendel, V. 1991. Statistical methods and insights for protein and DNA sequences. Annu. Rev. Biophys. Chem. 20:175203. Kruskal, J.B. 1983. An overview of sequence comparison. In Time Warps, String Edits, and Macromolecules: The Theory and Practice of Sequence Comparison (D. Sankoff and J.B. Kruskal, eds.) pp. 1-44. Addison-Wesley, Reading, Mass. Kruskal, J.B. and Sankoff, D. 1983. An anthology of algorithms and concepts for sequence comparison. In Time Warps, String Edits, and Macromolecules: The Theory and Practice of Sequence Comparison (D. Sankoff and J.B. Kruskal, eds.) pp. 265-310. Addison-Wesley, Reading, Mass.


Kyte, J. and Doolittle, R.F. 1982. A simple method for displaying the hydrophobic character of a protein. J. Mol. Biol. 157:105-132. Landau, G.M., Vishkin, U., and Nussinov, R. 1988. Locating alignments with k differences for nucleotide and amino acid sequences. CABIOS 4:19-24. Landau, G.M., Vishkin, U., and Nussinov, R. 1990. Fast alignment of DNA and protein sequences. Methods Enzymol. 183:487-502.

Pearson, W.R. and Miller, W. 1992. Dynamic programming algorithms for biological sequence comparison. Methods Enzymol. 210:576-610. Pevzner, P.A. 1992. Statistical distance between texts and filtration methods in sequence comparison. CABIOS 8:121-127. Pizzi, E.M., Attimonelli, M., Liuni, S., Frontali, C., and Saccone, C. 1991. A simple method for global sequence comparison. Nucl. Acids Res. 20:131-136.

Landes, C., Henaut, A., and Risler, J.-L. 1993. Dotplot comparisons by multivariate analysis (DOCMA): A tool for classifying protein sequences. CABIOS 9:91-196. Lipman, D.J. and Pearson, W.R. 1985. Rapid and sensitive protein similarity searches. Science 227:1435-1441. Livingstone, C.D. and Barton, G.F. 1993. Protein sequence alignments: A strategy for the hierarchical analysis of residue conservation. CABIOS 9:745-756.

Reich, J.G. and Meiske, W. 1987. A simple statistical significance test of window scores in large dot matrices obtained from protein or nucleic acid sequences. Comput. Appl. Biosci. 3:25-30. Rigoutsos, I. and Califano, A. 1994. Searching in parallel for similar strings. IEEE Computatl. Sci. Eng. 60-75. Robson, B. and Greaney, P.J. 1992. Natural sequence code representations for compression and rapid searching of human-genome-style databases. CABIOS 8:283-289.

Maizel, J.V. and Lenk, R.P. 1981. Enhanced graphic matrix analysis of nucleic acids and protein sequences. Proc. Natl. Acad. Sci. U.S.A. 78:76657669. McLachlan, A.D. 1971. Test for comparing related amino acid sequences: Cytochrome c and cytochrome c-551. J. Mol. Biol. 61:409-424. McLachlan, A.D. 1972. Repeating sequences and gene duplication in proteins. J. Mol. Biol. 72:417-437. Michaels, G.S., Taylor, R., Hagstrom, R., Price, M., and Overbeek, R. 1993. Searching for genomic organizational motifs: Explorations of the E. coli chromosome. Comp. Chem. 17:209-217.

Rohde, K. and Bork, P. 1993. A fast, sensitive pattern-matching approach for protein sequences. CABIOS 9:183-189.

Mrazek, J. and Kypr, J. 1993. UNIREP: A microcomputer program to find unique and repetitive nucleotide sequences in genomes. CABIOS 9:355-360. Murata, M., Richardson, J.S., and Sussman, J.L. 1985. Simultaneous comparison of three protein sequences. Proc. Natl. Acad. Sci. U.S.A. 82:3073-3077. Nedde, D.N. and Ward, M.O. 1993. Visualizing relationships between nucleic acid sequences using correlation images. CABIOS 9:331-335. Needleman, S.B. and Wünsch, C.D. 1970. A general method applicable to the search for similarities in the amino acid sequence of two proteins. J. Mol. Biol. 48:443-453. Panjukov, V.V. 1993. Finding steady alignments: Similarity and distance. CABIOS 9:285-290. Parry-Smith, D.J. and Atwood, T.K. 1992. ADSP— A new package for computational sequence analysis. CABIOS 8:451-459. Pearson, W.R. 1990. Rapid and sensitive comparison with FASTP and FASTA. Methods Enzymol. 183:63-98. Pearson, W.R. 1994. Using the FASTA program to search protein and DNA sequence databases. Methods Mol. Biol. 24:365-389.

Searls, D. 1993. The computational linguistics of biological sequences. In Artificial Intelligence and Molecular Biology (L. Hunter, ed.) pp. 47120. MIT Press, Cambridge, Mass. Sellers, P.H. 1974. On the theory and computation of evolutionary distances. SIAM J. Appl. Math. 26:787-793. Smith, R.F. and Smith, T.F. 1992. Pattern-induced multisequence alignment (PIMA) algorithm employing secondary structure–dependent gap penalties for use in comparative protein modeling. Protein Eng. 5:35-41. Smith, T.F. and Waterman, M.S. 1981. Comparative biosequence metrics. J. Mol. Evol. 18:38-46. Staden, R. 1994a. Statistical and structural analysis of protein sequences. Methods Mol. Biol. 24:125-130. Staden, R. 1994b. Searching for motifs in protein sequences. Methods Mol. Biol. 24:131-139. Staden, R. 1994c. Using patterns to analyze protein sequences. Methods Mol. Biol. 24:141-154. Staden, R. 1994d. Comparing sequences. Methods Mol. Biol. 24:155-170. States, D.J. 1992. Molecular sequence accuracy: Analyzing imperfect data. Trends Genet. 8:5255. States, D.J. and Boguski, M.S. 1990. Sequence Analysis Primer. Stockton Press, New York. Streletc, V.B., Shindyalov, I.N., Kolchanov, N.A., and Lim, H.A. 1991. Fast, statistically based alignment of amino acid sequences on the base of diagonal fragments of dot matrices. CABIOS 8:529-534. Taylor, W.R. 1988. Pattern matching methods in protein sequence comparison and structure prediction. Protein Eng. 2:77-86. Computational Analysis


Tyler, E., Horton, M.R., and Krause, P.R. 1991. A review of algorithms for molecular sequence comparison. Computers Biomed. Res. 24:72-96. Waterman, M.S. 1989. Sequence alignments. In Mathematical Methods for DNA Sequences (M.S. Waterman, ed.) pp. 53-90. CRC Press, Boca Raton, Fla. Waterman, M.S. 1990. Consensus patterns in sequences. In Mathematical Methods for DNA Sequences (M.S. Waterman, ed.) pp. 93-115. CRC Press, Boca Raton, Fla. Waterman, M.S. and Eggert, M. 1991. A new algorithm for best subsequence alignments with application to tRNA-rRNA comparisons. J. Mol. Biol. 197:723-728.

Waterman, M.S. and Jones, R. 1990. Consensus methods for DNA and protein sequence alignment. Methods Enzymol. 183:221-237. Wilbur, W.J. and Lipman, D.J. 1983. Rapid similarity searches of nucleic acid and protein data banks. Proc. Natl. Acad. Sci. U.S.A. 80:726-730.

Contributed by George Michaels and Robert Garian George Mason University Fairfax, Virginia



Hydrophobicity Profiles for Protein Sequence Analysis INTRODUCTION Seminal work by Anfinsen et al. (1961) showed that the three-dimensional structure of a protein is determined by its amino acid sequence. Thus, significant effort has been aimed at unraveling the complex relationship between sequence and structure. Although there is no immediate prospect of a general solution to the protein folding problem, stepwise empirical methods have been developed that analyze protein sequences and successfully provide predictive results concerning protein structure. The work of Kauzmann (1959) determined that hydrophobic interactions are a major force in protein folding. Accounting for the hydrophobicity of amino acids within a protein is thus the first step toward understanding protein folding (Tanford, 1980). Hydrophobicity studies have resulted in a wide range of hydropathy scales, indices that attempt to quantify the relative hydrophobicity of the amino acids. The actual hydrophobic content attributed to each amino acid is dependent on the methodology used to estimate it and may vary from scale to scale. Some scales are based solely on physicochemical measurements such as the free energy of side-chain transfer (Nozaki and Tanford, 1971; Wolfenden et al., 1981). Other scales are based on statistical analysis of threedimensional structures (Chothia, 1976; Rose et al., 1985). In addition, scales can be a combination of both procedures (Kyte and Doolittle, 1982). The general observation that hydrophobic residues tend to be buried in the interior of the protein whereas hydrophilic residues prefer to be exposed to solvent is predominant in all scales. Generating a hydropathy profile requires two things: (1) the primary structure and (2) the hydropathy scale. The primary structure can be obtained by translating the cDNA or by sequencing the protein. Hydropathy profiles for a protein are obtained by averaging or combining the hydrophobicity or hydrophilicity of an individual amino acid (taken from a particular hydropathy scale) with the values of several neighboring residues along the protein sequence, then plotting the local averages or sums against the amino acid sequence. In such hydrophobicity plots, regions of hydrophobicity and hydrophilicity appear as maxima and minima, respectively. A fundamental use of hy-

dropathy profiles, therefore, is predicting the overall hydropathy (hydrophobicity or hydrophilicity) of a protein from its amino acid sequence. The distribution of local hydropathy values can provide insight into the overall folding pattern of the protein (see Applications). Numerous procedures have been developed for calculating the hydropathy profile of a protein based on the free energy of side-chain transfer from water or vacuum to an organic solvent (Nozaki and Tanford, 1971; Tanford, 1980; Wolfenden et al., 1981; Fauchere and Pliska, 1983). The procedures differ primarily in the hydropathy scale used, and the different hydropathy scales have arisen, in part, to address specific applications. For example, the hydrophilicity scale of Hopp and Woods (1981) was developed to locate antigenic segments; the Goldman, Engleman, and Steitz (GES) scale (Engleman et al., 1986) was developed to identify transmembrane regions in a protein. Although individual scales were developed for different purposes or by alternative procedures, significant similarity exists in the resultant profiles. Most hydropathy plots show similar distributions of minima and maxima in the sequence profiles. In Figure 2.2.1, the result of applying the most widely used hydrophobicity scale, that of Kyte and Doolittle (1982), is compared to results of applying the scales of Hopp and Woods (1981) and Eisenberg (1984) to hen egg white lysozyme. The similarity in profiles is partially due to the fact that maxima and minima often reflect secondary structure elements; regions of hydrophobicity and hydrophilicity can be correlated with β sheets and β turns, respectively (Kuntz, 1972; Roy and Rose, 1980). Hydropathy profiles can be used to examine the surface features of proteins in order to generate hypotheses that can be confirmed experimentally. For example, such analyses received widespread use in the prediction of antigenic and other interaction sites on proteins. Furthermore, when used in conjunction with other secondary structure prediction algorithms (see UNIT 2.3), hydropathy profiles have been shown to correlate well with α-helical, turn, and β-sheet propensities. As such, they have proved to be invaluable for predicting protein secondary structure. The most extensive use of hydropathy profiles has been for

Contributed by Stanley R. Krystek, Jr., William J. Metzler, and Jiri Novotny Current Protocols in Protein Science (1995) 2.2.1-2.2.13 Copyright © 2000 by John Wiley & Sons, Inc.

UNIT 2.2


2.2.1 CPPS

20

Hydropathy value

10

0

–10

– 20

–30 0

50

100

150

Sequence position

Figure 2.2.1 Hydropathy profiles of hen egg white lysozyme. The three profiles were created with the Kyte and Doolittle, the Hopp and Woods, and the Eisenberg scales using a scanning window of 7 amino acids. The Hopp and Woods scale was modified as described in the text to match other hydropathy scales.

deducing transmembrane regions in membrane proteins. Previous reviews describe systematic evaluations of hydropathy procedures and provide criteria for selecting the most appropriate amino acid hydropathy scale (Engleman et al., 1986; Cornette et al., 1987; Banghan, 1988; Hopp, 1989; Rose and Dworkin, 1989). This unit describes the application of hydrophobicity plots to typical problems and provides suggested uses for a few selected scales.

METHODOLOGY Selecting Hydropathy Scales

Hydrophobicity Profiles for Protein Sequence Analysis

The first step in generating a hydropathy profile (or profiles) for an individual amino acid sequence is selecting the appropriate scale(s). Table 2.2.1 lists several widely used scales. This unit focuses primarily on four hydropathy scales: the Hopp and Woods, the Kyte and Doolittle, the Eisenberg, and the GES scales. Differences among the hydropathy indices for each amino acid in the four scales (Table 2.2.2) arise primarily from differences in the experimental data from which the scales were derived and from the methodologies by

which each scale was optimized. The hydrophilicity scale derived by Hopp and Woods is based on solubility values of hydrophobic and hydrophilic amino acids (Nozaki and Tanford, 1971; Levitt, 1976). In the Kyte and Doolittle and the Eisenberg scales, the hydropathy index is a measure of the relative affinity of an amino acid for hydrophobic phases. The GES scale is based on physicochemical considerations of helical structure and several energetic factors that reflect partitioning of an amino acid side chain between aqueous solution and a membrane bilayer. Both the Kyte and Doolittle and the Eisenberg scales are combinations of earlier scales. The Kyte and Doolittle scale combines values from the water-to-vapor transfers and the internal/external distributions of amino acid residues from known structures as determined by Chothia (1976). The Eisenberg scale is a consensus of the Wolfenden (Wolfenden et al., 1981), Chothia (Chothia, 1976), Janin (Janin, 1979), and von Heijne (von Heijne and Blomberg, 1979) scales and is designed to minimize effects of the outlying values observed for some amino acids in previous scales.


Table 2.2.1

Hydropathy Scales

Reference scale

Literature

Derivationa

Bigelow Tanford Chothia Levitt Janin von Heijne and Blomberg Roy and Rose Hopp and Woods Wolfenden Argos Fraga Kyte and Doolittle Fauchere and Pliska Eisenberg Hopp Rose GES Parker, Guo, and Hodges Esposti et al.

Bigelow, 1967 Nozaki and Tanford, 1971 Chothia, 1976 Levitt, 1976 Janin, 1979 von Heijne and Blomberg, 1979 Roy and Rose, 1980 Hopp and Woods, 1981 Wolfenden et al., 1981 Argos et al., 1982 Fraga, 1982 Kyte and Doolittle, 1982 Fauchere and Pliska, 1983 Eisenberg, 1984 Hopp, 1984 Rose et al., 1985 Engleman et al., 1986 Parker et al., 1986 Esposti et al., 1990

Experimental Experimental Theoretical Experimental Theoretical Theoretical Experimental Experimental Experimental Theoretical Theoretical Combination Experimental Combination Theoretical Theoretical Theoretical Experimental Theoretical

aExperimental, largely original data from physicochemical experiments; theoretical, derivation based on

analysis of structural data; combination, consensus of multiple scales.

There are several other interesting scales that deserve mention. The first hydropathy scale for membrane proteins was developed and revised by von Heijne (von Heijne and Blomberg, 1979) and corresponds rather closely to the GES scale described above (Engleman et al., 1986). A hydrophilicity scale by Parker, Guo, and Hodges (1986) provides new hydrophilicity values that are determined from HPLC retention times of model synthetic peptides. Another important set of scales includes the statistical scales developed by Argos and co-workers for prediction of membraneburied helices (Argos et al., 1982; Rao and Argos, 1986). These scales are derived from the statistical preference of specific amino acid residues to form the integral regions in transmembrane proteins, and they have been shown to correlate with highly resolved crystal structures much better than the physicochemical hydropathy scales (Esposti et al., 1990). Because all hydropathy scales are conceptually similar, the strategy of applying them to a given problem should be analogous regardless of the particular scale chosen. Moreover, although there may always be debates over which of the scales (methods) will prove to be best, experience has shown that application of

several different hydropathy schemes and alternative predictive rules will tend to build a consensus picture of protein hydrophobicity. For this unit the values of the Hopp and Woods and GES scales are multiplied by −1 so that these scales comply with others having peak maxima corresponding to hydrophobicity and peak minima corresponding to hydrophilicity (the scale values were not modified in Table 2.2.2). The Hopp and Woods and GES plots shown are therefore inverse plots.

Applying the Scanning Window Averaging Technique After the appropriate scale(s) has been selected, the next stage is generating the hydropathy profile for a protein. To create the profile, the numerical values derived from the hydropathy scales for each residue in the protein are repetitively averaged over the length of protein sequence. This procedure is commonly referred to as the scanning window averaging technique (Hopp and Woods, 1981; Kyte and Doolittle, 1982). In practice, the hydropathy is obtained by summing the individual hydropathy values (hydrophobicities or hydrophilicities) for a given number of contiguous amino acids along the protein chain. The sum of hy-



Table 2.2.2

Hydrophobicity Indices for Selected Scales

Amino acid

H & Wa

K & Db

Eisenbergc GESd

Neutral, hydrophobic, aliphatic −0.4 Gly 0.0 −0.5 Ala 1.8 −1.5 Val 4.2 −1.8 Ile 4.5 −1.8 Leu 3.8 −1.3 Met 1.9

0.16 0.25 0.54 0.73 0.53 0.26

−1.0 −1.6 −2.6 −3.1 −2.8 −3.4

Neutral, hydrophobic, aromatic −2.5 Phe 2.8 −2.3 −1.3 Tyr −3.4 −0.9 Trp

0.61 0.02 0.37

−3.7 0.7 −1.9

Neutral, hydrophilic Ser 0.3 −0.4 Thr Asn 0.2 Gln 0.2

−0.8 −0.7 −3.5 −3.5

−0.26 −0.18 −0.64 −0.69

−0.6 −1.2 4.8 4.1

Acidic, hydrophilic Asp 3.0 Glu 3.0

−3.5 −3.5

−0.72 −0.62

9.2 8.2

Basic, hydrophilic −0.5 His Lys 3.0 Arg 3.0

−3.2 −3.9 −4.5

−0.40 −1.1 −1.8

3.0 8.8 12.3

Thiol-containing −1.0 Cys

2.5

0.04

−2.0

−1.6

−0.07

0.2

Imino acid Pro

0.0

aHopp and Woods, 1981. bKyte and Doolittle, 1982. cEisenberg, 1984. dEngleman et al., 1986.


dropathy values is calculated progressively through the sequence, with each amino acid being the start of a new sum (Fig. 2.2.2). The resultant sums (or averages) are plotted versus the sequence position at the position of the central (or first) amino acid in the segment being summed. The choice of the center versus the first residue has no effect on the calculated profiles but shifts the sequence axis by half the window size. We prefer to plot versus the center position as it seems more intuitive. Similarly, the choice of summing or averaging has no effect on the profile but modifies the relative values plotted on the y-axis. We use sums for the profiles presented in this unit. This approach

displays the general hydropathic character of a protein. The ability to discern structurally interpretable features in the calculated hydropathy profile is critically dependent on the width of the scanning window, that is, the number of amino acids included in the summation. Figure 2.2.3 shows the effect of altering the size of the scanning window on hydropathy profiles generated for lysozyme. In Figure 2.2.3A, where the window size is set to 1 amino acid, the resultant hydropathy profile is featureless, providing essentially no structural information. On introducing a window size of 7 amino acids, well-resolved peaks appear in the hydropathy


10 20 protein sequence . . . RTYFCDEQASQDWLVNAR . . . sliding window (size of 7) average hydropathy . . . 4 5 6 7 8 9 10 11 12 13 14 . . . value mapped to center residue of window

Figure 2.2.2 Schematic depicting the scanning window averaging technique.

profile (Fig. 2.2.3B). Increasing the window size to 11 amino acids (Fig. 2.2.3C) results in a smoothing or broadening of observed peaks in the profile, and with window sizes of 15 or 19 amino acids (Fig. 2.2.3D,E), the profiles again have few well-resolved features. From the family of profiles in Figure 2.2.3, it is readily apparent that window sizes exceeding 10 amino acids result in a loss of local information. This decrease in local information leads to increased errors in predicting hydrophobic regions in globular proteins (Kyte and Doolittle, 1982). In contrast, significant correlations do exist between hydropathy profiles generated with window sizes of 5 to 9 amino acids and known structural elements (Roy and Rose, 1980; Kyte and Doolittle, 1982; Rao and Argos, 1986; Cornette et al., 1987). In selecting the appropriate window size for a particular problem, the length of the structural property being investigated should be used as a guide. For example, transmembrane helices are roughly 20 residues in length, but the helices of globular proteins tend to be much shorter. One might therefore opt for a larger window size if searching for transmembrane helices, whereas a length of 7 or 9 amino acids is more appropriate for predicting surface sites or secondary structure. In cases of uncertainty, analysis of profiles generated with various window sizes will afford the most reliable results.

Threshold or Cutoff Values To facilitate the analysis of hydropathy profiles, it is convenient to define a threshold value, that is, the amplitude above which the hydropathy value must remain for a particular amino acid to be considered part of a contiguous region. Although somewhat arbitrary, the actual choice of threshold value for a particular profile is often derived empirically or from

experience. For example, if one were interested in analyzing the sequences of known membrane proteins for transmembrane helices, then an appropriate threshold value might be one which delineates peaks in the hydropathy profile to approximately 20 amino acids (see section on Predicting Transmembrane Regions in Proteins). As with selecting scanning window size, analyzing profiles with different threshold values should yield more reliable results until experience is gained.

APPLICATIONS Predicting Interaction Sites Hydropathy profiles have found widespread use in investigations aimed at elucidating the antigenic structure and surface features of proteins for which there are no three-dimensional structures (Hopp and Woods, 1981; Krystek et al., 1985a,b). Antigenic sites tend to occur on protein surfaces, where predominantly hydrophilic residues are both exposed to solvent and accessible for interaction with antibody. One would therefore anticipate good correlation between antigenic sites and regions of high hydrophilicity. A test of this hypothesis carried out by Tanaka et al. (1985) reported a success rate of 56%. Figure 2.2.4 depicts the Hopp and Woods profiles for myoglobin generated with a window size of 7 amino acids. Regions of hydrophilicity are detected as peak minima in the profile. The lines at the bottom of Figure 2.2.4 indicate the locations of five major antigenic sites determined experimentally (Atassi, 1984). All of the antigenic sites are associated with regions of high hydrophilicity. In addition, there are two regions of high hydrophilicity (peaks at positions 40 to 45 and 78 to 84) that are not associated with antigenicity. It is plau-



A

10 5 0 –5 –10 20

B

10 0 –10 –20 –30 –40

C Hydrophobicity value

20

D

10 0 –10 –20 –30 –40 20 10 0 –10 –20 –30 –40

E

20 10 0 –10 –20 –30 –40 0

50

100

150

Sequence position


Figure 2.2.3 Effect of altering the scanning window size on the resolution of the hydropathy profile of lysozyme. All profiles were created with the Kyte and Doolittle scale. Window sizes were set to (A) 1, (B) 7, (C) 11, (D) 15, and (E) 19 amino acids. Note that the scale in A is reduced to show the profile more clearly.


15

Hydrophilicity value

10

5

0

–5

–10

–15 0

50

100

150

Sequence position

Figure 2.2.4 Hopp and Woods profile for sperm whale myoglobin using a scanning window of 7 amino acids. Lines below the profile indicate the locations of five experimentally determined antigenic sites (Atassi, 1984).

sible that these correspond to antigenic sites that have yet to be identified experimentally, or they may point to an alternate, functionally significant property of myoglobin. Whatever the case, these results serve to emphasize the point that hydropathy profiles should be used only as a guide for predicting antigenicity. Additional features of protein surfaces can be highlighted by hydropathy profiles. As an example, consider human hemoglobin, whose three-dimensional structure is known (Fermi, 1975). Hemoglobin is found as a tetramer in its native state, being composed of two α- and two β-subunits. In the native protein the β- and α-subunits are folded against one another, forming contact regions that more closely resemble the interior of a globular protein. Because these contact regions are hydrophobic, one might examine hydropathy profiles for regions of hydrophobicity as a means of identifying potential subunit interacting surfaces. The Hopp and Woods profile for human hemoglobin is shown in Figure 2.2.5. The regions of contact in the β-subunit of human hemoglobin (Yoshioka and Atassi, 1986) are indicated as bars above the profile. The two largest hydrophobic peaks correspond to regions of the β-subunit in contact with each of

the α-subunits. Analogous analysis of hydropathy profiles has been successfully used to identify the subunit interacting surfaces for the α-subunits of glycoprotein hormones lutropin and follitropin (Krystek et al., 1991, 1992). Human hemoglobin also provides a demonstration of the dangers of using hydropathy profiles for analyzing antigenic sites of proteins that possess significant quaternary structure. In Figure 2.2.5 the lines below the profile indicate the regions of five major continuous antigenic determinants. Although two sites correlate with regions of hydrophilicity, the others do not. In such cases the amino acid sequence alone is not sufficient for predictability. This may be due in part to the unique nature of proteins composed of multiple subunits.

Correlating Profiles with Secondary Structure An increasingly important application of hydropathy profiles is their use in conjunction with secondary structure prediction tools (see UNIT 2.3). Figure 2.2.6 shows the hydropathy profiles for interleukin 1β (IL-1β) generated with the Eisenberg scale (Fig. 2.2.6A) and the Kyte and Doolittle scale (Fig. 2.2.6B), using a scan window of 7 amino acids. The solid bars



15

Hydrophilicity value

10

5

0

–5

–10 0

50

100

150

Sequence position

Figure 2.2.5 Hopp and Woods profile for human hemoglobin β-subunit using a scanning window size of 7 amino acids. Bars above the profile mark the locations of β-subunit surfaces in contact with α-subunits (Yoshioka and Atassi, 1986). Lines below the profile indicate the locations of antigenic sites.


above the profiles indicate the locations of β strands found in the crystal structure of this all-β protein (Finzel et al., 1989). Although there is a difference in scale magnitude between the Kyte and Doolittle and Eisenberg methods, both profiles show exceptionally good correlation of hydrophobicity (peak maxima) with the location of β strands. The correlation of hydrophobicity with β-sheet structural elements is a direct reflection of the fact that sheets are most often located in the hydrophobic interior of a protein; thus, they comprise residues most compatible with the hydrophobic environment. There is also good correlation between hydrophilicity (peak minima) and the location of turns in proteins. For interleukin 1β essentially all of the turns identified in the structure correspond to minima. Figure 2.2.7 shows the hydropathy profiles for myoglobin, a protein composed solely of α helices. The larger α helices appear to correlate with peak maxima. Like β sheets, helices are frequently found in regions of hydrophobicity in the hydrophobic core of proteins. Because of the interspersed nature of hydrophilic and hydrophobic residues of amphipathic helices, however, hydropathy profiles are less useful in

predicting the location of these helices. Moreover, as both helices and sheets result in peak maxima, the hydropathy profile is not sufficient to discriminate between them. Therefore, correlations of secondary structure with protein hydropathic character are most successful when used in conjunction with protein secondary structure prediction algorithms (see UNIT 2.3).

Predicting Transmembrane Regions in Proteins Perhaps the most popular use of hydropathy profiles is predicting transmembrane regions in proteins. In addition to hydropathy scales, there are several membrane preference scales. Such scales have been developed independently from the hydropathy scales and are based on statistical distributions of each amino acid in membrane regions of proteins. The statistical scales have been previously compared to the hydropathy scales (Esposti et al., 1990); in general, they are equally efficient at predicting the location of the transmembrane and amphipathic helices. Examining the structure of known membrane proteins has enabled the predictive rules


6

A

4 2 0 –2

B

Hydropathy value

–4 –6 –8 20 10 0 –10 –20 –30 0

50

100

150

Sequence position

Figure 2.2.6 Hydropathy profiles for interleukin 1β using a scanning window size of 7 amino acids. Bars above the profile indicate locations of the β strands, and lines below the profile mark the locations of turns (Finzel et al., 1989). (A) Profile generated using the Eisenberg scale. (B) Profile generated using the Kyte and Doolittle scale.

for the various scales to be optimized. In addition, estimates of the hydrophobic thickness of the bilayer have provided guides to the minimum number of helical amino acids necessary to span the bilayer: the minimum number is estimated to be 20 (Engleman et al., 1986). This greatly facilitates the prediction of putative transmembrane helices by adding an additional criterion to the analysis of hydropathy profiles for transmembrane regions, namely, maxima in the profiles must span ∼20 residues. Figure 2.2.8A shows the Kyte and Doolittle profile for bacteriorhodopsin, a membrane protein whose structure is known (Henderson et al., 1990). For Kyte and Doolittle profiles, the best resolution is seen for window scans of 7 or 9 amino acids and threshold values of 0.7 × window size. For a window scan of 7, the

threshold value is 4.9. There are seven major peaks with threshold values greater than 4.9 that span at least 20 amino acids (indicated by brackets above the profile in Fig. 2.2.8A). Comparison of the predicted transmembrane regions with the location of the actual helices identified in the structure (bars above the profile) indicates that the prediction is quite good. Similar agreement is seen when the method of Engleman et al. (1986) is used to predict the transmembrane regions for bacteriorhodopsin (Fig. 2.2.8B). Although the location of the helices can be predicted reliably, neither the Kyte and Doolittle nor the GES method is able to predict accurately the boundaries of the transmembrane regions. Fortunately, visual inspection of the sequence, as well as general features of the



30

Hydrophobicity value

20

10

0

–10

–20

–30 0

50

100

150

Sequence position

Figure 2.2.7 Kyte and Doolittle hydropathy profile for sperm whale myoglobin generated with a scanning window size of 7 amino acids. Bars above the profile mark the locations of the helices in myoglobin.


transmembrane helices, can frequently be used to delineate transmembrane regions and the interfacial residues at each end of the helices. Some general features common to transmembrane regions to consider are as follows. (1) Approximately 75% of the amino acids in transmembrane regions are hydrophobic (Leu, Ile, Val, Met, Phe, Trp, Tyr, Cys, Ala, Pro, and Gly). (2) Proline residues are found in ∼20% of the transmembrane helices but in only 3% of globular proteins (Barlow and Thornton, 1988). (3) Aromatic residues are clustered near the interface of the transmembrane helix and bulk water. Positional preferences for amino acids in transmembrane regions have been identified for single-spanning membrane proteins (Landolt-Marticorena et al., 1993) and can serve as a further guide for delineating the ends of transmembrane regions. It can be expected that these preferences will be similar for polytopic membrane proteins. The positional preferences define a motif with the following characteristics: an extracellular terminal flanking region (Asp, Ser and Pro); an extracellular-interfacial region (Trp); a transmembrane domain composed primarily of hydrophobic residues (Leu, Ile, Val, Met, Phe, Trp, Cys, Ala, Pro, and Gly); an

intracellular-interfacial region (Tyr, Trp, and Phe); and an intracellular terminal flanking region (Lys and Arg). Application of these sequence analysis rules for transmembrane proteins greatly improves the results of using hydropathy profiles for predicting transmembrane regions.

CONCLUSIONS 1. Virtually all hydropathy scales are able to reveal structural information in an equivalent manner. Selection of the appropriate hydropathy scale is therefore a practical and/or personal decision. 2. Window scanning procedures should use an appropriate window size for the structural element being examined: 7 or 9 amino acids for predicting surface sites or secondary structure and 19 amino acids for predicting membranespanning regions. 3. Identification of interaction sites can be accomplished using peak minima (hydrophilicity). 4. Correlation of protein secondary structure with hydropathy profiles may be useful for protein structure prediction methodology (see UNIT 2.3).


A

30

Hydrophobicity value

20

10

0

–10

Free energy of transfer to water

B

–20 50 40 30 20 10 0 –10 –20 –30 –40 –50 0

50

100

150

200

250

300

Sequence position

Figure 2.2.8 Hydropathy profiles for bacteriorhodopsin. Bars above the profiles mark the locations of the helices. (A) Kyte and Doolittle profile generated using a scanning window size of 7 amino acids. Brackets indicate the helices predicted on the basis of the profile alone. (B) GES profile generated using a scanning window size of 19 amino acids.

5. Prediction of transmembrane regions of integral membrane proteins can be made using a variety of scales and protocols; analysis of the hydropathy profiles should be augmented with the applicable sequence analysis rules.

rithms can also easily be encoded into programming languages or spreadsheets and interfaced with numerous graphical software packages for plotting the hydropathy profiles (if graphics are not included in the particular prediction program). These are also discussed in UNIT 2.1.

ACCESSIBILITY OF SOFTWARE Most commercial personal computer or computational packages (Genetics Computer Group, 1994; Biosym Technologies, 1991; Tripos Associates, 1993) contain algorithms for hydropathy profile generation along with other secondary structure prediction tools (see Table 2.3.6). In addition, the authors of each scale or protocol have published analysis programs (Table 2.2.1), of which many are available from public domain sources on the Internet. Algo-

LITERATURE CITED Anfinsen, C.B., Haber, E., Sela, M., and White, F.H., Jr. 1961. The kinetics of formation of native ribonuclease during oxidation of the reduced polypeptide chain. Proc. Natl. Acad. Sci. U.S.A. 47:1309-1314. Argos, P., Rao, J.K.M., and Hargrave, P.A. 1982. Structural prediction of membrane-bound proteins. Eur. J. Biochem. 128:565-575. Atassi, M.Z. 1984. Antigenic structure of proteins. Eur. J. Biochem. 145:1-20.



Banghan, J.A. 1988. Data-sieving hydrophobicity plots. Anal. Biochem. 174:142-145. Barlow, D.J. and Thornton, J. 1988. Alpha helices in proteins. J. Mol. Biol. 210:601-619. Bigelow, C.C. 1967. On the average hydrophobicity of proteins and the relation between it and protein structure. J. Theor. Biol. 16:187-211. Biosym Technologies. 1991. Insight II/Homology Modules. Biosym Technologies, San Diego. Chothia, C. 1976. The nature of the accessible and buried surfaces of proteins. J. Mol. Biol. 105:114. Cornette, J.L., Cease, K.B., Margalit, H., Spouge, J.L., Berzofsky, J.A., and DeLisa, C. 1987. Hydrophobicity scales and computational techniques for detecting amphipathic structures in proteins. J. Mol. Biol. 195:659-685. Eisenberg, D. 1984. Three-dimensional structure of membrane and surface proteins. Annu. Rev. Biochem. 53:595-623. Engleman, D.M., Steitz, T.A., and Goldman, A. 1986. Identifying nonpolar transbilayer helices in amino acid sequences of membrane proteins. Annu. Rev. Biophys. Chem. 15:321-353. Esposti, M.D., Crimi, M., and Venturoli, G. 1990. A critical evaluation of the hydropathy profile in membrane proteins. Eur. J. Biochem. 190:207219. Fauchere, J.L. and Pliska, V. 1983. Hydrophobic parameters P of amino acid side chains from the partitioning of N-acetyl-amino acid amides. Eur. J. Med. Chem. 18:369-375. Fermi, G. 1975. Three-dimensional Fourier synthesis of human deoxyhaemoglobin at 2.5 angstroms resolution, refinement of the atomic model. J. Mol. Biol. 97:237-256. Finzel, B.C., Clancy, L.L., Holland, D.R., Muchmore, S.W., Watenpaugh, K.D., and Einspahr, H.M. 1989. Crystal structure of recombinant human interleukin-1β at 2.0 angstroms resolution. J. Mol. Biol. 209:779-791. Fraga, S. 1982. Theoretical prediction of protein antigenic determinants from amino acid sequences. Can. J. Chem. 60:2606-2610. Genetics Computer Group. 1994. GCG Program Manual for the Wisconsin Package. Genetics Computer Group, Inc., Madison, Wis. Henderson, R., Baldwin, J.M., Ceska, T.A., Zemlin, F., Beckmann, E., and Downing, K.H. 1990. Model for the structure of bacteriorhodopsin based on high resolution electron cryomicroscopy. J. Mol. Biol. 213:899-929. Hopp, T.P. 1984. Protein antigen conformation: Folding patterns and predictive algorithms; selection of antigenic and immunogenic peptides. Ann. Sclavo 2:47-60.


Hopp, T.P. 1989. Use of hydrophilicity plotting procedures to identify protein antigenic segments and other interaction sites. Methods Enzymol. 178:571-585.

Hopp, T.P. and Woods, K.R. 1981. Prediction of protein antigenic determinants from amino acid sequences. Proc. Natl. Acad. Sci. U.S.A. 78:3824-3828. Janin, J. 1979. Surface and inside volumes in globular proteins. Nature 277:491-492. Kauzmann, W. 1959. Some factors in the interpretation of protein denaturation. Adv. Prot. Chem. 14:1-63. Krystek, S.R., Jr., Dias, J.A., Reichert, L.E., and Andersen, T.T. 1985a. Prediction of antigenic sites in follicle stimulating hormones: Difference profiles enhance antigenicity prediction methods. Endocrinology 117:1125-1130. Krystek, S.R., Jr., Reichert, L.E., and Andersen, T.T. 1985b. Analysis of computer generated hydropathy profiles for human glycoprotein and lactogenic hormones. Endocrinology 117:11101117. Krystek, S.R., Jr., Dias, J.A., and Andersen, T.T. 1991. Identification of subunit contact sites on the α-subunit of lutropin. Biochemistry 30:18581864. Krystek, S.R., Jr., Dias, J.A., and Andersen, T.T. 1992. Identification of a subunit contact site on the α-subunit of follitropin. Pept. Res. 25:165168. Kuntz, I.D. 1972. Protein folding. J. Am. Chem. Soc. 94:4009-4012. Kyte, J. and Doolittle, R.F. 1982. A simple method for displaying the hydropathic character of a protein. J. Mol. Biol. 157:105-132. Landolt-Marticorena, C., Williams, K.A., Deber, C.M., and Reithmeier, A.F. 1993. Nonrandom distribution of amino acids in the transmembrane segments of human type I single span membrane proteins. J. Mol. Biol. 229:602-608. Levitt, M. 1976. A simplified representation of protein conformations for rapid simulation of protein folding. J. Mol. Biol. 104:59-107. Nozaki, Y. and Tanford, C. 1971. The solubility of amino acids and two glycine peptides in aqueous ethanol and dioxane solutions. J. Biol. Chem. 246:2211-2217. Parker, J.M.R., Guo, D., and Hodges, R.S. 1986. New hydrophobicity scale derived from highperformance liquid chromatography peptide retention data: Correlation of predicted surface residues with antigenicity and X-ray-derived accessible sites. Biochemistry 25:5425-5432. Rao, J.K.M. and Argos, P. 1986. A conformational preference parameter to predict helices in integral membrane proteins. Biochim. Biophys. Acta 869:197-214. Rose, G.D. and Dworkin, J.E. (1989). The hydrophobicity profile. In Prediction of Protein Structure and the Principles of Protein Conformation (G.D. Fasman, ed.) pp. 625-633. Plenum, New York. Rose, G.D., Geselowitz, A.R., Lesser, G.J., Lee, R.H., and Zehfus, M.H. 1985. Hydrophobicity of amino acid residues in globular proteins. Science 229:834-838.


Roy, S. and Rose, G.D. 1980. Hydrophobic basis of packing in globular proteins. Proc. Natl. Acad. Sci. U.S.A. 8:4643-4647. Tanaka, T., Slamon, D.J., and Cline, M.J. 1985. Efficient generation of antibodies to oncoproteins by using synthetic peptide antigens. Proc. Natl. Acad. Sci. U.S.A. 82:3400-3404. Tanford, C. 1980. The Hydrophobic Effect. John Wiley & Sons, New York. von Heijne, G. and Blomberg, C. 1979. Transmembrane translocation of proteins. The direct transfer model. Eur. J. Biochem. 97:175-181. Tripos Associates. 1993. Biopolymer/Composer Module. Tripos Associates, Inc., St. Louis, Mo. Wolfenden, R., Andersson, L., Cullis, P.M., and Southgate, C.C. 1981. Affinities of amino acid side chains for solvent water. Biochemistry 20:849-855.

Yoshioka, N. and Atassi, M.Z. 1986. Subunit interacting surfaces of human haemoglobin. Biochem. J. 234:457-461.

KEY REFERENCES Engleman et al., 1986. See above. Provides applications of hydropathy profiles to transmembrane region prediction. Kyte and Doolittle, 1982. See above. Describes the fundamental development and application of hydropathy profiles.

Contributed by Stanley R. Krystek, Jr., William J. Metzler, and Jiri Novotny Bristol-Myers Squibb Pharmaceutical Research Institute Princeton, New Jersey



Protein Secondary Structure Prediction It is widely accepted that knowledge of the structure of a protein can provide important information regarding its function and mechanism of action. However, determining the threedimensional structure of a protein, whether by X-ray crystallography (Wyckoff et al., 1985; Blundell and Johnson, 1976) or nuclear magnetic resonance (NMR) spectroscopy (Oppenheimer and James, 1989; Wuthrich, 1986), remains a formidable task, as it requires relatively large amounts of pure protein (generally greater than milligram quantities). Thus, the structures of many proteins will remain out of reach. At the same time, continual advances in molecular biology provide protein sequence information (primary structures) at a pace that far exceeds the speed with which higher-order protein structures can be determined, thus making the need for methods to derive structural information even more pressing. This unit describes procedures developed for predicting protein structure from the amino acid sequence. On the basis of function and location in the cell, proteins can be loosely divided into three groups: globular, structural, and membrane. To date, the majority of protein structural data has been derived from studies of globular (soluble) proteins because of the ease of chemical handling of globular proteins. Thus, a wide disparity has developed in the number of three-dimensional structures that have been experimentally determined in the three protein groups. As a result, current secondary structure prediction schemes apply almost solely to globular proteins. In the native structures of globular proteins, the hydrophobic residues are generally buried in the interior and the hydrophilic residues are exposed to the surrounding solvent (Kauzmann, 1959; Fisher, 1964; Dill, 1990). Disulfide bonds are often present and stabilize the structure. Sections of amino acid sequence rich in hydrophobic residues can be correlated with secondary structural elements such as α helices and β strands (Roy and Rose, 1980). Those rich in polar residues correspond to surface loops and β turns (Kuntz, 1972; UNIT 2.2). These secondary structural elements (Pauling and Corey, 1951; Pauling et al., 1951) are characterized by hydrogen bonding between the amino and carboxyl groups of the polypeptide backbone (Schulz and Schirmer, 1979). Several reviews of protein structure exist, and readers are referred to books by Schulz and

Schirmer (1979) and Branden and Tooze (1991) for thorough discussions of protein structure. Protein secondary structure elements associate to form compact domains. Levitt and Chothia (1976) and Richardson (1981) have classified protein domains according to the types of arrangements of secondary structures that form them. Illustrations for the four classes, referred to as all-α, all-β, α/β, and α + β, are available (Richardson, 1981; Lesk, 1991). This unit focuses on examining secondary structure predictions using specific examples from different protein classes; methods of predicting folding class are also discussed. This unit is divided into four sections. The first section is an overview and brief history of structure prediction schemes. The second section describes four distinct prediction schemes, with emphasis on their differences. In the third part each prediction scheme is used to evaluate three proteins that have different folding patterns. The final section is a comparison of the prediction results and suggestions for secondary structure prediction.

OVERVIEW OF PREDICTION SCHEMES Early Predictive Schemes Two of the earliest secondary structure prediction methods were developed by Guzzo (1965) and Prothero (1966). By analyzing a database of protein structures and corresponding sequences, they generated a set of rules for predicting structure. Because of the limited number of structures available at the time, their prediction methods were restricted to helices, the most abundant conformation in the structures available. Soon after, Schiffer and Edmunson (1967) developed the helical wheel method, a pictorial tool that is still widely used. The helical wheel is a circular plot that represents the view one obtains when looking down the helix axis (Fig. 2.3.1). The “spokes” of the wheel are precisely positioned on the basis of the Corey-Pauling α-helical geometry, having 3.6 residues per turn, which results in an angular shift of two neighboring residues of 100° (Pauling et al., 1951). A helical wheel is constructed by placing successive amino acids at the spokes of the wheel. Helical wheels have been particularly valuable for identifying proteins and peptides that contain amphipathic helices. In an amphipathic

Contributed by Stanley R. Krystek, Jr., William J. Metzler, and Jiri Novotny Current Protocols in Protein Science (1995) 2.3.1-2.3.20 Copyright © 2000 by John Wiley & Sons, Inc.

UNIT 2.3


2.3.1 CPPS

T

K

11

7

P

A 4

14 G

A

15

3

T 10

8

17

1

I

L

6

V

G

12 G

5

13 L

2 I

9 L

16

V

L

Figure 2.3.1 Helical wheel for a portion of the amphipathic helix of melittin. Residues in bold type form a hydrophobic side. Note that this is not a perfect amphipathic helix as the polar face includes Ala-4, Ala-15, and Val-8. This helix also contains several glycine residues, which are not commonly found in helices (Table 2.3.1).

Protein Secondary Structure Prediction

helix, hydrophobic residues will cluster to one side of the wheel and polar residues will be found on the opposite side. Figure 2.3.1 shows a helical wheel for the amphipathic helix of melittin. Helical wheels have often been used for analyzing helical dimerization proteins such as leucine zippers (Krystek et al., 1991; O’Shea et al., 1992), for designing helical multimers (Harbury et al., 1993; Kamtekar et al., 1993; Lovejoy et al., 1993), and for predicting and analyzing membrane proteins which contain helices that span the lipid bilayer (Baldwin, 1993). Other early predictive schemes include those developed by Dunhill (1968) and Lim (1974). Dunhill’s “helical net” is a pictorial method that is similar to the helical wheel and is still used in protein design and helix prediction (DeGrado, 1988; Goodman and Kim, 1991). Lim’s method relies on an extensive set of rules derived from the physicochemical properties of amino acids in conjunction with the analysis of a structural database. The large number of detailed rules, however, makes the Lim method unwieldy. When several methods were compared in a blind test organized by Schulz et al. (1974), the methods of Ptitsyn and Finkelstein (1970) and Chou and Fasman (1974a,b) performed better than others (Burgess et al., 1971; Lewis et al., 1971; Robson and Pain, 1971; Nagano, 1973).

Statistical Prediction Schemes The statistical methods developed by Chou and Fasman (1974a,b) and by Garnier et al. (1978) are the most prominent of the class of predictive schemes based on statistical amino acid preference rules. These secondary structure prediction schemes tacitly assume that only local sequence interactions influence the formation of local structure. This is almost certainly an oversimplification. Significant interactions occur between protein secondary structural elements (tertiary structure); these interactions may involve hydrophobic and/or electrostatic interactions between structural units. Therefore, the success and accuracy of prediction schemes that rely on empirical algorithms are necessarily limited and are unlikely to exceed a level of 70% to 75%. Reports of secondary structure prediction accuracy levels of ∼70%, however, are not uncommon using these methods (Fasman, 1985).

Other Predictive Schemes Two additional methodologies for structure prediction not based on statistics of amino acid occurrence have evolved. Of the neural network methods (Quinn and Sejnowski, 1988; Rost and Sander, 1993a,b; Kneller et al., 1990), that of Rost and Sander claims a level of predictive accuracy exceeding 70% for all proteins tested and is publicly available through the


Internet computer network (for details, see discussion of Applying the PHD Scheme and Table 2.3.6 therein). The method of Stultz et al. (1993) is based on state-space modeling with a set of predefined structural class models.

Selecting Prediction Tools Secondary structure prediction proceeds as follows. First, it should be confirmed that the three-dimensional structure has not yet been determined. This can be done by searching known X-ray and NMR structures in the Protein Data Bank (Brookhaven, 1994) for an amino acid sequence identical to that of interest. Next, identification of proteins with significantly similar sequences whose structures are already known should be attempted. Numerous software packages are available for performing similarity searches (see UNIT 2.1). If proteins of similar sequence (>20% identity) are found, the Protein Data Bank should again be searched to determine whether the structures of any of those proteins are known. If so, then comparative homology modeling (Greer, 1990; Bajorath et al., 1993) can be used to develop a three-dimensional model of the desired sequence. Only if there are no proteins in the structural database of sufficient similarity should protein secondary prediction schemes be used. It should be pointed out, however, that proteins may have the same fold without a detectable sequence similarity (Kabsch and Sander, 1984, 1985). In the future, sequence threading through known three-dimensional structures will be helpful (Jones et al., 1992; Sippl and Weitckus, 1992). It is recommended that protein secondary structure prediction be performed in conjunction with other sequence analysis approaches, such as hydrophobicity profiling, flexibility analysis, surface exposure experiments, charge distribution mapping, and motif searches (UNIT 2.2; Fasman, 1989b; Genetics Computer Group, 1994). For investigators using methods to predict protein secondary structure, at least two questions should be addressed: (1) which predictive scheme(s) should be used, and (2) what are the advantages of one method over others? This unit attempts to find answers to these questions by studying three proteins in detail: human hemoglobin β-subunit, human interleukin 1β (IL-1β), and human profilin I (see Application of Secondary Prediction Methods and Fig 2.3.2).

METHODS FOR SECONDARY STRUCTURE PREDICTION The Chou and Fasman Method The most popular secondary structure prediction scheme is the statistical method of Chou and Fasman (1974a). Using a database of 15 protein structures determined by X-ray crystallography, Chou and Fasman tabulated the number of occurrences of a given amino acid in α helix, β strand, or coil. From the tabulation, a set of conformational preference parameters for each amino acid type was calculated. The preference parameters were based on three functions: (1) the relative frequency of a specific amino acid type within each protein, (2) the occurrence of that amino acid in a given type of secondary structure, and (3) the fraction of amino acid residues occurring in each type of secondary structure. With the conformational preference parameters in hand, the authors then derived a set of empirical rules for predicting protein secondary structure (Chou and Fasman, 1974b). Chou and Fasman later published an extension of their analysis using a larger set of X-ray structures, totaling 29 proteins and over 4700 amino acids. In addition to refining the values for α helices and β strands, they included a conformational preference parameter for β turns as well (Chou and Fasman, 1977, 1978a,b). For β turns, a positional preference for residues was identified; for example, proline was found to occur more frequently in position 2 of the turn than in positions 1, 3, or 4. A further refined set of conformational preference parameters for helical and sheet residues was later derived from an expanded data set of 64 protein structures (Chou, 1989). These parameters are similar to those based on the 29 protein structures (differences are noted in Chou, 1989). An interesting addition to secondary structure prediction using the conformational preference parameters is the prediction of protein class. Chou presented a computerized algorithm that assigns a structural class to proteins on the basis of amino acid composition; however, these predictions have not been widely used. In the initial studies, Chou and Fasman used the conformational preference parameters to describe amino acids as being formers, breakers, or indifferent for helical and strand regions (Chou and Fasman, 1974a). Their analysis showed that residues with the highest prefer-



A T α, β

HB

B T α, β

HB

C T α, β

HB Residue number Figure 2.3.2 Secondary structure prediction using the method of Chou and Fasman as implemented by Novotny and Auffray (1984). Predicted secondary structure for (A) human hemoglobin β-subunit, (B) interleukin 1β, and (C) human profilin I. Below the amino acid sequence are given the turn prediction (T), the α helix (thin line) and β strand prediction (thick line), the location of positive (upward spikes) or negative charges (downward spikes) as derived specifically from amino acid side chains, and a hydrophobicity profile (HB).


ence for helices resided near the helix center, whereas residues with low helix preference could be found clustered at the helix termini. For helix and strand, residues were independently classified as strong formers, formers, weak formers, indifferent, breakers, and strong breakers. Table 2.3.1 contains a list of the conformational preference parameters and refined conformational assignments derived from the analysis of 64 proteins (Chou, 1989). These

values are combined with the turn preferences shown in Table 2.3.2, or values from the 29protein study (Chou and Fasman, 1978a,b; Prevelige and Fasman, 1989), to carry out sequence prediction as described below. An important contribution to helix prediction strategies is the recognition that helix start and stop signals in the amino acid sequence grammar are quite distinct (Presta and Rose, 1988; Richardson and Richardson, 1988). The


Table 2.3.1 β Stranda

Conformational Preference and Assignments for α Helix and

α helix

β strand

Residue

Pα

Assignmentb

Glu Ala Met Leu Lys His Gln Phe Asp Trp Arg Ile Val Cys Thr Asn Tyr Ser Gly Pro

1.44 1.39 1.32 1.30 1.21 1.12 1.12 1.11 1.06 1.03 1.00 0.99 0.97 0.95 0.78 0.78 0.73 0.72 0.63 0.55

Hα Hα Hα Hα hα hα hα hα hα Iα Iα iα iα iα iα iα bα bα Bα Bα

Residue

Pβ

Assignmentc

Val Ile Thr Tyr Trp Phe Leu Cys Met Gln Ser Arg Gly His Ala Lys Asp Asn Pro Glu

1.64 1.57 1.33 1.31 1.24 1.23 1.17 1.07 1.01 1.00 0.94 0.94 0.87 0.83 0.79 0.73 0.66 0.66 0.62 0.51

Hβ Hβ hβ hβ hβ hβ hβ hβ Iβ Iβ iβ iβ iβ iβ iβ bβ bβ bβ Bβ Bβ

aValues taken from Chou, 1989. bHα, strong helix former; hα, helix former; Iα, weak helix former; iα, indifferent; bα, helix

breaker; Bα, strong helix breaker. cHβ, strong β-strand former; hβ, β-strand former; Iβ, weak β-strand former; iβ, indifferent; bβ,

β-strand breaker; Bβ, strong β-strand breaker

α-helical structures in proteins are commonly stabilized by reciprocal hydrogen bond formation between the unpaired main-chain amino and carboxyl groups near the helix termini and the side chains of amino acids flanking the termini (Harper and Rose, 1993). The helix-terminal residues are referred to as the N-cap and C-cap. Because of the unique geometry of proline, this amino acid is frequently found at the N-cap + 1 position and has been referred to as being an α-helix initiator (Richardson and Richardson, 1988). Residues that typically occupy the N-cap and C-cap positions of helices are listed in Table 2.3.3.

The Garnier, Osguthorpe, and Robson Method The Garnier, Osguthorpe, and Robson (GOR) method (Garnier et al., 1978; Garnier and Robson, 1989) is a widely available secondary structure prediction method that has been incorporated into several computer packages,

such as that from Genetics Computer Group (GCG). Although the GOR method incorporates essentially all the elements of the ChouFasman statistics, it also uses information theory to reach the final decision. A detailed description of the GOR method can be found elsewhere (Garnier and Robson, 1989). In principle, the GOR method assigns the conformation of a given residue by considering the contribution of residues relatively distant to it in the sequence. On the basis of studies showing the interdependence of residue conformations up to eight amino acids away, the GOR method assigns single-residue information values to each amino acid along the polypeptide chain. The predicted conformational state for an amino acid is the state with the highest positive information value. An extension of this procedure, the GORIII method, has been developed to use pair information instead of directional information. The GORIII method extracts both single-residue information values



Table 2.3.2

Conformational Preference and Assignments for β Turnsa

Positional preferences for β turnsb

Residue Preference fi Asn Gly Pro Asp Ser Cys Tyr Lys Gln Thr Trp Arg His Glu Ala Met Phe Leu Val Ile

1.56 1.56 1.52 1.46 1.43 1.19 1.14 1.01 0.98 0.96 0.96 0.95 0.95 0.74 0.66 0.60 0.60 0.59 0.50 0.47

Asn Cys Asp His Ser Pro Gly Thr Tyr Trp Gln Arg Met Val Leu Ala Phe Glu Lys Ile

0.161 0.149 0.147 0.140 0.120 0.102 0.102 0.086 0.082 0.077 0.074 0.070 0.068 0.062 0.061 0.060 0.059 0.056 0.055 0.043

fi+2

fi+1 Pro Ser Lys Asp Thr Arg Gln Gly Asn Met Ala Tyr Glu Cys Val His Phe Ile Leu Trp

0.301 0.139 0.115 0.110 0.108 0.106 0.098 0.085 0.083 0.082 0.076 0.065 0.060 0.053 0.048 0.047 0.041 0.034 0.025 0.013

Asn Gly Asp Ser Cys Tyr Arg His Glu Lys Thr Phe Trp Gln Leu Ala Pro Val Met Ile

0.191 0.190 0.179 0.125 0.117 0.114 0.099 0.093 0.077 0.072 0.065 0.065 0.064 0.037 0.036 0.035 0.034 0.028 0.014 0.013

fi+3 Trp Gly Cys Tyr Ser Gln Lys Asn Arg Asp Thr Leu Pro Phe Glu Ala Ile Met His Val

0.167 0.152 0.128 0.125 0.106 0.098 0.095 0.091 0.085 0.081 0.079 0.070 0.068 0.065 0.064 0.058 0.056 0.055 0.054 0.053

aValues taken from Chou and Fasman (1977, 1978a,b). bThe positional preference f is defined as the relative occurence of a specific amino acid found in position i of a β turn. i

Table 2.3.3 Amino Acid Positional Preferences in α Helicesa

Effect

Residuesb

Amino-terminal stabilizing (N-cap)

Gly Ser, Thr Asp, Glu Asn, Gln His Gly Asn His(+), Lys, Arg His(+), Lys, Arg Val, Leu, Ile, Met Ser Asp Val, Leu, Ile, Met

Carboxyl-terminal stabilizing (C-cap)

Amino-terminal destabilizing Carboxyl-terminal destabilizing

aAmino acid preferences were compiled from Fersht and Serrone (1993),

Presta and Rose (1988), and Richardson and Richardson (1988).


bResidues listed in order of increasing magnitude of the effect.


and pair information values for each possible pair and for amino acids within eight amino acids of a central residue. The predictive accuracy of the GOR methods is similar to that of the Chou and Fasman method (Garnier et al., 1978). Thus, the GOR methods may serve as a check of the results obtained with the Chou and Fasman method.

The EMBL Profile Neural Network Method An important feature of the EMBL Profile Neural Network (PHD) prediction scheme is the use of multiple sequence alignments (Rost and Sander, 1993a,b). Previous studies have shown that homologous proteins (generally with at least 20% to 30% sequence identity but often less) frequently have identical three-dimensional folds (tertiary structure). Thus, multiple sequence alignments are grouped by structural family and used as input. By considering many aligned sequences, the algorithm makes use of more information about structure than is available from just the individual protein sequence for which the secondary structure is to be predicted. To determine which protein sequences are to be used in the prediction, the PHD method uses a weighted dynamic programming method, MaxHom (Sander and Schneider, 1991), in searching a database of sequences. A family profile for the input sequence is generated and the amino acid frequencies at each alignment position are calculated for use in the prediction. Results have shown that the multiple sequence alignments improve secondary structure prediction >6% compared to singlesequence predictions (Rost and Sander, 1993a,b). The PHD method uses a reference network or neural network that has been trained and tested on a database of nonhomologous protein structures. In a typical database there is an uneven distribution of secondary structure types, such that there are many more examples of turns than of helices or sheets (Rost and Sander, 1993a,b). Because of the uneven distribution, empirically based predictions of loops are more accurate than those of helices, which in turn are more accurate than those of sheets. The neural network approach of the PHD method eliminates this bias: Rost and Sander developed a network that is trained with each type of secondary structure in equal proportion, rather than the proportions present in any database of protein structures. This results in a more balanced prediction.

The network system for structure prediction consists of three layers. In the first layer, secondary structure is predicted on the basis of the multiple aligned sequences. The results are passed to the second layer, which collectively considers secondary structure elements that are adjacent in the protein sequence, thereby allowing nearest neighbor structure to influence the structure of a particular region of the protein. The third layer of the network averages the results of the independently trained networks. The output from the third layer is determined by a “winner take all” prediction of the secondary structure at a given sequence position (Rost and Sander, 1993a,b).

The PSA Method The Protein Sequence Analysis (PSA) method (Stultz et al., 1993) computes the probability that a given amino acid belongs to a particular secondary structure element when there are no homologous protein sequences or structural data. To do this, mathematical models were developed for 15 single-domain protein superclasses or macroclasses. These superclasses, patterned after the protein class categorization of Richardson (1981), are groups of proteins with similar secondary and tertiary structures. The mathematical models represent constraints on the patterns of secondary structure elements for a given protein class. To perform structure prediction, the PSA method uses the amino acid sequence to calculate the probability that a given residue is contained in each of the modeled secondary structure elements. Also calculated is the probability that the submitted sequence is a member of one of the predefined superclasses. Thus, in addition to evaluating the secondary structure prediction, the PSA method provides the bonus of the potential tertiary structural information contained in the predicted structural superclass.

The Helical Wheel The helical wheel is a plot of the amino acid residues around a potentially helical segment (see Fig. 2.3.1 and discussion in Early Predictive Schemes). The method was developed to find helices with a hydrophobic face buried away from a polar solvent, with the graphical representation showing the clustering of polar and/or nonpolar residues toward one face of a helix. In its original use the helical wheel facilitated the identification of potential helical segments in protein sequences (Schiffer and Edmundson, 1967), but its applicability has since been expanded to include designing proteins (e.g.,



leucine zipper proteins; Harbury et al., 1993; Kamtekar et al., 1993; Lovejoy et al., 1993) and studying transmembrane proteins (e.g., G-protein-coupled receptors; Baldwin, 1993).

APPLICATION OF SECONDARY STRUCTURE PREDICTION METHODS Applying the Chou and Fasman Method The wide use of the Chou and Fasman method is due in part to the simplicity of application, ease of understanding, and potential interactive nature of the analysis. It also ranks favorably in accuracy of prediction relative to other methods. Application of the method entails the following steps. 1. Conformational preference parameters for each type of secondary structure are locally smoothed by averaging techniques (see UNIT 2.2 for an example of the scanning window averaging method) in order to calculate α-helix, β-sheet, and turn probabilities along the polypeptide backbone. 2. The α-helix, β-strand, and turn probabilities are plotted versus sequence position for interpretation. Basically, the conformation is assigned according to the highest curve (Fig. 2.3.2, Fig. 2.3.3, and Fig. 2.3.4). 3. On the basis of conformational preference parameters, amino acids are assigned as formers, breakers, or indifferent for α-helical and β-sheet regions (Fig. 2.3.3A). 4. Program options usually include full predictions that are based on the original Chou and Fasman rules (Table 2.3.4). Predictions can also be made by the user in an interactive environment by combining the Chou and Fasman rules with sequence analysis using other empirical methods such as hydrophobicity profiling, flexibility prediction, charged residue mapping, and surface probability calculation (Fig. 2.3.3 and Fig. 2.3.4; see also Comparison of Prediction Schemes). We find that an interactive approach allows for the best interpretation of the prediction results (Fasman, 1989a; Schulz, 1988). Several other correlations can be used in this analysis and are listed in Table 2.3.5. Many programs can also generate sketches of predicted peptide structures (Fig. 2.3.5).

Applying the GOR Scheme Protein Secondary Structure Prediction

The GOR scheme is coded into many commercial and academic protein structure analysis packages. This method provides estimated

probabilities that a given residue is in a particular conformational state, and it predicts the conformation of an amino acid without any input from the user other than the sequence of interest. There are two outputs from the GOR method. The first is a plot of the turns, α helices, and β strands as predicted from the method. The second is a summary of the full protein structure prediction similar to that produced by the Chou and Fasman method. Unlike the Chou and Fasman method, assignment of each amino acid to a structural element is made automatically. Because many prediction schemes can be printed in the same output format, it is possible to print results of Chou and Fasman, GOR, and other structural predictions together for a more interactive analysis of the sequence (Fig. 2.3.4). Use of a combination of structural prediction schemes and the secondary structure correlations presented in Table 2.3.5 should increase the accuracy of secondary structure prediction.

Applying the PHD Scheme Access to the PHD scheme is available only through electronic mail or the WWW (World Wide Web). The Internet address for this service is given in Table 2.3.6. To perform sequence analysis with the PHD scheme, one merely needs to supply the sequence to the EMBL Predictprotein server. Additionally, one can submit prealigned sequences or a defined set of homolog sequences to be used in the predictive scheme. If a single sequence is submitted, the scheme will search the sequence database, and, if homologs are found, the sequence prediction will be performed on the multiple sequence alignments (Fig. 2.3.6). The output from the scheme consists of several pages describing the network and analysis, including the list of any homologous proteins along with the sequence alignment used in the analysis, the definition of accessibility, and a reliability index for the accuracy of prediction (Fig. 2.3.7). This method provides prediction of solvent accessibility as well as secondary structure. A comparison of this analysis to others described in this unit is described in the section on Comparison of Prediction Schemes.

Applying the PSA Scheme The PSA scheme, like the PHD scheme, has been established on the Internet (Table 2.3.6). The electronic mail server, called Protein Sequence Analysis (PSA) System, accepts as input the amino acid sequence of a protein. After sequence analysis the server returns four


A

PEPPLOT

0 A G W N A Y I D N L M A D G

50 T C Q D A A I V G Y K D S P S V W A A V P G K T F

V N I T P A E V G V L V G K D R S S F Y V N G L

100 T L G G Q K C S V I R D S L L Q D G E F S M D L

R T K S T G G A P T F N V T V T K T D K T L V L L

M G K E G V H G G L I N K K C Y E M A S H L R R

S Q Y

Basic Acidic

1.5

HPhobic Hphilic Beta Forming Beta Breaking

Beta Chou & Fasman

1.0 Alpha 0.5 Alpha Forming 15 Alpha Breaking Beta Alpha 0 Beta Alpha 5

NH2 End

15 COOH End 0 Turn Hydrophobic Moment

0

1

Beta Alpha

0

3

Goldman et al. Kyte-Doolittle

0

HPhobic HPhilic

–3

0

50

100

B

Alpha Alpha Beta Beta Alpha Beta Pos Res Stat Ave Stat Ave NH2 COOH NH2 COOH Turn ------------------------------------------------------------------ .. 1 A 1.42 0.94 0.83 0.96 0.43 0.17 0.29 0.56 0.30 2 G 0.57 0.94 0.75 0.96 0.73 0.04 2.33 0.42 0.15 3 W 1.08 0.97 1.37 1.14 0.57 0.25 0.47 0.76 0.28 4 N 0.67 0.97 0.89 1.20 1.08 0.78 0.75 1.41 0.78 5 A 1.42 1.05 0.83 1.11 0.51 1.25 0.16 1.75 0.04 6 Y 0.69 0.86 1.47 1.13 0.25 0.31 1.24 1.27 0.45 7 I 1.08 0.99 1.60 1.08 0.58 0.08 2.44 0.52 0.63 8 D 1.01 1.09 0.54 0.95 2.19 0.12 0.56 1.08 0.24 9 N 0.67 1.19 0.89 1.02 1.98 0.95 0.09 2.40 0.03 10 L 1.21 1.27 1.30 0.93 0.82 3.18 0.04 2.33 0.14 11 M 1.45 1.11 1.05 0.79 0.31 1.52 0.31 0.63 1.41 12 A 1.42 0.96 0.83 0.83 0.84 0.17 2.28 0.22 0.99 13 D 1.01 0.78 0.54 0.92 2.45 0.13 1.21 0.25 1.04 14 G 0.57 0.80 0.75 1.06 3.18 0.13 1.15 0.46 1.26 15 T 0.83 0.91 1.19 1.01 1.79 0.29 0.45 0.39 0.14 16 C 0.70 1.06 1.19 0.92 0.53 0.19 2.91 0.18 1.52 17 Q 1.11 1.24 1.10 0.83 0.27 0.13 2.86 0.10 0.17 18 D 1.01 1.23 0.54 0.95 0.15 0.40 1.29 0.25 0.22 19 A 1.42 1.25 0.83 1.24 0.12 0.71 0.30 0.76 0.03 20 A 1.42 1.03 0.83 1.22 0.14 3.45 0.11 2.52 0.09 21 I 1.08 0.85 1.60 1.38 0.75 0.78 0.15 1.95 0.49 22 V 1.06 0.87 1.70 1.17 1.03 0.56 0.18 4.19 0.57 23 G 0.57 0.86 0.75 0.88 4.24 0.87 0.31 3.57 0.39 24 Y 0.69 0.91 1.47 0.88 1.08 1.11 0.49 4.63 1.79 25 K 1.16 0.88 0.74 0.65 1.39 0.50 1.50 0.74 0.51

HPhob –0.56 –0.33 –0.08 0.22 0.22 –0.02 0.04 –0.18 –0.18 –0.18 –0.40 –0.41 –0.11 0.74 0 .74 0 .68 –0.03 –0.03 0.27 –0.11 –0.40 –0.43 –1.00 –0.76 –0.41

Figure 2.3.3 Secondary structure prediction for human profilin I using the method of Chou and Fasman as implemented by GCG:PepPlot algorithm. (A) Predicted secondary structure and (B) numerical output for residues 1 through 25.

output files to the requester: a text file summarizing the tertiary class probabilities and most probable superclass; a file containing plots of the class and superclass probabilities; a file containing the secondary structure probability distributions displayed on a contour plot; and a file containing the secondary structure probability distributions on x-y plots. The secondary structure and protein class are easily deter-

mined from the combination of plots (Fig. 2.3.8) and can be compared with those suggested by other structural prediction methods.

COMPARISON OF PREDICTION SCHEMES The secondary structures of hemoglobin (an all-α protein), interleukin 1β (an all-β protein), and profilin (an α/β protein) were predicted



A

PLOTSTRUCTURE

50

100

50

100

5.0 KD Hydrophilicity –5.0 10.0 Surface Prob. 0.0 1.2 Flexibility 0.8 1.7 Jameson-Wolf (Antigenic Index) –1.7 CF Turns CF Alpha Helices CF Beta Sheets GOR Turns GOR Alpha Helices GOR Beta Sheets Glycosyl Sites

B

PEPTIDESTRUCTURE Hydrophilicity (Kyte-Doolittle) averaged over a window of: 7 Surface Probability according to Emini Chain Flexibility according to Karplus-Schulz Secondary Structure according to Chou-Fasman Secondary Structure according to Garnier-Osguthorpe-Robson Antigenicity Index according to Jameson-Wolf Pos 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35


AA A G W N A Y I D N L M A D G T C Q D A A I V G Y K D S P S V W A A V P

GlycoS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

HyPhil

0.750 0.240 0.417 –0.286 0.471 0.914 0.243 –0.529 –0.529 –0.214 0.486 0.086 -0.771 0.271 1.043 1.043 0.286 –0.414 –1.114 –0.700 –1.014 –0.957 –0.200 0.171 1.043 1.757 1.100 1.043 0.229 –0.529 –1.243 –1.243 –1.300 –0.143 –0.171

SurfPr

FlexPr

0.633 0.500 0.613 0.426 0.718 1.099 0.563 0.552 0.356 0.848 0.502 0.451 0.293 0.513 0.848 0.513 0.523 0.254 0.352 0.201 0.189 0.374 0.618 1.181 2.460 3.331 1.578 0.830 0.502 0.378 0.182 0.210 0.279 0.531 0.759

1.000 1.000 1.000 1.000 0.921 0.927 0.946 0.962 0.959 0.958 0.960 0.983 1.016 1.039 1.049 1.036 1.018 0.994 0.963 0.938 0.921 0.927 0.949 0.989 1.030 1.056 1.068 1.042 1.000 0.951 0.908 0.906 0.929 0.975 1.029

CF-Pred GORPred . . . . . . h h h h h h t t t t T T B B B B B B T T . T T B B B B B T

. . . . . B B B B B T T T T T B B B B B B B B T T . . . . B B B B B .

AI-Ind .. 0.750 0.450 0.450 –0.150 0.300 0.750 0.300 –0.600 –0.600 –0.300 0.700 0.700 0.150 1.050 1.350 0.950 0.850 –0.200 –0.600 –0.600 –0.600 –0.600 –0.300 0.850 1.700 1.300 0.900 1.150 0.850 –0.600 –0.600 –0.600 –0.600 –0.300 0.250

Figure 2.3.4 Secondary structure prediction for human profilin I using the methods of Chou and Fasman and GOR as implemented by GCG: PeptideStructure algorithm and plotted using PlotStructure utility. (A) Predicted secondary structure and (B) numerical output for residues 1 through 35.


with the Chou and Fasman, GOR, PHD, and PSA methods (Fig. 2.3.9). Analysis of the Chou and Fasman profiles was enhanced by displaying smoothed hydrophobicity profiles and consulting the amino acid positional preferences of Table 2.3.3. An example of the complete

analysis of profilin is presented in Figure 2.3.3, Figure 2.3.4, and Figure 2.3.5 (or Fig. 2.3.2C). The GOR predictions were made with the GCG commercial package (Fig. 2.3.4). The PHD and PSA predictions were obtained from the electronic mail response to a structure query; ex-

Abbreviated Chou and Fasman Prediction Rulesa

Table 2.3.4

Feature

Rule

Helix

Helix is initiated by a cluster of four helix-promoting residues out of six amino acids along the polypeptide sequence. Helical segment is extended in both directions until sets of tetrapeptide breakers are reached. Prolines can occur only within the first three residues of the N-terminal end. A segment length of six or more amino acids with α probability greater than β probability is predicted as helical. A cluster of three β formers or a cluster of three β formers out of five amino acids along the polypeptide sequence will initiate β-strand formation. The β strand is extended in both directions until terminated by a set of tetrapeptide breakers. A segment of three or more amino acids with β probability greater than α probability is predicted as β strand. Turn probability is calculated as peaks of probability that are greater than α or β probability. The β turns are also identified by Pt > 0.75 × 10−4, where Pt = fi × fi+1 × fi+2 × fi+3. For a region containing overlapping α helix and β strand, the prediction would be for helix if the α probabilities are greater than β probabilities, and strand if β probabilities exceed α probabilities.

Strand

Turns

Overlap

aFor the complete set of prediction rules, see Chou and Fasman (1974a,b, 1978a,b).

Table 2.3.5

Secondary Structure Correlations

Structure

Characteristics

β strands

Often correspond to peaks of hydrophobicity β-branched amino acids (Val, Ile, Thr) are typical Positively charged residues are relatively abundant Negatively charged residues (Glu) virtually never occur Average residue length is 7, resulting in short and acute peaks in hydrophobicity plots (see UNIT 5.2) β bulges may occur in the edge strands. Their signature is small amino acids and glycines (e.g., Gly-Xxx-Gly sequence in immunoglobulin V fold, strand G). Hydrophobicity peaks are longer, with a fine structure (teeth). Ideally, fine structure shows n + 4 periodicity. Glu, Ala, Leu, are typical Amino acid preferences exist at the N-cap and C-cap positions (see Table 2.3.3) Prolines can occur in the first turn of helices in globular proteins, and most frequently occupy the N-cap + 1 position Prolines, which “kink” the helices, are overrepresented in transmembrane helices Glycines frequently occur at the C-cap position Hydrophobic residues often occupy positions N-cap + 4 and C-cap − 4 Helices may be completely buried and not amphipathic

α helices



PLOTSTRUCTURE NH2

50 0

100 0

COOH

Figure 2.3.5 Sketch of the human profilin secondary structure as predicted in Figure 2.3.3 and Figure 2.3.4 by Chou-Fasman method.


amples of the output for the profilin secondary structure prediction are provided in Figure 2.3.6, Figure 2.3.7, and Figure 2.3.8. Figure 2.3.9A summarizes the results of secondary structure prediction for the β subunit of human hemoglobin. Immediately below the amino acid sequence is the location of the eight helices found in the three-dimensional structure of hemoglobin (Brookhaven Protein Data Bank entry 2hhb; Fermi, 1975). The Chou and Fasman scheme predicts most of the helical residues in helices 1, 2, 3, 4, 6, 7, and 8. Helices 6 and 7 are predicted to be longer than they actually are, and helices 2 and 3 are not distinguished as independent helices. A turn is predicted in place of the fifth helix. The GOR method predicts most of the residues for helices 1, 2, 5, 6, and 8, but it predicts a turn for helix 3 as well as a turn and a strand for helix 7. The PHD program found over 400 sequence homologs of hemoglobin-β, with one of the sequences being the submitted sequence. The PHD prediction identified a homolog with a known three-dimensional structure in the protein database and also suggested structure prediction by homology modeling. The prediction by PHD did quite well, accounting for seven of the eight helices and missing only the small helix 4. The PSA procedure suggested the α/β class as the most probable. Most of the helical residues in helices 2, 3, 5, 6, and 8 and part of helix 7 were correctly predicted. Helix 1 was missed by the PSA method, however, and the

C- and N-terminal residues of helices 2 and 3, respectively, were predicted to be in sheet structure. Figure 2.3.9B summarizes the results of secondary structure prediction for interleukin 1β. IL-1β contains 12 β strands connected by intervening turns (Brookhaven Protein Data Bank entry 1i1b; Finzel et al., 1989). The Chou and Fasman method predicted most of the residues in 8 of the 12 strands in the β sheet (strands 1, 2, 4, 5, 6, 8, 11, and 12). Helical regions were predicted for β strands 3 (with the adjacent turn) and 7. The Chou and Fasman method correctly predicted several of the turns. The GOR method correctly predicted 4 of the 12 β strands, but it predicted helical regions for β strands 3 (with the adjacent turn), 5, 6, and 8. Interestingly, both the Chou and Fasman and GOR methods predict helix for the third strand. This may be due to the negatively charged glutamate, which is generally considered a strand breaker (see Table 2.3.1). For IL-1β, the PHD output listed the 10 sequence homologs that were used in the structure prediction and also identified a known three-dimensional structure in the protein database. The PHD method correctly predicted most of the residues in the 12 β strands; however, a deficiency in the method is that it does not predict the location of possible turns. The PSA method did not successfully predict either the class or the secondary structure of IL-1b: the method pre-


Table 2.3.6

Commercial Packages for Sequence Analysis

Software

Platforma

Vendor or Internet locationb

Sybyl/Biopolymerc InsightII/Homologyc GCG MacVector Geneworks PC/Gene DNASTAR

SGId SGId VAX Macintosh Macintosh IBM-compatible IBM-compatible or Macintosh

Tripos Associates Biosym Technologies Genetics Computer Group International Biotechnologies IntelliGenetics/Betagen IntelliGenetics/Betagen DNASTAR

Internet locations for sequence analysis PHD methode

PSA methode

EMBL Predictprotein Server [email protected] [email protected] http://www.embl-heidelberg.de/predict protein/predict protein.html Protein Sequence Analysis Server [email protected]

aSoftware may run on other platforms in addition to these listed here. For a comprehensive list, contact

the appropriate vendor. bFor suppliers addresses, see SUPPLIERS APPENDIX. cBiopolymer and Homology are optional modules that can be purchased to run with Sybyl and InsightII, respectively. dSilicon Graphics Inc. ePHD and PSA methods are currently accessible only through the Internet; thus, results can be accessed by any platform.

dicted the protein fold class to be all-α rather than all-β, and therefore incorrectly predicted α structure rather than β structure. It is noteworthy that several of the predicted α helices correspond to the location of β strands, namely, those of β strands 2, 3, 4, 5, 7, 9, 11, and 12. Further testing is required to determine whether there are any inherent problems with the predictions of all-β proteins by the PSA method. Figure 2.3.9C summarizes the results of secondary structure prediction for human profilin, an α/β protein whose high-resolution structure (Metzler et al., 1993, 1995) was not publicly known at the time of testing. Profilin is composed of a central seven-stranded, antiparallel β sheet and four helices (Fig. 2.3.10). Seven turns that connect the helical and sheet structural elements in the structure have been identified. Figure 2.3.2C, Figure 2.3.3A, and 2.3.4A show typical Chou and Fasman profiles as determined with the programs of Novotny (Novotny and Auffray, 1984), PepPlot (GCG, 1994), and Plotstructure (GCG, 1994). Figure 2.3.3B shows the numerical output from the modified Chou and Fasman prediction pro-

grams (PepPlot), and Figure 2.3.4B shows the automated prediction by the Peptidestructure program using the Chou and Fasman and the GOR methods. A schematic representation of the predicted protein secondary structure is shown in Figure 2.3.5. The Chou and Fasman method predicted six of the seven β strands and the two terminal helices of profilin (Fig. 2.3.8C). This method also predicted correct locations of several of the turns located between the strands. In comparison, the GOR method performed significantly worse, predicting only four of seven β strands and neither of the terminal helices. The PHD method identified six sequence homologs of profilin (Fig. 2.3.6) that it used in the secondary structure prediction (Fig. 2.3.6). The method correctly predicted all seven β strands and the two terminal helices. It did, however, predict additional β strands for the interior helices but did not predict the locations of any turns (owing to a deficiency in the method, as stated previously). The PSA method predicted the profilin sequence to be a member of the all-β protein fold class (Fig. 2.3.8). The method did



## PROTEINS : EMBL/SWISSPROT identifier and alignment statistics NR. ID STRID %IDE %WSIM IFIR ILAS JFIR JLAS LALI NGAP PROTEIN 1 : pro1_human 1.00 1.00 1 139 1 139 139 0 PROFILIN I. 2 : prof_mouse 0.96 0.97 1 139 1 139 139 0 PROFILIN. 3 : prof_bovin 0.95 0.97 1 139 1 139 139 0 PROFILIN. 4 : pro2_human 0.62 0.71 1 139 1 139 139 0 PROFILIN II. 5 : prof_varv 0.32 0.48 1 137 2 131 130 2 PUTATIVE PROFILIN. 6 : prof_vaccv 0.31 0.47 1 137 2 131 130 2 PUTATIVE PROFILIN. ## ALIGNMENTS 1 6 SeqNo PDBNo AA STRUCTURE BP1 .4....:....5....:....6....:....7 1 A U 0 2 G U 0 3 W U 0 4 N U 0 5 A U 0 6 Y U 0 7 I U 0 8 D U 0 9 N U 0 10 L U 0 11 M U 0 12 A U 0 13 D U 0 14 G U 0 15 T U 0 16 C U 0 17 Q U 0 18 D U 0 19 A U 0 20 A U 0 21 I U 0 22 V U 0 23 G U 0 24 Y U 0 25 K U 0

BP2 ACC 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

NOCC 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

VAR 0 30 0 35 50 41 5 15 29 21 53 56 24 33 51 47 24 6 0 0 0 0 24 0 27

LGAP LSEQ2 ACCNUM 0

139

P07737

0

139

P10924

0

139

P02584

0

139

P35080

7

133

P33828

7

133

P20844

....:....1....:....2....:....3....:... AAAAAA GGGGEE WWWWWW NNNQHH AAASKK YYYYII IIIVII DDDDEE NSNNDD LLLLII MMMMSS AAACKK DDDDNN GGGGNN TTTCNK CCCCFF QQQQEE DDDEDD AAAAAA AAAAAA IIIIII VVVVVV GGGGDD YYYYYY KKKCKK

Figure 2.3.6 Sequence homologs and a portion of the sequence alignment statistics used in secondary structure prediction for human profilin I with PHD, the EMBL neural network method. Secondary structure prediction given in Figure 2.3.7. For description of column labels see Rost and Sander (1993a,b) and EMBL Predictprotein Server (address given in Table 2.3.6).


correctly predict the location of five of the β strands, but it predicted additional β strands for the amino and carboxyl termini and the second interior helices. The PSA method correctly predicted several of the turns located between the β strands, as well. In summary, both the PHD and the interactive Chou and Fasman method performed well, correctly predicting most secondary structural elements (see Table 2.3.7). The interactive Chou and Fasman method generally outperformed the GOR method. However, it must be

pointed out that in predicting the secondary structure with the Chou and Fasman method, the correlations contained in Table 2.3.3 and Table 2.3.5 and the Chou and Fasman rules contained in Table 2.3.4 facilitated both locating helix ends and discriminating between helix and strand when ambiguity existed. In contrast, a strict interpretation of the GOR results was used. It is also possible to interpret the GOR results interactively; this may equalize the predictive power of the two methods. In testing the PHD method, we were unable


Table 2.3.7 Prediction

Summary of Secondary Structure

Structure Hemoglobin Interleukin 1β Profilin

Method (%)a C-F

GOR

PHD

PSA

78 53 60

62 41 44

86 67 87

60 — 48

aNumber of amino acids correctly predicted, divided by the total

number of amino acids known to be in helices, sheets, or turns. No value is provided for the PSA analysis of interleukin 1β because the class predicted was all-α.

protein:

2a20

length

139

....,....1....,....2....,....3....,....4....,....5....,....6 AA |AGWNAYIDNLMADGTCQDAAIVGYKDSPSVWAAVPGKTFVNITPAEVGVLVGKDRSSFYV| PHD sec | HHHHHHHHH EEEEE EEEEEE EEEEE EEEEEEEE EEEE| Rel sec |965679988733797356089964258635986257448665481777999726874787| detail: prH prE prL subset: SUB

sec sec sec sec

ACCESSIBILITY 3st: P_3 acc 10st: PHD acc Rel acc subset: SUB acc

|026778888763100210000000000110000000000000001000000000000000 |000000010000001211489876421156887521268777304788989752116788 |972110001136897567410013568622012368631222684111000147883111 |LLHHHHHHHH..LLL.LL.EEEE..LLL.EEEE.LL..EEEE.L.EEEEEEE.LLL.EEE |

|eebeebbebbbeeeebebbbbbbbee eebbbbbbeebbbeb be bbbbbeee eebbb| |980670060007777060000000775770000007700070507400000777576000 |842231810404525012897740440647597304515146213084247355141324 |ee....b..b.ee.e...bbbbb.ee.eebbbb..ee.b.eb....bb.bb.ee.e...b|

....,....7....,....8....,....9....,....10...,....11...,....1 AA |NGLTLGGQKCSVIRDSLLQDGEFSMDLRTKSTGGAPTFNVTVTKTDKTLVLLMGKEGVHG| PHD sec |EEEEE EEEEEEEEE EEEEEE EEEEEE EEEEEE | Rel sec |324627717999975224589315997625799988313999975525888882678767| detail: prH prE prL subset: SUB

sec sec sec sec


|000000000000000111000011000001000000000000001121100000011100 |656751148999986542200346887741100011345899982126888884100011 |342247851000012336788542001156898988653000016741000005788877 |...E.LL.EEEEEEE...LLL..EEEEE.LLLLLLL...EEEEELL.EEEEEE.LLLLLL |

|bbbebbeeebbbb bbbbeebebbbeb beeeeebbbbebbbbebbebbbbbbbeeebee| |000600776000050000770700060507779700006000060070000000776077 |201100442808311640440542725137454511011738111055997684450044 |......ee.b.b...bb.ee.eb.b.b..eeeee.....b.b....ebbbbbbbee..ee |

2...,....13...,....14...,....15...,....16...,....17...,....1 AA |GLINKKCYEMASHLRRSQY| PHD sec | HHHHHHHHHHHHHH | Rel sec |8424899999999972589 detail: prH prE prL subset: SUB

sec sec sec sec

|0146889999999985200 |0210000000000000000 |8632100000000014689 |L...HHHHHHHHHHH.LLL|


|eebeeeb ebbeeb eeee| |7606770570076057789 |4131359142931413469 |e....eb.e.b..b..eee|

Figure 2.3.7 Secondary structure prediction for human profilin I with PHD, the EMBL neural network method. Sequence homologs and a portion of the sequence alignment used in the analysis are given in Figure 2.3.6.



Superclass Probabilities for Sequence p using vpp6.m, 15-Dec-94 (10:50:50)

Probability

1 0.8 0.6 0.4 0.2 0 alpha-beta

alpha

irregular

beta

Probability of being in a strand for Sequence p (psp2.m), 15-Dec-94 (10:53:40)

1 0.5 0 Probability of being in a turn for Sequence p (psp2.m), 15-Dec-94 (10:53:40)

Probability

1 0.5 0 Probability of being in a helix for Sequence p (psp2.m), 15-Dec-94 (10:53:39)

1 0.5 0 0

20

40

60

80

100

120

130

Residue position

Figure 2.3.8 Class and secondary structure predictions for human profilin I with the PSA method.

to identify any protein whose structure is included in the Brookhaven Protein Data Bank for which there are no sequence homologs in sequence databases. If such a sequence and structure were available, the comparison of the predictive methods might have been less biased. However, this point highlights an advantage of the PHD method. Sequence homologs with known three-dimensional structures often help achieve an accurate prediction of secondary structure. The availability of homologous amino acid sequences imparts a significant predictive advantage to the PHD method in many cases.


analysis (which requires interpretation of the output data) and PHD neural network predictions (which are automatic) is recommended as the most accurate and preferable procedure. As three-dimensional protein structural databanks grow, the future of protein structure prediction may lie with methods that “thread” the polypeptide sequence onto known motifs and evaluate the resulting model by pseudoenergetic analysis (e.g., PROSA; Sippl and Weitckus, 1992). Such programs currently exist, and successful fold identifications are being reported. However, the probability of a successful sequence-structure match is currently 10% (Sippl and Weitckus, 1992).

RECOMMENDATIONS AND THE FUTURE

CONCLUSIONS

On the basis of the examples presented here and our experience in general, a combination of the Chou and Fasman interactive sequence

1. Empirically derived rules can aid sequence prediction using the GOR and Chou and Fasman statistical methods. Summaries of


A

1 Hem. Structure C-F GOR PHD PSA Hem. Structure

B

C

20

40

60

VHLTPEEKSAVTALWGKVNVDEVGGEALGRLLVVYPWTQRFFESFGDLSTPDAVMGNPKVKAHGKKVLGAFSDGLA

80

100

120

140

HLDNLKGTFATLSELHCDKLHVDPENFRLLGNVLVCVLAHHFGKEFTPPVQAAYQKVVAGVANALAHKYH

C-F GOR PHD PSA 1

20

40

60

Il-1b. Structure C-F GOR PHD PSA

APRVRSLNCTLRDSQQKSLVMSGPYELKALHLQGQDMEQQVVFSMSFVQGEESNEKIPVALGLKEKNLYLSCVLKDD

Il-1b. Structure C-F GOR PHD PSA

KPTLQLESVDPKNYPKKKMEKRFVFNKIEINNKLEFESSAQFPNWYISTSQAENMPVFLGGTKGGQDITDFTMQFVSS

80

100

1

120

20

140

40

60

Profilin. Structure C-F GOR PHD PSA

AGWNAYIDNLMADGTCQDAAIVGYKDSPSVWAAVPGKTFVNITPAEVGVLVGKDRSSFYVNGLTLGG

Profilin. Structure

QKCSVIRDSLLQDGEFSMDLRTKSTGGAPTFNVTVTKTDKTLVLLMGKEGVHGGLINKKCYEMASHLRRSQY

80

100

120

139

C-F GOR PHD PSA

Figure 2.3.9 Summary of the secondary structure predicted for (A) human hemoglobin β-subunit, (B) human interleukin 1β, and (C) human profilin I. Each panel lists, from top to bottom, the amino acid sequence, the secondary structure determined experimentally, and the secondary structure predicted by each of the following: C-F, the Chou and Fasman method; GOR, the GOR method; PHD, the EMBL neural network method; and PSA, the state-space method. The location of helices (solid bars), turns (open bars), and sheet structure (hatched bars) are indicated.

these correlations (rules) are contained in Tables 2.3.4 and 2.3.5. 2. Methods that are open to interpretation do not guarantee the same result when applied by different users. 3. The PHD and PSA methods allow no input from the user, other than the sequence of interest, and thus are the easiest to use. For the other methods described here, the user can interactively participate in the prediction process. 4. If sequence homologs and, in particular, three-dimensional data are available, the PHD method gives reliable secondary structure predictions.

5. The PSA method was found to be less accurate than the other predictive schemes tested.

ACCESSIBILITY OF SOFTWARE Several prediction schemes have been incorporated into commercial packages (e.g., GCG) that also contain algorithms for additional sequence analysis. Other packages have interactive graphics modules that combine sequence analysis with molecular displays (Tripos and Biosym); however, these packages do not easily incorporate additional user input. There have been numerous reports in the literature describing computerized implementation of various



Figure 2.3.10 Schematic representation of the three-dimensional structure of human profilin I (Metzler et al., 1995) as displayed by Molscript (Kraulis, 1991). A seven-stranded antiparallel β sheet bisects the molecule. The N- and C-terminal α helices are on the left-hand side, and the two interior α helices are on the right-hand side in the view shown.

prediction schemes (Fasman, 1989b). Several of these programs have been published in a variety of programming languages. There are also several sites available on the Internet for sequence analysis. A listing of some of the commercial and Internet programs is contained in Table 2.3.6.

Literature Cited Bajorath, J., Stenkamp, R., and Aruffo, A. 1993. Knowledge-based model building of proteins: Concepts and examples. Protein Sci. 2:17981810. Baldwin, J.M. 1993. The probable arrangement of helixes in G protein–coupled receptors. EMBO J. 12:1693-1703. Blundell, T.L. and Johnson, L.N. (eds.) 1976. Protein Crystallography. Academic Press, London.


acid residues and prediction of backbone topography in proteins. Isr. J. Chem. 12:239-286. Chou, P.Y. 1989. Prediction of protein structural class from amino acid compositions. In Prediction of Protein Structure and the Principles of Protein Conformation (G.D. Fasman, ed.) pp. 549-586. Plenum, New York. Chou, P.Y. and Fasman, G.D. 1974a. Conformational parameters for amino acids in helical, beta sheet and random coil regions calculated from proteins. Biochemistry 13:211-222. Chou, P.Y. and Fasman, G.D. 1974b. Prediction of protein conformation. Biochemistry 13:223-245. Chou, P.Y. and Fasman, G.D. 1977. β-turns in proteins. J. Mol. Biol. 115:135-175. Chou, P.Y. and Fasman, G.D. 1978a. Empirical predictions of protein conformation. Annu. Rev. Biochem. 47:251-276.

Branden, C. and Tooze, J. 1991. Introduction to Protein Structure. Garland Publishing, New York.

Chou, P.Y. and Fasman, G.D. 1978b. Prediction of the secondary structure of proteins from their amino acid sequence. Adv. Enzymol. 47:45-148.

Brookhaven. 1994. Protein Data Bank. Brookhaven National Laboratory, Upton, N.Y.

DeGrado, W.F. 1988. Design of peptides and proteins. Adv. Protein Chem. 39:51-124.

Burgess, A.W., Ponnuswamy, P.K., and Scheraga, H.A. 1971. Analysis of conformations of amino

Dill, K.A. 1990. Dominant forces in protein folding. Biochemistry 29:7133-7155.


Dunhill, P. 1968. The use of helical net diagrams to represent protein structures. Biophys. J. 8:865875.

Jones, D.T., Taylor, W.R., and Thornton, J.M. 1992. A new approach to protein fold recognition. Nature 358:86-89.

Fasman, G.D. 1985. A critique of the utility of the prediction of protein secondary structure. J. Biosci. 8:15-23.

Kabsch, W. and Sander, C. 1984. On the use of sequence homologies to predict different conformations. Proc. Natl. Acad. Sci. U.S.A. 81:10751078.

Fasman, G.D. 1989a. The development of the prediction of protein structure. In Prediction of Protein Structure and the Principles of Protein Conformation (G.D. Fasman, ed.) pp. 193-316. Plenum, New York. Fasman, G.D. (ed.) 1989b. Prediction of Protein Structure and the Principles of Protein Conformation. Plenum, New York. Fermi, G. 1975. Three-dimensional Fourier synthesis of human deoxyhaemoglobin at 2.5 angstroms resolution, refinement of the atomic model. J. Mol. Biol. 97:237-256. Fersht, A.R. and Serrone, L. 1993. Principles of protein stability derived from protein engineering experiments. Curr. Opin. Struct. Biol. 3:7583. Finzel, B.C., Clancy, L.L., Holland, D.R., Muchmore, S.W., Watenpaugh, K.D., and Einspahr, H.M. 1989. Crystal structure of recombinant human interleukin-1β at 2.0 angstroms resolution. J. Mol. Biol. 209:779.

Kabsch, W. and Sander, C. 1985. Identical pentapeptides with different backbones. Nature 317:207. Kamtekar, S., Schiffer, J.M., Xiong, H., Babik, J.M., and Hecht, M.H. 1993. Protein design by binary patterning of polar and non-polar amino acids. Science 262:1680-1685. Kauzmann, W. 1959. Some factors in the interpretation of protein denaturation. Adv. Protein Chem. 14:1-63. Kneller, D.G., Cohen, F.E., and Largridge, R. 1990. Improvements in protein secondary structure prediction by an enhanced neural network. J. Mol. Biol. 214:171-182. Kraulis, P.J. 1991. MOLSCRIPT: A program to produce detailed and schematic plots of protein structures. J. Appl. Cryst. 24:946-950. Krystek, Jr., S.R., Bruccoleri, R.E., and Novotny, J. 1991. Stabilities of leucine zipper dimers estimated by an empirical force energy method. Int. J. Pept. Protein Res. 38:229-236.

Fisher, M.F. 1964. A limiting law relating the size and shape of protein molecules to their composition. Proc. Natl. Acad. Sci. U.S.A. 51:1285-1291.

Kuntz, I.D. 1972. Protein folding. J. Am. Chem. Soc. 94:4009-4012.

Garnier, J. and Robson, B. 1989. The GOR method for predicting secondary structure in proteins. In Prediction of Protein Structure and the Principles of Protein Conformation (G.D. Fasman, ed.) pp. 417-466. Plenum, New York.

Levitt, M. and Chothia, C. 1976. Structural patterns in globular proteins. Nature 261:552-558.

Garnier, J., Osguthorpe, D.J., and Robson, B. 1978. Analysis of the accuracy and implications of simple methods for predicting the secondary structure of globular proteins. J. Mol. Biol. 120:97-120. Genetics Computer Group. 1994. (GCG) Wisconsin Sequence Analysis Package Program Manual. Genetics Computer Group, Inc., Madison, Wis. Goodman, E.M. and Kim, P.S. 1991. Periodicity of amide proton exchange rates in a coiled-coil leucine zipper peptide. Biochemistry 30:1161511620. Greer, J. 1990. Comparative modeling: Application to the family of the mammalian serine proteases. Proteins Struct. Funct. Genet. 7:317-334. Guzzo, A.V. 1965. The influence of amino acid sequence on protein structure. Biophys. J. 5:809822. Harbury, P.B., Zhang, T., Kim, P.S., and Alber, T. 1993. A switch between two-, three-, and fourstranded coiled coils in GCN4 leucine zipper mutants. Science 262:1401-1407. Harper, E.T. and Rose, G.D. 1993. Helix stop signals in proteins and peptides: The capping box. Biochemistry 32:7605-7609.

Lesk, A.M. 1991. Protein architecture. IRL Press, Oxford.

Lewis, P.N., Momany, F.A., and Scheraga, H.A. 1971. Folding of polypeptide chains in proteins: A proposed method of folding. Proc. Natl. Acad. Sci. U.S.A. 68:2293-2297. Lim, V.I. 1974. Algorithms for prediction of α helices and β-structural regions in globular proteins. J. Mol. Biol. 88:873-894. Lovejoy, B., Choe, S., Cascio, D., McRorie, D.K., DeGrado, W.F., and Eisenberg, D. 1993. Crystal structure of a synthetic triple-stranded α-helical bundle. Science 259:1288-1293. Metzler, W.J., Constantine, K., Friedrichs, M.S., Bell, A., Ernst, E., Lavoie, T.B., and Mueller, L. 1993. Characterization of the three-dimensional structure of human profilin: 1H, 13C and 15N assignments and global folding pattern. Biochemistry 32:13818-13829. Metzler, W.J., Farmer, B.T., II, and Mueller, L. 1995. Refined solution structure of human profilin I. Protein Sci. 4:450-459. Nagano, K. 1973. Logical analysis of the mechanism of protein folding. I. Prediction of helices, loops and β-structures from primary structure. J. Mol. Biol. 75:401-420.



Novotny, J. and Auffray, C. 1984. A program for prediction of protein secondary structure from nucleotide sequence data: Application to histocompatability antigens. Nucl. Acids Res. 12:243255. Oppenheimer, N.J. and James, T.L. (eds.) 1989. Nuclear magnetic resonance (Part B). Methods Enzymol. Volume 177.

Roy, S. and Rose, G.D. 1980. Hydrophobic basis of packing in globular proteins. Proc. Natl. Acad. Sci. U.S.A. 8:4643-4647. Sander, C. and Schneider, R. 1991. Database of homology-derived structures and the structural meaning of sequence alignment. Proteins Struct. Funct. Genet. 9:56-68.

O’Shea, E.K., Rutkowski, R., and Kim, P.S. 1992. Mechanism of specificity in the Fos-Jun oncoprotein heterodimer. Cell 68:699-708.

Schiffer, M. and Edmundson, A.B. 1967. The use of helical wheels to represent the structures of proteins and to identify segments with helical potential. Biophys. J. 7:121.

Pauling, L. and Corey, R.B. 1951. The pleated sheet, a new layer of configuration of polypeptide chains. Proc. Natl. Acad. Sci. U.S.A. 37:251-256.

Schulz, G.E. 1988. A critical evaluation of methods for prediction of protein secondary structures. Annu. Rev. Biophys. Chem. 17:1-21.

Pauling, L., Corey, R.B., and Branson, H.R. 1951. The structure of proteins: Two hydrogen-bonded helical configurations of the polypeptide chain. Proc. Natl. Acad. Sci. U.S.A. 37:205-211.

Schulz, G.E., Barry, C.D., Friedman, J., Chou, P.Y., Fasman, G.D., Finkelstein, A.V., Lim, V.I., Ptitsyn, O.B., Kabat, E.A., Wu, T.T., Levitt, M., Robson, B., and Nagano, K. 1974. Comparison of predicted and experimentally determined secondary structure of adenyl kinase. Nature 250:140-142.

Presta, L.G. and Rose, G.D. 1988. Helix signal in proteins. Science 240:1632-1641. Prevelige, P., Jr. and Fasman, G.D. 1989. Chou-Fasman prediction of the secondary structure of proteins. In Prediction of Protein Structure and the Principles of Protein Conformation (G.D. Fasman, ed.) pp. 391-416. Plenum, New York. Prothero, J.W. 1966. Correlation between the distribution of amino acids in alpha helices. Biophys. J. 6:367-370. Ptitsyn, O.B. and Finkelstein, A.V. 1970. Biofizika 15:757-768. Quinn, N. and Sejnowski, T.J. 1988. Predicting the secondary structure of globular proteins using neural network models. J. Mol. Biol. 202:865884. Richardson, J.S. 1981. The anatomy and taxonomy of protein structure. Adv. Protein Chem. 34:167339. Richardson, J.S. and Richardson, D.C. 1988. Amino acid preferences for specific locations at the ends of α-helices. Science 240:1648-1652. Robson, B. and Pain, R.H. 1971. Analyses of the code relating sequence to conformation in proteins: Possible implication for the mechanism of formation of helical regions. J. Mol. Biol. 58:237-259. Rost, B. and Sander, C. 1993a. Improved prediction of protein secondary structure by use of sequence profiles and neural networks. Proc. Natl. Acad. Sci. U.S.A. 90:7558-7562. Rost, B. and Sander, C. 1993b. Prediction of protein structure at better than 70% accuracy. J. Mol. Biol. 232:584-599.

Schulz, G.E. and Schirmer, R.H. 1979. Principles of Protein Structure. Springer-Verlag, New York and Heidelberg. Sippl, M.J. and Weitckus, S. 1992. Detection of native-like models for amino acid sequences of unknown three-dimensional structure in a data base of known protein conformations. Proteins Struct. Funct. Genet. 13:258-271. Stultz, C.M., White, J.V., and Smith, T.F. 1993. Structural analysis based on state-space modeling. Protein Sci. 2:305-314. Wuthrich, K. 1986. NMR of Proteins and Nucleic Acids. John Wiley & Sons, New York. Wyckoff, H.W., Hirs, C.H.W., and Timasheff, S.N. (eds.) 1985. Diffraction methods for biological molecules (Parts A and B). Methods Enzymol. Volumes 114 and 115.

Key Reference Fasman, 1989b. See above. Provides extensive descriptions and a historical perspective describing methods for protein structure prediction.

Contributed by Stanley R. Krystek, Jr., William J. Metzler, and Jiri Novotny Bristol-Myers Squibb Pharmaceutical Research Institute Princeton, New Jersey



Internet Basics

UNIT 2.4

With the explosion of sequence and structural information available to researchers, the field of bioinformatics is playing an increasingly large role in the study of fundamental biomedical problems. The challenge facing computational biologists will be to aid in gene discovery and in the design of molecular modeling, site-directed mutagenesis, and experiments of other types that can potentially reveal previously unknown relationships with respect to the structure and function of genes and proteins. This challenge becomes particularly daunting in light of the vast amount of data that has been produced by the Human Genome Project and other systematic sequencing efforts to date. Before embarking on any practical discussion of computational methods in solving biological problems, it is necessary to lay the common groundwork that will enable users to both access and implement the algorithms and tools discussed in this book. We begin with a review of the Internet and its terminology, also discussing major classes of Internet protocols, without becoming overly engaged in the engineering minutiae underlying these protocols. A more in-depth treatment on the inner workings of these protocols may be found in a number of well-written reference books intended for the lay audience (Krol and Klopfenstein, 1996; Rankin, 1996; Kennedy, 1999). This unit will also discuss matters of connectivity, ranging from simple modem connections to digital subscriber lines (DSL). Finally, we will address one of the most common problems that has arisen with the proliferation of Web pages throughout the world—i.e., finding useful information on the World Wide Web. INTERNET BASICS Despite the impression that the Internet is a single entity, it is actually a network of networks, composed of interconnected local and regional networks in more than 100 countries. While work on remote communications began in the early 1960s, the true origins of the Internet lie with a research project on networking at the Advanced Research Projects Agency of the U.S. Department of Defense in 1969 named ARPANET. The original ARPANET connected four nodes on the West Coast, with the immediate goal of being able to transmit information on defense-related research between laboratories. A number of different network projects subsequently surfaced, with the next landmark developments coming over ten years later. In 1981, BITNET (“Because It’s Time”) was introduced, providing point-to-point connections between universities for the transfer of electronic mail and files. In 1982, ARPA introduced the Transmission Control Protocol (TCP) and Internet Protocol (IP); TCP/IP allowed different networks to be connected to and communicate with one another, creating the system that is in place today. A number of references chronicle the development of the Internet and communications protocols in detail (Quarterman, 1990; Froehlich and Kent, 1991; Krol and Klopfenstein, 1996). Most users, however, are content to leave the details of how the Internet works to their systems administrators; the relevant fact to most is that it does work. Once the machines on a network are connected to one another, there needs to be some way to unambiguously specify a single computer so that messages and files find their intended recipient. To accomplish this, all machines directly connected to the Internet have an IP number. IP numbers are unique, identifying one and only one machine. The IP number is made up of four numbers separated by periods; for example, the IP number for the main file server at the National Center for Biotechnology Information (NCBI) at the National Institutes of Health (NIH) is 130.14.25.1. The numbers themselves represent, Contributed by Andreas D. Baxevanis and B.F. Francis Ouellette Current Protocols in Protein Science (2000) 2.4.1-2.4.16 Copyright © 2000 by John Wiley & Sons, Inc.


2.4.1 Supplement 20

from left to right, the domain (130.14 for the NIH), the subnet (.25 for the National Library of Medicine at NIH), and the machine itself (.1). While the use of IP numbers aids the computers in directing data, they are obviously very difficult for users to remember. Therefore, an IP number often has a fully-qualified domain name (FQDN) associated with it, which is dynamically translated in the background by a domain name server. Going back to the NCBI example, instead of using 130.14.25.1 to access the NCBI computer, a user could instead use ncbi.nlm.nih.gov and achieve the same result. Reading from left to right, the IP number goes from least to most specific, while the FQDN equivalent goes from most specific to least. The name of any given computer can then be thought of as taking the general form computer.domain, with the top-level domain (the portion coming after the last period in the FQDN) falling into one of the six broad categories shown in Table 2.4.1. Outside the United States, the top-level domain names may be replaced with a two-letter code specifying the country where the machine is located (e.g., .ca for Canada and .uk for the United Kingdom). In an effort to anticipate the needs of Internet users in the future, as well as to try to erase the arbitrary line between top-level domain names based on country, the now-dissolved International Ad Hoc Committee (IAHC) was charged with developing a new framework of generic top-level domains (gTLD). The new, recommended gTLDs were set forth in a document entitled The Generic Top Level Domain Memorandum of Understanding (gTLD-MOU); these gTLDs are overseen by a number of governing bodies and are also shown in Table 2.4.1. The most concrete measure of the size (and, thereby, the success) of the Internet lies in actually counting the number of machines physically connected to it. The Internet Table 2.4.1

Top-Level Domain Names

Top-level domain names Inside U.S. .com .edu .gov .mil .net .org

Commercial site Educational site Government site Military site Gateway or network host Private (usually not-for-profit) organizations

Examples of top-level domain names used outside the United States Canadian site .ca Academic site in the United Kingdom .ac.uk Commercial site in the United Kingdom .co.uk Generic top-level domains proposed by IAHC Firms or businesses .firm Businesses offering goods to purchase (stores) .shop Entities emphasizing activities relating to the World Wide Web .web Cultural and entertainment organizations .arts Recreational organizations .rec Information sources .info Personal names (e.g., yourlastname.nom) .nom

Internet Basics

2.4.2 Supplement 20


Software Consortium conducts an Internet Domain Survey twice each year to count these machines, otherwise known as hosts. In performing this survey, ISC considers not only how many hostnames have been assigned, but how many of those are actually in use; a hostname might be issued, but the requestor may be holding the name in abeyance for future use. To test for this, a representative sample of host machines are sent a probe (a “ping”), with a signal being sent back to the originating machine if the host was indeed found. The rate of growth of the number of hosts has been phenomenal; from a paltry 213 hosts in August 1981, the Internet has in excess of 60 million hosts now. The doubling time for the number of hosts is on the order of 18 months. Most of this growth has come from the commercial sector, capitalizing on the growing popularity of multimedia platforms for advertising and communications such as the World Wide Web. CONNECTING TO THE INTERNET Of course, before being able to use all of the resources that the Internet has to offer, one needs to actually make a physical connection between one’s own computer and “the information superhighway.” For purposes of this discussion, the elements of this connection have been separated into two discrete parts: the actual, physical connection (meaning the “wire” running from one’s computer to the Internet backbone) and the service provider, who handles issues of routing and content once connected. Keep in mind that, in practice, these are not necessarily treated as two separate parts—for instance, one’s service provider may also be the same company that will run cables or fibers right into one’s home or office. Copper Wires, Coaxial Cables, and Fiber Optics Traditionally, users who were attempting to connect to the Internet away from the office have had one and only one option—a modem, which uses the existing copper twisted-pair cables carrying telephone signals to transmit data. Data transfer rates using modems are relatively slow, allowing for data transmission in the range of 28.8 to 56 kilobits per second (Kbps). The problem with using conventional copper wire to transmit data lies not in the copper wire itself, but in the switches that are found along the way that route information to their intended destinations. These switches were designed for the efficient and effective transfer of voice data, but were never intended to handle the high-speed transmission of data. While most people still use modems from their homes, a number of new technologies are already in place and will become more and more prevalent for accessing the Internet away from hard-wired Ethernet networks. The maximum speeds at which each of the services that are discussed below can operate are shown in Figure 2.4.1. The first of these “new solutions” is the integrated services digital network, or ISDN. While the advent of ISDN was originally heralded as the way to bring the Internet into the home in a speed-efficient manner, it required that special wiring be brought into the home. It also required that users be within a fixed distance from a central office, on the order of 20,000 feet or less. The cost of running this special, dedicated wiring, along with a per-minute pricing structure, effectively placed ISDN out of reach of most individuals. While ISDN is still available in many areas, this type of service is quickly being supplanted by more cost-effective alternatives. In looking at alternatives that did not require new wiring, cable television providers began to look at ways in which the coaxial cable already running into a substantial number of households could be used to also transmit data. Cable companies are able to use bandwidth that is not being used to transmit television signals (effectively, unused channels) to push Computational Analysis


Supplement 20

Telephone modem 0.056 Cellular wireless 0.128 ISDN 0.128 0.4 Satellite T1 1.544 Cable modem ADSL Ethernet 0

2

4.0 7.1

4 6 Maximum speed (Mbps)

10

8

10

Figure 2.4.1 Performance of various types of Internet connections, by maximum throughput. The numbers indicated in the graph refer to peak performance; often the actual performance of any given method may be on the order of one-half slower, depending on configurations and system conditions.

data into the home at very high speeds, up to 4.0 megabits per second (Mbps). The actual computer is connected to this network through a cable modem, which uses an Ethernet connection to the computer and a coaxial cable to the wall. Homes in a given area all share a single cable, in a wiring scheme very similar to that by which individual computers are connected via the Ethernet in an office or laboratory setting. While this branching arrangement can serve to connect a large number of locations, there is one major disadvantage—as more and more homes connect through their cable modems, service effectively slows down as more signal attempts to pass through any given node. One way of circumventing this problem is the installation of more switching equipment and reducing the size of a given “neighborhood.” Since the local telephone companies were the primary ISDN providers, they quickly turned their attention to ways in which the existing, conventional copper wire already in the home could be used to transmit data at high speed. The solution here is the digital subscriber line, or DSL. By using new, dedicated switches that are designed for rapid data transfer, DSL providers can circumvent the old voice switches that slowed down transfer speeds. Depending on the user’s distance from the central office and whether a particular neighborhood has been wired for DSL service, speeds are on the order of 0.8 to 7.1 Mbps. The data transfers do not interfere with voice signals, and users can use the telephone while connected to the Internet; the signals are “split” by a special modem that passes the data signals to the computer and a microfilter that passes voice signals to the handset. There is a special type of DSL called asynchronous DSL, or ADSL. This is the variety of DSL service that is becoming more and more prevalent. Most home users download much more information than they send out, so systems are engineered to provide super-fast transmission in the “in” direction, with transmissions in the “out” direction being 5 to 10 times slower. Using this approach maximizes the amount of bandwidth that can be used without necessitating new wiring. One of the advantages of ADSL over cable is that ADSL subscribers effectively have a direct line to the central office, meaning that they do not have to compete with their neighbors for bandwidth. This, of course, comes at a price; at the time of this writing, ADSL connectivity options were on the order of twice as expensive as cable Internet. Internet Basics

2.4.4 Supplement 20


Some of the newer technologies involve wireless connections to the Internet. These include using one’s own cell phone or a special cell phone service (such as Ricochet) to upload and download information. These cellular providers can provide speeds on the order of 28.8 to 128 Kbps, depending on the density of cellular towers in the service area. Fixed-point wireless services can be substantially faster, since the cellular phone does not have to “find” the closest tower at any given time. Along these same lines, satellite providers are also coming on-line. These providers allow for data download directly to a satellite dish with a southern exposure, with uploads occuring through traditional telephone lines. While the satellite option has the potential to be amongst the fastest of the options discussed, current operating speeds are only on the order of 400 Kbps. Content Providers versus ISPs Once an appropriately fast and price-effective connectivity solution is found, users will then need to actually connect to some sort of service that will enable them to traverse the Internet space. The two major categories in this respect are on-line services or Internet service providers (ISPs). On-line services such as America Online (AOL) and CompuServe offer a large number of interactive digital services, including information retrieval, electronic mail (e-mail), bulletin boards, and “chat rooms” where users who are on line at the same time can converse with each other about any number of subjects. While the on-line services now provide access to the World Wide Web (see discussion of The World Wide Web), most of the features and services available through these systems reside within a proprietary, closed network; once a connection is made between the user’s computer and the on-line service, accessing the special features, or content, of these systems does not require ever leaving the on-line system’s host computer. Specialized content can range from access to on-line travel reservation systems to encyclopedias that are constantly being updated—items that would not be available to anyone unless they subscribed to that particular on-line service. Internet service providers, or ISPs, take the opposite tack. Instead of focusing on providing content, the ISPs provide the tools necessary for users to send and receive e-mail, upload and download files, and navigate around the World Wide Web to find information at remote locations. The major advantage of ISPs is connection speed; ISPs often provide faster connections than the on-line services. Most ISPs charge a monthly fee for unlimited use. The line between on-line services and ISPs has already begun to blur. AOL’s now monthly flat fee pricing structure allows users to obtain all of the proprietary content found on AOL as well as all of the Internet tools available through ISPs, often at the same cost as a simple ISP connection. The extensive AOL network puts access to AOL as close as a local phone call in most of the United States, providing access to e-mail no matter where the user is located—a feature that small, local ISPs cannot match. Not to be outdone, many of the major national ISP providers now also provide content through the concept of portals. Portals are Web pages that can be customized to the needs of the individual user and which serve as a jumping-off point to other sources of news or entertainment on the Net. In addition, many national firms such as Mindspring are able to match AOL’s ease of connectivity on the road, and both ISPs and online providers are becoming more and more generous in providing users the capacity to publish their own Web pages. Developments such as this, coupled with the move of local telephone and cable companies into providing Internet access through new, faster fiber optic networks, foretell major changes in how people will access the Net in the future, changes that should favor the end-user both in price and performance.



Supplement 20

ELECTRONIC MAIL Most people are introduced to the Internet through the use of electronic mail (e-mail). The use of e-mail has become practically indispensable in many settings owing to its convenience as a medium for sending, receiving, and replying to messages. Its advantages are many: It is much quicker than postal, or “snail mail.” Messages tend to be much clearer and more to the point than is the case in typical telephone or face-to-face conversations. Recipients have more flexibility in deciding whether a response needs to be sent immediately, relatively soon, or at all, giving individuals more control over workflow. It provides a convenient method by which messages can be filed or stored. There is little or no cost involved in sending an e-mail message. While these and other advantages have pushed e-mail to the forefront of interpersonal communication in both industry and the academic community, users should be aware of several major disadvantages. First is the issue of security. As mail travels towards its recipient, it may pass through a number of remote nodes. The message could be intercepted and read at any one of those nodes by someone with high-level access, such as a systems administrator. Second is the issue of privacy. In industrial settings, e-mail is often considered to be an asset of the company for use only in official communication and, as such, is subject to monitoring by supervisors. The opposite is often true in academic, quasi-academic, or research settings; for example, National Institutes of Health policy encourages personal use of e-mail within the bounds of certain published guidelines. The key words here are “published guidelines”; no matter what the setting, users of e-mail systems should always be informed as to their organization’s policy regarding the appropriate use and confidentiality of e-mail so that they may use the tool properly and effectively. An excellent, basic guide to the effective use of e-mail is highly recommended (Rankin, 1996). Sending E-Mail E-mail addresses take the general form [email protected], where user is the name of the individual user and computer.domain specifies the actual computer that the e-mail account is located on. Like a postal letter, an e-mail message is comprised of an envelope or header, showing the e-mail addresses of the sender and recipient, a line indicating the subject of the e-mail, and information about how the e-mail message actually travelled from the sender to the recipient. The header is followed by the actual message, or body, analogous to what would go inside the postal envelope. Figure 2.4.2 illustrates all the components of an e-mail message. E-mail programs vary widely, depending on both the platform and the needs of the users. Usually the characteristics of the local area network (LAN) dictate what types of mail programs can be used, and the decision is often left to systems administrators rather than individual users. Among the most widely used e-mail packages with a graphical user interface are Eudora for the Macintosh and both Netscape Messengerand Microsoft Exchange for Macintosh, Windows, and UNIX platforms. Text-based e-mail programs, which are accessed by logging into a UNIX-based account, include Elm and Pine.

Internet Basics

Bulk E-Mail As with postal mail, there has been an upsurge in “spam” or “junk e-mail,” where companies compile bulk lists of e-mail addresses for use in commercial promotions. Since most of these lists are compiled from on-line registration forms and similar sources, the

2.4.6 Supplement 20


Header Body

From: [email protected] (PredictProtein) To: [email protected] Subject: PredictProtein

Delivery details (Envelope)

Received: from dodo.cpmc.columbia.edu (dodo.cpmc.columbia.edu [156.111.190.78]) by members.aol.com (8.9.3/8.9.3) with ESMTP id RAA13177 for <[email protected]>; Sun, 2 Jan 2000 17:55:22 -0500 (EST) Received: (from phd@localhost) by dodo.cpmc.columbia.edu (980427.SGI.8.8.8/980728.SGI.AUTOCF) id RAA90300 for [email protected]; Sun, 2 Jan 2000 17:51:20 -0500 (EST) Date: Sun, 2 Jan 2000 17:51:20 -0500 (EST) Message-ID:

Sender, Recipient, and Subject

PredictProtein Help PHDsec, PHDacc, PHDhtm, PHDtopology, TOPITS, MaxHom, EvalSec Burkhard Rost Table of Contents for PP help 1. Introduction 1. What is it? 2. How does it work? 3. How to use it?

Figure 2.4.2 Anatomy of an e-mail message, with relevant components indicated. This message is an automated reply to a request for help file for the PredictProtein E-mail server.

best defense for remaining off of these bulk e-mail lists is to be selective as to whom e-mail addresses are provided. Most newsgroups keep their mailing lists confidential; if in doubt and this is a concern, one should ask. E-mail Servers Most often, e-mail is thought of as a way to simply send messages, whether it be to one recipient or many. It is also possible to use e-mail as a mechanism for making biological predictions or retrieving records from biological databases. Users can send e-mail messages in a predefined format, defining the action to be performed for remote computers known as servers; the servers will then perform the desired operation and e-mail back the results. While this method is not interactive (in that the user cannot adjust parameters or have control over the execution of the method in real time), it does place the responsibility of hardware maintenance and software upgrades on the persons maintaining the server, allowing users to concentrate on their results instead of on programming. For most servers, sending the message help to the server e-mail address will return a detailed set of instructions for using that server, including the way in which queries need to be formatted. Aliases and Newsgroups In the example in Fig. 2.4.2, the e-mail message is being sent to a single recipient. One of the strengths of e-mail is that a single piece of e-mail can be sent to a large number of people. The primary mechanism for doing this is through aliases; a user can define a group of people within their mail program and give the group a special name, or alias. Instead of using the individual e-mail addresses for all of the people in the group, the user can just send the e-mail to the alias name, and the mail program will handle broadcasting the message to each person in that group. Setting up alias names is a tremendous time-saver even for small groups; it also ensures that all members of a given group actually receive all e-mail messages intended for the group.



Supplement 20

The second mechanism for broadcasting messages is through newsgroups. This model works slightly differently in that the list of e-mail addresses is compiled and maintained on a remote computer through subscriptions, much like magazine subscriptions. For example, the BIOSCI newsgroups are amongst the most highly-trafficked, offering a forum for discussion or the exchange of ideas in a wide variety of biological subject areas. To begin receiving the messages posted to the automated sequencing discussion group within BIOSCI, a user would send a message to [email protected] with the wording subscribe autoseq in the body of the message. The user would then receive all future postings to that group and be able to participate in the discussions. If a user wished to be removed from the group, a message would be sent to the same address, but this time, the body of the message would read unsubscribe autoseq. For more information on BIOSCI, including a complete list of discussion groups, an e-mail message can be sent to [email protected]; in this case the subject line should be left blank and the words info faq typed in the body of the message. The BIOSCI server will then return a copy of the Frequently Asked Questions (FAQ) in response, with detailed information on each newsgroup overseen by BIOSCI. It is also possible to participate in newsgroups without having each and every piece of e-mail flood into one’s private mailbox. Instead, interested participants can use newsreading software, such as NewsWatcher for the Macintosh, which provides access to the individual messages making up a discussion. The major advantage is that the user can pick and choose which messages to read by scanning the subject lines; the remainder can be discarded by a single operation. NewsWatcher is an example of what is known as a client-server application—the client software (here, NewsWatcher) runs on a client computer (a Macintosh), which in turn interacts with a machine at a remote location (the server). Client-server architecture is interactive in nature, with a direct connection being made between the client and server machines. Once NewsWatcher is started, the user is presented with a list of newsgroups available to them (Fig. 2.4.3); this list will vary, depending on the user’s location, as systems administrators have the discretion to allow or block certain groups at a given site. From the rear-most window in the figure, the user double-clicks on the newsgroup of interest (here, bionet.genome.arabidopsis), which spawns the window shown in the center. At the top of the center window is the current unread message count, and any message within the list can be read by double-clicking on that particular line. This, in turn, spawns the last window (in the foreground), showing the actual message. If a user decides not to read any of the messages, or is done reading individual messages, the balance of the messages within the newsgroup (center) window can be deleted by first choosing Select All from the File menu, then selecting Mark Read from the News menu. Once the newsgroup window is closed, the unread message count is reset to zero. Every time NewsWatcher is restarted, it will automatically poll the news server for new messages that have been created since the last session. As with most of the tools that will be discussed in this unit, news-reading capability is built into Web browsers such as Netscape Navigator and Microsoft Internet Explorer. FILE TRANSFER PROTOCOL

Internet Basics

Despite the many advantages afforded by e-mail in transmitting messages, experienced e-mail users have no doubt experienced frustration in trying to transmit files (attachments) along with an e-mail message. The mere fact that a file can be attached to an e-mail message and sent does not mean that the recipient will be able to detach, decode, and actually use the attached file. While more cross-platform e-mail packages such as

2.4.8 Supplement 20


Figure 2.4.3 Using NewsWatcher to read postings to newsgroups. The list of newsgroups that the user has subscribed to is shown in the Subscribed List window (left). The list of new postings for the highlighted newsgroup (bionet.genome.arabidopsis) is shown in the center window. The window in the foreground shows the contents of the posting selected from the center window.

Microsoft Exchange are being developed, the use of different e-mail packages by people at different locations means that sending files via e-mail is not an effective, foolproof method, at least in the short term. One solution to this problem is through the use of a file transfer protocol (or FTP). The workings of FTP are quite simple—a connection is made between a user’s computer (the client) and a remote server, and that connection remains in place for the duration of the FTP session. File transfers are very fast, at rates on the order of 5 to 10 kilobytes per second, with speeds varying with time of day, distance between the client and server machines, and overall traffic on the network. In the ordinary case, making an FTP connection and transferring files requires that a user have an account on the remote server. However, there are many files and programs that the academic community makes freely available, and access to those files does not require having an account on each and every machine where these programs are stored. Instead, connections are made using a system called anonymous FTP. Under this system, the user connects to the remote machine, and instead of entering a username/password pair, types anonymous as the username and enters an e-mail address in place of a password. Providing one’s e-mail address allows the server’s systems administrator to compile



Supplement 20

$ ftp ftp.bio.indiana.edu Connected to magpie.bio.indiana.edu. 220 iubio.bio.indiana.edu FTP server ready. Name: anonymous 331 Guest login ok, send your complete e-mail address as password. Password: ******** 230-

Welcome to IUBio archive!

230230-

This is a user-supported archive for biology software and data.

230230-

See the file Archive.Doc for details of this archive.

230230-

See IUBio Bio-Mirror archive of large data sets at

230230-

ftp to iubio.bio.indiana.edu, user: iubio, password: iubio This includes GenBank, EMBL and DDBJ and other biosequence data.

230230-

Report problems, uploads and other matters via e-mail to

230-

[email protected].

230230 Guest login ok, access restrictions apply. Remote system type is UNIX. Using binary mode to transfer files. ftp> cd /molbio/align/clustal 250 CWD command successful. ftp> get clustalw1.75.unix.tar.Z local: clustalw1.75.unix.tar.Z remote: clustalw1.75.unix.tar.Z 200 PORT command successful. 150 Opening BINARY mode data connection for clustalw1.75.unix.tar.Z (230379 bytes). 226 Transfer complete. 230379 bytes received in 0.45 seconds (500.75 Kbytes/s) ftp> quit 221-You have transferred 230379 bytes in 1 files. 221-Total traffic for this session was 231859 bytes in 1 transfers. 221-Thank you for using the FTP service on iubio.bio.indiana.edu. 221 Goodbye.

Figure 2.4.4 Using UNIX FTP to download a file. An anonymous FTP session is established with the molecular biology FTP server at the University of Indiana to download the ClustalW alignment program. The user inputs are shown in boldface.

access statistics which may, in turn, be of use to those actually providing the public files or programs. An example of an anonymous FTP session using UNIX is shown in Figure 2.4.4. Although FTP occurs within the UNIX environment, Macintosh and PC users can employ programs that utilize graphical user interfaces (GUI, pronounced “gooey”) to navigate through the UNIX directories on the FTP server. Users need not have any knowledge of UNIX commands to download files; they can instead rely on pop-up menus and the ability Internet Basics

2.4.10 Supplement 20


Figure 2.4.5 Using Fetch to download a file. An anonymous FTP session is established with the molecular biology FTP server at the University of Indiana (top) to download the ClustalW alignment program (bottom). Notice the difference between this GUI-based program and the UNIX equivalent illustrated in Figure 2.4.4.

to point-and-click their way through the UNIX file structure. The most popular FTP program on the Macintosh platform for FTP sessions is Fetch. A sample Fetch window is shown in Figure 2.4.5 to illustrate the difference between using a GUI-based FTP program and the equivalent UNIX FTP in Figure 2.4.4. In the figure, notice that the Automatic radio button (near the bottom of the second window under the Get File button) is selected, meaning that Fetch will determine the appropriate type of file transfer to perform. This may be manually overridden by selecting either Text or Binary, depending on the nature of the file being transferred. As a rule, text files should be transferred as Text, programs or executables as Binary, and graphic format files such as PICT and TIFF files as Raw Data. Computational Analysis


Supplement 20

THE WORLD WIDE WEB While FTP is of tremendous use in the transfer of files from one computer to another, it does suffer from some limitations. When working with FTP, once a user enters a particular directory, they can only see the names of the directories or files. In order to actually view what is within the files, it is necessary to physically download them onto one’s own computer. This inherent drawback led to the development of a number of distributed document delivery systems (DDDS), interactive client-server applications that allowed information to be viewed without having to perform a download. The first generation of DDDS development led to programs like Gopher, which allowed plain text to be viewed directly through a client-server application. From this evolved the most widely known and widely used DDDS, namely the World Wide Web. The Web is an outgrowth of research performed at the European Nuclear Research Council (CERN) in 1989 that was aimed at sharing research data between several locations. That work led to a medium through which text, images, sounds, and videos could be delivered to users on demand, anywhere in the world. Navigation on the World Wide Web Navigation on the Web does not require advance knowledge of the location of the information being sought. Instead, users can navigate by clicking on specific text, buttons, or pictures. These clickable items are collectively known as hyperlinks. Once one of these hyperlinks is clicked, the user is taken to another Web location, which could be at the same site or halfway around the world. Each document displayed on the Web is called a Web page, and all of the related Web pages on a particular server are collectively called a Web site. Navigation strictly through the use of hyperlinks has been nicknamed “Web surfing.” Users can take a more direct approach to finding information by entering a specific address. One of the strengths of the Web is that the programs used to view Web pages (appropriately termed browsers) can be used to visit Gopher and FTP sites as well, somewhat obviating the need for separate Gopher or FTP applications. As such, a unified naming convention was introduced to indicate to the browser program both the location of the remote site and, more importantly, the type of information at that remote location so that the browser could properly display the data. This standard-form address is known as a uniform resource locator (URL), and takes the general form protocol://computer. domain, where protocol specifies the type of site and computer.domain specifies the location (Table 2.4.2). The http used for the protocol in World Wide Web URLs stands for hypertext transfer protocol, the method used in transferring Web files from the host computer to the client.

Table 2.4.2 Uniform Resource Locator (URL) Format for Each Type of Transfer Protocol

Site

URL format (example)

General form FTP site

protocol://computer.domain ftp://ftp.ncbi.nlm.nih.gov

Gopher site Web site

gopher://gopher.iubio.indiana.edu http://www.nhgri.nih.gov

Internet Basics



Browsers Browsers, which are used to look at Web pages, are client-server applications that connect to a remote site, download the requested information at that site, display the information on the user’s monitor, then disconnect from the remote host. The information retrieved from the remote host is in a platform-independent format called hypertext markup language (HTML). HTML code is strictly text-based, and any associated graphics or sound for that document exist as separate files in a common format. For example, images may be stored and transferred in GIF format, developed by CompuServe for the quick and efficient transfer of graphics; while GIF format is most commonly used for graphics, other formats such as JPEG and BMP may also be used. Because of this, a browser can display any Web page on any type of computer, whether it be a Macintosh, IBM-compatible, Linux, or UNIX machine. The text is usually displayed first, then the remaining elements are placed on the page as they are downloaded. With minor exceptions, a given Web page will look the same when the same browser is used on any of the above platforms. The two major players in the area of browser software are Netscape, with their Communicator product, and Microsoft, with Internet Explorer. As with many other areas where multiple software products are available, the choice between Netscape and Internet Explorer comes down to one of personal preference. While the computer literati will debate the fine points of difference between these two packages, for the average user, both packages perform equally well and offer the same types of features, adequately addressing the Web-browser needs of most users. It is worth mentioning that, while the Web is by definition a visually based medium, it is also possible to travel through Web space and view documents without the associated graphics. For users limited to line-by-line terminals, a browser called Lynx is available. Developed at the University of Kansas, Lynx allows users to use their keyboard arrow keys to highlight and select hyperlinks, using their return key the same way that Netscape and Internet Explorer users would click their mouse. Internet versus Intranet The World Wide Web is normally thought of as a way to communicate with people at a distance, but the same infrastructure can be used to connect people within an organization. Such intranets provide an easily accessible repository of relevant information, capitalizing on the simplicity of the Web interface. It also provides another channel for broadcast or confidential communication within the organization. Having an intranet is of particular value when members of an organization are physically separated, whether it be in different buildings or different cities. Intranets are protected in such a way that people who are not on the organization’s network are prohibited from accessing the internal Web pages; additional protections through the use of passwords are also common. Finding Information on the World Wide Web Most people find information on the Web the old fashioned way—by word of mouth either by using lists such as Table 2.4.3 or by simply following hyperlinks put in place by Web authors. Continuously clicking from page to page can be a highly ineffective way of finding information, especially when the information sought is of a very focused nature. One way of finding interesting and relevant Web sites is to consult virtual libraries, which are curated lists of Web resources arranged by subject. Virtual libraries of special interest to biologists include the WWW Virtual Library, maintained by Keith Robison at Harvard and the EBI BioCatalog, based at the European Bioinformatics Institutes. The URLs for these sites can be found in Table 2.4.3.



Supplement 20

Table 2.4.3

World Wide Web Sites of Interest

Site

URL

Domain names gTLD-MOU Internet Software Consortium

http://www.gtld-mou.org http://www.isc.org

Electronic mail and newsgroups BIOSCI Newsgroups http://www.bio.net/docs/biosci.FAQ.html Eudora http://www.eudora.com Microsoft Exchange http://www.microsoft.com/exchange/ NewsWatcher ftp://ftp.acns.nwu.edu/pub/newswatcher/ File Transfer Protocol Fetch 3.0/Mac FTP Voyager

http://www.dartmouth.edu/pages/softdev/fetch.html http://ftpvoyager.deerfield.com

Internet access America Online AT&T Bell Atlantic Bell Canada CompuServe MCI Ricochet Telus

http://www.aol.com http://www.att.com/worldnet http://www.bellatlantic.net http://www.bell.ca http://www.compuserve.com http://www.mci.com http://www.ricochet.net http://telus.com

Virtual libraries EBI BioCatalog Amos’ WWW Links Page NAR Database Collection WWW Virtual Library

http://www.ebi.ac.uk/biocat/biocat.html http://www.expasy.ch/alinks.html http://www3.oup.co.uk/nar/Volume_28/Issue_01/introduction/ http://mcb.harvard.edu/BioLinks.html

World Wide Web browsers Internet Explorer http://www.microsoft.com/insider/ie5/default.htm Lynx ftp://ftp2.cc.ukans.edu/pub/lynx Netscape Navigator http://home.netscape.com World Wide Web search engines AltaVista http://www.altavista.com Excite http://www.excite.com HotBot http://hotbot.lycos.com Infoseek http://infoseek.go.com Lycos http://www.lycos.com Northern Light http://www.northernlight.com World Wide Web meta-search engines MetaCrawler http://www.metacrawler.com Savvy Search http://www.savvysearch.com

Internet Basics



Table 2.4.4 Number of Hits Returned for Four Defined Search Queries on Some of the More Popular Search and Meta-Search Engines

Search term genetic mapping human genome positional cloning prostate cancer

Search engine

Meta-search engine

HotBot

Excite

Infoseek

Lycos

478 13,231 279 14,044

1,040 34,760 735 53,940

4,326 15,980 1,143 24,376

9,395 19,536 666 33,538

MetaCrawler SavvySearch 62 42 40 60

58 54 52 57

It is also possible to directly search the Web by using search engines such as Alta Vista and Excite, among others. A search engine is simply a specialized program that can perform full-text or keyword searches on databases that catalog Web content. The result of a search is a hyperlinked list of Web sites fitting the search criteria from which the user can visit any or all of the found sites. However, search engines use slightly different methods in compiling their databases. One variation is the attempt to capture most or all of the text of every Web page that the search engine is able to find and catalog (“Web crawling”). Another technique is to catalog only the title of each Web page rather than its entire text. A third is to consider words that must appear next to each other or only relatively close to one another. Because of these differences in search-engine algorithms, the results returned by issuing the same query to a number of different search engines can produce wildly different results (Table 2.4.4). It may also be noted from Table 2.4.4 that most of the numbers are exceedingly large, reflecting the overall size of the World Wide Web. Unless a particular search engine ranks its results by relevance (for example, by scoring words in a title higher than words in the body of the Web page), the results obtained may not be particularly useful. It is also necessary to keep in mind that, depending on the indexing scheme that the search engine is using, the found pages may actually no longer exist, leading the user to the dreaded “404 Not Found” error. Compounding this problem is the issue of coverage—the number of Web pages that any given search engine is actually able to survey and analyze. A comprehensive study by Lawrence and Giles (1998) indicates that the coverage provided by any of the search engines studied is both small and highly variable. For example, the HotBot engine produced 57.5% coverage of what was estimated to be the size of the “indexable Web,” while Lycos had only 4.41% coverage, a full order of magnitude less than HotBot. The most important conclusion from this study was the extent of coverage increased as the number of search engines are increased and the results from those individual searches are combined. Combining the results obtained from the six search engines examined in this study produced coverage approaching 100%. To address this point, a new class of search engines called meta-search engines have been developed. These programs will take the user’s query and poll anywhere from five to ten of the “traditional” search engines. The meta-search engine will then collect the results, filter out duplicates, and return a single, annotated list to the user. One big advantage is that the meta-search engines take relevance statistics into account, returning much smaller lists of results. Even though the hit list is substantially smaller, it is much more likely to contain sites that directly address the original query. Since the programs must poll a number of different search engines, searches conducted this way obviously take longer to perform, but the higher degree of confidence in the compiled results for a given query outweighs the extra few minutes (sometimes only seconds) of search time. Reliable and easy-to-use meta-search engines include SavvySearch and MetaCrawler (see Table 2.4.3).



Supplement 20

LITERATURE CITED Froehlich, F. and Kent, A. 1991. ARPANET, the Defense Data Network, and Internet. In Encyclopedia of Communications. Marcel Dekker, New York. Kennedy, A.J. 1999. The Internet: Rough Guide 2000. Rough Guides, London. Krol, E. and Klopfenstein, B.C. 1996. The Whole Internet User’s Guide and Catalog. O’Reilly and Associates, Sebastopol, Calif. Lawrence, S. and Giles, C.L. 1998. Searching the World Wide Web. Science 280:98-100. Rankin, B. 1996. Dr. Bob’s Painless Guide to the Internet and Amazing Things You Can Do with E-mail. No Starch Press, San Francisco. Quarterman, J. 1990. The Matrix: Computer Networks and Conferencing Systems Worldwide. Digital Press, Bedford, Mass.

Contributed by Andreas D. Baxevanis National Human Genome Research Institute, NIH Bethesda, Maryland B.F. Francis Ouellette Centre for Molecular Medicine and Therapeutics University of British Columbia Vancouver, British Columbia

Internet Basics



Sequence Similarity Searching Using the BLAST Family of Programs

UNIT 2.5

Database sequence similarity searching is carried out thousands of times each day by researchers worldwide. Scientists in traditional laboratories use search results, for example, to infer the functions of newly discovered cDNAs, predict new members of gene families, and explore evolutionary relationships between sequences. In turn, they populate sequence databases with biochemically and/or genetically characterized sequences. With the advent of whole-genome sequencing, a new breed of scientist is now using these characterized sequences to predict the location and function of coding and regulatory regions in large segments of genomic DNA. Their contribution to the sequence databases is the submission, at a rapid pace, of generally uncharacterized mRNA and genomic DNA sequences. These data can then be used by the first group of scientists to enhance the understanding of their sequences. Both types of sequencing efforts have greatly increased the size and quality of the sequence databases in recent years. Thus, sequence similarity searching has become a valuable research tool for all molecular biologists (Altschul et al., 1994; Schuler, 1998; also see UNIT 2.1). Over the years, a number of algorithms have been implemented that allow searching of sequence databases. The most useful of these tools should share the following characteristics: (1) Speed. Because today’s databases are so large, the programs must be fast in order to process megabases of sequence in seconds. (2) Sensitivity. The programs must report all potentially interesting similarities. (3) Rigorous statistics. The programs must provide a way to evaluate the significance of the results. (4) Ease of use. Scientists with no formal training in sequence-analysis algorithms should understand how to use the programs and interpret the results. Advanced users should have the option to tailor the programs to their needs. (5) Access to up-to-date databases. The doubling time of GenBank is currently ∼16 months. It is important to search the most recent version of the database. The BLAST (Basic Local Alignment Search Tool) family of sequence similarity search programs satisfies the above criteria. In short, users input either a nucleotide or amino acid query sequence, and search a nucleotide or amino acid sequence database. The program returns a list of the sequence “hits,” alignments to the query sequence, and statistical values. This unit describes how to choose an appropriate BLAST program and database, perform the search, and interpret the results. ACCESSING BLAST PROGRAMS AND DOCUMENTATION The National Center for Biotechnology Information (NCBI) currently supports two versions of BLAST free of charge: BLAST 2.0 (gapped BLAST) and Position-Specific Iterated BLAST (PSI-BLAST). BLAST 2.0 is the standard version of BLAST, which allows a user to search a sequence database with a nucleotide or protein sequence of interest. BLAST 2.0 places gaps into the query and target sequences so that separate areas of similarity between the two sequences can be returned as one hit. PSI-BLAST is an iterative BLAST search, which is optimized for finding distantly related sequences. Other BLAST programs are also available from the NCBI Web page. “BLAST 2 sequences” uses the BLAST search engine to produce an alignment of two sequences entered by the user. On the Specialized BLAST pages, researchers can use the BLAST engine to search sequences that are not in GenBank. At present, these databases include unfinished microbial genomes, P. falciparum (the human malaria parasite), and tentative Contributed by Tyra G. Wolfsberg and Thomas L. Madden Current Protocols in Protein Science (1999) 2.5.1-2.5.29 Copyright © 1999 by John Wiley & Sons, Inc.


2.5.1 Supplement 15

human consensus (THC) sequences from The Institute for Genomic Research (TIGR). The content of these pages is, however, subject to change. The easiest and most popular way to access the BLAST suite of programs is through the NCBI World Wide Web site, at http://www.ncbi.nlm.nih.gov/BLAST/. All versions of BLAST are accessible from this site, and can be used to query all sequence databases available at the NCBI. Documentation, which includes an overview of BLAST, BLAST frequently asked questions (FAQs), a “What’s New” page, the BLAST manual, and a list of references, is also available here. For users who want to run BLAST against private local databases or downloaded copies of NCBI databases, the NCBI offers a stand-alone version of the BLAST program. BLAST binaries and documentation are provided for the latest versions of IRIX, Solaris, DEC OSF1, and Win32 systems. BLAST 2.0 executables may be found on the NCBI anonymous FTP server at ftp://ncbi.nlm.nih.gov/blast/executables/. BLAST can also be run as a client-server program, in which the user installs client software on a local machine that communicates across the network with a server at NCBI. This setup is useful for researchers who run large numbers of searches on NCBI databases, because they can automate the process to run on their local computer. The BLAST client may be found on the NCBI anonymous FTP server at ftp://ncbi.nlm.nih.gov/blast/network/netblast. The NCBI BLAST e-mail server is the best option for people without convenient access to the Web. A similarity search can be performed by sending a properly formatted e-mail message containing the nucleotide or protein query sequence to [email protected]. The query sequence is compared against a specified database and the results are returned in an e-mail message. For more information on formulating e-mail BLAST searches, please send a message consisting of the word HELP to the same address, [email protected]. This unit concentrates on the BLAST searches which can be performed from the NCBI Web site. The other implementations are mainly for advanced users of BLAST, or those with special needs. For any of these services, questions should be directed to [email protected]. INTRODUCTION TO BLAST Basic Versus Advanced BLAST Searches BLAST 2.0 is available in a basic or advanced version. In both versions, the user can select the type of BLAST program and the database to be searched, and choose whether to filter the query sequence to mask low-complexity regions (see below). The advanced version allows the user to change parameters as well. For most researchers, the Basic version, which uses the default parameters, is adequate. For a discussion of BLAST parameters, see Appendix A at the end of this unit. BLAST Programs Five types of the BLAST program have been developed to support sequence similarity searching using a variety of nucleotide and protein sequence queries and databases. These programs are listed and described in Table 2.5.1. The type of BLAST search to be carried out is dependent on the type of information that is desired. Sequence Similarity Searching Using BLAST

2.5.2 Supplement 15


Table 2.5.1

BLAST Search Programs

Program

Query sequence Database sequence Comments

BLASTP

Protein

Protein

BLASTN

Nucleotide (both strands)

Nucleotide

BLASTX

Nucleotide (six-frame translation)

Protein

TBLASTN Protein

Nucleotide (six-frame translation)

TBLASTX Nucleotide (six- Nucleotide (sixframe translation) frame translation)

Can be run in standard mode or in a more sensitive iterative mode (PSI-BLAST), which uses the previous search results to build a profile for subsequent rounds of similarity searching. Parameters optimized for speed, not sensitivity; not intended for finding distantly related coding sequences. Automatically checks complementary strand of query. Very useful for preliminary data containing potential frameshift errors (ESTs, HTGs, and other “single-pass” sequences). Essential for searching protein queries against EST database. Often useful for finding undocumented open reading frames or frameshift errors in database sequences. Should be used only if BLASTN and BLASTX produce no results. Restricted for search against EST, STS, HTGS, GSS, and Alu databases.

NCBI Databases One frequent mistake in sequence similarity searching is failure to search an up-to-date database. The NCBI produces GenBank (Benson et al., 1998; ftp://ncbi.nlm.nih.gov/genbank/gbrel.txt) and updates it daily. It also shares data on a daily basis with the DNA Data Bank of Japan (DDBJ; Tateno et al., 1998) and the European Molecular Biology Laboratory (EMBL; Stoesser et al., 1998). A search of the NCBI databases using the BLAST Web page, client, or e-mail server guarantees access to the most recent database. The NCBI supports a number of databases for sequence similarity searching. These databases are subject to change, and a current list and description are available on the NCBI Web site at http://www.ncbi.nlm.nih.gov/BLAST/blast_databases.html. This section describes some of the commonly used NCBI databases. Examples demonstrating the utility of various BLAST programs and NCBI databases are given in the next section. Peptide sequence databases for BLASTP and BLASTX nr. The nr (nonredundant) database is the most comprehensive. GenBank, DDBJ, and EMBL are nucleotide sequence databases. If any nucleotide sequence in any of the databases is annotated with a coding sequence (CDS), this CDS appears in nr. nr also contains protein sequences obtained from PDB (sequences associated with 3-dimensional structures in the Brookhaven Protein Data Bank), Swiss-Prot (a curated database of protein sequences; Bairoch and Apweiler, 1998), PIR (Protein Identification Resource, a comprehensive collection of protein sequences; Barker et al., 1998), and PRF (Protein Research Foundation). Although nr may contain multiple copies of similar sequences, identical sequences are merged into one entry. To be merged, two sequences must have identical lengths and every residue at every position must be the same. There are nr databases for both peptide and nucleotide sequences. The peptide database is automatically selected for BLASTP and BLASTX searches.



Supplement 15

month. The month database receives its sequences from the same sources as nr, but contains only those sequences released within the last 30 days. All sequences in month are also present in nr. Alu. The Alu database contains six-frame translations of representative Alu repeats from all Alu subfamilies (Claverie and Makalowski, 1994). If a query sequence containing an Alu repeat is used in a BLAST search of nr or month, many of the resulting high-scoring hits will also contain Alu sequences. It may be useful, especially with a genomic sequence query, to perform a search of the Alu database to identify the location of any Alu repeats that might produce high-scoring and potentially misleading hits in queries of other databases. Nucleotide sequence databases for BLASTN, TBLASTN, and TBLASTX nr. The nr (nonredundant) database contains all nucleotide sequences present in GenBank, EMBL, and DDBJ. It also contains nucleotide sequences obtained from PDB (sequences associated with 3-dimensional structures in the Brookhaven Protein Data Bank). nr comprises only sequences that are normally well annotated, so it does not contain expressed sequence tag (EST), sequence-tagged site (STS), genome survey sequence (GSS), or high-throughput genomic (HTG) sequences. Although nr may contain multiple copies of similar sequences, identical sequences are merged into one entry. To be merged, two sequences must have identical lengths and every nucleotide at every position must be the same. month. The month database contains all nucleotide sequences present in GenBank, EMBL, DDBJ, and PDB that were released within the last 30 days. Unlike nr, it also contains EST, STS, GSS, and HTG sequences released within the last month. EST. EST accesses a nonredundant copy of all ESTs present in GenBank, EMBL, and DDBJ (Boguski et al., 1993). ESTs are short sequences, a few hundred nucleotides in length, which are derived by partial, single-pass sequencing of inserts of randomly selected cDNA clones (Adams et al., 1991). Since the number of ESTs is increasing rapidly, it is an important database to search for novel cDNAs. As of August, 1998, ∼70% of the sequences in GenBank were ESTs; of these, 61% were from human, 20% from mouse. STS. STS contains a nonredundant copy of all STSs present in GenBank, EMBL, and DDBJ. An STS is a short unique genomic sequence that is used as a sequence landmark for genomic mapping efforts (Olson et al., 1989). As of August, 1998, 83% of the sequences in the STS database were from human.

Sequence Similarity Searching Using BLAST

HTGS. HTGS contains “unfinished” DNA sequences generated by the high-throughput sequencing centers (Ouellette and Boguski, 1997). A typical HTG record might consist of all the first-pass sequence data generated from a single cosmid, BAC, YAC, or P1 clone. The record is composed of two or more sequence fragments that have a total length of ≥2 kb and contain one or more gaps. The sequences are normally updated by the sequencing centers as more data become available. A single accession number is assigned to this collection of sequences. The accession number does not change as the record is updated, and only the most recent version of the record remains in GenBank. Phase 1 HTG sequences are unordered, unoriented contigs with gaps. Phase 2 HTG sequences are ordered, oriented contigs with or without gaps. All HTG records contain a prominent warning that the sequence data is unfinished and may contain errors. When a record is considered finished, it becomes a Phase 3 HTG and is moved to the nr database with the same accession number. HTGS is a valuable source of new genomic sequences not yet in nr.

2.5.4 Supplement 15


GSS. GSS includes short, single-pass genomic data identified by various means (Smith et al., 1994). Many of the sequences have been mapped. As of August, 1998, 80% of the sequences in GSS were from human, 14% from Arabidopsis thaliana. Alu.The Alu database contains representative Alu repeats from all Alu subfamilies (Claverie and Makalowski, 1994). If a query sequence containing an Alu repeat is used in a BLAST search of the above nucleotide databases, many of the resulting high-scoring hits will also contain Alu sequences. It may be useful, especially with a genomic sequence query, to perform a search of the Alu database to identify the location of any Alu repeats that might produce high-scoring and potentially misleading hits in queries of other databases. Vector. The Vector database contains nucleotide sequences of a number of standard cloning vectors. New sequences should be screened against the Vector database to assure that they do not contain any vector contamination. Mito. The Mito database contains representative mitochondrial sequences from many families. Nuclear-derived sequences may be screened against the Mito database to assure that they do not contain any mitochondrial contamination. Formatting the Query Sequence Users of the BLAST Web page can initiate a search either by entering the sequence itself, or, if the sequence is already in the sequence database, by entering the accession number or gi (see Appendix B at the end of this unit). The preferred format for entering new sequences is the so-called FASTA format; however, if the sequence is not in FASTA format or the sequence is interspersed with numbers and spaces, BLAST will still accept the query. A sequence in FASTA format begins with a single-line description, followed by lines of sequence data. The description line is distinguished from the sequence data by a > symbol (greater than) in the first column. An example sequence in FASTA format is: >aaseq Human choroideremia protein MADNLPTEFDVVIIGTGLPESILAAACSRSGQRVLHIDSRSYYGGNWASFSFSGLLSWLKEYQQNNDIGE ESTVVWQDLIHETEEAITLRKKDETIQHTEAFPYASQDMEDNVEEIGALQKNPSLGVSNTFTEVLDSALP EESQLSYFNSDEMPAKHTQKSDTEISLEVTDVEESVEKEKYCGDKTCMHTVSDKDGDKDESKSTVEDKAD EPIRNRITYSQIVKEGRRFNIDLVSKLLYSQGLLIDLLIKSDVSRYVEFKNVTRILAFREGKVEQVPCSR ADVFNSKELTMVEKRMLMKFLTFCLEYEQHPDEYQAFRQCSFSEYLKTKKLTPNLQHFVLHSIAMTSESS CTTIDGLNATKNFLQCLGRFGNTPFLFPLYGQGEIPQGFCRMCAVFGGIYCLRHKVQCFVVDKESGRCKA IIDHFGQRINAKYFIVEDSYLSEETCSNVQYKQISRAVLITDQSILKTDLDQQTSILIVPPAEPGACAVR VTELCSSTMTCMKDTYLVHLTCSSSKTAREDLESVVKKLFTPYTETEINEEELTKPRLLWALYFNMRDSS GISRSSYNGLPSNVYVCSGPDCGLGNEHAVKQAETLFQEIFPTEEFCPPPPNPEDIIFDGDDKQPEAPGT NNVVMAKLESSEESKNLESPEKHLQN

Alternatively, the user can enter the database accession number or gi. The above sequence is in fact already in the database with a gi identifier of 116365 and a Swiss-Prot accession number of P26374. Either of these sequence identifiers can be entered on the BLAST page. The type of identifier (for nucleotide or protein sequence) must match the type of query sequence used in that search (Ostell and Kans, 1998; Ouellette, 1998). Sequences in manuscripts are often referred to by their GenBank accession number; however, this accession number refers to a nucleotide (not protein) sequence, and cannot be used to initiate a BLASTP or TBLASTN search, both of which require a protein sequence query. The identifier of the protein sequence encoded by a nucleotide accession number can be obtained by searching the NCBI’s Entrez nucleotide database, at http://www.ncbi.nlm. nih.gov/Entrez. Computational Analysis


Supplement 15

Filtering Sequences Both nucleotide and protein sequences may contain regions of low complexity, i.e., regions with homopolymeric tracts, short-period repeats, or segments enriched in one or only a few residues. Such low-complexity regions commonly give spuriously high BLAST scores that reflect compositional bias rather than significant position-by-position alignment (Altschul et al., 1994). For example, two protein sequences that contain low-complexity regions rich in the same amino acids may produce high-scoring alignments in those regions even though other parts of the proteins are entirely dissimilar. Because these alignments do not reflect a common ancestry, no functional inference is justified despite their high statistical significance. Filtering the query sequence (that is, replacing the repeated sequence with strings of n for nucleotide sequence or X for protein sequence) can eliminate potentially confounding matches, such as hits to low-complexity, proline-rich regions or poly(A) tails present in the database. By default, all searches performed through the NCBI BLAST Web page automatically filter the query sequence, as do the BLAST clients, e-mail server, and stand-alone programs. However, filtering can be turned off, even on the Basic BLAST page. Filtered sequence is represented in the final BLAST report as a string of n or X (e.g., nnnnnnnnnn or XXXXXXXXX). BLASTN queries are filtered with DUST (R.L. Tatusov and D.J. Lipman, pers. comm.). Other BLAST queries use SEG (Wootton and Federhen, 1993, 1996). Viewing the BLAST Results The NCBI BLAST Web page can return results in one of three ways. The default is to display the results in the browser window from which the user initiated the search. The document will have hypertext links that make it easier to analyze the results. BLAST queries are processed in the order in which they are received, except that less computationally intensive jobs (e.g., BLASTN, BLASTP, and smaller databases such as month) are given priority. In the middle of the afternoon, the BLAST server may be busy. Thus, it is sometimes more efficient to receive the BLAST results by e-mail. E-mail results are sent either as plain text or in HTML format. The HTML-formatted results must be opened in a Web browser, and the resulting document contains hypertext links. EXAMPLES OF BLAST SEARCHES In this section, interpretation of BLAST results is explained using examples for BLASTP, BLASTX, TBLASTN, BLASTN, and PSI-BLAST. An example is not shown for TBLASTX, as it is a tool of last resort and not useful for the majority of users. The first example, for BLASTP, contains general information, and should be read by all BLAST users. GenBank increases in size by thousands of sequences every week; thus, results obtained from running the same searches again will differ from those shown in the examples. BLASTP The BLASTP program compares a protein query to a protein database. A typical search is shown in Figure 2.5.1. The Swiss-Prot database has been selected and the program has been changed to BLASTP. A query in FASTA format has been entered in the input box. The query is the human choroideremia protein, implicated in hereditary blindness (Seabra et al., 1993). Sequence Similarity Searching Using BLAST

The top of the results page for this search is shown in Figure 2.5.2. It begins with some header information about the type of program (BLASTP), the version (2.0.5), and a release

2.5.6 Supplement 15


Figure 2.5.1 Submitting a BLASTP search using the NCBI’s World Wide Web interface.

Figure 2.5.2 Example of the top portion of a BLASTP report.



Supplement 15

Figure 2.5.3 Example of the graphical view of a BLASTP report. The bars are color coded by the strength of the database match. The strongest matches (those with a bit score >200) are red, followed by pink (bit score 80 to 200), green (50 to 80), blue (40 to 50), and black (300 genes NA NA

http://binddb.org http://dip.doe-mbi.ucla.edu http://biodata.mshri.on.ca/grid/ http://daisy.bio.nagoya-u.ac.jp/golab/hetpdbnavi.html http://surya.bic.nus.edu.sg/mpid/

15,141 interactions 22,229 interactions 13,819 interactions NA 90 complexes

AARSDB ABCdb AraC/XylS database ASPD Breast Cancer Gene CSDBase DExH/D Family Database EF-hand CaBP EcoCyc ENZYME Gene Ontology (GO) GPCRDB (receptors) Homeobox Page MEROPS (peptidase) MHCPEP (peptides) Nuclear Protein Database (NPD) O-GlycBase PDD PROCAT Protein Kinase Resource (PKR) RNase P Sentra TransportDB Tumor Gene WIT2 Wnt gene Homepage Binding: BIND DIP GRID Het-PDB Navi MHC-Peptide Interaction Database

continued


Supplement 33

Table 2.6.1

Web Addresses and Sizes of Selected Protein Databases, continued

Database

Web site

Sizea

Protein-Protein Interface ReLiBase Energetics: ProTherm References:

http://www-lmmb.ncifcrf.gov:80/~tsai/

NA

http://relibase.rutgers.edu/

11,938 proteins

http://www.rtc.riken.go.jp/jouhou/Protherm/protherm.html

13046 entries

http://www.infobiogen.fr/services/dbcat/

511 databases

http://www.ncbi.nlm.nih.gov/PubMed/ http://www.expasy.ch/seqanalref/

>12,000,000 citations NA

http://www.rtc.riken.go.jp/jouhou/3dinsight/3dinsight.html http://www.ncbi.nlm.nih.gov/Entrez/ http://srs.ebi.ac.uk

NA NA 264 databases

DBCAT (Catalog of Databases) MEDLINE SeqAnalRef Combined: 3DinSight Entrez SRS

aNA, size not available at time of printing. The data are as of March, 2003.

(Bairoch and Apweiler, 1999) contains all the translations of EMBL nucleotide sequence entries (Rodriguez-Tom et al., 1996) that have not been integrated into SwissProt. Another protein sequence database, the Protein Identification Resource (PIR; Barker et al., 1999), attempts to build a complete and nonredundant database from a number of protein and nucleic acid sequence databases. Identical and highly similar sequences from the same species are merged into a single entry. Each entry in PIR provides bibliographic and annotated information for the protein. The nr protein database is used for BLAST searching (Altschul et al., 1997), which is described in UNIT 2.5 of this book. It includes entries from the nonredundant GenBank (Benson et al., 1999) translations, SwissProt, PIR, Protein Research Foundation (PRF) in Japan, and the Protein Data Bank (PDB). Only entries with absolutely identical sequences are merged. Most of the sequence databases have a sequence search tool and cross-references to entries of other protein and gene databases. Many sequence databases, such as SwissProt and PIR, also provide text searching using, for instance, protein names or key words. To study a new protein, the authors recommend first performing a sequence search using BLAST in nr if the protein sequence is available. The search often gives entry names in the protein databases included in nr. Even when the protein is not found in nr, it is likely that a homologous protein will be hit, which can often lead to some useful information, such as the function of the query protein. If the sequence of the query protein is unavailable, doing a text search in SwissProt or PIR usually identifies the protein. SwissProt is probably the place to obtain the most information about a protein if it can be found in SwissProt. However, some additional information may be found by checking other sequence databases. For example, PIR provides some useful information on protein family classification for each entry, and the Kyoto Encyclopedia of Genes and Genomes (KEGG; Ogata et al., 1999) annotates some gene entries with information about metabolic and regulatory pathways. If the query protein is not available in nr, one may find a gene model (predicted protein sequence) by searching genetic sequence databases that contain predicted genes from nucleotide sequences (e.g., the Genome Channel; Mural et al., 1999). Although predicted sequences generated by computational gene-finding tools may contain errors, such databases cover a large number of proteins and are often reliable enough to provide useful information. Protein Databases on the Internet

2.6.4 Supplement 33


SwissProt, as a curated protein sequence database, offers a wide range of annotations, covering areas such as function, domain parsing, post-translational modifications, and variants. SwissProt can be accessed at http://www.expasy.ch/sprot/sprot-top.html (home server, Switzerland), http://us.expasy.org/sprot/sprot-top.html (USA), http://ca.expasy. org/sprot/sprot-top.html (Canada), http://cn.expasy.org/sprot/sprot-top.html (China), http://kr.expasy.org/sprot/sprot-top.html (Korea), and http://tw.expasy.org/sprot/sprot-top. html (Taiwan). SwissProt is integrated into many other databases, such as SRS (see SRS below). Human vitronectin is used here as an example for searching protein sequence databases. To locate the SwissProt entry for this protein, one can search either the entry name (VTNC_HUMAN) or the accession number (P04004) obtained from a BLAST search. Alternatively, one can use the full-text search at the SwissProt Web page to search by protein name (human vitronectin) or key words (e.g., serum spreading, as vitronectin is also called serum spreading factor s-protein). A combination of several entries can be used in a search. The entry name in SwissProt has the general format X_Y, where X is a mnemonic code of up to four characters indicating the protein name (in this case, VTNC), and Y is a mnemonic species identification code of up to five characters for the biological source of the protein. Some codes used for Y are full English names, e.g., HORSE, HUMAN, MAIZE, MOUSE, PIG, RAT, SHEEP, YEAST (baker’s yeast, Saccharomyces cerevisiae), and WHEAT. Some are abbreviations, including BOVIN (bovine), CHICK (chicken), ECOLI (Escherichia coli), PEA (garden pea, Pisum sativum), RABIT (rabbit), SOYBN (soybean, Glycine max), and TOBAC (common tobacco, Nicotina tabacum). An entry name may have several accession numbers if they have been merged. An accession number is always conserved from release to release, and therefore allows unambiguous citation. Each entry contains the following items shown in table format in the NiceProt View layout: (1) general information about the entry, (2) name and origin of the protein, (3) references providing the protein sequence, (4) comments (e.g., annotated functions, subunits, similar proteins), (5) cross-references (links to other databases), (6) key words, (7) features, and (8) sequence information. The text in the comments entry provides a function annotation for the protein (e.g., “Vitronectin is a cell adhesion and spreading factor found in serum and tissues. Vitronectins interact with glycosaminoglycans and proteoglycans...”). Cross-references lists the annotations of the protein by other databases, such as GeneCards (Rebhan et al., 1998) and ProDom (Corpet et al., 1999). GeneCards, a database of human genes, shows chromosomal location and the involvement of the protein in certain diseases (if applicable). ProDom contains protein domain families by automated sequence comparisons. Clicking the link to ProDom from SwissProt leads to a nice graphic view for domain parsing, as shown in Figure 2.6.1 for vitronectin.

0

100

200

300

400

500

100

200

300

400

500

VTNC_HUMAN

0

Figure 2.6.1 Domain parsing of human vitronectin by ProDom in SwissProt.



Supplement 33

Various research results are given under features. Some of the features items for VTNC_HUMAN are as follows: SIGNAL CHAIN CHAIN PEPTIDE DOMAIN DOMAIN SITE SITE MOD_RES CARBOHYD DISULFID BINDING CONFLICT

1 20 399 20 150 288 64 398 75 86 293 362 50

19 398 478 63 287 478 66 399 75 86 430 395 50

V65 SUBUNIT. V10 SUBUNIT. SOMATOMEDIN B. HEMOPEXIN-LIKE 1. HEMOPEXIN-LIKE 2. CELL ATTACHMENT SITE. CLEAVAGE. SULFATATION.

HEPARIN. C -> N (IN REF. 5).

where SIGNAL represents the extent of a signal sequence (prepeptide), MOD_RES indicates a post-translationally modified residue (sulfatation, in this case), CARBOHYD shows the glycosylation site, DISULFID means that a disulfide bond exists between the two indicated residues (293 and 430), and CONFLICT shows that different papers report differing sequences. PROTEIN STRUCTURAL DATABASES Searching structure databases is becoming more and more popular in molecular biology. The three-dimensional structures of proteins not only define their biological functions, but also hold a key in rational drug design. Traditionally, protein structures were solved at a low-throughput mode. However, recent advances in new technologies, such as synchrotron radiation sources and high-resolution nuclear magnetic resonance (NMR), accelerate the rate of protein structure determination substantially. There is an overwhelming consensus in the structural biology community that protein structures can be solved en masse (an effort called structural genomics), in a similar fashion as for determining DNA sequences, and that impact of this approach can be compared with that of the Human Genome Project (see Chapter 17). The only international repository for the processing and distribution of protein structures is the PDB (Bernstein et al., 1977). The structures in the PDB were determined experimentally by X-ray crystallography (∼86%) and NMR (∼14%). Theoretical models have been removed from PDB, effective July 2, 2002, based on the new PDB policy. The PDB also contains some structures of chemical ligands and nucleotides. Each PDB entry is represented by a four-character identifier (PDB ID), where the first character is always a number from 0 to 9 (e.g., 1cau, 256b). The PDB can be accessed through the home server (http://www.rcsb.org/pdb/ or http://www.pdb.org in the USA) or through one of many mirror sites from around the world, as listed at the home server. Some of the mirror sites are as follows: Rutgers University, Piscataway, NJ, United States: http://rutgers.rcsb.org/ NIST, Gaithersburg, MD, United States: http://nist.rcsb.org/ Cambridge Crystallographic Data Centre, United Kingdom: http://pdb.ccdc.cam.ac.uk/ National University of Singapore, Singapore: http://pdb.bic.nus.edu.sg/ Osaka University, Japan: http://pdb.protein.osaka-u.ac.jp/ Universidade Federal de Minas Gerais, Brazil: http://www.pdb.ufmg.br/ Max Delbrück Center for Molecular Medicine, Germany: http://www.pdb.mdc-berlin.de/ Protein Databases on the Internet

2.6.6 Supplement 33


The PDB offers three search methods: search by PDB ID, by SearchLite, and by SearchFields. SearchLite is a simple key-word search that uses, for instance, protein name or author’s name. A search using SearchFields, as an advanced search engine, allows a user to specify features of the protein, such as Enzyme Commission (EC) number, name of binding ligand, range of protein size, range of resolution in the X-ray structure, and secondary structure content. The PDB stores structural information in two formats: the PDB file format (Bernstein et al., 1977) and the macromolecular crystallographic information file (mmCIF) format (Bourne et al., 1997). The PDB file format is still the dominant format used in the protein community. It contains three parts: annotations, coordinates, and connectivities. The connectivity part, which shows chemical connectivities between atoms, is optional. It is listed at the end of the PDB file, beginning the line with the key word CONECT. The coordinate part uses each line for a three-dimensional coordinate of an atom, starting from ATOM (for standard amino acids) or HETATM (for nonstandard groups). The following shows an example of the PDB file format: HEADER

OXIDOREDUCTASE

COMPND

GLYCOLATE OXIDASE (E.C.1.1.3.1)

(OXYGEN(A))

14-JUN-89

1GOX

1GOX

3

1GOX

4

... ATOM

232

N

ALA

29

54.035

4.332

19.352

1.00

23.93

1GOX

374

ATOM

233

CA

ALA

29

52.992

65.356 19.569

1.00

24.74

1GOX

375

ATOM

234

C

ALA

29

53.519

66.762 19.309

1.00

25.43

1GOX

376

ATOM

235

O

ALA

29

54.648

67.179 19.655

1.00

25.66

1GOX

377

ATOM

236

C

BALA 29

52.433

65.340 20.993

1.00

24.54

1GOX

378

HETATM

3165

O

HOH 658

62.480

62.480 0.000

0.50

65.79N

1GOX

3170

CONECT

2837 2838

1GOX

3171

... 2854

Each line shows the atom serial number, atom type, residue type, chain identifier (in case of multi-chain structure), residue serial number, orthogonal coordinates (three values), occupancy, temperature factor, and segment identifier. The annotation part of the PDB file format contains dozens of possible record types, including: HEADER (name of protein and release date), COMPND (molecular contents of the entry), SOURCE (biological source), AUTHOR (list of contributors), SSBOND (disulfide bonds), SLTBRG (salt bridges), SITE (groups comprising important sites), HET (nonstandard groups or residues [heterogens]), MODRES (modifications to standard residues), SEQRES (primary sequence of backbone residues), HELIX (helical substructures), SHEET (sheet substructures), and REMARK (other information and comments). The PDB allows a user to view a molecule structure interactively through a Virtual Reality Modeling Language (VRML) viewer, RasMol (Sayle and Milner-White, 1995), Chime, or QuickPDB (a Java applet for viewing sequence and structure) when the browser is configured to support these free rendering tools. The PDB provides related information about the protein, such as secondary structure assignment and geometry. Each PDB entry also links to a wide range of annotations from secondary databases, including (1) summary and display databases such as Graphical Representation and Analysis of Structure Server (GRASS; Nayal et al., 1999), Image Library (Shnel, 1996), Molecular Modelling Database (MMDB; Marchler-Bauer et al., 1999) in Entrez, PDBsum (Laskowski et al., 1997), and Sequence to and within Graphics (STING); (2) domain partition information from 3Dee (Siddiqui and Barton, 1995); (3) the MEDLINE bibliography; (4) structure quality assessment in PDBREPORT from WHAT IF (Vriend, 1990); (5) protein movements recorded in Database of Macromolecular Movement (MolMovDB; Gerstein and Krebs,



Supplement 33

1998); (6) structure families (CATH, CE, FSSP, SCOP, and VAST, as discussed later in this unit); and (7) geometry analyses of the protein, such as CSU Contacts of Structural Units (Sobolev et al., 1999) and castP Identification of Protein Pockets & Cavities (Liang et al., 1998). Several structure databases that are not linked by the PDB can also provide useful information. WPDB (Shindyalov and Bourne, 1995) can be used to visualize and analyze a PDB entry from Microsoft Windows. BioMagResBank (University of Wisconsin, 1999) is a repository for NMR spectroscopy data on proteins, peptides, and nucleic acids. Particularly, it provides partial NMR data (e.g., chemical shifts) before the full structure is solved. PROTEIN FAMILY DATABASES Introduction Proteins can be classified according to their evolutionary, structural, or functional relationships. A protein in the context of its family is much more informative than the single protein itself. For example, residues conserved across the family often indicate special functional roles. Two proteins classified in the same functional family may suggest that they share similar structures, even when their sequences do not have significant similarity. There is no unique way to classify proteins into families. Boundaries between different families may be subjective. The choice of classification system depends in part on the problem; in general, the authors suggest looking into classification systems from different databases and comparing them. Three types of classification methods are widely adopted, based upon the similarity of sequence, structure, or function. Sequence-based methods are applicable to any proteins whose sequences are known, while structure-based methods are limited to the proteins of known structures, and function-based methods depend on the functions of proteins being annotated. Sequence- and structure-based classifications can be automated and are scalable to high-throughput data, whereas function-based classification is typically carried out manually. Structure- and function-based methods are more reliable, while sequence-based methods may result in a false positive result when sequence similarity is weak (i.e., two proteins are classified into one family by chance rather than by any biological significance). In addition, since protein structure and function are better conserved than sequence, two proteins having similar structures or similar functions may not be identified through sequence-based methods. Databases for Sequence-Based Protein Families Sequence-based protein families are classified according to a profile derived from a multiple-sequence alignment. The profile can be shown across a long domain (typically 100 residues or more) or can be revealed in short sequence motifs. Classification methods based on profiles across long domains tend to be more reliable but less sensitive than those based on short sequence motifs.

Protein Databases on the Internet

Several sequence-based methods focus more on profiles across long domains, including Pfam (Bateman et al., 1999), ProDom (Corpet et al., 1999), SBASE (Murvai et al., 1999), and Clusters of Orthologous Group (COG; Tatusov et al., 1997). These methods differ in the techniques used to construct families. Pfam builds multiple-sequence alignments of many common protein domains using hidden Markov models. The ProDom protein domain database consists of homologous domains based on recursive PSI-BLAST searches (UNIT 2.5). SBASE is organized through BLAST neighbors and is grouped by standard protein names that designate various functional and structural domains of protein sequences. COG aims toward finding ancient conserved domains by delineating families of orthologs across a wide phylogenetic range.

2.6.8 Supplement 33


The following shows an example of Pfam for the GRIP domain (accession number PF01465). Pfam lists some useful information for the entry as follows: The GRIP (golgin-97, RanBp2alpha, Imh1p and p230/golgin245) domain is found in many large coiled-coil proteins. It has been shown to be sufficient for targeting to the Golgi. The GRIP domain contains a completely conserved tyrosine residue. The references of the above annotation are also given. In addition, Pfam gives the alignment between the family members: 015045/1511-1558

SAANLEYLKNVLLQFIFLKPG--SERERLLPVINTMLQLSPEEKGKLAAV

O15045

YNP9_CAEEL/633-681

NEKNMEYLKNVFVQFLKPESVP-AERDQLVIVLQRVLHLSPKEVEILKAA

P34562

Q06704/864-909

KNEKIAYIKNVLLGFLEHKE----QRNQLLPVISMLLQLDSTDEKRLVMS

Q06704

Q92805/691-737

REINFEYLKHVVLKFMSCRES---EAFHLIKAVSVLLNFSQEEENMLKET

Q92805

O42657/703-748

MLIDKEYTRNILFQFLEQRD----RRPEIVNLLSILLDLSEEQKQKLLSV

O42657

O70365/1161-1205

EPTEFEYLRKVMFEYMMGR-----ETKTMAKVITTVLKFPDDQAQKILER

070365

Q21071/692-741D

DPAEAEYLRNVLYRYMTNRESLGKESVTLARVIGTVARFDESQMKNVISS

Q21071

Q18013/574-623

STSEIDYLRNIFTQFLHSMGSPNAASKAILKAMGSVLKVPMAEMKIIDKK

Q18013

The alignment shows accession numbers and the range of each sequence. One can identify some features of the family through this pattern (i.e., from particularly conserved residues at specific alignment positions). Some methods are based on “fingerprints” of small conserved motifs in sequences, as with PROSITE (Hofmann et al., 1999), PRINTS (Attwood et al., 1999), and BLOCKS (Heniko et al., 1999). In protein sequence families, some regions have been better conserved than others during evolution. These regions are generally important for the function of a protein or for the maintenance of its three-dimensional structure, and hence are suitable for fingerprinting. The fingerprints can be used to assign a newly sequenced protein to a specific family. Fingerprints are derived from gapped alignments in PROSITE and PRINTS, but are derived from ungapped alignments (corresponding to the highly conserved regions in proteins) in BLOCKS. A fingerprint in PRINTS may contain several motifs from PROSITE, and thus may be more flexible and powerful than a single PROSITE motif. Therefore, PRINTS can provide a useful adjunct to PROSITE. It should be noted that some functionally unrelated proteins may be classified together due to chance matches in short motifs. Other sequence-based protein family databases consist of multiple sources. The ProClass database (Wu et al., 1999) is a nonredundant protein database organized according to family relationships as defined collectively by PROSITE patterns and PIR superfamilies. The MEGACLASS server (States et al., 1993) provides classifications by different methods, including Pfam, BLOCKS, PRINTS, ProDom, and SBASE. The MOTIF search engine at http://motif.genome.ad.jp/ includes PROSITE, BLOCKS, ProDom, and PRINTS. Databases for Structure-Based Protein Families The hierarchical relationship among proteins can be clearly revealed in structures through structure-structure comparison. Structure families often provide more information on the relationship between proteins than what sequence families can offer, particularly when two proteins share a similar structure but no significant sequence identity. Figure 2.6.2 shows an example of a structure-structure alignment between two proteins. Sometimes, sequence similarity between two proteins exists but is not strong enough to produce an unambiguous alignment. In this case, the alignment between two structures can generate Computational Analysis


Supplement 33

Figure 2.6.2 Structure superposition between glycolate oxidase (1gox, in black) and inosine monophosphate dehydrogenase (1ak5, in gray). This figure was made using MOLSCRIPT (Kraulis, 1991).

better alignment in terms of biological significance, and thus may pinpoint the active sites more accurately. Different structure-structure comparison methods yield different structure families. CATH (Class, Architecture, Topology and Homologous superfamily; Orengo et al., 1997) is a hierarchical classification of protein domain structures. CE (Combinatorial Extension of the optimal path; Shindyalov and Bourne, 1998) provides structural neighbors of the PDB entries with structure-structure alignments and three-dimensional superpositions. FSSP (Fold classification based on Structure-Structure alignment of Proteins; Holm and Sander, 1996) features a protein family tree and a domain dictionary, in addition to whole-chain-based classification, sequence neighbors, and multiple structure alignments. SCOP (Structural Classification of Proteins; Murzin et al., 1995) uses augmented manual classification, class, fold, superfamily, and family classification. VAST (Vector Alignment Search Tool; Gibrat et al., 1996) contains representative structure alignments and threedimensional superpositions. Among these five databases, SCOP provides more functionrelated information. However, due to the manual work involved, SCOP is not updated as frequently as the others (as of June, 2003, it was last updated for the PDB release on March 1, 2003), whereas FSSP and CATH follow the PDB updates closely. SCOP is used here as an example to show the features of structure-based families. SCOP can be accessed through its home server in the UK (http://scop.mrc-lmb.cam.ac.uk/scop/). It is also widely mirrored around the world, including: http://pdb.wehi.edu.au/scop/ (Australia) http://mdl.ipc.pku.edu.cn/scop/ (China) http://www.cdfd.org.in:5555/scop/ (India) http://pdb.weizmann.ac.il/scop/ (Israel) http://loki.polito.it/scop/ (Italy) http://scop.protres.ru/ (Russia) http://scop.bic.nus.edu.sg/ (Singapore) http://scop.life.nthu.edu.tw/ (Taiwan) http://scop.berkeley.edu/ (USA) Protein Databases on the Internet



SCOP describes the hierarchical relationship among proteins through the major levels of (homologous) family, superfamily, and fold. Proteins are clustered together into a (homologous) family if they have significant sequence similarity. Different families that have low sequence similarity but whose structural and functional features suggest a common evolutionary origin are placed together in a superfamily. Different superfamilies are categorized into a fold if they have the same major secondary structures in the same arrangement and with the same topological connections (the peripheral elements of secondary structure and turn regions may differ in size and conformation). Two superfamilies in the same fold may not have a common evolutionary origin. Their structural similarities may arise from the physics and chemistry of proteins favoring certain packing arrangements and chain topologies (Murzin et al., 1995). Figure 2.6.3 shows the SCOP interface using an example of protein 1gox in the PDB. Databases for Function-Based Protein Families There are various protein functional families classified from different perspectives. The ENZYME data bank (Bairoch, 1993) contains the following data for each enzyme: EC number, recommended name, alternative names, catalytic activity, cofactors, pointers to the SwissProt entry, and pointers to any disease associated with a deficiency of the enzyme. PROCAT is a database of three-dimensional enzyme active site templates (Wallace et al., 1996). PDD (Protein Disease Database; Lemkin et al., 1995; Merril et al., 1995) correlates diseases with proteins observable in serum, urine, and other common human body fluids based on biomedical literature. There are also a growing number of databases dedicated to special types of proteins, such as antibodies, G-protein-coupled receptors, HIV proteases, glycoproteins, and RNases, as shown in Table 2.6.1. OTHER DATABASES Protein Binding Databases Protein binding includes protein-substrate docking and protein-protein association. ReLiBase (Hendlich, 1998) is a database system for analyzing receptor-ligand complexes in the PDB. DIP (Database of Interacting Proteins) records protein pairs that are known to bind with each other. The information in DIP may provide information related to signaling pathways, multiple interactions, and complex systems. Protein Energetics Databases There are few databases for protein energetics, due to the low-throughput nature of the data source. One useful energetics database can be found in ProTherm (Thermodynamic Database for Proteins and Mutants; Gromiha et al., 1999). It contains thermodynamic data on mutations, including Gibbs free energy, enthalpy, heat capacity, and transition temperature. These data are important for understanding the structure and stability of proteins. Bibliographic Databases Searching for protein information through traditional bibliographic databases, such as MEDLINE or Grateful Med, can be rewarding. In addition, some bibliographic reference databases dedicated to proteins may provide certain information more directly. For example, SeqAnalRef stores papers dealing with sequence analysis.



Supplement 33

Figure 2.6.3 An example of the SCOP interface when searching the structure of 1gox in the PDB.

COMBINED DATABASES Introduction By integrating different types of protein databases together, a database of databases (or a data warehouse) can be built. Such combined databases not only serve as “one-stop shopping,” but also provide cross-references between entries in different databases. Two combined databases, Entrez and SRS, have been very successful. Entrez Entrez (Schuler et al., 1996) is a combined database consisting of literature, protein sequence and structure, nucleotide sequence, and taxonomy. Different types of information are interconnected through the grouping of sequences/structures and references by computed similarity scores. Entrez can be used through a variety of media, including CD-ROM, a custom Graphical Interface client, a World Wide Web browser, a command line browser (CLEVER), and the National Center for Biotechnology Information’s (NCBI’s) toolkit written in C. SRS SRS (Sequence Retrieval System; Etzold et al., 1996) is the most comprehensive database for molecular biology. The home server at http://srs.ebi.ac.uk supports 264 biological databases (as of March, 2003), including almost all the major protein/genetic databases. As an indexing system, it provides fast access to different databases through searches by sequence or by key words from various data fields. SRS also builds indices using cross-references between databases. An entry from one database can be linked to other databases that contain the entry. However, it should be noted that the contents of SRS may Protein Databases on the Internet



lag behind the other databases in updating (i.e., some new entries in the original databases may not be included in SRS). SUMMARY This unit reviews several major protein databases on the Internet, and shows what kind of information users can expect from protein databases. Although all technical procedures cannot be described here, most of the protein databases are easy to use and provide detailed on-line manuals so that even users with little computer skill can learn them quickly. Protein databases may not always be easily accessible or usable through the Internet. Sometimes a database server may be down or the Internet connection may be interrupted. Some structures or image files are very large (several megabytes), and the download time may be long. It can be helpful to use a mirror site of the database at a close location in order to accelerate the access speed. For a frequent user, it may be worthwhile to install the database on a local machine. On the other hand, it must be kept in mind that a mirror site or a local copy may contain an older version of the database than the one on the home server. It is important to assess the quality of the data. There are three types of data in protein databases. (1) Experimental data are generally very reliable. However, some entries may contain errors (e.g., some protein sequences) or may be based on low-resolution data (e.g., some protein structures determined by NMR). (2) Annotation data uses computational techniques on experimental data, for example, secondary structure assignment and domain partition in structure. These data depend on the quality of the experimental data and the computational methods used. Different methods may yield different results. (3) Prediction data includes, for example, sequence domain parsing and three-dimensional structure prediction. No matter how good the method, the results are still predictions and should be subjected to experimental verification. In addition, different methods typically give different predictions. In summary, caution is needed when using the data from databases to draw a conclusion. It is worthwhile to check the same type of data from different databases and compare them. It is sometimes necessary to use additional computational tools (e.g., tools to assess the quality of a structure) for further analysis. LITERATURE CITED Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., and Lipman, D.J. 1997. Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucl. Acids Res. 25:3389-3402. Attwood, T.K., Flower, D.R., Lewis, A.P., Mabey, J.E., Morgan, S.R., Scordis, P., Selley, J., and Wright, W. 1999. PRINTS prepares for the new millennium. Nucl. Acids Res. 27:220-225. Bairoch, A. 1993. The ENZYME data bank. Nucl. Acids Res. 21:3155-3156. Bairoch, A. and Apweiler, R. 1999. The SwissProt protein sequence data bank and its supplement TrEMBL in 1999. Nucl. Acids Res. 27:49-54. Barker, W.C., Garavelli, J.S., McGarvey, P.B., Marzec, C.R., Orcutt, B.C., Srinivasarao, G.Y., Yeh, L.L., Ledley, R.S., Mewes, H., Pfeiffer, F., Tsugita, A., and Wu, C. 1999. The PIR-international protein sequence database. Nucl. Acids Res. 27:39-42. Bateman, A., Birney, E., Durbin, R., Eddy, S.R., Finn, F.D., and Sonnhammer, E.L.L. 1999. Pfam 3.1: 1313 multiple alignments match the majority of proteins. Nucl. Acids Res. 27:260-262. Benson, D.A., Boguski, M.S., Lipman, D.J., Ostell, J., Ouellette, B.F., Rapp, B.A., and Wheeler, D.L. 1999. Genbank. Nucl. Acids Res. 27:12-17. Bernstein, F.C., Koetzle, T.F., Williams, G.J.B., Meyer, E.F., Brice, M.D., Rodgers, J.R., Kennard, O., Shimanouchi, T., and Tasumi, M. 1977. The protein data bank: A computer based archival file for macromolecular structures. J. Mol. Biol. 112:535-542. Bourne, P., Berman, H., Watenpaugh, K., Westbrook, J., and Fitzgerald, P. 1997. The macromolecular crystallographic information file (mmCIF). Methods Enzymol. 277:571-590.



Supplement 33

Corpet, F., Gouzy, J., and Kahn, D. 1999. Recent improvements of the ProDom database of protein domain families. Nucl. Acids Res. 27:263-267. Etzold, T., Ulyanov, A., and Argos, P. 1996. SRS: Information retrieval system for molecular biology data banks. Methods Enzymol. 266:114-128. Gerstein, M. and Krebs, W. 1998. A database of macromolecular motions. Nucl. Acids Res. 26:4280-4290. Gibrat, J.F., Madej, T., and Bryant, S.H. 1996. Surprising similarities in structure comparison. Curr. Opinion Struct. Biol. 6:377-385. Gromiha, M.M., An, J., Kono, H., Oobatake, M., Uedaira, H., and Sarai, A. 1999. Protherm: Thermodynamic database for proteins and mutants. Nucl. Acids Res. 27:286-288. Hendlich, M. 1998. Databases for protein-ligand complexes. Acta Crystallogr., Sect. D 1:1178-1182. Heniko, J.G., Heniko, S., and Pietrokovski, S. 1999. New features of the blocks database servers. Nucl. Acids Res. 27:226-228. Hofmann, K., Bucher, P., Falquet, L., and Bairoch, A. 1999. The PROSITE database, its status in 1999. Nucl. Acids Res. 27:215-219. Holm, L. and Sander, C. 1996. Mapping the protein universe. Science 273:595-602. Kraulis, P. 1991. MOLSCRIPT—a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24:946-950. Laskowski, R.A., Hutchinson, E.G., Michie, A.D., Wallace, A.C., Jones, M.L., and Thornton, J.M. 1997. PDBsum: A web-based database of summaries and analyses of all PDB structures. Trends Biochem. Sci. 22:488-490. Lemkin, P.F., Orr, G.A., Goldstein, M.P., Creed, G.J., Myrick, J.E., and Merril, C.R. 1995. The protein disease database of human body fluids: I. Computer methods and data issues. Appl. Theor. Electrophor. 5:55-72. Liang, J., Edelsbrunner, H., and Woodward, C. 1998. Anatomy of protein pockets and cavities: Measurement of binding site geometry and implications for ligand design. Protein Science 7:1884-1897. Marchler-Bauer, A., Addess, K.J., Chappey, C., Geer, L., Madej, T., Matsuo, Y., Wang, Y., and Bryant, S.H. 1999. MMDB: Entrez’s 3D structure database. Nucl. Acids Res. 27:240-243. Merril, C.R., Goldstein, M.P., Myrick, J.E., Creed, G.J., and Lemkin, P.F. 1995. The protein disease database of human body fluids: I. Rationale for the development of this database. Appl. Theor. Electrophor. 5:49-54. Mural, R.J., Parang, M., Shah, M., Snoddy, J., and Uberbacher, E.C. 1999. The Genome Channel: A browser to a uniform first-pass annotation of genomic DNA. Trends Genet. 15:38-39. Murvai, J., Vlahovicek, K., Barta, E., Szepesvari, C., Acatrinei, C., and Pongor, S. 1999. The SBASE protein domain library, release 6.0: A collection of annotated protein sequence segments. Nucl. Acids Res. 27:257-259. Murzin, A.G., Brenner, S.E., Hubbard, T., and Chothia, C. 1995. SCOP: A structural classification of proteins database for the investigation of sequences and structures. J. Mol. Biol. 247:536-540. Nayal, M., Hitz, B.C., and Honig, B. 1999. GRASS: A server for the graphical representation and analysis of structures. Protein Sci. 8:676-679. Ogata, H., Goto, S., Sato, K., Fujibuchi, W., Bono, H., and Kanehisa, M. 1999. KEGG: Kyoto encyclopedia of genes and genomes. Nucl. Acids Res. 27:29-34. Orengo, C.A., Michie, A.D., Jones, D.T., Swindells, M.B., and Thornton, J.M. 1997. CATH—a hierarchic classification of protein domain structures. Structure 5:1093-1108. Rebhan, M., Chalifa-Caspi, V., Prilusky, J., and Lancet, D. 1998. GeneCards: A novel functional genomics compendium with automated data mining and query reformulation support. Bioinformatics 14:656-664. Rodriguez-Tom, P., Stoehr, P.J., Cameron, G.N., and Flores, T.P. 1996. The European Bioinformatics Institute (EBI) databases. Nucl. Acids Res. 24:6-13. Sayle, R.A. and Milner-White, E.J. 1995. RASMOL: Biomolecular graphics for all. Trends Biochem. Sci. 20:374-376. Schuler, G.D., Epstein, J.A., Ohkawa, H., and Kans, J.A. 1996. Entrez: Molecular biology database and retrieval system. Methods Enzymol. 266:141-162. Shindyalov, I.N. and Bourne, P.E. 1995. WPDB: A PC-based tool for analyzing protein structure. J. Appl. Crystallogr. 28:847-852. Shindyalov, I.N. and Bourne, P.E. 1998. Protein structure alignment by incremental combinatorial extension (CE) of the optimal path. Protein Eng. 11:739-747. Shnel, J. 1996. Image library of biological macromolecules. Comput. Appl. Biosci. 12:227-229. Protein Databases on the Internet

Siddiqui, A.S. and Barton, G.J. 1995. Continuous and discontinuous domains: An algorithm for the automatic generation of reliable protein domain definitions. Protein Sci. 4:872-884.



Sobolev, V., Sorokine, A., Prilusky, J., Abola, E.E., and Edelman, M. 1999. Automated analysis of interatomic contacts in proteins. Bioinformatics. 15:327-332 States, D.J., Harris, N.L., and Hunter, L. 1993. Computationally efficient representation of classes in protein sequence megaclassification. Proc. Intel. Syst. Mol. Biol. 1:387-394. Tatusov, R.L., Koonin, E.V., and Lipman, D.J. 1997. A genomic perspective on protein families. Science 278:631-637. University of Wisconsin. 1999. BioMagResBank. University of Wisconsin, Madison, Wis. Vriend, G. 1990. WHAT IF: A molecular modelling and drug design program. J. Mol. Graphics 8:52-56. Wallace, A.C., Laskowski, R.A., and Thornton, J.M. 1996. Derivation of 3D coordinate templates for searching structural databases: Application to Ser-His-Asp catalytic triads in the serine proteinases and lipases. Protein Sci. 5:1001-1013. Wu, C., Shivakumar, S., and Huang, H. 1999. Proclass protein family database. Nucl. Acids Res. 27:272-274.

INTERNET RESOURCES The Web addresses of the databases mentioned in this unit are listed in Table 2.6.1. Readers can find more protein databases and their related tools in the following Web pages, which collect a large number of useful links. http://compbio.ornl.gov/structure/resource/ Oak Ridge National Laboratory’s resources of protein modeling tools. http://www.public.iastate.edu/~pedro/research_tools.html Pedro’s biomolecular research tools. http://www.expasy.org/alinks.html Amos’ WWW links page.

Contributed by Dong Xu and Ying Xu Oak Ridge National Laboratory Oak Ridge, Tennessee

The authors thank Drs. Edward C. Uberbacher, Michael A. Unseren, Jay Snoddy, and Gwo-liang Chen for helpful discussions. This work is supported by the Office of Biological and Environmental Research, U.S. Department of Energy, under Contract DE-AC05-00OR22725, managed by UT-Battelle, LLC.



Supplement 33

Protein Tertiary Structure Prediction

UNIT 2.7

Predicting the three-dimensional structures of proteins from their amino acid sequences using computational methods is a very important and challenging problem in contemporary molecular biology. The three-dimensional structures of proteins not only define their biological functions, but also provide a key to engineering proteins and to designing drugs that target proteins related to disease. Protein tertiary structure prediction provides timely tools on account of rapid progress being made in genome sequencing, which is proceeding at a rate many times faster than experimental determination of protein structures. In conjunction with supporting experimental data, these tools can provide valuable insights into protein structure and function. In particular, predicted structures often give experimentalists some direction for further studies. Protein tertiary structure prediction is not only important, but also becomes more and more practical. Many nontrivial structure predictions (Nilges and Brünger, 1993; Hu et al., 1995; Madej et al., 1995) obtained prior to experimental structures turned out to be fairly accurate. Most notably, the success of protein structure prediction has been demonstrated in community-wide experiments on the Critical Assessment of Techniques for Protein Structure Prediction (CASP; see CASP, 1995, 1997, 1999). In each of the CASP experiments, crystallographers and NMR spectroscopists were solicited for the names of proteins whose structures were likely to be solved before a given deadline. The sequences of these target proteins were made available, and predictors submitted their models by the deadline. These blind predictions were then assessed through comparison with the experimental structures. It has been shown that many predictions in CASP have good qualities. There are three types of tertiary structure predictions: homology or comparative modeling, fold recognition, and ab initio structure prediction. Homology modeling constructs the coordinates of all the atoms in a query protein based on sequence alignment between the query protein and another protein of known three-dimensional structure. Fold recognition identifies a suitable fold for the query sequence from a structure library and provides an alignment between the query protein and the fold. It consists of two approaches, the sequence profile approach and the sequence-structure comparison (threading) approach. The ab initio structure prediction attempts to predict protein structure based directly on physicochemical principles. Homology modeling has been widely used. It can provide relatively reliable atomic coordinates with a low root mean square deviation (RMSD) between a model and a high-quality experimental structure. Fold recognition, which often generates useful backbone structures, also starts to gain ground. The ab initio prediction is still far from general application, although a few isolated successful predictions have been reported. The boundaries between the different types of predictions are becoming blurred as researchers start to combine them together. For example, one can use the alignment derived from fold recognition in homology modeling, or assemble partial structures predicted by threading in ab initio prediction. Both homology modeling and fold recognition rest on the foundation that the three-dimensional structures of proteins have been better conserved during evolution than their sequences. The relationship between a query protein and its template in a structure database can be classified at different hierarchical levels according to their structural and evolutionary relationship. A widely used classification consists of family, superfamily, and fold (Murzin et al., 1995). A query protein and its template are clustered into a family if they have clear evolutionary relationship with significant sequence identity between them (often 25% or higher). If two proteins have low sequence identity, but their structural Contributed by Dong Xu and Ying Xu Current Protocols in Protein Science (2000) 2.7.1-2.7.17 Copyright © 2000 by John Wiley & Sons, Inc.

Quantitation of Protein Interactions

2.7.1 Supplement 19

and functional features suggest that a common evolutionary origin is probable, then they are placed in the same superfamily. If two proteins have the same major secondary structures in the same arrangement and with the same topological connections, but they have no significant sequence similarity and may not have a common evolutionary origin, they are defined as having the same fold. It is estimated that the number of folds in all genomes ranges from 650 to 15,000, or less than one thousandth of the number of proteins (Wang, 1998). Different methods of structure predictions are better suited to proteins with different relationship to their templates. When a query protein and its template belong to the same family and they have high sequence identity to produce an unambiguous sequence alignment, homology modeling is the best choice and it typically produces a high-quality model including sidechain atoms. About 17% of identified genes fall into this category (Sanchez and Sali, 1998). The sequence profile approach is suitable for the following two cases: (1) a query protein and its template belong to the same family, but they do not have sufficient sequence identity to produce an unambiguous sequence alignment, or (2) they belong to the same superfamily, and the multiple-sequence alignment of the query protein has a strong pattern. For these two cases, which are also referred to as “remote homologs,” the sequence profile approach typically identifies the structure template with high confidence and good alignment. It extends the genome-wide coverage from 17% to ∼30% to 40% (Gerstein, 1998). The threading approach becomes the major tool for the following two cases: (1) a query protein and its template belong to the same superfamily, but the query protein has few proteins with significant sequence similarities in sequence databases, or the multiple-sequence alignment of the query protein has a weak pattern, or (2) they belong to the same fold, and threading gives a good confidence level for the prediction. This category currently accounts for an additional ∼10% to 20% of identified genes (Jones, 1999). If a query protein and its template belong to the same fold, but threading cannot recognize the template with a good confidence level, or the template does not exist in the structure database (i.e., the query protein is a new fold), it can only be predicted by ab initio methods. In total, structure predictions for about half of the genes do not have to use ab initio prediction and can provide useful information reliably. This number is growing as more and more structures are being solved. This unit provides an overview for homology modeling, fold recognition, and ab initio prediction. It introduces several major protein structure prediction tools. For each tool, it addresses (1) the method, (2) the technical protocol, (3) the expected result, and (4) the confidence level of the prediction. The Web addresses of the major structure prediction tools are listed in Table 2.7.1. The reader can find more detailed information about the tools in their Web pages or manuals. The examples used in this unit are all from the third CASP experiment (CASP-3) held in 1998. This blind test provides an up-to-date benchmark regarding the results that a user can expect from these prediction tools. Among the examples used, the models for t0053, t0067, t0068, and t0074 are from the authors’ own predictions submitted to CASP-3 (Xu et al., 1999). All the results are based on the programs and databases as of August, 1999. HOMOLOGY MODELING


A homology modeling method typically consists of three steps: (1) identification of protein templates with known three-dimensional structures and production of an alignment between the query sequence and its templates, (2) building a model for the query protein based on its alignment with the template structures, and (3) evaluation of the quality of the model.

2.7.2 Supplement 19


Table 2.7.1

Selected Protein Structure Prediction Tools

Program

Web site

Typea

Sequence alignment: ALIGN

http://www2.igh.cnrs.fr/bin/align-guess.cgi

Server

BLAST FASTA

http://www.ncbi.nlm.nih.gov/BLAST/ http://www.embl-heidelberg.de/cgi/fasta-wrapper-free

Server Server

KESTREL SSEARCH

http://www.cse.ucsc.edu/research/kestrel/ http://vega.igh.cnrs.fr/bin/ssearch-guess.cgi

Server Server

Homology modeling: COMPOSER

http://www-cryst.bioc.cam.ac.uk

SGI executable

CONGEN

http://www.tripos.com/software/composer.html http://www.congenomics.com/congen/congen toc.html

SGI module (GUI) Executables

CPHmodels DRAGON

http://www.cbs.dtu.dk/services/CPHmodels/ http://www.nimr.mrc.ac.uk/∼mathbio/a-aszodi/dragon.html

Server SGI executable

LOOK MODELLER

http://www.mag.com/products/look.html http://guitar.rockefeller.edu/modeller/

Module (GUI) Executables

SwissModel

http://www.msi.com/solutions/products/insight/modules/Modeler.html SGI module (GUI) Server http://www.expasy.ch/swissmod/SWISS-MODEL.html

WHAT IF

http://www.sander.embl-heidelberg.de/whatif/

Executables

http://www.ncbi.nlm.nih.gov/BLAST/ http://www.cse.ucsc.edu/research/compbio/sam.html

Server Server

Sequence profile methods: PSI-BLAST SAM-T99

Singleton-potential threading: 123D http://www-lmmb.ncifcrf.gov/∼nicka/123D.html

Server

SAS TOPITS

http://www.biochem.ucl.ac.uk/bsm/sas/ http://dodo.cpmc.columbia.edu/predictprotein/

Server Server

UCLA-DOE

http://www.doe-mbi.ucla.edu/people/frsvr/frsvr.html

Server

Pairwise-potential threading: NCBI Package PROFIT

http://www.ncbi.nlm.nih.gov/Structure/ http://lore.came.sbg.ac.at

Executables Executables

PROSPECT THREADER

http://compbio.ornl.gov/structure/prospect/ http://globin.bio.warwick.ac.uk/∼jones/threader.html

Executables Executables

ToPLign

http://cartan.gmd.de/ToPLign.html

Server

aExecutables without a specified machine type can be run on multiple platforms. Abbreviations: GUI, Graphic User Interface; SGI, Silicon Graphic,

Inc.

Template Search and Alignment Conventional homology modeling requires that the sequence similarity between a query sequence and its template sequence be significant, so that using a pairwise sequence-sequence alignment can identify the template and produce an unambiguous alignment. The most widely used sequence-sequence alignment tool is BLAST (UNIT 2.5), which is very fast. Using BLAST to search against the Protein Data Bank (PDB; Bernstein et al., 1977) is probably the best starting point for homology modeling. If the identified templates or alignments have any uncertainty, more sensitive but slower methods (e.g., FASTA; Pearson and Lipman, 1988) can be used. An even slower but much more sensitive method



Supplement 19

is the Smith-Waterman alignment (Smith and Waterman, 1981), which has been implemented in SSEARCH and KESTREL (Hughey, 1996). Templates and alignments using fold recognition methods, as described in the following two sections, can also be used for homology modeling. In fact, it has been shown in CASP-3 that profile-based alignment methods and threading methods can produce more accurate alignments than single-sequence alignment methods even when the query protein and its template have some sequence similarity (CASP, 1999). It is sometimes important to find as many templates as possible with a multiple-sequence alignment. Template search and alignment are essential for the correctness and the quality of a homology model. Homology modeling programs always generate a structure for any query sequence using the conformation of the template structures and the alignments between the query protein and its templates. If the templates or the alignments are incorrect, the output model will certainly be a result of “garbage in, garbage out.” Building Atomic Models Different homology modeling methods use different ways to construct a three-dimensional model from given templates and alignments. Some automated homology modeling servers, e.g., SwissModel (UNIT 2.8; Peitsch, 1996) and CPHmodels (Lund et al., 1997), provide interfaces for submitting a sequence and obtaining the model interactively or through e-mail. The use of SwissModel is described in detail in UNIT 2.8. These servers are fast and easy to use. The program WHAT IF (Vriend, 1990) provides the option to construct a crude model quickly or to build a structure using a better, but much slower, method (several hours for a large protein, e.g., >300 amino acids). Another program is COMPOSER (Srinivasan and Blundell, 1993), which has a free academic version and is also integrated into the commercial molecular modeling package SYBYL (Tripos, 1999). It has a specific tool for dealing with loop regions, which contain gaps in the alignment. COMPOSER under SYBYL also provides an interactive Graphic User Interface (GUI) for model building, which allows a user to edit at each step. The most widely used homology modeling program is MODELLER (Sali and Blundell, 1993), which has a free academic version as well as a commercial version integrated into the molecular modeling package INSIGHT (Molecular Simulations, 1998). The model structure starts with an extended strand, and then folds into a compact one by satisfying spatial restraints derived from the alignment between the query sequence and its templates of known structures. In particular, it tries to preserve main chain dihedral angles or hydrogen bonding features from the template structures. Meanwhile, MODELLER uses physical force fields to prevent interatomic clashes. In loop regions with gaps in the alignment, MODELLER uses statistical information derived from the alignment of many proteins of known three-dimensional structure. The final three-dimensional model is obtained by optimization through conjugate gradients and molecular dynamics with simulated annealing. MODELLER provides a wide range of options for structure construction and refinement.


t0074 (the second EH domain of eps15), which was predicted by the authors, can be used as a CASP-3 example to show some basic features of MODELLER. Constructing a model requires at least three input files: the three-dimensional coordinate file for the template, the alignment file between the query protein and the template, and a script file to run MODELLER. The template found for t0074 is the PDB entry 1ahr (calcium-binding protein), which has three-dimensional coordinates recorded in the file 1ahr.atm (the same file as 1ahr.pdb in the PDB database). The alignment between t0074 and 1ahr is saved in the file 1ahr.ali using the PIR database format as follows:

2.7.4 Supplement 19


>P1;t0074 sequence:t0074: 1 : : 72 : : casp-t0074: : PWAVKPE--DKAKYDAIFDSLSP-VNGFLSGDKVKPVLLNS--KLPVDILGRVWELSDIDHDGML DRDEFAVAMFLV* >P1;1ahr structureX:1ahr: 73 : : 148 : : template ARKMK-D--SEEEIREAFRVFDKDGNGFISAAELRHVMTNLGEKLTDEEVDEMIREADIDGDGQV NYEEFVTMMTSK*

where the first line of each entry specifies the protein code after the “>P1", the second line describes the type, the name, and the residue ranges, and the third line provides the sequence. The following is a sample script file, run.top, for building the model for the query protein t0074: INCLUDE

# Include predefined routines

SET TOPLIB = ’${LIB}/top_allh.lib’ # topology library SET PARLIB = ’${LIB}/par.lib’ # parameters library SET HYDROGEN_IO = ’ON’ SET HETATM_IO = ’ON’

# include hydrogen atoms # include HETATM entries in template

SET ALNFILE = ’1ahr.ali’ SET KNOWNS = ’1ahr’ SET SEQUENCE = ’t0074’

# alignment filename # template name # target sequence name

CALL ROUTINE = ’model’

# do homology modeling

The text following “#” is commentary that will be ignored when the script is run. To run the script, simply type mod run.top

It typically takes several minutes to an hour to run a MODELLER job, depending on the protein sizes and the machine used. After the job mod run.top is finished, a file (t0074.B99990001) containing the three-dimensional coordinates for the atomic model of t0074 is generated, as shown in Fig. 2.7.1A. MODELLER can easily incorporate more sophisticated modeling protocols. For example, several more templates can be used at the same time by changing the SET KNOWNS entry in run.top. For example, to add three templates in addition to “1ahr,” SET KNOWNS would be changed to: SET KNOWNS = ’1ahr’ ’4cln’ ’1cmg’ ’1cdla’

and the alignments of all four templates with the query sequence would be provided in 1ahr.ali. MODELLER can also use specified information (such as NMR data, secondary structure types, and disulfide bonds) as spatial restraints. The following shows an example for building a model with the constraint of disulfide bonds between certain residues: SUBROUTINE ROUTINE = ’special_patches’ # A disulfide between residues 5 and 60: PATCH RESIDUE_TYPE = ’DISU’, RESIDUE_IDS = ’5’ # A disulfide between residues 22 and 38: PATCH RESIDUE_TYPE = ’DISU’, RESIDUE_IDS = ’22’ RETURN END_SUBROUTINE

’60’ ’38’ Quantitation of Protein Interactions


Supplement 19

A

B

Figure 2.7.1 A comparison between the predicted structure using MODELLER (thick lines) and the experimental structure (thin lines) for CASP-3 targets t0074 (A) and t0068 (B). This figure was made using visual molecular dynamics (VMD; Humphrey et al., 1996).

Model Assessment and Refinement The quality of a model depends primarily on the sequence identity between the query protein and the template. The higher the sequence identity, the more accurate a structure homology modeling can provide. As shown in Figure 2.7.1A, there are some conformational differences between the model and the experimental structure, mainly due to the structural differences between the template and the query protein. In this case, the sequence identity between the query protein and the template is only 15%. For another CASP-3 target, t0068, the sequence identity between the query protein and the template is 25%. As shown in Figure 2.7.1B, the model superimposes the experimental structure better than that of t0074. For high sequence identity, homology modeling often produces models with an all-atom RMSD of 60% pure when analyzed by SDS-PAGE (UNIT 6.3). Extracting Protein The washed pellets are extracted with high concentrations of protein denaturants such as 6 to 8 M guanidine⋅HCl or urea. It should be noted that some proteins are resistant to denaturation with high concentrations of these reagents, especially urea. Some washed pellets extracted with 8 M guanidine⋅HCl can be viscous and unsuitable for direct chromatography. In these cases, pre-extraction of the washed pellets with a limiting concentration (0.5 to 2.0 M) of guanidine⋅HCl can often overcome this problem. Solubilization with the anionic detergent N-lauroylsarcosine (Nguyen et al., 1993; Burgess and Knuth, 1996) and with 10% to 20% acetic acid has also been useful (UNIT 6.5); other denaturants for extracting inclusion bodies are described elsewhere (UNIT 6.3; Marston and Hartley, 1990). For background information on the mode of action of protein denaturants, readers should consult the reviews of Tanford (1968) and Creighton (1993). If the protein contains cysteine residues, it is essential to include a reducing agent, preferably 5 to 10 mM dithiothreitol (DTT). Even in the presence of strong protein denaturants, it may be necessary to sonicate or heat samples briefly to completely disperse and solubilize the protein. The extraction process should completely disaggregate and denature the protein into unfolded monomers. Urea is not recommended for the initial extraction. For example, even if it is known that a native version of protein can be unfolded with 4 M urea, the same protein in an E. coli inclusion body will almost certainly not be completely extracted as unfolded monomers with that same concentration of urea (or in most cases, even with 8 M urea). Initial extraction trials should be carried out with guanidine⋅HCl, which is more effective than urea. Most proteins will be extracted with 6 to 8 M guanidine⋅HCl. There should be adequate reductant present to maintain sulfhydryl groups in the reduced state, and thus prevent artificial disulfide bond formation. The presence of EDTA and a slightly acidic pH of 6.0 to 6.5 will help minimize cysteine oxidation. The extract may require clarification by filtration or centrifugation. Choosing Purification or Folding The extracted protein can be further purified, or it can be directly folded and then purified. Protein folding appears to be unaffected by the protein background in bacterial extracts (London et al., 1974), however, removal of nonproteinaceous material prior to folding has been reported to be beneficial (Darby and Creighton, 1990). Based on recent work, it is worth noting that high concentrations of background bacterial protein may promote aggregation of the unfolded recombinant protein by macromolecular crowding effects (Ellis, 2001). If purification of protein in the denatured state is possible, use the purified material to develop a folding protocol. Then use this protocol with clarified protein extracts, or better still with protein partially purified by DEAE-Sepharose, to observe if the presence of contaminants has any effect on the yield of folded protein. Finally, there may be specific reasons for purifying proteins in the denatured state. For example, some proteolytic enzymes, such as HIV-1 protease, self-digest (undergo auto-

Purification of Recombinant Proteins


Supplement 30

proteolysis) in the uninhibited state (Mildner et al., 1994, and references cited therein) but can be purified intact in the denatured (inactive) state, then refolded when required. Other proteins once folded may have low solubilities and be especially susceptible to aggregation, resulting in poor behavior on column matrices (see VP26 purification below). However, in general, unfolded proteins are more susceptible to chemical and proteolytic modifications. Purifying Denatured Proteins If the protein is extracted with guanidine⋅HCl, gel filtration is a useful first purification method; often protein >80% pure can be obtained (UNIT 6.3; Wingfield et al., 1997). The proteins exist as random coils in the denaturant and their elution from the column should be in order of their molecular weight and not be influenced by shape. If the protein is located in several peaks there may have been incomplete solubilization during the extraction. In this case, 8 M guanidine⋅HCl should be used for the extraction and the protein dispersed by sonication or by heating if necessary. Another possibility is intermolecular disulfide bond formation, in which case the DTT concentration in the sample and column buffers should be increased. It is worth noting that the column can often be equilibrated and eluted with lower guanidine⋅HCl concentrations (e.g., 4 M) than those used for the actual extraction process. Only monomeric protein should be selected for further processing. The protein at this stage can be stored frozen, ideally at −80°C. The partially purified protein in guanidine⋅HCl can be directly folded (see Performing Protein Folding), or the denaturant can be exchanged by dialysis or gel filtration for 1% to 5% (v/v) acetic or formic acids (acetonitrile at 5% to 10% v/v can also be included) and then lyophilized. Alternatively, the protein can be acidified with trifluoroacetic acid (TFA; ≤0.1% v/v) and further purified by reversed-phase chromatography (Wingfield, 1997; Wingfield et al., 1999). Useful high-flow matrices (Source 15RPC from Amersham Biosciences) can be purchased as bulk media. These matrices may not have the resolution of traditional prepackaged silica-based reversed-phase columns, but they have high capacity, can be eluted at higher flow rates, and are stable over a wide range of pH. Proteins eluted with acetonitrile/TFA are also suitable for lyophilization. Proteins tagged with histidine residues can be purified in guanidine⋅HCl-, urea-, or even SDS-containing buffers, using metal chelate chromatography (UNIT 6.5). There are many reports of “on-column protein folding” by binding the unfolded protein in guanidine⋅HCl or urea and then accomplishing folding using a reverse urea gradient (e.g., Gulnik et al., 2001). Proteins in urea and non-ionic or zwitterionic detergents (e.g., CHAPS) can be purified by ion-exchange chromatography (e.g., Wingfield et al., 1990). For ion-exchange chromatography, better results have been reported using protein that has been first extracted with guanidine⋅HCl, and then exchanged into urea (Shire et al., 1984). If urea is used either for extraction or for maintaining solubility during refolding, a cyanate scavenger such as a glycine- or Tris-based buffer should be included to prevent carbamylation of the protein (Stark et al., 1960). For critical work, urea can be deionized with a mixed bed ion-exchange resin (see discussion of Protein Folding Reagents in APPENDIX 3A). Performing Protein Folding

Purification of Recombinant E. coli Proteins

Protocols for folding proteins basically involve controlled removal of the denaturant under conditions that minimize aggregation and allow correct formation of disulfide bonds. For overviews of the practical aspects of protein folding, see UNIT 6.4; Wetzel (1992); Thatcher



et al. (1996); Rudolph et al. (1997); Lilie et al. (1998); De Bernardez Clark et al. (1999); and De Bernardez Clark (2001). To minimize nonproductive aggregation, folding is normally carried out at low protein concentrations (e.g., 0.01 to 0.10 mg/ml); for small, single-domain proteins, higher concentrations (e.g., 0.1 to 1.0 mg/ml) can often be tolerated. Dilution and dialysis are the most common methods for removing the denaturant. Solubility during folding can be maintained with co-solvents such as nondenaturing concentrations of urea (1 to 4 M; London et al., 1974; UNIT 6.5) or guanidine⋅HCl (0.1 to 1.5 M; Orsini and Goldberg, 1978), arginine (0.4 to 0.8 M; De Bernardez Clark et al., 1999), nonionic detergents and lipids (Zardeneta and Horowitz, 1994), cationic detergents (Puri et al., 1992), and polyethylene glycol (PEG; Cleland et al., 1992). These various additives function by minimizing intermolecular associations between “sticky” hydrophobic surfaces present in folding intermediates. For further discussion of aggregation versus folding, see Goldberg et al. (1991) and Kiefhaber et al. (1991). Additives such as ammonium sulfate, glycerol, sucrose, enzyme substrates or inhibitors, and ligands have also been used to improve protein folding (see Table 1 in De Bernardez Clark et al., 1999, for a useful list of additives used in folding). Protein expressed in the cytoplasm of E. coli is in the reduced state; this is true for both soluble and insoluble proteins. Once insoluble protein is solubilized, it needs to be maintained in a reduced state by the presence of reductant until protein folding is initiated. The oxidative formation of disulfide bonds (one of the rate-limiting steps in protein folding) can be catalyzed by low-molecular-weight thiol and disulfide pairs such as reduced and oxidized glutathione (GSH/GSSG). Redox buffers facilitate oxidation through thiol/disulfide exchange reactions (reviewed by Wetlaufer, 1984; Creighton, 1984; Gilbert, 1995). Normally GSH/GSSG ratios of 5 to 10 are used with a total glutathione concentration of 1 to 5 mM (Wetlaufer, 1984). To reduce the rate of GSH loss due to air oxidation, 1 mM EDTA should be included in the buffer (Wetlaufer et al., 1987). The optimal concentrations and ratios of reagents must be established in an empirical manner. Folding and oxidation are normally carried out concurrently (for further details, see Rudolph et al., 1997). Analogous to the approach commonly used to optimize conditions for protein crystallization, various screens have been developed to establish initial conditions for protein renaturation and oxidation (Hofmann et al., 1995; Armstrong et al., 1999) and kits are commercially available (FoldIt Screen from Hampton Research at http://www.hamptonresearch.com). For examples of preparative protein folding, see UNIT 6.5. In addition, some recent examples from the author’s laboratory are given below. The refolding of Fab fragments expressed in E. coli (Buchner and Rudolph, 1992) is illustrative of the systematic and empirical approach used to optimize folding conditions. Other examples of interest are described by Kohno et al. (1990) and Grunfeld et al. (1992). Protein-assisted folding and oxidation Protein folding in vivo is assisted in both eukaryotes and prokaryotes by two classes of accessory proteins: folding catalysts (for a review, see Schiene and Fischer, 2000) and molecular chaperones (Eisenberg, 1999; Feldman and Frydman, 2000). Folding catalysts accelerate rate-limiting steps in protein folding such as disulfide bond formation (protein disulfide isomerases) and the rotation of X-Pro bonds (peptidyl prolyl cis-trans isomerase) during protein folding. Chaperones bind denatured or unfolded proteins thus preventing misfolding and aggregation. The cytoplasm of E. coli is maintained in the reduced state by thioredoxin and the glutathione/glutaredoxin pathways. In hosts where the reduction of thioredoxin and glutathione is impaired by mutations to the thioredoxin



Supplement 30

reductases and glutathione reductase genes, the resultant oxidizing conditions allow the formation of disulfide bonds in expressed proteins located in bacterial cytoplasm (Bessesste et al., 1999; cells and expression kits are commercially available from Novagen). The periplasm of E. coli also contains protein disulfide isomerases, the Dsb enzymes, which have thioredoxin-like folds and act as strong thiol:disulfide oxidants (Missiakkas and Raina, 1997; Braun et al., 1999). Secretion of proteins into the periplasmic space has been the traditional approach for producing oxidized proteins in vivo, but with the aforementioned advances in cytoplasmic oxidations, this approach is probably best suited for proteins that are toxic to the cell when expressed in the cytoplasm (Cornelis, 2000). As mentioned, molecular chaperones prevent aggregation by interacting transiently with hydrophobic patches on unfolded proteins and suppressing aggregation and promoting folding (UNIT 6.4; reviewed by Jaenicke, 1993; Ellis and Hart, 1999; Feldman and Frydman, 2000). There are now many examples of chaperone-assisted protein expression in which the endogenous levels of the bacterial chaperones GroES and GroEL (∼1%) are increased up to ten-fold by co-expression with a target protein (Cole, 1996; Goenka and Rao, 2001). Often, increases in soluble protein expression are observed, but this is not always the case. Chaperones have also been used in vitro as protein folding reagents and some examples of folding in the presence of protein disulfide isomerase, peptidyl prolyl cis-trans isomerase and GroES/GroEL are given in Rudolph et al. (1997). Protocols for the high-level expression and rapid purification of E. coli GroEL and GroES are described by Kamireddi et al. (1997). Purifying Folded Protein Once the protein has been folded, any of the purification methods discussed in Chapters 8 and 9 can be used. The number of purification steps required should be fewer than those for a protein expressed in a soluble state because of the purification factor obtained by preparation of washed inclusion bodies (UNIT 6.3). One of the purification methods that should be included is gel filtration, which may be the only one required. A correctly selected matrix should remove any remaining E. coli proteins and separate aggregated and misfolded protein from the native folded protein. Misfolded protein may be expected to have a larger molecular radius (higher apparent mass) than the corresponding native protein. Monitoring Protein Folding The restoration of function (e.g., enzymatic or biological activity) is perhaps the best criterion for detecting successful folding. However, it is not always practical to use activity measurements to monitor folding. It is also worth mentioning that an unfolded protein may become activated following the dilution required for many activity measurements. Conversely, native proteins can be denatured or inactivated during prolonged incubation at 37°C or by adsorption to microtiter plates. The use of antibodies to monitor protein folding is briefly reviewed by Goldberg (1991), and reviews of common spectroscopic methods, such as circular dichroism and fluorescence, are provided in Chapter 7 and by Schmid (1997). BACTERIAL EXPRESSION OF PROTEINS NORMALLY GLYCOSYLATED


Because E. coli lacks glycosylation machinery, expression of glycoproteins in E. coli systems results in the synthesis of nonglycosylated variants. Glycoproteins expressed in E. coli are often, but not always, insoluble. In vitro folding studies with glycosylated and nonglycosylated forms of proteins indicate that the carbohydrate can stabilize folding



intermediates, and thus enhance folding, while not necessarily affecting the stability of the native state (Kern et al., 1993, and references cited therein). In eukaryotic cells, interference with protein glycosylation can lead to the formation of misfolded, aggregated, and degraded protein. This indicates that in vivo glycosylation (N-linked) may also prevent the aggregation of folding intermediates (reviewed by Helenius, 1994). Detailed NMR studies on glycoproteins have clearly shown that carbohydrates stabilize folded proteins and even prevent marginally stable proteins from unfolding (for a review, see Wyss and Wagner, 1996). Despite potential pitfalls, many nonglycosylated protein variants have been successfully folded from E. coli inclusion bodies. Examples include cytokines of biomedical importance such as granulocyte/macrophage colony-stimulating factor (GM-CSF; Diederichs et al., 1991) and interleukin 5 (IL-5; Milburn et al., 1993). Inclusion body formation was avoided in some studies by using secretion vectors; examples include GM-CSF (Walter et al., 1992) and the extracellular domain of the human growth hormone receptor (deVos et al., 1992). The aforementioned proteins have been crystallized and their structures determined by X-ray crystallography, supporting the view that the structural integrity and conformation of the proteins were not affected by the lack of glycosylation and their respective preparative histories. If a glycoprotein of interest is available from a eukaryotic recombinant expression system or if the natural protein is available, then before investing time with E. coli expression, it may be worthwhile to determine whether the protein can be denatured and refolded in vitro. Pilot experiments can be carried out on intact protein and on protein enzymatically deglycosylated with glycosidases and, if disulfides are present, with and without reduction. Of course, if the protein can be secreted to the periplasm, aggregation and the necessity for in vitro folding may be avoided. The production of deglycosylated proteins in E. coli expression systems for in vitro biochemical and structural studies is obviously of great value; however, the proteins may not always be suitable for in vivo studies due to low biological activity. Compared to authentic proteins, nonglycosylated variants can have a reduced circulatory lifetime and can exhibit increased immunogenicity and protease sensitivity (Rasmussen, 1992). SOME EXAMPLES OF PROTEIN EXPRESSION AND PURIFICATION Examples of protein expression and purification can be found in most biochemical journals, two which may be especially useful: Protein Expression and Purification (http://www.academicpress.com/pep), which covers advances in the expression and purification of recombinant proteins mainly from E. coli although other expression systems are often included; and Current Opinion in Biotechnology, which regularly provides updates on various aspects of recombinant protein production as well as useful reference lists. Detailed protocols are also given in the units of this Chapter and a few recent examples of protein expression and purification are discussed below to illustrate some of the general approaches used to deal with soluble and insoluble E. coli protein expression. Soluble Proteins HIV Nef Nef is a 205-residue myristolylated protein expressed at high levels in the early stages of HIV infection. The protein is important for the induction of AIDS and is being actively researched as a potential drug target. Unlike most HIV-1 and related proteins expressed in bacteria, Nef is recovered from the soluble fraction of E. coli extracts. The purification



Supplement 30

protocol adopted following cell breakage and low-speed centrifugation is fairly straightforward comprising two stages of ion-exchange chromatography using DEAE-Sepharose (weak exchanger) followed by Q Sepharose (strong exchanger) and finally gel filtration using Superdex 75. Characterization of the purified protein yielded the following information. 1. Nef has a maximum solubility of ∼0.5 to 0.6 mM (∼10 mg/ml) in low-ionic strength buffers at pH 7.5 to 8.0, (e.g., 5 mM Tris⋅Cl). The solubility can be increased by the inclusion of nondenaturing concentrations (2 M) of urea, as established by titration studies monitored by far-UV circular dichroism. Acetonitrile (5% to 10%) also increases the solubility of protein. 2. The protein contains three cysteines (positions 54, 141, and 205), none of which are involved in native disulfide bond formation. The cysteines at positions 54 and 205 are solvent-exposed. 3. Digestion of the purified protein with proteases indicated rapid digestion of the N-terminal region (residues 1-38). For example, digestion was complete with a few minutes using relatively low concentration of trypsin (1% w/w). The above information was exploited to increase the robustness of the purification protocol. Low solubility was a major issue during purification and this was improved by including 4 M urea in the extraction buffer and 2 M urea in the two anion exchange column buffers. For the final gel filtration step, 10% acetonitrile was included to help maintain both the solubility of Nef and fortuitously cause aggregation of some E. coli contaminants that eluted in the void volume. Neither the urea nor the acetonitrile at the concentrations used resulted in Nef denaturation. The problem of cysteine oxidations was circumvented by mutating cysteines 54 and 205 to alanines. Mutation of cysteine 205 alone and including 5 mM DTT in all the column buffers was also a satisfactory solution. The high susceptibly of the N-terminal region to proteolytic processing indicates that it is solventaccessible and likely to be unstructured. In the case of Nef, this region can be deleted without affecting the folding of the protein and removes the potential for heterogeneity due to partial processing by E. coli proteases. The NMR structure of HIV Nef was determined with protein prepared as described above (Grzesiek et al., 1997). MAP30 MAP30 is a plant protein obtained from bitter melon that has anti-HIV and anti-tumor activities. The 30-kDa protein is well expressed in E. coli as a soluble protein and is purified by two stages of exchange chromatography followed by gel filtration. The clarified extract is first applied to a DEAE-Sepharose column at pH 8.0; the majority of MAP30 does not bind or weakly binds the exchange resin. The column flow-through and early eluting fractions are dialyzed against pH 6.5 buffer then fractionated using SPSepharose (strong cation exchanger). The final step is gel filtration using a Superdex 200 column at pH 8.0.


There are clear similarities between the MAP30 purification scheme and the one developed for the Nef protein; both utilize an initial clean-up step using DEAE-Sepharose followed by a second more discriminating ion-exchange step and finally a “polishing” step using gel filtration. For Nef, the second ion-exchange step employs an anion-exchange resin while the MAP30 method uses a cation-exchange resin. The choice of resin for the second step reflects the difference in the isoelectric points of these proteins. Nef has a calculated pI of ∼5.95 and is positively charged at pH values greater than this. MAP30 has a slightly basic pI of 9.00 and is negatively charged at pH values below this. Thus, Nef binds to DEAE-Sepharose and Q-Sepharose at pH 7.4 and 8.0, respectively.



On the other hand, MAP30 does not bind to DEAE-Sepharose at pH 7.4 but binds strongly to a cation exchanger at pH 6.5. Apart from purification, there is also another similarity between Nef and MAP30, namely susceptibility to proteolytic processing during purification. As previously mentioned, the N-terminal region (residues 1-38) of Nef is at risk for proteolysis, and to maintain the structural integrity, especially during cell breakage and the initial processing, protease inhibitor cocktails must be included in the buffers. MAP30 also has a region susceptible to processing, namely, the ∼20 residues at the C-terminal end of the protein. Again, this is due to the fact that this region is largely unstructured in an otherwise folded and stable molecule (Wang et al., 1999). When purifying MAP30, standard protease inhibitors are included during the early stages of purification and, in addition, α-macroglobulin (15 to 2.0 µg/mg protein) is added to the protein prior to the gel-filtration step. The macroglobulin inhibits a wide range of proteases by a trapping mechanism (Sottrup-Jensen, 1989). If proteins are to be used for structural studies, deletion mutants can eliminate unstructured regions at the N- and C- terminal regions. Deletions of such regions from either Nef or MAP30 do not significantly change the pI of either protein, so the same purification procedures can be applied to the deletion mutants. Although incremental structural determination is an important strategy in structural biology, one should always be aware that regions deleted, even those that appear unstructured, may have important functional roles. There are many examples of disordered proteins and protein domains that adopt folded structures upon binding to their biological targets (for a review, see Dyson and Wright, 2001), and in the case of Nef, it appears that the apparently unstructured N-terminal region (residues 1-57) mediates binding to the tumor suppressor protein p53, possibly enhancing HIV-1 replication (Greenway et al., 2002). A dual vector co-expression system for producing heteromeric complexes in E. coli (Johnson et al., 2000) may be particularly useful for producing proteins requiring binding partners for folding and stability. Insoluble proteins HIV-1 gp41 ectodomain The membrane-associated glycoproteins of HIV-1 include gp120 and gp41, the latter mediating membrane fusion with the host cell. These viral envelope proteins have been the subject of intense structural analysis over the last several years as inhibition of membrane fusion, hence viral entry, is a potential drug target in the development of therapeutics for AIDS. A basic strategy in tackling membrane-associated proteins is to remove the membrane-spanning region by expressing the non-membrane-associated region or ectodomain. The gp41 ectodomain is a 150-residue protein that is recombinantly expressed in E. coli as an insoluble protein. The protein can be extracted from inclusion bodies with 8 M guanidine⋅HCl and purified by one step of gel filtration in the presence of 4 M guanidine⋅HCl. The guanidine is removed by preparative reversed-phase HPLC and the protein folded upon dialysis against 50 mM sodium formate at pH 3.0. The yield of folded protein is >90%. Characterization of the protein indicates that its solubility decreases dramatically below pH 4.0. Between pH 3.0 and 4.0, the protein has an all α-helical secondary structure with a trimeric subunit structure. The protein was demonstrated to have folded by determining its full structure at pH 3.5 using multidimensional NMR (Caffrey et al., 1998). The protein was also crystallized from a buffer at pH 3.5 and its structure determined by X-ray crystallography (Yang et al, 1999). Purification of Recombinant Proteins


Supplement 30

Other insoluble proteins expressed in E. coli that exhibit acid stability similar to the gp41 ectodomain can be processed and folded using a similar scheme as described above. For example, the HIV protease can be purified and folded with this method. The HIV protease, after folding at pH 3.5, exhibits fair solubility up to pH 5.0, with solubility decreasing at higher pH values. Other proteins may only be partially folded or unfolded at acidic pH values; in these cases, the reversed-phase HPLC step could be used to simply remove the denaturant, then the protein can be freeze dried from TFA-acetonitrile solvent and used for folding trials. The gp41 ectodomain contains two cysteine residues in a loop region connecting N- and C-terminal helical domains. These cysteines do not form intramolecular disulfides and can be substituted by alanine residues. This is a common theme. If a protein contains free solvent-accessible cysteines that play no structural or functional role, it is often a good idea to substitute them (usually with Ala), especially if structural studies are planned. Human Tissue Inhibitor of Metalloprotease-2 (TIMP-2) and hepatocyte growth factor isoforms (NK1 and NK2) The TIMP families of proteins are inhibitors of the matrix metalloproteases and are critical effectors of extracellular matrix turnover. The hepatocyte growth factor (HGF) is a multifunctional protein stimulating a wide range of cellular targets. The HGF gene codes for three distinct proteins: the full-length form and two truncated isoforms that include an N-terminal domain (N) and one-kringle (NK1) or two-kringle domains (NK2). TIMP-2 (21 kDa), NK1 (21 kDa), and NK2 (30 kDa) contain multiple disulfides that stabilize the folded conformations. For example, TIMP-2, apart from having 12 cysteines that form 6 disulfides, contains a cysteine as the N-terminal residue. All three proteins were expressed in E. coli as insoluble proteins, extracted with guanidine⋅HCl and reductant, and the unfolded protein separated by gel filtration in a similar manner to that previously discussed. The partially purified proteins can be conveniently stored frozen in guanidine⋅HCl at −80°C for several years without deleterious effects on folding or recovery of active protein. The folding and oxidation of the proteins are detailed in the respective publications, Stahl et al. (1997) and Wingfield et al. (1999), but briefly, the protocols involve equilibrium dialysis incorporating urea as a co-solvent to maintain solubility during folding, and a glutathione-based oxido-shuffling system (redox buffer) to promote formation of disulfide bonds (this approach is also detailed in Basic Protocol 1 in UNIT 6.5). The final stage of the purification process is gel filtration of the folded proteins, which, apart from removing host contaminates, separates folded monomers from any misfolded and aggregated protein.


When recombinant expressed proteins are insoluble in E. coli, the purification scheme can be very simple as illustrated above where one or two steps of gel filtration may be all that is required; the challenge is determining a method to fold and oxidize the protein. In all three examples discussed above, the key to efficient folding is maintaining solubility, whether it be by taking advantage of the acid stability of the protein and working at pH 3.5, or by including the co-solvent urea. As mentioned above, TIMP-2 has an N-terminal cysteine residue. When this protein was originally expressed, an alanine was appended to the N-terminus since it had been observed that partial N-terminal processing occurred when cysteine was the terminal residue. The alanine residue was added in an effort to produce homogeneous protein for structural studies. The purified Ala+ TIMP-2 appeared monomeric and folded, yet was devoid of its normal inhibitory activity (Wingfield et al., 1999). It was determined that the coordination of a zinc atom by the N-terminal cysteine stabilized substrate binding and required a free amino terminal group. This was demonstrated by exopeptidase digestion (using aminopeptidase 1) of Ala+ TIMP-2, which



removed the N-terminal alanine making cysteine the N-terminal residue and, thus, restoring biological activity. A GST fusion protein The protein VP26 is a 12-kDa capsid protein of the herpes simplex virus and initial attempts to directly express this protein in E. coli failed. It was possible, however, to produce this protein at fairly high levels in E. coli as a GST fusion (Wingfield et al., 1997a). The insoluble protein was treated in the usual manner: solubilized with guanidine⋅HCl and partially purified by gel filtration also in guanidine⋅HCl. The usual purification for GST fusion proteins is affinity chromatography using immobilized glutathione, which requires that the GST moiety be folded (UNIT 6.6). Due to the low solubility of VP26 and its high propensity for aggregation, the following approach was used. First, the VP26-GST fusion was folded from the guanidine⋅HCl solution by equilibrium dialysis against buffer containing 2.5 M urea, 10 mM CHAPS, and 0.25 M NaCl, and then against the same buffer lacking the urea. The buffer additives were included to maintain protein solubility (solubility is improved with >0.25 M NaCl, but the cleavage of the GST moiety by thrombin is inhibited by high salt concentrations). Following cleavage of GST and VP26, the proteins were denatured again with guanidine⋅HCl, separated by gel filtration and the purified VP26 refolded from urea and CHAPS as described above. As an aside, the GST moiety is readily refolded from guanidine⋅HCl and does not require high salt or CHAPS to maintain solubility during the dialysis steps. The purification approach used here may appear inelegant, but the fusion system was used not to facilitate purification, but to facilitate expression of the protein. PROTEIN HANDLING Storing Purified Proteins Purified protein should be filter-sterilized prior to storage. Millex-GV 0.22-µm filters (Millipore) employ hydrophilic membranes with low binding capacities and are recommended for most proteins. Proteins are best stored at −80°C or may be stored on ice; freezing at −20°C is not recommended. Rapid freezing in small aliquots using dry ice/ethanol mixtures is preferred to slow freezing at −20°C . The addition of sucrose or glycerol often increases protein stability during storage and during freezing and thawing cycles (Arakawa and Timasheff, 1985; Timasheff and Arakawa, 1997). Lyophilization is best for long-term storage; however, care should be taken in choosing the protein solvent (Franks, 1993). Promoting Protein Solubility and Stability If the recombinant protein contains reactive unpaired sulfhydryl groups in the native conformation, 1 to 5 mM DTT should be included in the column buffers during purification. However, reductant should not be used gratuitously, as the native protein may contain intra- or intermolecular disulfide bonds, disruption of which can reduce the stability and solubility of the protein. Reductants should be included, for example, during gel filtration if dimers or higher aggregates need to be converted to active monomeric protein. The presence of intermolecular (and occasionally intramolecular) disulfide bonds can be determined analytically by SDS-PAGE under nonreducing conditions (UNIT 6.5) by pretreating proteins sequentially with iodoacetamide (to prevent artificial disulfide exchange) and then with SDS in the absence of reductant. The use of reductants can best be rationalized once the native protein has been characterized. Purification of Recombinant Proteins


Supplement 30

EDTA (1 to 5 mM) is often included in buffers to remove heavy metals that can catalyze oxidative processes and inhibit certain proteases. It should be noted that EDTA will bind to anion exchange resins (Scopes, 1994). Other components often added to buffers to promote protein solubility during purification include nonionic or zwitterionic detergents, low concentrations of urea (1 to 2 M), and salt (0.5 to 1 M NaCl). These additives are compatible with ion-exchange chromatography, except for high-salt concentrations, which are compatible with hydrophobic-interaction chromatography (UNIT 8.4), affinity chromatography (Chapter 9), and gel-filtration chromatography (UNIT 8.3). Solvent pH is one of the most important variables for maintaining protein solubility; in general, proteins are least soluble at or near their isoelectric points. Preventing Contamination Precautions to prevent contamination of the protein of interest are as follows: 1. To avoid cross-contamination, especially from other recombinant proteins, dedicate one set of chromatography resins for the purification of each protein. If this is not possible, or if expensive prepackaged matrices are used, be sure to clean resins thoroughly after each use. Check the manufacturer’s recommendations and be aware of the chemical stability of the resin, especially for extremes of pH. 2. Store resins with preservatives (e.g., 1 mM sodium azide) and avoid storage in phosphate buffers, which provide a good medium for bacterial growth. 3. To generate reproducible protocols using ion-exchange methods, monitor the pH and conductivity of all buffers and column effluents (the latter ideally in-line). 4. Avoid protein cross-contamination in concentration equipment such as stirred ultrafiltration cells with ultrafiltration membranes. 5. Keep pH and conductivity probes scrupulously clean, especially when used with solutions containing proteases. Likewise, use care when using cuvettes for UV measurements. 6. Avoid vigorous stirring of protein solutions to prevent shear denaturation, and handle soft agarose-based column matrices carefully to prevent bead fragmentation. Removing Pyrogens Recombinant proteins used for in vivo studies should be free of endotoxins (pyrogenic lipopolysaccharide derived from the bacterial outer membrane of Gram-negative bacteria). Yeast and mammalian cell hosts do not contain endotoxins; however, exogenous contamination from water and others must be avoided. Pyrogens can be detected using the sensitive Limulus amoebocyte lysate (LAL) assay kits available from Sigma and other suppliers. As endotoxins are negatively charged, they will be removed by anion-exchange chromatography. Other methods are reviewed in detail by Petch and Anspach (2000). SCALE OF OPERATIONS AND AIMS OF PURIFICATION Determining Scale Purification of Recombinant E. coli Proteins

The amount of protein required and the level of purity will vary dramatically from laboratory to laboratory and study to study. The following guidelines will help in planning a strategy.



If a Coomassie blue–stained band corresponding to the expressed protein is observed on one-dimensional SDS-PAGE analysis of a whole-cell extract, then the protein constitutes at least 0.5% to 1% of the total protein. Wet E. coli cell paste contains ∼10% to 15% protein by weight (reviewed by Neidhardt, 1987). If the level of expression is low to average (e.g., 5%), then 1 g wet weight of cells will contain ∼5 mg recombinant protein. Hence, a cell paste of 20 to 50 g (a typical yield from a 1- to 2-liter benchtop fermentation) will contain 100 to 250 mg recombinant protein, and often two- to five-fold more. Shaker-flask fermentations of equivalent volumes might yield 5% or 10% of these amounts. Thus, for soluble proteins, or insoluble ones that can be refolded (with ≥5% yield), significant amounts of protein can be obtained from relatively small fermentations. For proteins secreted into the periplasm or medium, fermentations on larger scales may be required, as expression levels are usually considerably lower than that for direct expression. Deciding the Aims of the Purification There are many reasons, both scientific and commercial, for producing purified recombinant proteins. The development of laboratory-scale purification schemes that produce pure protein (a single band on SDS-PAGE) should be relatively straightforward given the relatively high abundance of recombinant proteins in cell extracts. Protein present at 1% of the total cell extract requires only a 100-fold purification compared to the several thousand-fold sometimes required for the purification of nonrecombinant proteins (reviewed by Stein, 1991). The widely used method of affinity tagging proteins allows the nonspecialist to rapidly purify protein for biochemical and activity studies without investment in some of the specialized equipment mentioned below. However, far more time and expertise is required to develop protocols that produce purified recombinant proteins having the physical and chemical homogeneity required for clinical use and for structural determinations. Furthermore, only after detailed characterization of the isolated protein will chemical and physical heterogeneities be revealed in enough detail for steps to be taken to either prevent their occurrence or rationalize fractionation of modified species. Therapeutic proteins Mammalian cells are the production host for many current protein therapeutics, however, E. coli, is also used to produce major biotechnological products including insulin and bovine growth hormone. Some advances in E. coli production of therapeutic proteins and methods used to fold solubilized protein for industrial processes have been recently reviewed (De Bernardez Clark, 2001; Swartz, 2001). Proteins used for clinical studies must be manufactured according to applicable FDA guidelines that include Good Manufacturing Procedures (GMP). Sofer and Hagel (1997) provide practical coverage of modern process development, including process chromatography and its scale-up. The physiochemical characterization of protein pharmaceuticals can be especially challenging and many of the methods and approaches used rely on mass spectrometry (see Chapter 16). Structure Determination For many investigators, a primary goal is to correlate the structure of a protein with its function (and vice versa). Many proteins produced by recombinant DNA technology are present only in trace amounts in nature (e.g., interferons and other cytokines; Ealick et al., 1991), and authentic material is not available for detailed molecular characterization. Knowledge of the 3-D structure allows a rational approach to protein engineering and the design of drugs that modulate the biological activity of the protein. The substitution,



Supplement 30

deletion, and insertion of residues allow a structure-function hypothesis to be tested and new, sometimes improved protein variants (or mutants) to be produced. NMR Spectroscopy It is a major challenge to produce proteins suitable for structural determination, not only in terms of quality, but in terms of the quantity which may be required, especially for NMR (UNIT 17.5). Many proteins, although they have native-like structure and biological activity, are not suitable for structural determination due to, e.g., limited solubility, conformational flexibility (floppy regions/domains), and heterogeneity of posttranslational modifications (especially carbohydrate). Often, these problems can be resolved by a combination of protein biochemistry and protein engineering approaches and requires a close collaboration between the structural biologist and the molecular/protein chemist. The NMR determination of the HIV-1 Nef structure is an example of this integrated approach (see above). Structural determination in solution by multidimensional NMR is presently limited to proteins 30 to 40 kDa (reviewed by Clore and Gronenborn, 1994). One of the largest proteins solved to date is the 44-kDa trimeric SIV gp41 ectodomain (Caffrey et al., 1998). Larger proteins can be studied incrementally (using the dissection approach) if information on the domain boundaries is known (Campbell and Downing, 1998). The sample demands can be as high as several hundred milligrams, and larger proteins (>10 kDa) must be uniformly labeled with various combinations of 2H, 13C, and 15N. Some of the labeling scenarios required to solve the HIV Nef structure are presented in Table 1 of Grzesiek et al. (1997). New developments in isotope labeling strategies are reviewed by Goto and Kay (2000). Labeling in E. coli is achieved by growing the bacteria in minimal medium containing one or more of the following stable (nonradioactive) isotopes: 15 NH4Cl (sole nitrogen source), [13C] glucose (sole carbon source), and 2H2O (UNIT 5.3). The 15N and 13C labeling of the HIV protease using a 2-liter fermentor is detailed by Yamazaki et al. (1996). Label incorporation is conveniently monitored by mass spectrometry of the purified protein. Over the lifetime of the structural study (4 to 12 months), because of the multisample requirements, a reliable and robust purification method is essential. It should also be noted that the labeling requirement usually dictates that the protein be produced in bacteria, although labeled proteins have been produced in yeast and insect cells (Goto and Kay, 2000). For NMR purposes, the recombinant protein must be homogeneous and soluble at 1 to 3 mM concentrations, preferably with solvents below pH 7.0 and at temperatures >30°C. As measurements take many hours to complete, the presence of trace amounts of proteases can ruin the experiment. In addition, particular attention must be paid to maintaining solvent-accessible and reactive cysteines (unpaired) in the reduced state (usually by including DTT or TCEP) and often cysteines are mutated to alanine residues (Wingfield et al., 1997). Protein crystallization and X-ray crystallography


The rate-limiting step in structure determination using X-ray crystallography is production of crystals that diffract to high resolution (UNIT 17.4). The scientists involved in the production and characterization of the protein are often best situated to crystallize the protein. Furthermore, once crystallization conditions have been optimized, it can be quite easy to interest structural groups in collaboration.



Determining optimal crystallization conditions may require as little as a few milligrams or as much as 100 mg of pure protein. The protein itself must usually be physically and chemically homogeneous; small amounts of protein impurities may significantly interfere with crystallization. In general, physical homogeneity is more critical than chemical homogeneity. Some of the methods used to establish physical and chemical homogeneity are discussed elsewhere (Chapter 7; Jones et al., 1994). Many investigators use the sparse matrix sampling technology to screen for initial crystallization conditions and commercial kits are available for this purpose (e.g., Hampton Research: http://www.hamptonresearch.com/index.html. The company site also has useful tips and protocols). The phase problem in crystallographic analysis has traditionally been solved by isomorphous replacement with heavy atoms, but over the last 10 years, multiwavelengh anomalous diffraction (MAD) has gained popularity (Hendrickson et al., 1990). For the latter approach, selenium is incorporated into recombinant proteins via selenomethionine (seleno-L-methionine; available from Sigma and others) using a methionine-requiring auxotroph. In studies of the gp41 protein (mentioned above), the T7 expression system (Novagen, http://www.novagen.com) and the host strain B834/DE3 (Novagen) were used. Briefly, transformed cells were grown overnight in a 0.5-liter shaker flask containing minimal media plus 1 mM methionine. Cells were collected and resuspended in 0.5 liters of media minus methionine. The cells were grown at 37°C in a small fermentor and fed 5 ml of 10 mg/ml selenomethionine. The cells were induced for 3 hr with IPTG and fed an additional 50 mg of selenomethionine (total feed: 100 mg). Cells were collected (∼7.0 g wet weight) and 130 mg of pure gp41 ectodomain was isolated as described above. Mass spectrometry indicated that the single methionine was >98% labeled. The protein was then crystallized as previously described (Wingfield et al., 1997). For more details on labeling using E. coli, see Chapter 5 (UNIT 5.3). Selenomethionine incorporation into eukaryotic systems is not as successful as in E. coli; incorporation can be as high as 90% in baculovirus, but only ∼60% in yeast. The production of well-defined protein complexes for structural studies can be straightforward. For example, monomeric proteins can be expressed in bacteria which self-associate into stable complexes ranging from simple dimers (e.g., γ-IFN) and trimers (e.g., α-TNF) to complex structures such as viral nucleocapsids (e.g., Hepatitis B Virus core antigen, 180-mer). These stable (tightly associated) homopolymers are well suited for structural studies. Heteroprotein complexes can be made by either co-expression of protein subunits (Johnson et al., 2000; Kholod and Mustelin, 2001) or by in vitro assembly of individual components. The former approach may be required in the case where individual subunits are unstable (Nash et al., 1987). In contrast to stable complexes, there are many biologically significant complexes characterized by weak association. Many protein-protein interactions of interest, e.g., signal transduction pathways, may be somewhat transitory and involve weak interactions. In these complexes, the dissociation constants (Kd) between proteins are 500 ml), the Manton-Gaulin-APV homogenizer (APV Gaulin) is recommended. For further processing of cells and cell lysates (e.g., UNITS 6.2 & 6.3), an ultrasonic homogenizer is required. An instrument with a 400-W (or higher) capacity is recommended (Branson, http://www.bransonultrasonics.com). After low-speed centrifugation using standard preparative centrifuges (Beckman Coulter Preparative Centrifuge, Avanti Series can be found at http://www.beckman.com), highspeed centrifugation is a convenient and rapid cleanup step before column chromatography (Fig. 6.1.3). With Beckman ultracentrifuges, the 45 Ti rotor is recommended. This six-place rotor has a maximum speed of 235,000 × g; with thick-walled polycarbonate tubes, its capacity is ∼400 ml. Chromatographing Proteins


Most chromatography is carried out at 4°C either in a cold room or, more conveniently, in a cold cabinet in the laboratory. The basic components of a chromatography system are as follows: column, column matrix, pumps, a gradient-making device, UV/visible or other detection system, and a fraction collector. These components can be bought as units such as the AKTA Explorer or FPLC chromatograph systems (Amersham Bioscience, http://www.apbiotech.com), which can be used for laboratory-scale to large-scale work. Systems can also be custom assembled from individual components from Amersham and other vendors. Column matrices can be purchased prepacked or as bulk media that are packed in columns by the user. Ion-exchange separations, using standard low- to mediumpressure resins (agarose/dextran/cellulose-based), require at least one narrow (2.5-cm) and one wide (5.0-cm) column with adjustable flow adapters so that the resin height can be varied between 5 and 30 cm. Gel filtration requires columns with diameters of 1.25



and 2.5 cm (5 cm for larger-scale work) and lengths of 60 to 100 cm. Simple gradient makers with capacities of 150 ml to 2 liters are generally available. Concentrating Proteins Stirred ultrafiltration cells are recommended for laboratory-scale work. The cells range in size from 3 ml to 2 liters and are used in conjunction with variable molecular weight cutoff membranes (Millipore, http://www.millipore.com). For larger volumes, Millipore also sells various systems. For smaller volumes (0.5 to 15 ml), centrifugational concentrators are available (Millipore and others). For a review of the equipment used for protein concentration, see Harris (1989). Making Analytical Measurements A protein purification laboratory should have a dependable scanning UV/visible spectrophotometer, ideally an instrument with computerized data collection and analysis. Hewlett Packard (Agilent) instruments with diode array detectors are recommended for most routine work (http://www.chem.agilent.com). For laboratories specializing in purifying recombinant proteins from E. coli, access to a spectropolarimeter (e.g., Jasco J-810, http://www.jascoinc.com) will be helpful for monitoring and developing folding protocols. For rapid chemical characterization and identity check of proteins, access to a mass spectrometer is also desired (Chapter 16). Most of the companies mentioned above have excellent Web sites where technical information is posted. The series of handbooks on chromatographic separations published by Amersham Biosciences can be conveniently downloaded as pdf files. Literature Cited Allet, B., Payton, M., Mattaliano, R.J., Gronenborn, A.M., Clore, G.M., and Wingfield, P.T. 1988. Purification and characterization of the DNA-binding protein Ner of bacteriophage Mu. Gene 65:259-268. Arakawa, T. and Timasheff, S.N. 1985. Theory of protein solubility. Methods Enzymol. 114:49-77. Armstrong, N., De Lencastre, A., and Gouaux, E. 1999. A new protein folding screen: Application to the ligand binding domain of a glutamate and kainite receptor and to a lysozyme and carbonic anhydrase. Protein Sci. 8:1475-1483. Asenjo, J.A. and Patrick, I. 1990. Large-scale protein purification. In Protein Purification Applications: A Practical Approach (E.L.V. Harris and S. Angal, eds.) pp. 1-27. IRL Press, Oxford. Baneyx, F. 1999. Recombinant protein expression in Escherichia coli. Curr. Opin. Biotechnol. 10:411-421. Beacham, I.R. 1979. Periplasmic enzymes in Gram-negative bacteria. Int. Biochem. 10:877-883. Ben-Bassat, A., Bauer, K., Chang, S.-Y., Myambo, K., Boosman, A., and Chang, S. 1987. Processing of the initiation methionine from proteins: Properties of the E. coli methionine aminopeptidase and its gene structure. J. Bacteriol. 169:751-757. Bessette, P.H., Aslund, F., Beckwick, J., and Georgiou, G. 1999. Efficient folding of proteins with multiple disulfide bonds in the Escherichia coli cytoplasm. Proc. Natl. Acad. Sci. U.S.A. 96:13703-13708. Bibi, E. and Beja, O. 1994. Membrane topology of multidrug resistant protein expressed in E. coli. J. Biol. Chem. 31:19910-19915. Bowden, G.A., Paredes, A.M., and Georgiou, G. 1991. Structure and morphology of protein inclusion bodies in E. coli. Biotechnology 9:725-730. Braun, P., Gerritse, G., van Dijl, J.-M., and Quax, W.J. 1999. Improving protein secretion by engineering components of the bacterial translocation machinery. Curr. Opin. Biotechnol. 10:376-381. Buchner, J. and Rudolph, R. 1992. Renaturation and characterization of recombinant Fab fragments produced in Escherichia coli. Biotechnology 9:157-162. Burgess, R.R. and Jendrisak, J.J. 1975. A procedure for the rapid, large-scale purification of E. coli DNA-dependent RNA polymerase involving polymin P precipitation and DNA-cellulose chromatography. J. Biol. Chem. 14:4634-4638. Purification of Recombinant Proteins


Supplement 30

Burgess, R.R. and Knuth, M.W. 1996. Purification of a recombinant protein overproduced in Escherichia coli. In Strategies for Protein Purification and Characterization: A Laboratory Course Manual (D.R. Marshak, J.T. Kadonaga, R.R. Burgess, M.W. Knuth, W.A. Brennan, and S.-H. Lin, eds.) pp. 205-217 and 245-262. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Caffrey, M., Cai, M., Kaufman, J., Stahl, S., Wingfield, P.T., Covell, D.G., Gronenborn, A.M., and Clore, G.M. 1998. Three dimensional solution structure of the 44 kDa ectodomain of SIV gp41. EMBO J. 17:4572-4584. Campbell, I. and Downing, A.K. 1998. NMR of modular proteins. Nat. Struct. Biol. 5:496-499. Cereghino, G.P.L. and Clegg, J.M. 1999. Applications of yeast in biotechnology: Protein production and genetic analysis. Curr. Opin. Biotechnol. 10:422-427. Cheng, Y.-S.E., McGowan, M.H., Kettner, C.A., Schloss, J.V., Erickson-Viitanen, S., and Yin, F.H. 1990. High synthesis of recombinant HIV-1 protease and the recovery of active enzyme from inclusion bodies. Gene 87:243-248. Cleary, S., Mulkerrin, M.G., and Kelley, R.F. 1989. Purification and characterization of tissue plasminogen activator kringle-2 domain expressed in E. coli. J. Biol. Chem. 28:1884-1891. Cleland, J.L., Builder, S.E., Swartz, J.R., Winkler, M., Chang, J.Y., and Wang, D.I.C. 1992. Polyethylene glycol enhanced protein refolding. Biotechnology 10:1013-1019. Clore, G.M. and Gronenborn, A.M. 1994. Multidimensional heteronuclear nuclear magnetic resonance of proteins. Methods Enzymol. 239:249-363. Cole, P.A. 1996. Chaperone-assisted protein expression. Structure 4:239-242. Colon, W. 1999. Analysis of protein structure by solution optical spectroscopy. Methods Enzymol. 309:605632. Cornelis, P. 2000. Expressing genes in different Escherichia coli compartments. Curr. Opin. Biotechnol. 11:450-454. Creighton, T.E. 1984. Disulfide bond formation in proteins. Methods Enzymol. 107:305-329. Creighton, T.E. 1993. Proteins: Structures and Molecular Properties, 2nd ed. pp. 292-296. Freeman, New York. Dale, G.E., Broger, C., Langen, H., D’Arcy, A., and Struber, D. 1994. Improving protein stability through rationally designed amino acid replacements: Solubilization of the trimethoprim-resistant type S1 dihydrofolate reductase. Protein Eng. 7:933-939. Danley, D.E., Strick, C.A., James, L.C., Lanzetti, A.J., Otterness, I.G., Grenett, H.E., and Fuller, G.M. 1991. Identification and characterization of a C-terminally extended form of recombinant murine IL-6. FEBS Lett. 283:135-139. Darby, N.J. and Creighton, T.E. 1990. Folding proteins. Nature 344:715-716. De Bernardez Clark, E. 2001 Protein folding for industrial processes. Curr. Opin. Biotechnol. 12:202-207. De Bernardez Clark, E., Schwartz, E., and Rudolph, R. 1999. Inhibition of aggregation side reactions during in-vitro protein folding. Methods Enzymol. 309:217-236. deVos, A.M., Ultsch, M., and Kossiakoff, A.A. 1992. Human growth hormone and extracellular domain of its receptor: Crystal structure of the complex. Science 255:306-312. Diederichs, K., Boone, T., and Karplus, A. 1991. Novel fold and putative receptor binding site of granulocyte-macrophage colony stimulating factor. Science 254:1779-1782. DiRienzo, J.M., Nakamura, K., and Inouye, M. 1978. The outer membrane proteins of Gram-negative bacteria: Biosynthesis, assembly and function. Annu. Rev. Biochem. 47:481-532. Dyson, H.J. and Wright, P.E. 2001. Coupling of folding and binding for unstructured proteins. Curr. Opin. Struct. Biol. 12:54-60. Ealick, S.E., Cook, W.J., Vijay-Kumar, S., Carson, M., Nagabhushan, T.L., Trotta, P.P., and Bugg, C.E. 1991. Three-dimensional structure of recombinant human interferon-gamma. Science 252:698-702. Eisenberg, D. 1999. How chaperones protect virgin proteins. Science 285:1021-1022. Ellis, R.J. 1994. Role of chaperones in protein folding. Curr. Opin. Struct. Biol. 4:117-122. Ellis, R.J. 2001. Macromolecular crowding: An important but neglected aspect of the intracellular environment. Curr. Opin. Struct. Biol. 11:114-119. Ellis, R.J. and Hart, F.U. 1999. Principles of protein folding in the cellular environment. Curr. Opin. Struct. Biol. 9:102-110. Purification of Recombinant E. coli Proteins

Feldman, D.E. and Frydman, J. 2000. Protein folding in vivo: The importance of molecular chaperones. Curr. Opin. Struct. Biol. 1026-33.



Fersh, A. 1999. Structure and Mechanism in Protein Science: A Guide to Enzyme Catalysis and Protein Folding. W.H. Freeman and Company. New York. Franks, F. 1993. Storage stabilization of proteins. In Protein Biotechnology (F. Franks, ed.) pp. 486-531. Humana Press, Totowa, N.J. Georgiou, G. and Valax, P. 1999. Isolating inclusion bodies from bacteria. Methods Enzymol. 309:48-58. Gilbert, H.F. 1995. Thiol/disulfide exchange equilibria and disulfide bond stability. Methods Enzymol. 251:8-28. Goenka, S. and Rao, C.M. 2001. Expression of recombinant ζ-crystallin in Escherichia coli with the help of GroEL/ES and its purification. Protein Expr. Purif. 21:260-267. Goff, S.A. and Goldberg, M.E. 1985. Production of abnormal proteins in E. coli stimulates transcription of lon and other heat shock genes. Cell 41:587-595. Goldberg, M.E. 1991. Investigating protein conformation dynamics and folding with monoclonal antibodies. Trends Biochem. Sci. 16:358-362. Goldberg, M.E., Rudolph, R., and Jaenicke, R. 1991. A kinetic study of the competition between renaturation and aggregation during the refolding of denatured-reduced egg white lysozyme. Biochemistry 30:27902797. Goto, N.K. and Kay, L.E. 2000. New developments in isotope labeling strategies for protein solution NMR spectroscopy. Curr. Opin. Struct. Biol. 10:585-592. Greenway, A.L., McPhee, D.A., Allen, K., Johnson, R., Holloway, G., Mills, J., Azad, A., Sankovich, S., and Lambert, P. 2002. Human immunodeficiency virus type 1 Nef binds to tumor suppressor p53 and protects cells against p53-mediated apoptosis. J. Virol. 76:2692-2702. Grunfeld, H., Patel, A., Shatzman, A., and Nishikawa, A.H. 1992. Effector-assisted refolding of recombinant tissue-plasminogen activator produced in Escherichia coli. Appl. Biochem. Biotechnol. 33:117-138. Grzesiek, S., Bax, A., Hu, J.-S., Kaufman, J.D., Palmer, I., Stahl, S.J., Tjandra, N., and Wingfield, P.T. 1997. Refined solution structure and backbone dynamics of HIV-1 Nef. Protein Science 6:1248-1263. Grzesiek, S., Stahl, S.J., Wingfield, P.T., and Bax, A. 1996. The CD4 determinant for downregulation by HIV-1 Nef directly binds to Nef: Mapping of the Nef binding surface by NMR. Biochemistry 35:1025610261. Guisz, Y., Fache, I., Campfield, L.A., Smith, F.J., Farid, A., Plaetinck, G., Van der Heydon, J., Tavernier, J., Fiers, W., Burns, P., and Devos, R. 1998. Efficient secretion of biological active recombinant OB protein (leptin) in Escherichia coli, purification from the periplasm and characterization. Protein Expr. Purif. 12:249-258. Gulnik, S.V., Afonina, E.I., Gustchina, E., Yu, B., Silva, A.M., Kim, Y., and Erickson, J.W. 2001. Utility of (His)6 Tag for purification and refolding of proplasmepsin-2 and mutants with altered activation properties. Protein Expr. Purif. 24:412-419. Hammarstrom, M., Hellgren, N., Van Den Berg, S., Berglund, H., and Hard, T. 2002. Rapid screening for improved solubility of small human proteins produced as fusion proteins in Escherichia coli. Protein Sci. 11:313-321. Harris, E.L. 1989. Concentration of the extract. In Protein Purification Methods: A Practical Approach (E.L.V. Harris and S. Angal, eds.) pp. 125-172. IRL Press, Oxford. Helenius, A. 1994. How N-linked oligosaccharides affect glycoprotein folding in the endoplasmic reticulum. Mol. Biol. Cell. 5:253-265. Hendrickson, W.A., Horton, J.R., and LeMaster, D.M. 1990. Selenenomethionyl proteins for analysis by multiwavelengh anomalous diffraction (MAD): A vehicle for direct determination of three-dimensional structure. EMBO J. 9:1665-1672. Heppel, L.A. 1967. Selective release of enzymes from bacteria. Science 156:1451-1455. Hoffman, A., Tai, M., Wong, W., and Glabe, C.G. 1995. A sparse matrix screen to establish initial conditions for protein renaturation. Anal. Biochem. 230:8-15. Holland, I.B., Kenny, B., Steipe, B., and Pluckthun, A. 1990. Secretion of proteins in E. coli. Methods Enzymol. 182:132-143. Hopkins, T.R. 1991. Physical and chemical cell disruption for the recovery of intracellular proteins. In Purification and Analysis of Recombinant Proteins (R. Seetharam and S.K. Sharma, eds.) pp. 57-83. Marcel Dekker, New York. Hwang, D.D.W., Liu, L.-F., Kuan, I.-C., Lin, L.-Y., Tam, T.-C.S., and Tam, M.F. 1999. Co-expression of glutathione S-transferase with methionine aminopeptidase: A system of producing enriched N-terminal processed proteins in E.coli. Biochem J. 338:335-342. Jaenicke, R. 1993. Role of accessory proteins in protein folding. Curr. Opin. Struct. Biol. 3:104-112.



Supplement 30

Janson, J.-C. and Ryden, L. 1989. Protein Purification: Principles, High Resolution Methods, and Applications. VCH Publishers, New York. Johnson, B.H. and Hecht, M.H. 1994. Recombinant proteins can be isolated from E. coli by repeated cycles of freezing and thawing. Biotechnology 12:1357-1360. Johnson, K., Clements, A., Venkataramani, R.N., Trievel, R.C., and Marmorstein, R. 2000. Coexpression of proteins in bacteria using a T7-based expression plasmid: Expression of heteromeric cell cycle and transcriptional regulatory complexes. Protein Expr. Purif. 20:435-443. Jones, C., Mulloy, B., and Thomas, A.H. 1994. Microscopy, optical spectroscopy, and macroscopic techniques. Methods Mol. Biol. 22:1-245. Jones, D.H., Ball, E.H., Sharpe, S., Barber, K.R., and Grant, C.W.M. 2000. Expression and membrane assembly of a transmembrane region from Neu. Biochemistry 39:1878-1878. Kaback, H.R. 1971. Bacterial membranes. Methods Enzymol. 22:99-120. Kamireddi, M., Eisenstein, E., and Reddy, P. 1997. Stable expression and rapid purification of Escherichia coli GroEL and GroES chaperones. Protein Expr. Purif. 11:47-52. Kelley, W.S. and Stump, K.H. 1979. A rapid procedure for isolation of large quantities of E. coli DNA polymerase 1 utilizing a λ polA transducing phage. J. Biol. Chem. 254:3206-3210. Kern, G., Kern, D., Jaenicke, R., and Seckler, R.L. 1993. Kinetics of folding and association of differentially glycosylated variants of invertase from Saccharomyces cerevisiae. Protein Sci. 2:1862-1868. Kholod, N. and Mustelin, T. 2001. Novel vectors for co-expression of two proteins in E.coli. Biotechniques 31:322-328. Kiefhaber, T., Rudolph, R., Kohler, H.-H., and Buchner, J. 1991. Protein aggregation in vitro and in vivo: A quantitative model of the kinetic competition between folding and aggregation. Biotechnology 9:825-829. Kohno, T., Carmichael, D.F., Sommer, A., and Thompson, R.C. 1990. Refolding of recombinant proteins. Methods Enzymol. 185:187-195. Kost, T.A. and Condreay, J.P. 1999. Recombinant baculovirus as expression vectors for insect and mammalian cells. Curr. Opin. Biotechnol. 10:428-433. Laue, T.M., Senear, D.F., Eaton, S., and Ross, J.B.A. 1993. 5-Hydroxytryptophan as a new intrinsic probe for investigating protein-DNA interactions by analytical ultracentrifugation. Study of the effect of DNA on the self-assembly of the bacteriophage λ cI repressor. J. Biol. Chem. 32:2469-2472. LaVallie, E.R., DiBlasio, E.A., Kovacic, S., Grant, K.L., Schendel, P.F., and McCoy, J.M. 1993. A thioredoxin gene fusion expression system that circumvents inclusion body formation in the E. coli cytoplasm. Biotechnology 11:187-193. Lilie, H., Schwartz, E. and Rudolph, R. 1998 Advances in refolding of proteins produced in E.coli. Curr. Opin. Biotechnol. 9:497-501. Lindwall, G., Chau, M.-F., Gardner, S.R., and Kohlstaedt, L.A. 2000. A sparse matrix approach to the solubilization of overexpression proteins. Protein Eng. 13:67-71. London, J., Skrzynia, C., and Goldberg, M.E. 1974. Renaturation of Escherichia coli tryptophanase after exposure to 8 M urea. Eur. J. Biochem. 47:409-415. Lu, H.S., Fausset, P.R., Sotos, L.S., Clogston, C.L., Rohde, M.F., Stoney, K.S., and Herman, A.C. 1993. Isolation and characterization of three recombinant human granulocyte colony stimulating factor His to Gln isoforms produced in E. coli. Protein Expr. Purif. 4:465-472. Markrides, S. 1996. Strategies for achieving high-level expression of genes in Escherichia coli. Microbiol.l Rev. 60:512-538. Marston, F.A.O. and Hartley, D.L. 1990. Solubilization of protein aggregates. Methods Enzymol. 182:264276. Matthew, J.B., Friend, S.H., Botelho, L.D., Lehman, L.D., Hanania, G.I., and Gurd, F.R.H. 1978. Discrete charge calculations of potentiometric titrations for globular proteins. Biochem. Biophys. Res. Commun. 81:416-421. Maurizi, M.R. 1992. Proteases and protein degradation in Escherichia coli. Experientia 48:178-201. Milburn, M.V., Hassel, A.M., Lambert, M.H., Jordon, S.R., Proudfoot, A.E.I., Graber, P., and Wells, T.N.C. 1993. A novel dimer configuration revealed by the crystal structure at 2.4 angstrom resolution of human interleukin-5. Nature 363:172-176. Mildner, A.M., Rothrock, D.J., Leone, J.W., Bannow, C.A., Lull, J.M., Reardon, I.M., Sarcich, J.L., Howe, W.J., Tomich, C.-S.C., Smith, C.W., Heinrikson, R.L., and Tomasselli, A.G. 1994. The HIV-1 protease as enzyme and substrate: Mutagenesis of autolysis sites and generation of a stable mutant with retained kinetic properties. Biochemistry 33:9405-9413. Purification of Recombinant E. coli Proteins



Miller, C.G., Strauch, K.L., Kurral, A.M., Miller, J.L., Wingfield, P.T., Mazzei, G.J., Werlen, R.C., Graber, P., and Movva, N.R. 1987. N-Terminal methionine-specific peptidase in Salmonella typhimurium. Proc. Natl. Acad. Sci. U.S.A. 84:2718-2772. Miller, H.I., Henzel, W.J., Ridgway, J.B., Kuang, W.-J., Chrisholm, V., and Liu, C.-C. 1989. Cloning and expression of a yeast ubiquitin-protein cleaving activity in E. coli. Biotechnology 7:698-704. Missiakas, D. and Raina, S. 1997. Protein folding in the bacterial periplasm. J. Bacteriol. 179:2465-2471. Murby, M., Uhlen, M., and Stahl, S. 1996. Upstream strategies to minimize proteolytic degradation upon recombinant protein in Escherichia coli. Protein Expr. Purif. 7:129-136. Nagata, K., Kikuchi, N., Ohara, O., Teraoka, H., Yoshida, N., and Kawade, Y. 1986. Purification and characterization of recombinant murine immune interferon. FEBS Lett. 205:200-204. Nash, H.A., Robertson, C.A., Flamm, E., Weisberg, R.A., and Miller, H. 1987. Overproduction of Escherichia coli integration host factor, a protein with noidentical subunits. J. Bacteriol. 169:4124. Neidhardt, F.C. 1987. Chemical composition of Escherichia coli. In Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (F.C. Neidhardt, ed.) pp. 3-6. American Society for Microbiology, Washington, D.C. Nguyen, L.H., Jenson, D.B., and Burgess, R.R. 1993. Overproduction and purification of sigma-32, the Escherichia coli heat shock transcription factor. Protein Expr. Purif. 4:425-433. Nilsson, J., Stahl, S., Lundeberg, J., Uhlen, M., and Nygren, P.-A. 1997. Affinity fusion strategies for detection, purification, and immobilization of recombinant proteins. Protein Expr. Purif. 11:1-16. Orsini, G. and Goldberg, M.E. 1978. The renaturation of reduced chymotrypsin A in guanidine⋅HCl. J. Biol. Chem. 253:3453-3458. Pace, N.C, Vajdos, F., Fee, L., Grimsley, G., and Gray, T. 1995. How to measure and predict the molar absorption coefficient of a protein Protein Sci. 4:2411-2423. Patel, D. 1993. Chromatographic fractionation media. In Biochemistry Labfax (J.A.A. Chambers and D. Rickwood, eds.) pp. 49-68. BIOS Scientific Publishers and Academic Press, Oxford. Petsch, D. and Anspach, F.B. 2000. Endotoxin removal from protein solutions. J. Biotechnol. 76:97-119. Janson, J.-C. and Ryden, L. 1998. Protein purification: Principles, high resolution methods, and applications (2nd ed.). Wiley-LISS, New York. Puri, N.K., Crivelli E., Cardamome, M., Fiddes, R., Bertoloini, J., Ninham, B., and Brandon, M.R. 1992. Solubilization of growth hormone and other recombinant proteins from Escherichia coli by using a cationic surfactant. Biochem. J. 285:871-879. Rasmussen, J.R. 1992. Effect of glycosylation on protein function. Curr. Opin. Struct. Biol. 2:682-686. Ren, Z. and Schaefer, T.S. 2001. Isopropyl-b-D-thiogalactosidase (IPTG)-inducible tyrosine phosphorylation of protein in E.coli. Biotechniques 31:1254-1258. Rudolph, R., Bohm, G., Lilie, H., and Jaenicke, R. 1997. Folding proteins. In Protein Function: A Practical Approach. Second Edition. (T.E. Creighton, ed.) pp. 57-99. IRL Press, Oxford. Schein, C.H. 1989. Production of soluble recombinant proteins in bacteria. Biotechnology 7:1141-1147. Schiene, C. and Fisher, G. 2000. Enzymes that catalyze the restructuring of proteins. Curr. Opin. Struct. Biol. 10:40-45. Schmid, F.X. 1997. Optical spectroscopy to characterize protein conformation and conformational changes. In Protein Structure: A Practical Approach. Second Edition. (T.E. Creighton, ed.) pp. 261-296. IRL Press, Oxford. Schneider, C., Newman, R.A., Sutherland, D.R., Asser, U., and Greaves, M.F. 1982. A one step purification of membrane proteins using a high affinity immunomatrix. J. Biol. Chem. 257:10766-10769. Scopes, R.K. 1994. Protein Purification: Principles and Practice, 3rd ed. Springer-Verlag, New York and Heidelberg. Sherman, F., Stewart, J.W., and Tsunasawa, S. 1985. Methionine or not methionine at the beginning of a protein. Bioessays 3:27-31. Sherman, P.A. and Fyfe, J.A. 1990. Human immunodeficiency virus integration protein expressed in E. coli possesses selective DNA cleaving activity. Proc. Natl. Acad. Sci. U.S.A. 87:5119-5123. Shire, S.J., Bock, L., Ogez, J., Builder, S., Kleid, D., and Moore, D.M. 1984. Purification and immunogenicity of fusion VP1 protein of foot and mouth disease virus. Biochemistry 23:6474-6480. Skerra, A., Pfitzinger, I., and Pluckthun, A. 1991. The functional expression of antibody Fv fragments in Escherichia coli: Improved vectors and a generally applicable purification technique. Biotechnology 9:273-278. Sofer, G. and Hagel, L. 1997. Handbook of process chromatography: A guide to optimization, scale-up, and validation. Academic Press, San Diego, Calif.



Supplement 30

Sottrup-Jensen, L. 1989. Alpha-macroglobulins: Structure, shape, and mechanism of proteinase complex formation. J. Biol. Chem. 264:11539-11542. Stahl, S.J., Wingfield, P.T., Kaufman, J.D., Pannell, L.K., Cioce, V., Sakata, H., Taylor, W.G., Rubin, J.S., and Bottaro, D.P. 1997. Functional and biophysical characterization of recombinant human growth factor isoforms produced in Escherichia coli. Biochem. J. 326:763-772. Stark, G.R., Stein, W.H., and Moore, S. 1960. Reactions of cyanate present in aqueous urea with amino acids and proteins. J. Biol. Chem. 235:3177-3181. Stein, S. 1991. Isolation of natural proteins. In Fundamentals of Protein Biotechnology (S. Stein, ed.) pp. 137-160. Marcel Dekker, New York. Sutcliffe, J.G., Shinnick, T.M., Green, N., and Lerner, R.A. 1983. Antibodies that react with predetermined sites on proteins. Science 219:660-665. Swartz, J.P. 2001. Advances in Escherichia coli production of therapeutic proteins. Curr. Opin. Biotechnol. 12:195-201. Tanford, C. 1968. Protein denaturation. Adv. Protein Chem. 23:122-275. Thatcher, D.R. 1996. Industrial scale purification of proteins. In Proteins LabFax (N.C. Price, ed.) pp. 131-137. BIOS Scientific Publishers, Oxford. Thatcher, D.R. and Panayotatos, N. 1986. Purification of recombinant IFN-2. Methods Enzymol. 119:166177. Thatcher, D.R., Wilks, P., and Chaudhuri, J. 1996. Inclusion bodies and refolding. In Proteins LabFax (N.C. Price, ed.) pp. 119-130. BIOS Scientific Publishers, Oxford. Timasheff, S.N. and Arakawa, T. 1997. Stabilization of protein structure by solvents. In Protein Structure: A Practical Approach. Second Edition (T.E. Creighton, ed.) pp. 349 -363. IRL Press, Oxford. Uhlen, M., Forberg, G., Moks, T., Hartmanis, M., and Nilsson, B. 1992. Fusion proteins in biotechnology. Curr. Opin. Biotechnol. 3:363-369. Ultsch, M., deVos, A.M., and Kossiakoff, A.A. 1991. Crystals of the complex between human growth hormone and the extracellular domain of its receptor. J. Mol. Biol. 222:865-868. Van Reis, R. and Zydney, A. 2001. Membrane separations in biotechnology. Curr. Opin. Biotechnol. 12:208-211. Walter, H. and Johansson, G. 1986. Partitioning in aqueous two-phase systems: An overview. Anal. Biochem. 155:215-242. Walter, M.R., Cook, W.J., Ealick, S.E., Nagabhushan, T.L., Trotta, P.P., and Bugg, C.E. 1992. Three-dimensional structure of human recombinant granulocyte-macrophage colony stimulating factor. J. Mol. Biol. 224:1075-1085. Wang, Y.-X., Neamati N, Jacob, J., Palmer, I., Stahl, S.J., Kaufman, J.D., Huang, P.L., Huang, P.L., Winslow, H.E., Pommier, Y., Wingfield, P.T., Lee-Huang, S., Bax, A., and Torchia, D.A. 1999. Solution structure of anti-HIV-1 and anti-tumor protein MAP30: Structural insights into its multiple functions. Cell 99:433-442. Watanabe, E., Tsoka, S., and Asenjo, J.A. 1994. Selection of chromatographic protein purification operations based on physicochemical properties. Ann. N.Y. Acad. Sci. 721:348-364. Wetlaufer, D.B. 1984. Nonenzymatic formation and isomerization of protein disulfides. Methods Enzymol. 107:301-304. Wetlaufer, D.B., Branca, P.A., and Chen, G.-X. 1987. The oxidative folding of proteins by disulfides plus thiol does not correlate with redox potential. Protein Eng. 1:141-146. Wetzel, R. 1992. Principles of protein stability. Part 2—Enhanced folding and stabilization of proteins by suppression of aggregation in vitro and in vivo. In Protein Engineering: A Practical Approach (A.R. Rees, M.J.E. Sternberg, and R. Wetzel, eds.) pp. 191-216. IRL Press, Oxford. Widmann, M. and Christen P. 2000. Comparison of folding rates of homologous prokaryotic and eukaryotic proteins. J. Biol. Chem. 275:18619-18622. Wingfield, P.T., Mattaliano, R.J., MacDonald, H.R., Craig, S., Clore, G.M., Gronenborn, A.M., and Schmeissner, U. 1987a. Recombinant-derived interleukin 1α stabilized against specific deamidation. Protein Eng. 1:413-417. Wingfield, P.T., Graber, P., Rose, K., Simona, M.G., and Hughes, G.J. 1987b. Chromatofocusing of N-terminally processed forms of proteins: Isolation and characterization of two forms of interleukin 1β and bovine growth hormone. J. Chromatogr. 387:291-300. Wingfield, P.T., Payton, M., Graber, P., Rose, K., Dayer, J.-M., Shaw, A.R., and Schmeissner, U. 1987c. Purification and characterization of human interleukin 1α produced in Escherichia coli. Eur. J. Biochem. 165:537-541. Purification of Recombinant E. coli Proteins



Wingfield, P.T., Stahl, S.J., Payton, M.A., Venkatesan, S., Misra, M., and Steven, A. 1990. HIV-1 Rev expressed in recombinant Escherichia coli: Purification polymerization and conformational properties. Biochemistry 30:7527-7534. Wingfield, P.T., Stahl, S.J., Williams, R.W., and Steven, A.C. 1995. Hepatitis core antigen produced in E. coli: Conformational analysis, and in vitro assembly. Biochemistry 34:4919-4932. Wingfield, P.T., Stahl, S.J., Kaufman, J., Zlotnick, A., Hyde, C.C., Gronenborn, A.M., and Clore G.M. 1997. The extracellular domain of immunodeficiency virus gp41 protein: Expression in Escherichia coli, purification and crystallization. Protein Sci. 6:1653-1660. Wingfield, P.T., Stahl, S.J., Thomsen, D.R., Homa, F.L., Booy, F.P., Trus, B.L., and Steven, A.C. 1997a. Hexon-only binding of VP26 reflects differences between the hexon and penton conformations of the VP5, the major capsid protein of Herpes Simplex Virus. J. Virology 71:8955-8961. Wingfield, P.T., Stahl, S.J., Kaufman, J., Palmer, I., Chung, V., Sax, J.K., Kleiner, D.E., and Stetler-Stevenson, G.W. 1999. Functional and biophysical characterization of full length, recombinant human TIMP-2 produced in Escherichia coli: Comparison of wild type and N-terminal alanine substituted variant. J. Biol. Chem. 274:21362-21368. Wurm, F. and Bernard, A. 1999. Large–scale transient expression in mammalian cells for recombinant protein production. Curr. Opin. Biotechnol. 10:156-159. Wyss, D.F. and Wagner, G. 1996. The structure of sugars in glycoproteins. Curr. Opin. Struct. Biol. 7:409-416. Yamazaki, T., Hinck, A.P., Wang, Y-X., Nicholson, L.K., Torchia, D.A., Wingfield, P.T., Stahl, S.J., Kaufman, J.D., Chang, C.-H., Domaille, P.J., and Lam, P.Y.S. 1996. Three dimensional solution structure of the HIV-1 protease complexed with DMP 323, a novel cyclic urea-type inhibitor, determined by nuclear magnetic resonance spectroscopy. Protein Sci. 5:495-506. Yang, Z.-N., Mueser, T.C., Kaufman, J., Stahl, S.J., Wingfield, P.T., and Hyde, C. 1999. The structure of the SIV gp41 ectodomain at 1.47 A. J. Struct. Biol. 126:133-144. Yarranton, G.T. and Mountain, A. 1992. Expression of proteins in prokaryotic systems—Principles and case studies. In Protein Engineering: A Practical Approach (A.R. Rees, M.J.E. Sternberg, and R. Wetzel, eds.) pp. 303-324. IRL Press, Oxford. Zardeneta, G. and Horowitz, P.M. 1994. Detergent, liposome, and micelle-assisted protein refolding. Anal. Biochem. 223:1-6. Zhang, Y., Olsen, D.R., Nguyen, K.B., Olson, P.S., Rhodes, E.T., and Mascarenhas, D. 1998. Expressoin of eukaryotic proteins in soluble form in Escherichia coli. Protein Expr. Purif. 12:159-165.

Contributed by Paul T. Wingfield National Institutes of Health Bethesda, Maryland



Supplement 30

Preparation of Soluble Proteins from Escherichia coli

UNIT 6.2

Once a suitable protein expression system involving Escherichia coli is developed and optimized (UNITS 5.1 & 5.2), large-scale production of recombinant proteins (UNIT 5.3) generates large quantities of culture material from which the protein of interest must be purified. Harvesting (UNIT 5.3) produces cell concentrate or culture medium, depending on the subcellular localization of the protein. Cell paste is the starting material for purification of proteins expressed in soluble form inside cells, such as interleukin 1β (IL-1β). Human IL-1β is a 153-residue (17.4-kDa) protein cytokine of biomedical importance that plays a central role in immune and inflammatory responses. Purification of human IL-1β is used as an example of the preparation of soluble proteins from E. coli. Bacteria containing IL-1β are lysed, and the resulting supernatant is clarified to remove ribosomes and other particulate matter. The sample is then applied to an anion-exchange column to separate recombinant IL-1β from cellular contaminants, such as E. coli proteins, nucleic acids, and lipopolysaccharides. The sample is further purified through salt precipitation and cation-exchange chromatography, then concentrated. Finally, the IL-1β protein is applied to a gel-filtration column to separate it from remaining higherand lower-molecular-weight contaminants, the purified protein is stored frozen or is lyophilized. The purification protocol described is typical for a protein that is expressed in fairly high abundance (i.e., >5% total protein) and accumulates in a soluble state. With these expression levels, only about a 20-fold overall purification is required to obtain pure protein. Therefore, conventional chromatographic methods can be used, and normally only three or four purification stages are required (see Table 6.2.1 for an outline of the procedure as applied to IL-1β, including time considerations). The process can be shortened somewhat through the use of the Pharmacia Biotech BioPilot or FPLC systems (see Time Considerations). SDS-PAGE (UNIT 10.1) is used to monitor column fractions for the presence of IL-1β, which is detected as a stained band (UNIT 10.5) at the expected 17.4-kDa location. This is common practice in recombinant protein purification, as the specific assays for many proteins, including IL-1β, are relatively complex. The original expression and fermentation experiments involving IL-1β (Wingfield et al., 1986) followed procedures similar to those given in UNIT 5.3. CAUTION: Avoid direct contact with IL-1β-containing solutions. Trace material from aerosols or from hand contact can cause severe inflammation of the eyes. Safety glasses and gloves should be worn. PURIFICATION OF A PROTEIN EXPRESSED IN ESCHERICHIA COLI IN A SOLUBLE STATE: INTERLEUKIN 1β Materials DEAE Sepharose CL-4B resin (Pharmacia Biotech) Anion-exchange buffer (see recipe) 0.26% (w/v) sodium hypochlorite/70% ethanol or 5% (v/v) bleach (e.g., Clorox)/70% ethanol E. coli cells (∼50 g wet weight) from fermentation (UNIT 5.3) containing IL-1β Lysis buffer (see recipe) Contributed by Paul T. Wingfield Current Protocols in Protein Science (1995) 6.2.1-6.2.15 Copyright © 1995 by John Wiley & Sons, Inc.

BASIC PROTOCOL


6.2.1 CPPS

Bovine pancreas DNase I and RNase A (Worthington; optional, for reducing solution viscosity) 2 N sodium hydroxide Ammonium sulfate, ground with mortar and pestle Cation-exchange buffer (see recipe) CM Sepharose CL-4B (Pharmacia Biotech) Cation-exchange buffer/250 mM NaCl (see recipe) Tris base Gel-filtration buffer (see recipe) Ultrogel AcA54 gel-permeation resin (BioSepra) Lyophilization buffer (see recipe; optional) 2- or 3-liter sintered glass funnel with fritted disc (coarse porosity) and 5-liter filter flask Chromatography columns (preferably glass) with adjustable flow adapters: one (or optionally two) 5 × 50 cm and one 2.5 × 100 cm (Pharmacia Biotech, Amicon, or equivalent) RK50 packing reservoir (Pharmacia Biotech) Peristaltic pump, UV monitor, and fraction collector (Pharmacia Biotech or equivalent) 16 × 150–mm culture tubes 40-ml French pressure cell and rapid-fill kit (SLM-AMINCO) Aminco laboratory press (SLM-AMINCO) 1-liter Waring commercial blender

Table 6.2.1

Day

Outline of Interleukin β Purificationa

Steps (low pressure)

1

Preparation of DEAE Sepharose column As for low pressure (steps 1-17) (steps 1-3) Cell breakage (steps 4-8) Clarification of lysate (steps 9 and 10) DEAE Sepharose chromatography (step 11) SDS-PAGE of DEAE Sepharose fractions (step 12) (NH4)2SO4 fractionation (steps 13-15) Dialysis (steps 16 and 17)

2

CM Sepharose chromatography (steps 18-22) SDS-PAGE of CM Sepharose fractions (step 22)

Fast Flow (substitute for CM Sepharose; steps 18-21) SDS-PAGE of Fast Flow fractions (steps 22) Concentration (step 23a) Superdex 75 gel filtration (substitute for Ultrogel; steps 24-26)

3

Concentration (step 23a or 23b)

SDS-PAGE of Superdex 75 fractions (step 26) Concentration (step 27)

Ultrogel gel filtration (steps 24-26) 4 Preparation of Soluble E. coli Proteins

Steps (BioPilot or FPLC)

SDS-PAGE of gel-filtration fractions (step 26) Concentration (step 27)

aStep numbers in parentheses refer to the Basic Protocol. Chromatography materials as well as the BioPilot and FPLC

systems are from Pharmacia Biotech.


250, 500, and 1000-ml stainless steel beakers Ice bucket, ∼4 liter Tissue-grinder homogenizer (Polytron Model PT 10/35, Brinkmann) Ultrasonic homogenizer, ≥400 W, with sound enclosure (Branson or equivalent) Preparative centrifuge: Beckman J2-21M Rotors for preparative centrifuge: Beckman JA-14 (capacity 6 × 250 ml) or JA-20 (capacity 8 × 50 ml) Ultracentrifuge: Beckman Optima XL-90 Rotors for ultracentrifuge: Beckman 45Ti (capacity 6 × 100 ml) or 35Ti (capacity 6 × 94 ml) Conductivity meter (Radiometer America) Spectra/Por 1 dialysis tubing (Spectrum) Gradient maker: Model 2000 (working volume 0.65 to 2 liters; Life Technologies) 200- or 400-ml stirred ultrafiltration cell and Diaflo ultrafilter PM10 or YM3 membranes (Amicon; optional) Millex-GV 0.22-µm-pore-size filter units (Millipore) 10- or 20-ml syringe Additional materials and equipment for SDS-PAGE (UNIT 10.1) and dialysis (APPENDIX 3B) NOTE: All protocol steps are carried at 4°C unless otherwise stated. Forces for centrifugation steps refer to the maximum × g (i.e., centrifugal force at the bottom of the tubes). Prepare anion-exchange column 1. Pour 400 to 500 ml DEAE Sepharose CL-4B ion-exchange resin into a sintered-glass funnel and wash with several liters water followed by 1 liter anion-exchange buffer (pH 8.5). Measure the conductivity of the starting buffer and eluted buffer to make sure they are the same before proceeding to the next step. The resin is supplied in 500-ml bottles as a slurry in 20% ethanol. When washing the resin, do not allow it to run dry on the filter funnel. Laboratory vacuum (e.g., water aspirator) is adequate for filtering.

2. Suspend the washed resin in anion-exchange buffer to 75% settled gel/25% buffer by volume, per manufacturer’s recommendations. Degas in a filter flask and pour into a 5 × 50–cm chromatography column fitted with a filling reservoir. After settling, the height of the resin should be ∼20 to 25 cm (390 to 490 ml packed resin). For details on packing columns, see UNIT 8.4. Because the solubility of gases decreases with increases in temperature, it is usual practice to pack the column at room temperature and then run it in a cold room or cold box.

3. Elute column with anion-exchange buffer at 100 to 150 ml/hr using a peristaltic pump. Make sure there is no compression of column contents. Monitor the absorbance of the effluent at 260 or 280 nm with a UV detector. Collect ∼15-ml fractions in 16 × 50–mm culture tubes using a fraction collector. Check that the pH and conductivity of the column effluent are the same as the for anion-exchange buffer applied to the column (this indicates that the column matrix is correctly equilibrated). The bed height should not change significantly once the column is packed. Compression indicates that the pressure applied to the column is too high (see manufacturer’s recommendations for maximum flow rates).

Break cells with a French press 4. Clean bench areas that may come in direct contact with cells with 0.26% sodium hypochlorite/70% ethanol.



5. Assemble the French pressure cell and and chill to ∼4°C either by incubation in ice or by refrigeration. Install the cell (first dried with paper towels if necessary) in the Aminco laboratory press. It is important to cool the equipment because pressurizing will generate heat. The 20K rapid-fill French pressure cell (1-in.-diameter piston) has a capacity of 40 ml and can be continuously filled while installed on the press. Before using the pressure cell, replace the nylon ball at the end of the flow valve assembly or, at the very least, check it for distortion. For small-scale work, a miniature French pressure cell (3/8-in.-diameter piston) with a 3.7-ml capacity is available.

6. Suspend thawed E. coli cells (∼50 g wet weight) with 150 ml lysis buffer using a Waring blender. Place the suspension in a stainless steel beaker and homogenize with the Polytron tissue-grinder homogenizer until clumps are no longer detected. IMPORTANT NOTE: Wear disposable gloves and safety glasses while working with E. coli. The high-pressure homogenization may generate aerosols. The E. coli cells are stored frozen at −80°C as a flattened paste in heat-sealable plastic bags (UNIT 5.3). The cells are thawed at room temperature. Complete suspension of the cells with the blender is important, as any visible clumps of bacteria will block the French pressure cell. A clogged cell may have to be disassembled to clear the blockage.

7. Lyse the cells with two passes through the French press operated at 16,000 to 18,000 lb/in2 (with the high-ratio setting, pressure gauge readings between 1011 and 1135). Chill the cell suspension to 4°C after each pass through the pressure cell by incubation on ice. When filling the pressure cell, avoid drawing air into the cylinder to prevent foaming. If a French press is not available, the cells can be broken by including 200 ìg/ml lysozyme (Worthington) and 0.05% (w/v) sodium deoxycholate (Calbiochem) in the lysis buffer and incubating cells ∼20 min at 20° to 25°C with intermittent homogenization using the tissue grinder (Burgess and Jendrisak, 1975). Cell breakage by lysozyme treatment and sonication is described in Basic Protocol 3 of UNIT 6.5.

8. Place the suspension (contained in a steel beaker) on an ice bath and, using an ultrasonic homogenizer, sonicate 5 min at full power with 50% duty cycle (on for 0.5 sec then off for 0.5 sec). While sonicating, stir the suspension using a magnetic stirrer. IMPORTANT NOTE: Wear sound-protection earmuffs to protect ears from ultrasonic noise. Because sonication will generate some aerosol, use the sonicator in a microbiological hood if possible. High viscosity reduces the rate of sedimentation of the various contaminating cellular material and thus longer (sometimes much longer) centrifugation times are required. Sonication reduces the viscosity of the suspension prior to centrifugation by shearing the released DNA and RNA. The viscosity can also be reduced by digesting the lysate 15 to 30 min at 4° to 10°C with bovine pancreas DNase I (25 to 50 ìg/ml) and RNase A (50 ìg/ml). If the nucleases are used, EDTA in the lysis buffer should be replaced by 5 mM MgCl2 as DNase requires Mg2+.

Clarify the lysate 9. Transfer sample from the beaker to centrifuge bottles and centrifuge the cell lysate 40 min at 22,000 × g (e.g., in a Beckman J2-21M preparative centrifuge at 12,000 rpm using JA-14 rotor or at 13,500 rpm using JA-20 rotor), 4°C. Decant the supernatants, pool, and recentrifuge 90 min at ∼100,000 × g (30,000 rpm in Beckman Optima XL-90 ultracentrifuge using Ti45 rotor), 4°C. Preparation of Soluble E. coli Proteins

Low-speed centrifugation removes unbroken cells and large cellular debris. Highspeed centrifugation removes smaller particles such as ribosomes and membrane


vesicles; the Beckman 70Ti rotor (capacity 8 × 39 ml) can be used in the ultracentrifuge for smaller-scale work. Clarification of the lysate can also be carried out by salt fractionation (see Background Information, section on determining solubility, for further details). Pellets are usually discarded immediately; but see Critical Parameters and Troubleshooting, section on protein purification for further comments.

10. Dilute the supernatant from step 6 (∼160 ml) 1:2 (three-fold) with anion-exchange buffer and adjust to pH 8.5 (if necessary) with 2 N NaOH. Using a conductivity meter, measure the conductivity of the diluted supernatant. If it is higher than 5.0 to 5.3 mS/cm, reduce by dilution with water. The conductivity of the protein solution is carefully adjusted to ensure that the proteins (in this case contaminants) are bound to the matrix. Too high an ionic strength will reduce or prevent binding. The DEAE Sepharose column is normally prepared (steps 1 to 3) before lysis of the cells is initiated. If there are any delays in applying sample to column, store the clarified lysate at 0° to 4°C (for example, in a covered beaker or flask embedded in an ice bucket). The same applies to the other chromatographic stages.

Chromatograph cleared lysate on anion-exchange resin 11. Apply the clarified lysate (480 to 500 ml) to the DEAE Sepharose column (step 3) at 150 ml/hr and elute the column with anion-exchange buffer. Continue the elution, collecting 15-ml fractions, until the effluent absorbance is close to the baseline value. In solutions above ∼pH 7.0, IL-1β (pI 6.8) is negatively charged. Therefore, the protein should, in principle, bind to an anion exchanger (positively charged matrix) buffered at pH 8.5. In fact, under the conditions described, IL-1β is only weakly bound to the matrix, which allows for partial resolution from proteins that do not bind to the matrix. Much of the unbound protein is of high molecular weight (or highly aggregated) and is partially separated from IL-1β by the gel filtration effect of the matrix. IL-1β is therefore usually located in the latter two-thirds (∼500 ml) of the column flowthrough volume.

12. Assay every second or third column fraction by SDS-PAGE (UNIT 10.1). See Figure 6.2.1A for an example of results from SDS-PAGE. For rapid analysis, use precast gels or the Hoefer Pharmacia Phast system. Alternatively, if speed of purification is important, the entire flowthrough can be used, eliminating the need to analyze separate fractions. Most protein contaminants bound to the column can be removed by step elution with 1 M NaCl in column buffer. After use, the resin should be unpacked from the column and washed on a sintered-glass funnel with 1 liter of 2 M NaCl/0.5% (w/v) Triton X-100, followed by 10 liters water. If the resin is to be stored, suspend it in 5% ethanol or 5 mM sodium azide and store at 4°C. In order to avoid potential cross-contamination, dedicate the used resin for repeat purifications of IL-1β only.

Fractionate sample with ammonium sulfate 13. Pool IL-1β-containing fractions from the DEAE Sepharose column, record the volume (∼500 ml), and transfer the solution to a 1-liter beaker (preferably stainless steel). Add 30.2 g (NH4)2SO4 per 100 ml solution at 0°C (51.3% saturation or 2 M final concentration): add powdered (NH4)2SO4 slowly over 30 min, mixing gently with a magnetic stirrer, then allow a further 30 min of mixing. Methods for calculating the percent saturation of ammonium sulfate solutions, which specifically refer to 0°C, have been described by Wood (1976). When using solid ammonium sulfate, calculations must include volume increases on addition of the solid to fixed volumes (as carried out above). Purification of Recombinant Proteins


A

B

a

1.0

b

c

d

e

g

f

h

1.0 .75 .50

0.75

V0

Vi

200 300 400 Elution volume (ml)

500

A280

.25

0 100

0.50

V0

Vi

0.25

P 0 100

200

300

400

500

Elution volume (ml)

Figure 6.2.1 Purification of IL-1β. (A) SDS-PAGE analysis of samples at various stages. Analysis was conducted on a gel of dimensions 12 cm × 16 cm × 1.5 mm. Lane a, purified protein (100 µg loaded); lane b, purified protein (10 µg loaded); lane d, CM Sepharose pool (80%); lane e, DEAE Sepharose pool after ammonium sulfate fractionation (56%); lane f, high-speed supernatant (starting material for DEAE Sepharose column; 13.5%); lane g, cell lysate (12.0%). The percentages refer to specific IL-1β contents of the fractions determined by densitomeric scanning of the Coomassie blue–stained gel lanes. Lanes c and h contain the following protein standards (low-range standards supplied by Bio-Rad) in order of increasing migration distance: phosphorylase b (97.4 kDa), bovine serum albumin (66.2 kDa), hen egg white ovalbumin (45 kDa), bovine carbonic anhydrase (31 kDa), soybean trypsin inhibitor (21.5 kDa), and hen white lysozyme (14.4 kDa). (B) Analysis of results from gel filtration on Ultrogel AcA54. The excluded volume (V0) and the fully included volume (Vi) are indicated. Inset, analytical rechromatography of the protein from the pooled fractions (indicated P in larger chromatogram).

Preparation of Soluble E. coli Proteins


14. Centrifuge the slightly cloudy solution 30 min at 22,000 × g (12,000 rpm in JA-14), 4°C. Decant the supernatant into a beaker and add an additional 17 g (NH4)2SO4 per 100 ml solution (77% saturation or 3 M final concentration). Equilibrate with stirring and centrifuge 30 min at 22,000 × g, 4°C. For the addition of (NH4)SO4 follow the same method as described in step 13.

15. Decant the supernatant and drain the pellets by inverting the tubes on a paper towel. Save the pellets. Dialyze the fractionated sample 16. Suspend the pellets in ∼300 ml cation-exchange buffer and dialyze, using Spectra/Por 1 dialysis tubing, against 5 liters cation-exchange buffer. Change the dialysis buffer at least once. The dialysis step is conveniently performed overnight; the CM Sepharose column used in step 18 can be prepared during this period. The dialysis tubing is prepared by heating 30 to 60 min at 90° to 95°C in 5 mM EDTA. The tubing is then washed well with water and stored in 10% ethanol at 4°C prior to use. A suitable length of tubing is filled to about one-half to three-quarters capacity with solution (to allow for expansion) and sealed with two knots at each end. Use gloves when handling the tubing, check for leaks before use, and make sure the magnetic stir-bar does not rub against the tubing. See APPENDIX 3B for further information concerning dialysis.

17. After dialysis, remove the slightly cloudy solution from the tubing and centrifuge 30 min at 22,000 × g (12,000 rpm in JA-14), 4°C. Save the supernatant. Chromatograph dialyzed sample on cation-exchange resin 18. Prepare ∼200 to 225 ml CM Sepharose CL-4B resin by washing on a sintered-glass funnel first with water, then with cation-exchange buffer (pH 5.7; wash as in step 1 except using different buffer). Pack the degassed resin into a 5 × 50–cm column as in step 2. The packed column will have a bed height of ∼11 to 12 cm. The comments made in the annotation to step 3 also apply here.

19. Elute the column using cation-exchange buffer at 100 to 150 ml/hr with a peristaltic pump. Monitor the column effluent at 280 or 260 nm using a suitable UV detector. Check that the pH and conductivity of the column effluent are the same as for the buffer applied to the column. 20. Check the pH and conductivity of the dialysate supernatant from step 17 and, if necessary, dilute with water so that the conductivity is in the range 1.0 to 1.2 mS/cm (at 4° to 6°C). Apply the clear solution to the CM Sepharose column at a flow rate of 150 ml/hr. When the UV absorbance of the column effluent approaches baseline, proceed to the next step. 21. Prepare a 0 to 250 mM NaCl gradient in cation-exchange buffer by adding 500 ml buffer to the inner chamber of the gradient maker and 500 ml buffer/250 mM NaCl to the outer chamber. Apply the gradient to the column at 150 ml/hr. Collect 15-ml fractions. The total volume of the gradient is 1 liter (∼4.5 column volumes). At pH 5.7, IL-1β (pI 6.8) is positively charged and binds to the negatively charged cation-exchange resin. IL-1β is eluted from the column with ∼100 mM NaCl, and it will be located in the major absorbance peak (see Wingfield et al., 1986, for figure of typical elution profile).



22. Monitor the progress of the gradient using an in-line conductivity meter positioned after the absorbance flow cell. Assay column fractions for IL-1β by SDS-PAGE (UNIT 10.1) and pool fractions containing IL-1β. Deciding what fractions to pool is dictated by the fact that the remaining purification stage is gel filtration, a method that will not remove contaminants with sizes close to that of IL-1β (17.4 kDa). As IL-1β is a well-expressed protein (>5% total protein), one can afford to be conservative and pool for purity rather than yield. The used CM-Sepharose matrix can be cleaned up and stored as described in the annotation to step 12. CM Sepharose Fast Flow or SP Sepharose FF (a strong cation exchanger; both resins are also from Pharmacia Biotech) can be used instead of CM Sepharose CL-4B. Similar results are obtained with either matrix, with the advantage of faster flow rates.

Concentrate the proteins 23a. To concentrate proteins by ultrafiltration: Adjust the CM Sepharose pool (150 to 250 ml) to pH 7.5 with Tris base. Assemble a 200- or 400-ml stirred ultrafiltration cell containing a washed ultrafilter membrane, either YM3 (3-kDa cutoff) or PM10 (10-kDa cutoff). Pressurize the cell with nitrogen according to manufacturer’s recommendations. Collect the effluent from the cell into a measuring cylinder and occasionally record the absorbance at 280 nm to check that the membrane is not leaking. When the volume remaining in the cell is ∼15 ml, depressurize the cell and carefully remove the solution using a Pasteur pipet with a small length of polyethylene tubing attached to the end so as not to scratch the membrane. Use ∼3 ml gel-filtration buffer to wash the membrane, adding washings to the main concentrate. Exposure to solutions below pH 4.5 and above pH 10.5 to 11.0 often causes protein denaturation. Hence, when adjusting the pH of protein solutions, there is less chance of overshooting the required pH by using concentrated buffer components instead of pure acid or base. For example, Tris base is used rather than dilute NaOH. Select the ultrafiltration membrane pore size based on the size of the protein. The membrane can be reused; store in 5% ethanol at 4°C. It is good practice to reserve a membrane for use with a particular protein (compare comment on resin usage) and to wash the concentration cell carefully after use.

23b. To concentrate proteins by salt precipitation: Adjust the CM Sepharose pool to pH 7.5 with Tris base and slowly add 53.9 g (NH4)2SO4 per 100 ml solution at 0°C (82.2% saturation or 3.2 M final concentration). Follow the basic salt precipitation method (steps 13 to 15) and suspend the pellets in 18 to 20 ml gel-filtration buffer. Conduct gel-filtration chromatography 24. Prepare ∼480 ml Ultrogel AcA54 resin by washing on a sintered-glass funnel with water, then gel-filtration buffer (as in step 1); suspend washed resin in gel-filtration buffer to 75% settled gel/25% buffer by volume and degas (step 2). Pour a slurry of degassed resin into a 2.5 × 100–cm chromatography column fitted with a filling reservoir. Pack the column at ∼35 ml/hr. UNITS 6.3 & 8.3 should be consulted for further details on the preparation and elution of gel-filtration columns. The column should be prepared in advance so that the protein samples can be applied as soon as the concentration step is completed.


The resin should be free-flowing yet concentrated enough to produce a packed bed with one pouring. A freshly packed gel-filtration column can be checked for packing irregularities by prerunning the column with a few colored markers. Blue dextran will be excluded from the gel matrix and will elute at the void volume (V0, which equals 30% to 35% of the total column volume); cytochrome c (red; 12.4 kDa) will elute close


to the expected position of IL-1β; and potassium dichromate (yellow) will be fully included in the gel matrix and will elute at the included volume (Vi, ∼480 ml). Superdex 75 gel-filtration matrix (Pharmacia Biotech) can be substituted for Ultrogel AcA54. The former allows higher flow rates and thus is more compatible with the FPLC and BioPilot systems. Both Superdex and the cation-exchange Fast Flow resins mentioned above can be purchased in various prepacked columns that are useful for method development.

25. Filter the concentrated protein using a Millex-GV 0.22-µm filter unit attached to a 10- or 20-ml syringe and apply to the gel-filtration column. The sample can be applied either directly to the top of the column using a Pasteur pipet (care is required to prevent breaking the tip and contaminating the column) or via a three-way valve and syringe without removing the top flow adapter. The volume of sample applied to the column (20 ml) represents ∼4% of the total column volume. For columns with different dimensions, apply the same proportionate volume of sample. The column should be ≥60 cm long. For analytical separations, the sample volume should not exceed 2.5% of the column volume.

26. Elute the column at 35 ml/hr with gel-filtration buffer and collect 10-ml fractions. The major eluting peak contains the IL-1β; monitor the fractions by SDS-PAGE (UNIT 10.1) and pool fractions that contain pure protein. Save side fractions that contain small amounts of contaminants; this material can be rechromatographed after concentration as described in step 23a or 23b. See Figure 6.2.1A for an example of results from SDS-PAGE. Once the sample has been run into the column, it can be eluted with any buffer or even with a suitable column storage solvent such as 5 mM sodium azide in water.

Concentrate and store purified protein 27. Concentrate the purified protein, if required, by ultrafiltration (see step 23a). Determine the protein concentration of purified IL-1β by measuring absorbance at 280 nm. A molar absorbance coefficient (ε) of 10.61 mM−1 cm−1 is used. This corresponds to an absorbance of 0.63 for a 1 mg/ml solution using a 1-cm-path length cell.

28. For short-term storage (≤12 months), filter the protein with a Millex-GV 0.22-µm filter unit, divide the solution into sterile plastic vials, and freeze aliquots rapidly with dry ice/ethanol. Store at −80°C. For long-term storage of IL-1β (>12 months), lyophilize the protein. Dialyze the sample using a volatile buffer such as 50 mM ammonium bicarbonate or a nonvolatile buffer such as lyophilization buffer. To circumvent the dialysis step, the phosphate-based lyophilization buffer can be used for gel filtration instead of the Tris⋅Cl gel-filtration buffer.

REAGENTS AND SOLUTIONS Use Milli-Q-purified water or equivalent in all recipes and protocol steps. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX. Buffer pH and conductivities are for solutions at 4° to 6°C. Units of conductance are given in siemens (S = Ω−1).

Anion-exchange buffer (50 mM Tris⋅Cl, pH 8.5) Dilute 1 M Tris⋅Cl, pH 8.0 20-fold with water and adjust to pH 8.5 with NaOH. Make immediately before use. Conductivity of the solution is 1.57 mS/cm.



Cation-exchange buffer, 10× (15 mM sodium phosphate, pH 5.7/1 mM sodium azide) 19.4 g NaH2PO4⋅H2O 1.4 g Na2HPO4⋅2H2O 0.65 g sodium azide H2O to 1 liter Store up to 1 month at 4°C Dilute 10-fold immediately prior to use Conductivity of the solution is 0.9 mS/cm. Sodium azide is an antibacterial agent.

Cation-exchange buffer (1×)/250 mM NaCl 14.61g NaCl 100 ml 10× cation-exchange buffer (see recipe) H2O to 1 liter Gel-filtration buffer (100 mM Tris⋅Cl, pH 7.5/1 mM sodium azide) Dilute 1 M Tris⋅Cl, pH 8.0 10-fold with water and adjust to pH 7.5 with HCl. Add sodium azide from a 1 M stock solution (65 g/liter). Make immediately before use. Lyophilization buffer (25 mM sodium phosphate, pH 7.5/0.5 mM sodium azide) 2.2 g NaH2PO4⋅H2O 11.9 g Na2HPO4⋅2H2O 1.3 g sodium azide H2O to 4 liters Make fresh and use immediately Usually, 2 liters is required for dialysis, with one buffer change (4 liters total). Gel filtration requires smaller volumes.

Lysis buffer 100 mM Tris⋅Cl, pH 8.0 2 mM EDTA, pH 8.0 5 mM benzamidine⋅HCl (780 mg/liter) Make immediately prior to use; alternatively, make ahead of time and store up to several days at 4°C. Conductivity of the solution is 1.57 mS/cm. It should be noted that a 1°C decrease in temperature increases the pH of the Tris buffer by ∼0.03 pH units. Both Tris⋅Cl and EDTA stock solutions are commercially available (e.g., Life Technologies). Benzamidine⋅HCl is a water-soluble serine protease inhibitor. An alternative is 50 ìM 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride (AEBSF; Perfabloc SC, Boehringer Mannheim), a water-soluble inhibitor with the same spectrum of activity as phenylmethylsulfonyl fluoride (PMSF).

COMMENTARY Background Information


Expression of soluble proteins The basic stages involved in preparing soluble proteins from E. coli are achieving protein expression, harvesting and breaking cells (or collecting culture medium, in the case of secreted proteins), purifying proteins, and characterizing the purified protein. Choices that must be made at each stage are briefly reviewed.

Optimizing the expression system The basic requirements for protein expression in E. coli are discussed in UNIT 5.1. Experience has shown that it is usually possible to express any given protein in E. coli, but whether the protein will be expressed in a soluble and unmodified state is unpredictable. Tinkering with the various elements of the expression plasmid, changing the E. coli host strains, and altering the fermentation conditions are carried out to optimize the expression


of unmodified soluble protein (UNIT 5.1). Use of secretion vectors and fusion protein constructs to prevent aggregation has been discussed elsewhere (UNITS 5.1 & 6.1). Even if efforts to express soluble protein fail and inclusion bodies are produced, there is often a good chance that the protein can be extracted and folded into native protein (see UNITS 6.4 & 6.5). There are situations where the intrinsic properties of a protein can preclude its expression in a soluble and stable form. Examples include the following: (1) individual subunits of a heterodimer, which may be unstable and may degrade or aggregate when expressed separately (coexpression of the subunits may solve the problem; Nash et al., 1987); (2) proteins derived from activated precursors (UNIT 6.4; Hlodan and Hartl, 1994); and (3) proteins in which the native (authentic) sequences have been modified by site-directed mutagenesis (Chrunyk et al., 1993). It is worth noting that the solubility of proteins that are otherwise insoluble or poorly soluble can, conversely, sometimes be improved by site-directed mutagenesis (Dyda et al., 1994). Once conditions have been found that result in the expression of soluble protein, the next hurdle is to recover the protein in an unmodified form. Avoiding proteolysis and chemical modification during the isolation process is crucial. Breaking cells Methods for breaking cells or selectively extracting proteins are reviewed in UNIT 6.1. In the protocol for purifying human IL-1β, cells are disrupted with a French press. This method is the most efficient way to break E. coli cells and is the method of choice. The equipment, however, requires dedicated laboratory space and may be expensive for some laboratories. Cell breakage with lysozyme (UNIT 6.5) and/or sonication is also frequently used; such methods are especially suited to small-scale work (reviewed by Hopkins, 1991). It is important to consider whether the cell breakage method influences the final solubility and stability of the recombinant protein in the cell lysates. A soluble recombinant protein constituting ∼12% of the total bacterial protein will be present in the cytoplasmic space at a concentration of at least 30 mg/ml, assuming a total protein and RNA concentration in the cytoplasm of ∼340 mg/ml in nontransformed cells (Zimmerman and Trach, 1991) and a 2.7:1 weight ratio of protein to RNA (Neidhart, 1987). Cell breakage can be expected to in-

crease the solubility of the recombinant protein by a dilution effect (discussed by Zimmerman and Trach, 1991). Cell lysis procedures, however, especially when mechanical shear is used (e.g., French press and sonication), can produce local heating that may have a denaturing effect and lead to aggregation. Mechanical shearing will also be expected to decrease the size of nucleic acids and release polyanionic lipopolysaccharides; both classes of coumpounds can bind nonspecifically to proteins, especially basic proteins, thereby reducing their solubility. Including detergents in the lysozyme treatment may also release potent membrane-bound proteases. Despite these potential pitfalls, the recovery of soluble protein is usually not dramatically influenced by the cell breakage method, but there is always the potential that it might be. Protein recovery as a function of cell breakage methodology is rarely systematically studied or reported. Johnson and Hecht (1994) reported that freezing and thawing of cells selectively releases recombinant proteins located in the cytoplasm. The freeze-thaw procedure, in principle, should be the best method for releasing soluble proteins in an unmodified state, but its general applicability remains to be seen. Determining solubility The solubility of recombinant protein in the cell lysate or extract is usually established by differential centrifugation (UNITS 6.1 & 5.3). Alternatively, the extract can be clarified by filtering and applied to a small gel-filtration column. An HPLC or FPLC gel-filtration separation in conjunction with a rapid screening method for assaying column fractions (e.g., SDS-PAGE using PhastSystem from Hoefer Pharmacia) can produce a determination of both solubility and size distribution in ∼2 hr or less. Purifying protein There are many approaches to purifying proteins, such as expressing fusion proteins that contain engineered affinity handles or tags (see UNITS 5.1 & 6.1 for further details). Most soluble proteins that are well behaved—i.e., do not contain excessive charge heterogeneity—are non-self-associating and can often be purified by two stages of ion-exchange chromatography and one or more polishing steps, one of which should be gel filtration. The basic approach is outlined in Figure 6.1.3. The purification of human IL-1β is a typical example. Specific accounts of the various chromatographic meth-



ods are found in Chapters 8 and 9; Scopes (1993) particularly emphasizes first principles. Characterizing protein Once a protein has been purified, it is usually characterized (Chapter 7) in order to establish chemical homogeneity (primary sequence) and physical homogeneity (size and conformation). Not only is this information required for a fundamental description of the protein, but it will also give insights on how best to rationalize the purification process. Moreover, the detection of certain heterogeneities may give clues on how to avoid their occurrence. For example, chemical modifications such as deamidation may be prevented by changing one or more purification steps that expose the protein to extremes of pH. Other modifications that arise from construction of the expression plasmid itself, once recognized, can often be fixed.


Characteristics of recombinant and authentic interleukin 1β The cytokine IL-1β and the closely related IL-1α are synthesized mainly by monocytes and mononuclear cells as 31-kDa cytoplasmic precursor proteins (reviewed by Dinarello, 1989). A unique cytoplasmic cysteine protease (IL-1β-converting enzyme, or ICE) cleaves the IL-1β precursor in half, generating the mature form of the protein comprising the C-terminal 153 residues. The biological activities of IL-β (and IL-1α) are initiated by interaction with either type I or type II cellular receptors (McMahan et al., 1991). Both receptors have three immunoglobulin-like extracellular ligand-binding domains and single membranespanning segments. Another form of IL-1, the interleukin 1 receptor antagonist (IL-1ra), is a natural competitive antagonist of IL-1β. Authentic IL-1β been isolated from human monocyte cell cultures by HPLC using anionexchange and gel-filtration (or reversed-phase) matrices. A 3000-fold purification was required to obtain 16 µg protein from 5 liters of cell culture extract (Gery and Schmidt, 1985). The authentic protein was established to be a nonglycosylated monomer (∼18 kDa), with a pI of 6.8 and containing a single N-terminal sequence (AlaProValArg) as predicted by the DNA coding sequence. The recombinant protein produced in E. coli has the same properties as the natural product except that N-terminal heterogeneity is often observed. The recombinant protein usually contains a mixture of forms with Met (20%), Ala (67%), or Pro (13%) as the N terminus. The

Met form is derived by incomplete processing of the initiating Met, and the Pro form is due to cleavage of the authentic N-terminal Ala by E. coli protease. Despite the slight differences among the recombinant variants, partial or complete separations have been achieved (Wingfield et al., 1987; Yem et al., 1988). Both the crystal structure of IL-1β (Priestle et al., 1988) and its nuclear magnetic resonance (NMR) structure in solution (Clore et al., 1991) have been determined using material purified by the protocol described here. For some NMR studies it was necessary to produce material with a single N terminus, and this was fortuitously achieved by replacing the N-terminal Ala residue with a Cys residue. Purified protein from cells expressing the mutant (N-terminal CysProValArg) contain only N-terminal Pro. It should be noted that deletion of the first four residues of IL-1β has no effect on activity; however, the presence of the unprocessed Met reduces receptor binding 10-fold. Interleukin 1β does not contain a disulfide— the cysteines are either buried (Cys-71) or partially buried (Cys-8)—so there is no need to include reductants in column buffers. As mentioned earlier, the authentic protein is nonglycosylated even though there is a potential Nglycosylation site (-Asn7-Cys8-Thr9-) near the N terminus. Interestingly, when IL-1β is expressed in yeast, equal amounts of N-glycosylated (21 kDa) and nonglycosylated protein (17 kDa) are produced (Livi et al., 1991). Other approaches to purifying recombinant interleukin 1β Because of the biomedical importance of IL-1β, several reports have described expression and purification of the protein. Some examples are discussed. Kronheim et al. (1986) acidified E. coli cell lysates to pH 4 and thus precipitated ∼70% of the total protein while most of the IL-1β remained soluble. IL-1β was further purified by cation-exchange, anion-exchange, and dyematrix chromatographies (UNITS 8.2 & 9.2); ∼200 mg protein was recovered from 2.5 liters of culture. The noteworthy feature of this method is the acid precipitation stage. Extracts of E. coli contain predominately negatively charged (anionic) proteins: 60% have pI values between 5.0 and 6.0 and 80% between 4.5 and 6.7 (Sherwood, 1992). Adjustments of cell extracts to pH 5.0 and below thus result in isoelectric precipitation of substantial amounts of E. coli protein. This is a useful purification step as long


as the recombinant protein remains soluble (or does not coprecipitate) under the conditions. In another approach, Meyers et al. (1987) carried out ammonium sulfate fractionation directly on the E. coli extract. The IL-1β protein was recovered in the fraction corresponding to 50% to 80% saturation. After desalting, the protein was further purified by ion-exchange (cation and anion) and gel-filtration chromatographies; ∼400 mg protein was recovered from 250 g wet weight cells. The high-speed centrifugation step used to remove particulate material (see Basic Protocol, step 6) can often be replaced by salt fractionation. For IL-1β, advantage can be taken of the fact that a relatively high (NH4)2SO4 concentration is required to precipitate the protein. The disadvantage of using this approach early in the purification is that the protein must be desalted before being applied to an ion-exchange column. Finally, it is of interest to note that IL-1β can be released from E. coli cells by osmotic shock treatment (Joseph-Liauzun et al., 1990). It is well known that certain E. coli cytoplasmic proteins, including thioredoxin, can be released by osmotic shock. Scale of procedure The Basic Protocol described starts with 50 g cell paste and yields ∼400 mg pure IL-1β. For smaller (1 to 5 g) or larger (100 g) cell quantities, the method can be adapted by simply reducing or increasing, respectively, the crosssectional area of the chromatography columns. For example, for 5 g cells, the diameters of the DEAE and CM Sepharose CL-4B columns are reduced to 2.5 cm, with column lengths of 20 and 10 cm, respectively. This results in 1:4 reductions in the resin volumes compared to those for the protocol utilizing 50 g cells. The original gel-filtration column can be used (2.5 × 100 cm); however, the volume of ∼1:4 sample applied should be reduced to 8 to 10 ml. Alternately, a column of about the same length (60 to 100 cm) but smaller diameter (e.g., 1.25 cm) should be used. For discussion on scaling up column chromatography procedures, see UNIT 8.3 (Strategic Planning and Scopes, 1993).

Critical Parameters and Troubleshooting High-level protein expression with good cell yield (UNITS 5.1-5.3) is required to obtain reasonable amounts of pure protein. This is obviously true no matter how efficient (or inefficient) the purification protocol. Once the host/vector sys-

tem and the fermentation conditions have been optimized and, equally important, standardized, the level and solubility of recombinant protein in the starting material should be consistent. However, it is sensible to check both the expression level and solubility before starting each protein purification. The expression level can usually be monitored by SDS-PAGE; UNIT 5.2 describes how to prepare samples for analysis by SDS-PAGE, and UNIT 10.1 gives electrophoresis procedures. Solubility can be easily monitored by breaking a small amount of cells (≤1.0 g) by sonication (UNIT 5.3) or by treating cells with lysozyme (UNIT 6.5) and microcentrifuging the resulting cell lysate to judge whether the protein is highly aggregated. There are many reasons why protein expression may not be as high as anticipated and, worse, why the expressed protein may switch from being soluble to insoluble. For example, faulty pH control during the fermentation may cause insolubility. Whatever the situation, one should be in a position to rationally decide whether to continue with the purification or repeat the initial fermentation. If the protein expression level is normal (>5%) and the protein has accumulated in a soluble state, implementing the Basic Protocol, which uses only standard purification methods, should be trouble-free. Common mistakes involve buffer preparation and other trivial errors such as pooling the wrong column fractions. As discussed later, do not discard column fractions until they have been completely analyzed. A classic purification table (see for example, Table III.1 in Dixon and Webb, 1979), has not been presented, as the biological activities of IL-1β could not be measured reliably in crude E. coli cell extracts by assays available to the author when the method was developed (Wingfield et al., 1986). As is often the case with proteins requiring complex biological assays, protein size analysis by SDS-PAGE is used to follow purification. During the initial stages, it is of course necessary to confirm the identity of the particular band one is following, and this can be accomplished by eluting or blotting the protein from SDS-polyacrylamide gels and performing N-terminal amino acid sequence analysis. Immunoblotting with specific antibody can be used to monitor protein purification when the recombinant protein is in low abundance (e.g., for secreted proteins in the cell medium). Finally, it should be noted that several manufacturers (e.g., R&D Systems and Genzyme) now supply assay kits for many cytokines, including IL-1β.



Cell breakage Efficient cell breakage should be troublefree as long as the French press is operated in accordance with the manufacturer’s instructions. It takes a little practice to operate the flow valve. The aim is to generate as high a flow rate as possible while maintaining a pressure gauge reading of ∼1000. If the flow rate is too fast, the pressure reading will drop and unbroken cells will pass into the flow stream. Toward the end of the run, the flow rate should be reduced, as it becomes difficult to control the pressure. After use the French pressure cell should be cleaned and dried, and the flow valve ball should be replaced. Store the cell at 4°C. Protein purification In general, troubleshooting a purification method will be much easier if fractions are not discarded until the appropriate monitoring of the purification steps is complete. When using an established method, fractions from a chromatographic run are frequently pooled on the basis of absorbance only and the remainder quickly discarded. Ammonium sulfate supernatants or pellets are often discarded on the basis of previous fractionation behavior or pilot-scale work. As stated, do not discard any fractions until they have been checked, usually by SDS-PAGE. If there has been a problem, it can usually be easily sorted out if all the fractions from the various stages are still available. If in doubt about conditions for storing fractions, freeze selected fractions at as low a temperature as possible (ideally −80°C) and discard when appropriate. When freezing material, it is worthwhile to take the extra effort to dispense small samples (44 kDa in size, and these are easily separated from the IL-1β protein (17.4 kDa) by gel filtration. Analytical rechromatography of the protein from the pooled fractions (indicated P in Fig. 6.2.1B) reveals a single symmetrical peak indicative of purity and physical homogeneity (Fig. 6.2.1B inset). A typical cation-exchange chromatogram for the purification appears elsewhere (Wingfield et al., 1986).

Time Considerations Because IL-1β appears to be stable against proteolytic degradation and other chemical modications during purification, the speed of purification was not critical in this case. The low-pressure chromatographic method described in the Basic Protocol requires ∼4 days, which can be shortened to ∼3 days using the Pharmacia Biotech BioPilot or FPLC systems in conjunction with matrices that allow faster flow rates. The times required for purification whether using low-pressure chromatography (as described in the Basic Protocol) or mediumpressure chromatography in the FPLC or BioPilot systems are summarized in Table 6.2.1.

Literature Cited Burgess, R.R. and Jendrisak, J.J. 1975. A procedure for the rapid, large-scale purification of E. coli DNA-dependent RNA polymerase involving polymin P precipitation and DNA-cellulose chromatography. J. Biol. Chem. 14:4634-4638. Chrunyk, B.A., Evans, J., Lillquist, J., Young, P., and Wetzel, R. 1993. Inclusion body formation and protein stability in sequence variants of Interleukin-1β. J. Biol. Chem. 268:18053-18061. Clore, G.M., Wingfield, P.T., and Gronenborn, A.M. 1991. High-resolution structure of interleukin 1β in solution by three- and four-dimensional nuclear magnetic resonance spectroscopy. Biochemistry 30:2315-2323.


Dinarello, C.A. 1989. Interleukin-1 and its biologically related cytokines. Adv. Immunol. 44:153205. Dixon, M. and Webb, E. 1979. Enzyme isolation. In Enzymes (3rd ed.) pp. 23-46. Academic Press, New York. Dyda, F., Hickman, A.B., Jenkins, T.M., Engelman, A., Craigie, R., and Davies, D.R. 1994. Crystal structure of the catalytic domain of HIV-1 integrase: Similarity to other polynucleotidyltransferases. Science 266:1981-1986. Gery, I. and Schmidt, J.A. 1985. Human interleukin 1. Methods Enzymol. 116:456-467. Hlodan, R. and Hartl, F.U. 1994. How the protein folds in the cell. In Mechanisms of Protein Folding (R.H. Pain, ed.) pp. 194-228. IRL Press, Oxford. Hopkins, T.R. 1991. Physical and chemical cell disruption for the recovery of intracellular proteins. In Purification and Analysis of Recombinant Proteins (R. Seetharam and S.K. Sharma, eds.) pp. 57-83. Marcel Dekker, New York. Johnson, B.H. and Hecht, M.H. (1994) Recombinant proteins can be isolated from E. coli by repeated cycles of freezing and thawing. Bio/Technology 12:1357-1360. Joseph-Liauzun, E., Legoux, R., Guerveno, V., Marchese, E., and Ferra, P. 1990. Human recombinant interleukin-1β isolated from E. coli by simple osmotic shock. Gene 86:291-295. Kronheim, S.R., Cantrell, M.A., Deeley, M.C., March, C.J., Glackin, P.J., Anderson, D.M., Hemenway, T., Merriam, J.E., Cosman, D., and Hopp, T.P. 1986. Purification and characterization of human interleukin-1 expressed in Escherichia coli. Bio/Technology 4:1078-1082. Livi, G.P., Lillquist, J.S., Ferrara, A., Sathe, G.M., Simon, P.L., Meyers, C.A., Gorman, J.A., and Young, P.R. 1991. Secretion of N-glycosylated interleukin-1β in Saccharomyces cerevisiae using a leader peptide from Candida albicans. Effect of N-linked glycosylation on biological activity. J. Biol. Chem. 266:15348-15348. McMahan, C.J., Slack, J.L., Mosley, B., Cosman, D., Lupton, S.D., Brunton, L.L., Grubin, C.E., Wignall, J.M., Jenkins, N.A., Brannan, C.I., Copeland, N.G., Huebner, K., Croce, C.M., Cannizzarro, L.A., Benjamin, D., Dower, S.K., Spriggs, M.K., and Sims, J.E. 1991. A novel IL-1 receptor, cloned from B cells by mammalian expression, is expressed in many cell types. EMBO J. 10:2821-2832. Meyers, C.A., Johanson, K.O., Miles, L.M., McDevitt, P.J., Simon, P.L., Webb, R.L., Chen, M.-J., Holskin, B.P., Lillquist, J.S., and Young, P.R. 1987. Purification and characterization of human recombinant interleukin-1β. J. Biol. Chem. 262:11176-11181.

Nash, H.A., Robertson, C.A., Flamm, E., Weisberg, R.A., and Miller, H. 1987. Overproduction of Escherichia coli integration host factor, a protein with nonidentical subunits. J. Bacteriol. 169:4124-4127. Neidhardt, F.C. 1987. Chemical composition of Escherichia coli. In Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (F.C. Neidhardt, J.L. Ingraham, K.B. Low, B. Magasanik, M. Schaechter, and H.E. Umbarger, eds.) pp. 3-6. American Society for Microbiology, Washington, D.C. Priestle, J.P., Schar, H.-P., and Grutter, M.G. 1988. Crystal structure of the cytokine interleukin 1β. EMBO J. 7:339-343. Scopes, R.K. 1994. Protein Purification: Principles and Practice, 3rd ed. Springer-Verlag, New York and Heidelberg. Sherwood, R.F. 1992. Making bacterial extracts suitable for chromatography. Meth. Mol. Biol. 11:287-305. Wingfield, P., Payton, M., Tavernier, J., Barnes, M., Shaw, A., Rose, K., Simona, M.G., Demczuk, S., Williamson, K., and Dayer, J.M. 1986. Purification and characterization of human interleukin1β expressed in recombinant Escherichia coli. Eur. J. Biochem. 160:491-497. Wingfield, P.T., Graber, P., Rose, K., Simona, M.G., and Hughes, G.J. 1987. Chromatofocusing on N-terminally processed forms of proteins. J. Chromatogr. 387:291-300. Wood, W.I. 1976. Tables for the preparation of ammonium sulfate solutions. Anal. Biochem. 73:250-257. Yem, A.W., Richard, K.A., Staite, N.D., and Deibel, M.R. 1988. Resolution and biological properties of three N-terminal analogues of recombinant human interleukin-1β. Lymphokine Res. 7:8592. Zimmerman, S.B. and Trach, S. 1991. Estimation of macromolecular concentrations and excluded volume effects in the cytoplasm of Escherichia coli. J. Mol. Biol. 222:599-620.

Key Reference Wingfield et al., 1986. See above. The original publication on which Basic Protocol 1 is based.

Contributed by Paul T. Wingfield National Institutes of Health Bethesda, Maryland



Preparation and Extraction of Insoluble (Inclusion-Body) Proteins from Escherichia coli

UNIT 6.3

High-level expression of many recombinant proteins in Escherichia coli leads to the formation of highly aggregated protein commonly referred to as inclusion bodies (UNITS 5.1 & 6.1). Inclusion bodies are normally formed in the cytoplasm; alternatively, if a secretion vector is used, they can form in the periplasmic space. Inclusion bodies recovered from cell lysates by low-speed centrifugation are heavily contaminated with E. coli cell wall and outer membrane components. The latter are largely removed by selective extraction with detergents and low concentrations of either urea or guanidine⋅HCl to produce so-called washed pellets. These basic steps result in a significant purification of the recombinant protein, which usually makes up ∼60% of the washed pellet protein. The challenge, therefore, is not to purify the recombinant-derived protein, but to solubilize it and then fold it into native and biologically active protein. Basic Protocol 1 describes preparation of washed pellets and solubilization of the protein using guanidine⋅HCl. The extracted protein, which is unfolded, is either directly folded as described in UNIT 6.5 or further purified by gel filtration in the presence of guanidine⋅HCl as in Basic Protocol 2. A support protocol describes the removal of guanidine⋅HCl from column fractions so they can be monitored by SDS-PAGE (UNIT 10.1). PREPARATION AND EXTRACTION OF INSOLUBLE (INCLUSION-BODY) PROTEINS FROM ESCHERICHIA COLI

BASIC PROTOCOL 1

Bacterial cells are lysed using a French press, and inclusion bodies in the cell lysate are pelleted by low-speed centrifugation. The pellet fraction is washed (preextracted) with urea and Triton X-100 to remove E. coli membrane and cell wall material. Guanidine⋅HCl (8 M) and dithiothreitol (DTT) are used to solubilize the washed pellet protein. Extraction with the denaturant simultaneously dissociates protein-protein interactions and unfolds the protein. As a result, the extracted protein consists (ideally) of unfolded monomers, with sulfhydryl groups (if present) in the reduced state. Materials E. coli cells from fermentation (UNIT 5.3) containing the protein of interest Lysis buffer (see recipe) Wash buffer (see recipe), with and without urea and Triton X-100 Extraction buffer (see recipe) 250- and 500-ml stainless steel beakers 0.22-µm syringe filters (e.g., Millex from Millipore) 20-ml disposable syringe Additional equipment for breaking cells, homogenizing cells and pellets and centrifuging at low and high speeds (UNIT 6.2) Break cells and prepare clarified lysate 1. Place thawed E. coli cells in a stainless steel beaker. Add 4 ml lysis buffer per gram wet weight of cells. Keep bacterial cells cool by placing the beaker on ice in an ice bucket. The cells can be pretreated with lysozyme prior to lysis in the French press. Lysozyme treatment involves incubating cells ∼20 min at 20° to 25°C in lysis buffer supplemented Contributed by Ira Palmer and Paul T. Wingfield Current Protocols in Protein Science (1995) 6.3.1-6.3.15 Copyright © 2000 by John Wiley & Sons, Inc.


6.3.1 CPPS

with 200 ìg/ml lysozyme, with intermittent homogenization using a tissue grinder. It should be emphasized that this optional step is carried out before French press breakage and is not simply an alternative method of cell breakage (compare the comments made in the annotation to step 4 of UNIT 6.2). Its purpose is to aid removal of the peptidoglycan and outer membrane protein contaminants during the washing steps (steps 6 to 9; for further details see UNIT 6.1 and Fig. 6.1.5). An example of this approach is given in Basic Protocol 1 of UNIT 6.5.

2. Suspend cells using a Waring blender and homogenize using the Polytron tissuegrinder homogenizer until all clumps are disrupted, as described in UNIT 6.2, step 3. 3. Lyse cells with two passes through the French pressure cell operated at 16,000 to 18,000 lb/in2 (with the high-ratio setting, pressure gauge readings between 1011 and 1135), chilling the cell suspension to 4°C after each pass, as described in UNIT 6.2, steps 2 and 4. 4. Reduce the viscosity of the suspension by sonicating 5 min at full power with 50% duty cycle (on for 5 sec, off for 5 sec) using an ultrasonic homogenizer, as described in UNIT 6.2, step 5. 5. Clarify the lysed cell suspension by centrifuging 1 hr at 22,000 × g (12,000 rpm in a JA-14 rotor in a Beckman J2-21M centrifuge), 4°C. Unbroken cells, large cellular debris, and the inclusion body protein will be pelleted. The JA-14 rotor uses 250-ml centrifuge bottles. For processing smaller volumes the Beckman JA-20 rotor (or equivalent) with 50-ml tubes can be used, at 13,500 rpm (22,000 × g). The procedure for dealing with insoluble inclusion-body proteins now diverges from that for purifying soluble proteins (UNIT 6.2).

Prepare washed pellets 6. Carefully pour off the supernatant from the pellet. Using a tissue homogenizer, suspend the pellet with 4 to 6 ml wash buffer per gram wet weight cells. Complete homogenization of the pellet is important to wash out soluble proteins and cellular components. Removal of cell wall and outer membrane material can be improved by increasing the amount of wash solution to 10 ml per gram cells. The concentration of urea and Triton X-100 in the wash buffer can be varied. The urea concentration is usually between 1 and 4 M; higher concentrations may result in partial solubilization of the recombinant proteins. The usual detergent concentration is 0.5% to 5%. Triton X-100 will not solubilize inclusion body proteins; it is included to help extract lipid and membrane-associated proteins.

7. Centrifuge the suspension 30 min at 22,000 × g (12,000 rpm in JA-14), 4°C. Discard supernatant and, using the tissue homogenizer, suspend the pellet in 4 to 6 ml wash buffer per gram wet weight of cells. 8. Repeat step 7 two more times. If the supernatant is still cloudy or colored, continue washing the pellet until the supernatant is clear.

9. Suspend the pellet with wash buffer minus the Triton X-100 and urea, using 4 to 6 ml buffer per gram wet cells. Centrifuge 30 min at 22,000 × g (12,000 rpm in JA-14), 4°C. Preparation and Extraction of Inclusion Bodies

The final wash removes excess Triton X-100 from the pellet.


If necessary the washed pellets can be stored at −80°C. It is better to store material at this stage rather than after the extraction stage (see comments to step 13).

Extract recombinant protein from washed pellets with guanidine⋅HCl 10. Using the tissue homogenizer, suspend the pellet with guanidine⋅HCl-containing extraction buffer. Use 0.5 to 1.0 ml buffer per gram wet weight of original cells if the extract will be subjected to gel filtration, and 2 to 4 ml buffer if the extract will be used in protein folding procedures. Perform this step at room temperature. To estimate the amount of recombinant protein in the washed pellets, use the following guidelines. (1) An expression level of 1% corresponds to ∼1 mg recombinant protein per 1 g wet cells. (2) The recovery of highly aggregated recombinant protein in the washed pellets is ∼75% that originally present in the cells. (3) About 60% of the total washed pellet protein is recombinant-derived. Thus, if 50 g cells is processed and the expression level is 5%, the washed pellets contain ∼200 mg recombinant protein. The total amount of recombinant-derived protein in washed pellets can be directly determined by measuring the total protein concentration or by analyzing the washed pellets via SDS-PAGE (see Support Protocol and UNIT 10.1) to determine the proportions of the protein constituents. For gel-filtration purposes, the pellets from 50 g wet weight E. coli cells are solubilized with 40 to 50 ml extraction buffer (see Basic Protocol 2); the concentration of recombinant protein in the extract will be 4 to 5 mg/ml. For direct protein folding (UNIT 6.5), the pellets are extracted with 100 to 200 ml buffer, and the concentration of recombinant protein 1 to 2 mg/ml. If the washed pellet is heavily contaminated with outer cell wall and peptidoglycan material, the extract must be diluted further with extraction buffer (usually 1:1 to 1:3) to reduce the viscosity before it can be used for chromatography.

11. Centrifuge the suspension 1 hr at 100,000 × g (30,000 rpm in Ti45 rotor in a Beckman Optima XL-90 ultracentrifuge), 4°C. For volumes 50% recombinant protein and are used as the starting material for purification of the protein of interest by gel-filtration chromatography. Superdex 200 gel-filtration medium, which allows high flow rates, is washed and packed into a column. The column is equilibrated at 4°C and the sample is applied. Assay of column fractions by gel electrophoresis in the presence of SDS is complicated by the fact that guanidine⋅HCl forms a precipitate with SDS. Therefore, preparing samples for gel analysis involves selective precipitation of protein from guanidine⋅HCl prior to



SDS-PAGE (see Support Protocol). The purified (or partially) purified protein is used as the starting material for procedures (e.g., UNIT 6.5) in which the denatured protein is folded into a native and biologically active structure. Materials Gel-filtration medium: Superdex 200 PG (preparative grade; Pharmacia Biotech) 5% (v/v) ethanol Gel-filtration buffer (see recipe) Guanidine⋅HCl extract of E. coli cells containing the protein of interest (see Basic Protocol 1) 4- to 6-liter plastic beaker Chromatography column: Pharmacia Biotech XK 16/100, 26/100, or 50/100 Packing reservoir: Pharmacia Biotech RK 16/26 (for 16- and 26-mm-i.d. columns) and RK 50 (for 50-mm-i.d. column) Chromatography pump: Pharmacia Biotech P-6000 or P-500 Injection valve (to select between sample loop and pump) UV monitor and fraction collector Sample loop (volume determined by size of column) NOTE: The various components of the chromatography system (pumps, valves, monitors, and sample loops) listed separately above are supplied as components of the BioPilot chromatography system (Pharmacia Biotech), which is used to run the XK 50/100 column. The smaller XK columns (2.6 and 2.5 cm i.d.) are run using the FPLC chromatography system (also from Pharmacia Biotech), which is designed for small- to mediumscale work. For further details on this equipment see the manufacturer’s literature (e.g., Process Products, Pharmacia Biotech). NOTE: Perform steps 1 to 11 at room temperature. After the column is packed, equilibrate and elute at 4°C. Pack the column 1. Wash the gel-filtration medium in a large plastic beaker with 5% ethanol. Let the medium settle and adjust the volume of liquid to give a gel slurry concentration of 65% to 75%. The XK 16/100, 26/100, and 50/100 columns are 100 cm long and have inner diameters of 16, 26, and 50 mm, respectively. Hence, for an XK 50/100 column, column volume = radius (2.5 cm)2 × 3.1416 × bed height (97 cm) ≅ 1900 ml, and ∼2 liters preparative-grade Superdex 200 is required. To pack this column, the gel medium is suspended in 5% ethanol to give a total volume of 3 liters which corresponds to ∼70% gel slurry (it should be noted that the RK 50 reservoir has a capacity of 1 liter, so the 3 liters of gel slurry can be poured in a single operation).

2. Fix the chromatography column in an upright position, using a level to adjust the position. Attach the packing reservoir. 3. Add sufficient 5% ethanol to displace the air from a few centimeters of the bottom of the column. Clamp off the bottom of the column. 4. Gently mix the gel-filtration medium in the plastic beaker to an even slurry of 70% medium suspended in 5% ethanol. 5. Degas the suspension 5 to 10 min using a vacuum flask and laboratory vacuum. Preparation and Extraction of Inclusion Bodies

The ethanol is included to reduce the surface tension and density of the solvent, thus allowing air bubbles that form to rise to the surface more quickly.


6. Carefully pour the slurry of medium into the column, introducing material along the side of the column to avoid creating air bubbles. 7. Let the column stand 5 min and then unclamp the bottom of the column. 8. Attach the chromatography pump to the packing reservoir and pump 5% ethanol (degassed) into the column at an appropriate flow rate (based on manufacturer’s instructions). Pack the column at a pressure greater than the pressure at which the column will be run (up to twice as high), but not greater than the maximum pressure rating of the column. The XK 50/100 column (rated to 0.5 MPa) is packed at ∼20 to 30 ml/hr and ∼0.4 MPa.

9. After the medium has settled, turn off the pump and close the bottom of the column. Pipet fluid from the reservoir and remove the reservoir. Once the column has been packed, be careful to prevent air from entering the column bed. Air will disturb the bed and reduce the column separation resolution.

10. Attach the column top adapter to the column. Place the top of the adapter onto the top of the packed medium and gently compress the medium. 11. Reattach the pump to the column and wash the column with water at a flow rate that will generate the maximum pressure to be used. If the medium continues to settle, readjust the top adapter to maintain a firm fit against the gel. From this point onward, perform all steps at 4°C. Equilibrate the column 12. Equilibrate the column with at least 1 column volume of gel-filtration buffer. Although the proteins were extracted with buffer containing 8 M guanidine⋅HCl (see Basic Protocol 1), the gel-filtration buffer contains only 4 M guanidine⋅HCl. The concentration is reduced to allow faster flow rates and for reasons of economy. Most proteins remain unfolded at the lower guanidine⋅HCl concentration. If, however, the protein elutes in an anomalous manner (e.g., in more than one peak or at an elution position not consistent with its size), and assuming there is adequate reducing agent present, then try increasing the guanidine⋅HCl concentration in the gel-filtration buffer.

13. Measure the actual flow rate while running the column at a flow rate that generates a back pressure about one-half of that generated when packing the column (step 8). For an XK 50/100 column packed using Superdex 200 at 0.4 MPa, a running pressure of ∼0.2 MPa is used, which generates flow rates of 5 to 10 ml/min that are equivalent to linear flow rates of 15.3 to 30.6 cm/hr. The linear flow rate equals the flow rate (ml/hr)/cross-sectional area (cm2). At these flow rates it takes between 3 and 6 hr to complete the chromatography.

14. Connect tubing from the end of the column to the UV monitor and the fraction collector. Apply the sample 15. Load the sample loop with the guanidine⋅HCl extract to be separated. Avoid loading a sample volume >5% of the total column volume; the optimum sample size is 2% (∼40 ml for the XK 50/100 column). The sample consists of washed pellets extracted with guanidine⋅HCl (see Basic Protocol 1). A sample size of 40 to 50 ml is usually derived from ∼50 g wet weight cells. With smaller sample sizes, use columns with proportionally smaller diameters (e.g., XK 16/100 or 26/100 columns). If purchase of only one column is possible, a 2.5 × 100–cm size is a good compromise for variable sample loading.



16. Monitor column effluent with the UV monitor and collect fractions with the fraction collector. For an XK 50/100 column, collect 15- to 20-ml fractions in 16 × 20–mm culture tubes. The eluent from the column is usually monitored at 280 nm or, if the protein has a particularly low extinction coefficient, at 230 nm (guanidine⋅HCl strongly absorbs below 225 nm). For an XK 50/100 column, fractions need only be collected after ∼500 ml of elution. The excluded volume (void volume) is ∼570 ml. Run one column volume (1900 ml) to ensure all of the load material is eluted from the column.

17. Prepare the fractions to be assayed for SDS-PAGE (see Support Protocol and UNIT 10.1). SUPPORT PROTOCOL

PREPARATION OF SAMPLES CONTAINING GUANIDINE HYDROCHLORIDE FOR SDS-PAGE Because guanidine⋅HCl forms a precipitate with SDS, it is necessary to remove the former before carrying out SDS-PAGE. Protein in column fractions is separated from guanidine⋅HCl by precipitation using 90% ethanol (Pepinsky, 1991). Materials Sample containing the protein of interest 100% ethanol, 0° to 4°C 1× SDS sample buffer (UNIT 10.1) Gilson Pipetman (Rainin Instrument) Additional reagents and equipment for gel electrophoresis (UNIT 10.1) 1. Pipet 25 µl sample containing the protein of interest into a 1.5-ml microcentrifuge tube. 2. Add 225 µl cold (0° to 4°C) ethanol to the sample in the tube. The final ethanol concentration is 90% by volume.

3. Mix the sample and ethanol well. Chill 5 to 10 min at −20°C or colder (e.g., −80°C). 4. Microcentrifuge the sample 5 min at maximum speed (∼15,000 × g), 4°C. Carefully withdraw the supernatant and retain the pellet. The pellet may be difficult to see. Be careful not to draw the pellet out of the microcentrifuge tube with the supernatant.

5. Suspend the pellet with 250 µl cold 90% (v/v) ethanol. Mix thoroughly using a vortex mixer. The 90% ethanol is made by mixing 225 ìl ethanol and 25 ìl H2O.

6. Microcentrifuge the sample 5 min at maximum speed, 4°C. Carefully pipet off the supernatant and suspend the pellet in 25 µl of 1× SDS sample buffer. Some proteins are more difficult than others to suspend from an ethanol precipitate. Electrophoresis sample buffer containing 8 M urea is helpful for such proteins (UNIT 10.1). Sonication with a microtip probe can also be used to disperse the sample. A volume of sample buffer >25 ìl may be required in this case (e.g., 50 ìl), and great care must be taken to prevent foaming of the sample caused by excessive sonication power. Preparation and Extraction of Inclusion Bodies


7. Heat the sample 3 to 5 min at 90° to 100°C. Load on an SDS-polyacrylamide gel (UNIT 10.1). REAGENTS AND SOLUTIONS Use Milli-Q-purified water or equivalent in all recipes and protocol steps. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Extraction buffer 50 mM Tris⋅Cl, pH 7.0 5 mM EDTA 8 M guanidine⋅HCl (764 g/liter) 5 mM DTT (770 mg/liter) If the buffer is cloudy, filter through a 0.45- to 0.5-µm filter (the solution should be clear if high-quality guanidine⋅HCl—e.g., ultrapure grade, ICN Biomedicals—is used; see APPENDIX 3A). Buffer can be stored minus DTT at least 1 month at 4°C. Gel-filtration buffer 50 mM Tris⋅Cl, pH 7.5 4 M guanidine⋅HCl (382 g/liter; ultrapure, ICN Biomedicals) 5 mM DTT (770 mg/liter) Buffer can be stored minus DTT at least 1 month at 4°C. Filter (as for extraction buffer; see recipe) and degas before use. Higher concentrations of guanidine⋅HCl (up to 8 M) may be required for some proteins (see comment at step 12).

Lysis buffer 100 mM Tris⋅Cl, pH 7.0 5 mM EDTA 5 mM DTT (770 mg/liter) 5 mM benzamidine⋅HCl (780 mg/liter) Prepare immediately before use The Tris⋅Cl and EDTA are diluted from concentrated stock solutions. The other components are added to the diluted buffer before use.

Wash buffer 100 mM Tris⋅Cl, pH 7.0 5 mM EDTA 5 mM DTT (770 mg/liter) 2 M urea (120 g/liter; ultrapure, ICN Biomedicals) 2% (w/v) Triton X-100 (20 g/liter; Calbiochem-Novabiochem) Add DTT, urea, and Triton X-100 to the other components directly before use. Prepare this buffer in two forms: one with and one without the urea and Triton X-100 (the latter for use in Basic Protocol 1, step 9).



COMMENTARY Background Information The decision of whether to work with insoluble recombinant protein or to put more effort into generating soluble protein (e.g., by modifying the expression vector or changing the host strain and fermentation conditions) can be dictated by the nature of the protein. A small protein (10 to 17 kDa) with only one or two cysteine residues might be expected to fold in reasonable yield from extracted inclusion bodies. Larger proteins (>25 kDa) with many cysteine residues may be more problematical, and lower folding yields can normally be expected. In the latter case, if only small amounts of material are needed then yield is not such an important issue. It should be emphasized that, unless proved otherwise, a protein folded from insoluble inclusion bodies can be expected to have the same structural and conformational integrity as the same protein directly purified from soluble extracts (also see UNIT 6.1). It is similarly true that a purified soluble protein can be denatured and renatured (reversible denaturation) without structural or conformational modifications (reviewed by Anfinson, 1973; Ghelis and Yon, 1982).

Preparation and Extraction of Inclusion Bodies

How inclusion bodies are formed Inclusion bodies are noncrystalline, amorphous structures; however, there is some evidence that the constituent densely packed proteins may have nativelike secondary structures (Oberg et al., 1994). This suggests that the aggregates are formed by the association of partially folded protein or so-called folding intermediates. Furthermore, the aggregation appears to take place late, rather than early, in the folding pathway. Although inclusion body formation and its prevention are of great academic and commercial interest, the pragmatic situation is clearly summarized by Seckler and Jaenicke (1992), who state: “That inclusion bodies are aggregates of otherwise intact polypeptides in nonnativelike conformations has been proven repeatedly by the successful refolding of active proteins after dissociation of the aggregates by chemical dissociation.” It must be noted that the work of Oberg et al. (1994) does not invalidate this statement. It is simply a matter of terminology: the protein in aggregates may have nativelike secondary structures, but compared with soluble folded protein it must still be considered nonnative.

The propensity of a protein to form inclusion bodies is not related to the presence of sulfhydryl residues, as proteins without sulfhydryls still form such aggregates. Furthermore, where carefully studied, it has been found that aggregates derived from proteins that in their native state contain disulfides consist mainly of reduced protein (Langley et al., 1987). This also supports the view that inclusion bodies are not aggregates of completely unfolded protein because, if so, random disulfide bond formation would occur despite the reducing environment of the E. coli cytoplasm. Regardless of the sulfhydryl state in the aggregates, once the protein has been solubilized, reducing agents must be included to prevent the formation of nonnative disulfide bonds. Extracting inclusion body protein Proteins are extracted from inclusion bodies using strong protein denaturants. Protein denaturation can be induced by the following solvent conditions or reagents. 1. pH. Protein denaturation occurs because of the ionization of side chains. Generally, proteins retain less residual structure (are more denatured) when exposed to high pH (e.g., >10.5) compared to low pH (1 M) often denature proteins. The denaturing power increases in the following order for anions: SO42− < CH3COO− < Cl− < Br− < ClO4− < SCN−; for cations, the order is (CH3)4N+, NH4+, K+, Na+ < Li+ < Ca2+ < Gdn+ (reviewed by von Hippel and Schleich, 1969). The organic guanidinium ion (Gdn) is included for comparative purposes. Salts have not been widely used for solubilizing inclusion bodies. On the other hand, they are frequently used for selective extraction of extrinsic membrane proteins and in principle should be useful for preextracting inclusion bodies. Denaturing salts such as KSCN or LiBr are sometimes referred to as chaotropes (tending to disorder). 6. Temperature. Thermally induced extraction is rarely used as it often results in irreversible protein denaturation (Zale and Klibanov, 1986). However, denaturants such as guanidine⋅HCl and urea are more effective at elevated temperatures (e.g., 37° to 60°C). Care must be taken when heating urea solutions because of the increased rate of cyanate formation, which will covalently modify amino groups on the protein, especially at pH >6. A comparison of some protein denaturants and their relative effectiveness is described by Pace and Marshall (1980, and references cited therein). Monera et al. (1994) provide an explanation of why guanidine⋅HCl is usually two to three times per mole (2- to 3-fold) more effective than urea at unfolding proteins. The mechanism of protein denaturation is reagent-specific; this topic has been reviewed by Tanford (1968), Ghelis and Yon (1982), and Creighton (1993). The extraction conditions can be empirically determined on a small scale (50% of the total (see Fig. 6.3.1, lanes h and i), only a 2-fold purification is required to obtain pure protein. Protein extracted with guanidine⋅HCl in the presence of reductant will ideally be in a random coil conformation with all sulfhydryl residues in the reduced state. Under such conditions, the order in which proteins elute from a gel-filtration matrix in guanidine⋅HCl can be directly correlated with molecular size (Mann and Fish, 1972). Selection of the proper chromatography resin is critical for success (for detailed discussion, see Critical Parameters and Troubleshooting). The main disadvantage of gel filtration in



guanidine⋅HCl, especially when using some of the (soft) agarose-based resins (e.g., Sepharoses and Bio-Gels), is that flow rates can be very slow due to the high viscosity of the gel-filtration buffer. Basic Protocol 2 uses a Superdex matrix that permits fast flow rates in the presence of high guanidine⋅HCl (or urea). A column size of 5 × 100 cm allows ∼40 to 50 ml guanidine⋅HCl-containing extract to be processed. For smaller sample sizes, columns with proportionally smaller diameters are used. Because the proteins separated by gel filtration in the presence of guanidine⋅HCl are unfolded, they have no biological activity, so the column fractions are usually assayed by SDSPAGE. Direct addition of SDS to samples containing guanidine⋅HCl results in the formation of the insoluble guanidinium salt of dodecyl sulfate. The guanidine⋅HCl must therefore be removed before addition of the sample buffer used for SDS-PAGE analysis (UNIT 10.1). The approach used in Support Protocol 1 takes advantage of the fact that guanidine⋅HCl is soluble in 90% ethanol whereas the protein is not. An alternative is to simply remove the guanidine⋅HCl by dialysis. Small samples (10 to 100 µl) can be dialyzed using one of various microdialyzing systems commercially available (e.g., the Microdialyzer System 100 from Pierce, which allows simultaneous dialysis of up to 12 samples) or a simple homemade device that uses modified microcentrifuge tubes (Falson, 1992).

Critical Parameters and Troubleshooting


Breaking cells and preextracting inclusion bodies Once it has been established that the recombinant protein is insoluble (UNIT 6.1), extra care should be taken to ensure complete cell lysis. If unbroken cells are present in the low-speed pellets, they will leach contaminants when the pellets are extracted with strong protein denaturants. Proper operation of the French press is described in UNIT 6.2. Other major sources of contamination are outer membrane and cell wall material, most of which can be preextracted with detergents such as Triton X-100 (0.5% to 5%) and urea (1 to 4 M). The proper concentration of urea in the wash buffer is determined empirically in small-scale extractions. It is probably safe to use 2 M urea for most aggregates. The effectiveness of the wash process at removing the cell wall and outer membrane material can be improved by

increasing the volume of wash buffer (see Basic Protocol 1, step 6: use 10 ml instead of 6 ml buffer per gram cells). Treatment of the cells with lysozyme prior to homogenization with the French press will also help remove this material (UNIT 6.5, Basic Protocols 1 and 2). Although insoluble proteins are less susceptible than native proteins to proteolysis, inclusion of EDTA and a serine protease inhibitor such as benzamidine⋅HCl or AEBSF is recommended. The choice of cell lysis buffer is not critical; however, the buffer pH should be >6.5, as many soluble E. coli proteins precipitate at slightly acidic pH values. After preextraction, wash the aggregates with buffer alone to remove excess detergent. When attempting to develop a reproducible folding protocol, note that varying amounts of detergent carryover may influence the outcome. In fact, nonionic detergents can be used as cosolvents to aid folding (UNIT 6.1); even so, it is important to know how much detergent is present. Extracting inclusion bodies with guanidine⋅HCl The washed pellets containing the insoluble inclusion body proteins are extracted with guanidine⋅HCl (see Basic Protocol 1, step 10). To obtain complete dissolution of the pellets, some form of mechanical dispersion is often required. Homogenization at room temperature with a tissue grinder (as described) is often adequate; however, sonication should be used if the pellet is especially recalcitrant. Heating the solution will also aid protein solubilization; 10 to 15 min at 50° to 60°C is usually a good starting point. Excess heating, whether direct or caused by sonication without adequate cooling, should be avoided. It is worth remembering that if extraction involves extremes of heat and or pH, such conditions will favor deamidation of Asn and may also promote cleavage at the labile Asp-Pro bond and other chemical modifications of the protein (Zale and Klibanov, 1986). It is best to use the extract directly after preparation rather than storing it frozen. In general, an unfolded protein in solution is much more susceptible to proteolytic degradation and chemical modification than its native (folded) counterpart. Selecting medium and column for gel filtration in guanidine⋅HCl Selecting the proper gel-filtration resin is one of the most critical steps. First of all, the


Table 6.3.1 Gel-Filtration Matrices Suitable for Use with Solutions Containing Guanidine Hydrochloride

Mass range (kDa)

Matrixa

Native proteins Sepharose CL-6B Bio-Gel A-5m Sepharose CL-4B Sephacryl S-100 HR Sephacryl S-200 HR Sephacryl S-300 HR Sephacryl S-400 HR Superdex 75 Superdex 200

Unfolded proteinsb

10-4,000 10-5,000 60-20,000 1-100 5-250 10-1,500 20-8,000 3-70 10-600

1-80 1-80 10-300 100c 4 M) guanidine⋅HCl concentrations.

4. Oxidation of sulfhydryl residues. The formation of intermolecular disulfide bonds is indicated by the appearance of protein eluting earlier than expected and corresponds to the formation of multimers (dimers, trimers, etc.). Disulfide interchange occurs more readily at high pH; thus, use of a slightly acidic buffer (pH 6.0 to 5.0) and inclusion of fresh reductant in both the sample and column buffer are recommended. For analytical separations, cysteine residues are normally capped by alkylation with iodoacetamide; cysteines can also be reversibility modified by sulfitolysis (Glazer et al., 1975). 5. Proteolysis. (a) The sample contains partially proteolyzed protein, with proteolysis by E. coli proteases occurring before or after cell breakage: In fully denatured protein, it should be possible to separate clipped protein from unmodified protein if there is sufficient mass difference. Once the protein has been separated by gel filtration, it should be checked by mass spectrometry to confirm its integrity. If the protein is not fully denatured, clipped protein may have similar elution properties as unmodified protein; however, the processing will be revealed when the protein is boiled in SDS and analyzed by gel electrophoresis. (b) Where the recombinant protein is itself a protease: As pointed out by Fish et al. (1969), “Since denaturation is a kinetic process, the dissolution of proteases in guanidine⋅HCl might result in a condition where the fraction of the protein not denatured at a given instance may digest the unfolded protein leading to low molecular weight estimate.” The converse of this situation is also true: if proteases are extracted under conditions where they are fully denatured and, thus, inactive (e.g., 8 M guanidine⋅HCl), changes in the solvent conditions (e.g., reducing the guanidine⋅HCl concentration or exchanging for urea) might lead to a fraction of the protein folding into an active conformation. If autolytic processing is suspected, include an appropriate inhibitor or choose a solvent pH where the enzyme has minimal activity.

Ancipitated Results Cell lysis and preparation of washed pellets Pelleted aggregates after washing contain ∼30% dry weight, of which 90% is protein. SDS-PAGE of a typical washed pellet preparation (Fig. 6.3.1, lanes h and i) indicates that recombinant bovine growth hormone (21 kDa) makes up >60% of the total protein. The


a

b

c

d

e

f

g

h

i

Figure 6.3.1 Analysis by SDS-PAGE of fractions from low-speed centrifugation of E. coli cell lysates containing aggregated bovine growth hormone. A 12.5% acrylamide gel of dimensions 12 cm × 16 cm × 1.5 mm was used with the Laemmli buffer system (UNIT 10.1). Lanes a and g contain standard proteins (low-range standards, Bio-Rad) in order of increasing migration distance: phosphorylase b (97.4 kDa), bovine serum albumin (66.2 kDa), hen egg white ovalbumin (45 kDa), bovine carbonic anhydrase (31 kDa), soybean trypsin inhibitor (21.5 kDa), and hen egg white lysozyme (14.4 kDa). After low-speed centrifugation of the clarified lysate and of the washed pellet homogenate (see Basic Protocol 1, steps 5 and 7), the supernatants will be cloudy (lane f) and the pellets usually consist of two layers (see Fig. 6.1.5). The bottom layer is inclusion body protein plus unbroken cells (lanes b and c) and the top layer consists of outer membrane and peptidoglycan fragments (lanes d and e). The outer membrane proteins OmpA (35 kDa) and OmpF/C (38 kDa) are indicated by ω and ο, respectively. After the washing steps, the growth hormone (marked β, 21 kDa) is the major constituent (lanes h and i) together, in this example, with another plasmid-encoded protein, namely kanamycin phosphotransferase (marked α, 30.8 kDa), the product of the gene conferring resistance to the antibiotic kanamycin.

washed pellets analyzed in Figure 6.3.1 are typical starting materials for the protein folding and purification described in UNIT 6.5 (Basic Protocol 1). The multiple bands in the background are either derived from unbroken cells (most likely) or are E. coli cytoplasmic proteins coprecipitated or trapped during aggregate formation. The gel analysis indicates the presence of another plasmid-encoded protein, namely, kanamycin phosphotransferase (30.8 kDa), a product of the gene conferring resistance to the antibiotic kanamycin (Kane and Hartley, 1991). The washing procedure described only partially extracts the phosphotransferase but removes most of the outer membrane proteins. For further details, see the legend to Figure 6.3.1.

Gel filtration in guanidine⋅HCl Fractionation of human immunodeficiency virus type 1 (HIV-1) protease illustrates expected results from gel-filtration chromatography. The protein is expressed in E. coli as a 17-kDa precursor that at some stage undergoes autolytic processing to form the mature-sized protease (10.7 kDa). The protease and undigested precursor accumulate as insoluble inclusions in the E. coli cytoplasm. Washed pellets prepared and extracted with 6 M guanidine⋅HCl as described (see Basic Protocol 1) were applied to a Superdex 200 column. The protease eluted in a single peak (Fig. 6.3.2, fractions 66 to 72), well separated from unprocessed protein (e.g., fraction 60) and larger molecular mass proteins (e.g., fractions 50 to 55). The fractions



2.00

A280

1.50

S 50 55 60 65 70 75 1.00

0.50

P 0.00 0

20

40

60

80

100

Fraction number

Figure 6.3.2 Gel filtration using Superdex 200 in 4 M guanidine⋅HCl. Column dimensions, 6 × 60 cm; buffer, 50 mM Tris⋅Cl (pH 7.5)/4 mM guanidine⋅Cl/2 mM DTT; flow rate, 5 ml/min (300 ml/hr).The sample was an extract containing HIV-1 protease, which has a mass of 10 kDa. Protein fractions 66 to 72 (pool P) was further purified under the same conditions using a Superdex 75 matrix. The inset shows SDS-PAGE analysis of selected fractions. The protein markers (lane S) correspond to standards with mass values of 66.2, 45, 30, 21.5, and 14.4 kDa, respectively (migration order top to bottom).


indicated in Figure 6.3.2 by P were pooled and the protein further purified by repeat chromatography on a Superdex 75 column under the same conditions. The protein at this stage is >95% pure and after solvent exchange is folded into active protein. (As it happens, the native protease is a 20-kDA homodimer that is very susceptible to autolytic digestion; hence, purification in the denatured, inactive state is highly advantageous). In the inset of Figure 6.3.2, analysis by SDS-PAGE of fraction 60 illustrates incomplete dissociation of the sample, a problem commonly observed when electrophoresing samples precipitated with ethanol and trichloroacetic acid (TCA). If necessary, repeat the analysis using SDS sample buffer (UNIT 10.1; see Support Protocol). The fractionation of HIV-1 protease is somewhat of a best-case scenario due to the small size of the protein. Usually it is not possible to purify the recombinant protein completely, especially for proteins close in size to the main

E. coli contaminants (30 to 45 kDa). In general, small proteins (80% of the protein in the inclusion bodies was already in the reduced state (P.T.W., unpub. observ.). In the extraction buffer used here, therefore, only the relatively weak reductant glutathione is included. This compound serves to maintain thiols in the reduced state, and is a component of the redox buffer that will be used later to oxidize the protein. It cannot, however, be assumed that other inclusion-body proteins contain only free thiols.

Fold protein 6. Dilute the clear amber-colored solution with an equal volume of BGH folding buffer A and pour into prewashed dialysis tubing. Use two or three pieces of tubing and fill dialysis bags only three-quarters of the way to allow for any volume increase during dialysis. The BGH concentration from step 5 should be 2 to 4 mg/ml, and after dilution should not exceed 1.0 to 2.0 mg/ml. If higher concentrations are found, the solution must be diluted further with folding buffer A. BGH concentration can be estimated by diluting the sample with 4 M guanidine⋅HCl in water and measuring the absorbance at 280 nm and 260 nm in a cell with a 1-cm path length. The total protein concentration (mg/ml) is estimated as 1.55 A280 − 0.76 A260 nm (Stoscheck, 1990). The BGH content may either be assumed to be 60% of the total protein or, more accurately, be estimated by performing SDS-PAGE and densitometry using the washed pellet from step 4. Urea is included in folding buffer A as a cosolvent to maintain solubility of the protein during refolding. Removal of guanidine⋅HCl by dialysis or dilution results in precipitation if no cosolvent is used. The urea concentration chosen (in this case 4 M) should be low enough to allow the native structure to form (see UNIT 6.4). Urea unfolding/folding profiles for BGH (i.e., equilibrium-denaturation curves in which protein conformation is measured as a function of denaturant concentration; see UNIT 6.4) were available in the literature prior to development of this method (Edelhoch and Burger, 1966). The urea concentration that induces protein unfolding can be determined rapidly by urea gradient electrophoresis (Goldenberg, 1989).

7. Dialyze solution 12 to 16 hr (or overnight) against 5 liters of BGH folding buffer A, then dialyze an additional 6 to 8 hr against 5 liters of BGH folding buffer B. The second dialysis can be continued overnight if necessary.



a

b

c

d

e

f

g

Figure 6.5.1 SDS-PAGE of bovine growth hormones on 12.5% polyacrylamide gel. Lane A, BGH-expressing E. coli cells minus the expression vector; lanes B and C, BGH-expressing E. coli cells with ∆A-4 and A-9 BGH expression vectors, respectively; lane D, purified recombinant A-9 BGH; lane E, BGH purified from pituitary (supplied by A.F. Parlow, UCLA); lane F, purified recombinant A-9 BGH with no reductant in sample buffer; lane G, BGH purified from pituitary with no reductant in sample buffer. In lane E (pituitary BGH), the two bands correspond to full-length protein and protein truncated at the N-terminus by 4 residues. It can be seen that the bottom band has the same mobility as E. coli extracts containing the ∆A-4 BGH construct. In lane G, it it may be noted that the two bands are not resolved under nonreducing conditions.

Oxidation of the protein during dialysis can be monitored by SDS-PAGE (UNIT 10.1). SDS-denatured oxidized BGH is more compact than SDS-denatured reduced BGH and thus migrates faster as a result of the lower apparent molecular weight (i.e., 18 kDa for the oxidized form versus 22 kDa for the reduced form). SDS-PAGE band patterns of oxidized and reduced BGH are shown in Figure 6.5.1. Any free thiol groups in the sample are quenched by addition of 20 mM iodoacetamide. SDS sample buffer (UNIT 10.1) minus reductant is then added. The pH of the SDS-treated sample may have to be readjusted with dilute alkali. The proportion of oxidized versus reduced protein is finally determined by densitometry of the Coomassie blue–stained gel. It should be noted that this approach does not prove whether or not the correct disulfide bond(s) have been formed. The shift in mobility upon reduction occurs because of the formation of the disulfide bond linking Cys-51 to Cys-163. BGH in which the second disulfide bond (linking Cys-180 to Cys-188) has been selectively reduced (Graf et al., 1975) still exhibits the gel shift. Despite these potential pitfalls, the gel method is useful and correlates with other approaches—e.g., direct monitoring of disulfide formation using 2-nitro-5-thiosulfobenzoate in the presence of sodium sulfite after the free sulfhydryls have been quenched and the buffer components removed (Thannhauser et al., 1984). Sulfhydryl groups can be assayed using Ellmans reagent (Riddles et al., 1979).

8. Centrifuge the slightly cloudy solution 30 min at 20,000 × g. Discard pellet and adjust supernatant to pH 9.0 with 2 M HCl. The volume of the supernatant should be 300 to 320 ml.

Folding and Purification of Insoluble Proteins from E. coli

Purify protein by ion-exchange and gel filtration chromatography 9. Pack a 5 × 50–cm column with DEAE Sepharose CL-4B to give a bed height of ∼10 cm and equilibrate with BGH column buffer A. Apply the solution from step 8 at a flow rate of 60 ml/hr and begin collecting 15-ml fractions. See UNIT 8.2 for further information on preparing an ion-exchange column.


10. Elute with BGH column buffer A at the same flow rate and continue fraction collection until A280 or A260 of eluant approaches a baseline value. Assay fractions by SDS-PAGE and pool fractions containing BGH. Under the ion-exchange conditions used, BGH does not bind to the matrix and is located in the flowthrough fractions. The more acidic E. coli contaminants bind tightly to the top portion of the column, which turns brown. Some of the earlier flowthrough fractions may be slightly contaminated with aggregated protein that separates from soluble protein as a result of the gel-filtration effect of the matrix. A BGH concentration of 0.2 to 0.3 mg/ml in ∼400 to 450 ml of pooled eluant should be obtained. This should exhibit a single band on the SDS-PAGE gel. Pooled fractions can be stored 1 to 2 days at 4°C and for longer periods at −80°C.

11. Concentrate pooled BGH-containing fractions to 50 to 60 ml (i.e., to ∼2 to 3 mg/ml protein concentration) using a stirred cell with Diaflo PM 10 ultrafiltration membrane. Centrifuge filtrate 15 min at 20,000 × g to remove any precipitated protein. 12. Pack a 5 × 100–cm column with Sephadex G-100 to a bed height of 95 cm and equilibrate with BGH column buffer B. Apply the filtrate from step 11 at a flow rate of 40 ml/hr and begin collecting 15-ml fractions. See UNIT 8.3 for additional information on preparing gel-filtration columns. If a 5 × 100–cm column is not available, a 2.5 × 100–cm column can be used. Depending on the amount of sample, the smaller-diameter column may be run several times, loading 18 to 20 ml of sample per run.

13. Elute with BGH column buffer B at the same flow rate and continue fraction collection until A280 or A260 of eluant approaches a baseline value. Assay fractions by SDS-PAGE and pool fractions containing BGH. The protein elutes in a single but slightly asymmetrical peak. The gel-filtration elution peak has a sharp front edge, whereas the descending portion is more diffuse and trailing. This elution behavior results from the fact that BGH is a rapidly associating/dissociating monomer/dimer system (with a Kd of ∼0.8 to 1.0 × 10−5 M; see Ackers, 1970, for discussion of theory). The apparent molecular weight estimated by gel filtration is 30 to 35 kDa (with a monomer mass of ∼22 kDa). The slightly alkaline pH of the column buffer and folding buffers is required to maintain solubility of the protein.

14. Repeat step 11, then filter sterilize filtrate using a Millex-GV 0.22-µm filter unit. Store purified BGH in aliquots at −80°C. For long-term storage the protein can be lyophilized. If this is to be done, the pooled fractions from step 10 should be directly dialyzed (APPENDIX 3B) against 0.1 M ammonium bicarbonate (pH 9.2 to 9.4, adjusted with ammonium hydroxide), and the dialysate filter-sterilized and freeze-dried. If an essentially salt-free protein is required, two cycles of freeze-drying should be performed and the protein reconstituted with water alone after the first drying cycle. The concentration of the purified BGH can be conveniently determined by UV absorbance measurement—i.e., 1 mg/ml native BGH has an A280 of 0.7 in a 1-cm quartz cuvette. Protein concentration may also be estimated in crude extracts using the Bio-Rad Protein Assay Kit (based on the Bradford method) and in partially purified fractions as described in the annotation to step 6. Biological activity of the protein is measured as described in Wingfield et al. (1987a). Purification of Recombinant Proteins


BASIC PROTOCOL 2

FOLDING AND PURIFICATION OF HUMAN INTERLEUKIN 2 Human interleukin 2 (hIL-2) is a hydrophobic, acid-stable polypeptide with a molecular weight of 15 kDa. Recombinant hIL-2 expressed in E. coli is deposited in inclusion bodies. Extraction of the inclusion bodies with denaturants followed by folding in aqueous solution often results in formation of insoluble hIL-2 aggregates. In this protocol, inclusion bodies containing recombinant hIL-2 are first extracted with acetic acid and the soluble monomeric hIL-2 is separated from other hIL-2 aggregates by gel-filtration chromatography. The soluble protein is then folded and oxidized by dialysis against water. Correctly folded hIL-2 is finally separated from unfolded hIL-2 and other E. coli contaminants by reversed-phase high-performance liquid chromatography (RP-HPLC). Materials hIL-2 break buffer (see recipe), 4°C E. coli cells expressing hIL-2: ∼20 g wet weight from a 3-liter fermentation (UNIT 5.3), and stored as a flattened paste in a sealed polyethylene bag at −80°C Sucrose Lysozyme (Worthington) hIL-2 wash buffer: 0.75 M guanidine⋅HCl/1% (w/v) Tween 20 (prepare immediately before use), 4°C PBS (APPENDIX 2E), 4°C 10% and 20% (v/v) acetic acid, 4°C (prepare fresh from glacial acetic acid) Sephadex G-100 gel-filtration resin (Pharmacia Biotech) Acetonitrile (HPLC grade) Trifluoroacetic acid (TFA; HPLC grade) 7 × 250–mm 300-Å octyl Aquapore RP-300 semiprep column (Brownlee column; Thomson Instrument) RP-HPLC solvent A (see recipe), room temperature RP-HPLC solvent B (see recipe), room temperature 25 mM acetic acid, 4°C Tissue homogenizer (e.g., Polytron, Brinkmann) 30°C water bath Sorvall RC-5C centrifuge with SS-34 rotor (or equivalent) 2.6 × 100–cm glass chromatography column Spectra/Por 3 dialysis tubing, 11.5- and 45-mm diameters (MWCO 3500; Spectrum) Sterivex-GS 0.22-µm filter units (Millipore) HPLC system with pumps, UV detector, and fraction collector (Waters) Additional reagents and equipment for cell breakage using a French press (UNIT 6.2), gel-filtration chromatography (UNIT 8.3), and dialysis (APPENDIX 3B) Break cells 1. Add 60 ml hIL-2 break buffer to 20 g (wet weight) of E. coli cells expressing hIL-2 in a 250-ml glass beaker. Mix well using tissue homogenizer. UNIT 6.2

describes use of a tissue homogenizer to resuspend E. coli cells.

2. Add 21 g sucrose and mix well with cell paste using tissue homogenizer. Add 34 mg lysozyme and mix again using tissue homogenizer. Incubate 30 min in a 30°C water bath, then dilute with 100 ml hIL-2 break buffer and cool on ice. Folding and Purification of Insoluble Proteins from E. coli

3. Break cells by passing suspension through a French press twice as described in UNIT 6.2. Centrifuge 30 min at 13,000 × g (10,400 rpm in SS-34 rotor), 4°C. Save pellet. The wet weight of the pellet should be ∼1.5 g.


4. Wash pellet by resuspending in 30 ml hIL-2 wash buffer and centrifuging 30 min at 13,000 × g. Repeat wash once with hIL-2 wash buffer and again using 30 ml PBS in place of the wash buffer. Washed pellets should be used immediately or stored at −70°C until required.

Extract protein with acid and prepare gel-filtration column 5. Add 12.5 ml of 20% acetic acid to the washed pellet and mix using tissue homogenizer. Centrifuge 30 min at 13,000 × g, 4°C, and save the supernatant. 6. Pack a 2.6 × 100–cm column with Sephadex G-100 to give a bed height of 95 cm and equilibrate with 10% acetic acid. For convenience the column may be packed the day before use.

7. Repeat step 5 and pool the two clear supernatants (total volume should be ∼25 ml). Isolate hIL-2 monomer from solubilized protein by gel filtration 8. Immediately apply pooled supernatants to column prepared in step 6. Perform gel filtration chromatography at 4°C as described in UNIT 8.3 and construct a chromatogram, pooling the fractions that make up the third peak. The solution should be applied to the column immediately after extraction to prevent any covalent modification of the protein. The column eluate usually shows three major peaks in A280. The first peak (in the void volume) contains aggregates, the second peak contains dimeric hIL-2, and the third peak contains monomeric hIL-2 (S.M.L., unpub. observ.). A T cell proliferation assay shows that the third peak contains the highest biological activity of hIL-2 (Liang et al., 1986; Bottomly et al., 1991).

9. Using 45-mm diameter Spectra/Por 3 dialysis tubing (MWCO 3500), dialyze the pooled fractions (∼60 ml) making up the third peak at 4°C overnight against 5 liters of Milli-Q water, then for an additional 3 to 4 hr against 5 liters of fresh Milli-Q water. The pooled eluate fractions should be extensively dialyzed against water to remove acetic acid. The dialysate should have a volume of ∼60 ml with an hIL-2 concentration of 0.2 mg/ml. After dialysis, the hIL-2 monomer can be stored at −20°C until required. As the acetic acid is removed; sample pH slowly increases, thereby allowing hIL-2 thiols to form disulfide bonds.

Purify protein by RP-HPLC 10. Filter the dialysate through a 0.22-µm filter unit, then add 30 ml acetonitrile (33% v/v final) and 0.09 ml TFA (0.1% v/v final). 11. Prepare and test the HPLC system according to manufacturer’s instructions. Place a 300-Å octyl Aquapore RP-300 reversed-phase chromatography column in line. Other reversed-phase columns; e.g., the 2.2 × 15–cm TMS-250 (TosoHaas) can also be used (Kato, 1985).

12. Fill the reservoirs of the HPLC system gradient maker with RP-HPLC solvent A and RP-HPLC solvent B. Carry out a blank run (at room temperature) with the following gradient program: 0 min: 0% solvent B/100% solvent A 10 min: 0% solvent B/100% solvent A 60 min: 50% solvent B/50% solvent A 90 min: 50% solvent B/50% solvent A 150 min: 70% solvent B/30% solvent A



160 min: 70% solvent B/30% solvent A 170 min: 100% solvent B/0% solvent A 180 min: 100% solvent B/0% solvent A 190 min: 0% solvent B/100% solvent A 210 min: 0% solvent B/100% solvent A. Repeat blank run as necessary to ensure that the baseline profile has stabilized.

13. Pump sample from step 10 into column at a flow rate of 4 ml/min. Run the gradient program described in step 12 at a flow rate of 1 ml/min at room temperature. Collect 8-ml fractions at a rate of 1 ml/min and pool the fractions making up the correctly folded hIL-2 peak, which elutes at ∼70 min. Care should be taken to avoid introducing air bubbles while pumping the sample into the column, as these will generate false peaks in the chromatogram. If the peaks in the HPLC profile are too broad, the gradient program should be varied to improve the separation (see Chapter 8).

14. Using 11.5-mm diameter Spectra/Por 3 dialysis tubing (MWCO 3500), dialyze the pooled hIL-2 fractions against Milli-Q water or 25 mM acetic acid at 4°C. Filter the purified protein solution using a 0.22-µm filter unit and store at −20°C or below. SUPPORT PROTOCOL

RESOLUTION OF NATIVE AND MISFOLDED FORMS OF hIL-2 BY RP-HPLC Human interleukin 2 (hIL-2) contains three cysteine residues—at positions 58, 105, and 125. In the correctly folded (native) protein, the cysteines at positions 58 and 105 form a disulfide bond that is important for biological activity of the protein (Robb et al., 1984; Wang et al., 1984; Liang et al., 1985). As hIL-2 contains three cysteines, three possible intramolecular disulfide–linked forms can exist, only one of which is the native form. RP-HPLC can separate these isomers, and is used to verify that the purified IL-2 contains the native disulfide. Additional Materials (also see Basic Protocol 2) 0.46 × 10–cm SynChropak RP-P C18 column (Thomson Instrument) RP-HPLC solvent C (see recipe) 25 mM acetic acid Purified folded hIL-2 solution (see Basic Protocol 2) 1. Prepare and test HPLC system according to manufacturer’s instructions. Place a 0.46 × 10–cm SynChropak RP-P C18 column in line. 2. Fill the reservoirs of the HPLC system gradient maker with RP-HPLC solvent A and RP-HPLC solvent C. Apply 20 µl of 25 mM acetic acid as a blank sample. Perform a blank run at room temperature using a gradient program starting with 100% solvent A for 5 min, followed by a linear gradient starting at 100% solvent A/0% solvent C and ending at 0% solvent A/100% solvent C, at a flow rate of 1 ml/min. Repeat as necessary to check that the baseline profile is reproducible. 3. Prepare a solution of ∼5 mg/ml purified folded hIL-2 solution in 25 mM acetic acid and filter through a 0.22-µm filter.


4. Apply 20 µl of purified folded hIL-2 solution in 25 mM acetic acid to the column and carry out RP-HPLC at room temperature using the gradient program described in step 2. Collect the fractions containing the properly folded hIL-2.


A

B

A214

0.3 0.2 0.1 0.0 C

D 1

0.3 0.2 0.1 0.0

2

0

40

80

120

0

40

80

120

Time (min)

Figure 6.5.2 Chromatograms illustrating peaks produced by (A) correctly folded hIL-2 (in absence of denaturant); (B) correctly folded hIL-2 (in presence of denaturant but at pH 3.5, which is too low for disulfide-bond exchange); (C) scrambled hIL-2 isomers (resulting from denaturant treatment at pH 8.5); (D) unfolded hIL-2 (resulting from denaturant/reductant treatment).

Properly folded hIL-2 usually elutes at ∼54% acetonitrile and unfolded hIL-2, which has no intramolecular disulfide bond, elutes at 57% acetonitrile. In the presence of denaturants such as guanidine⋅HCl, hIL-2 rapidly scrambles into a mixture of three disulfide-linked isomers. The incorrectly folded hIL-2 elutes at a lower acetonitrile concentration (2.0, dilute with 4 M guanidine⋅HCl in water. The scale of operation at this step is limited by the large dilution of the protein. This step can be scaled up if equipment suitable for concentration of large volumes is available (see annotation to step 12). Otherwise, the step can be repeated if more protein is needed. Protein folding is usually carried out at a low concentration to avoid aggregation, which is thought to arise from partially folded protein in the folding pathway (i.e., folding intermediates; Goldberg et al., 1991; Jaenicke, 1991; also see UNIT 6.4). Because aggregation is an intermolecular event, it is concentration-dependent. In this step the protein concentration is increased very slowly, allowing the protein already in solution to fold before additional unfolded protein is added. It is assumed that folded protein does not participate in aggregation.

12. Concentrate solution to ∼1 mg/ml protein using a stirred cell with Diaflo PM 10 ultrafiltration membrane. Filter concentrated solution (25 to 30 ml volume) through a Millex-GV 0.22-µm filter. If it is not be be used immediately, the concentrated solution can be stored at −80°C for several months. The largest stirred cell produced by Amicon has a 2-liter maximum capacity (i.e., the Model 2000). This cell can be used to concentrate the solution to ∼100 ml, and this must be further concentrated using a smaller cell (e.g., with a 200- or 400-ml capacity). For volumes >2 liters, or for a more rapid process, the Minitan tangential-flow membrane system (Millipore) can be used. This system allows a concentration rate of 0.5 to 1.0 liter/hr.

Cleave histidine tag with thrombin 13. Dilute folded protein from step 12 1:1 with 50 mM Tris⋅Cl (pH 7.5)/10 mM CHAPS (∼0.5 mg/ml final protein concentration). This dilution is carried out because thrombin digestion (in step 14) works best at NaCl concentrations 1 mg ml) and a redox buffer is used. For further information on in vitro folding of BGH and the closely related pGH, consult Brems and Havel (1989) and Bastiras and Wallace (1992). Havel et al. (1989) present a concise and useful review of the some of the spectroscopic methods used to monitor protein conformation and structure (e.g., UV absorption, circular dichroism, and fluorescence) with examples of studies on BGH.

Comparison of authentic and recombinant BGH Natural BGH is purified from bovine pituitaries, and examples of purifications leading to crystallizable protein are described by Bell et al. (1985) and Spitsberg (1987). The method that appears to give the most homogeneous product (Wood et al., 1989) involves extraction of pituitaries with 4.5 M urea (pH 8.8) followed by purifications employing anion- and cationexchange chromatography in the presence of urea. The urea maintains solubility of the protein during purification; at the concentration used (4.5 M) it does not perturb or unfold the protein. The urea is finally removed from the purified protein by dialysis at pH 10. A single bovine pituitary yields ∼16 mg of pure protein by this method. Preparations of the natural protein often show varying degrees of N-terminal heterogeneity, depending on the method of purification. This may involve truncations of either one (∆A- 1) or four (A-4) residues—e.g., (Ala)PheProAlaMetSerLeuSerGlyLeuCys AlaPhe…—where the residue in parentheses is lost in a ∆ A-1 truncation and the four boldface residues are lost in a A-4 truncation. An example of a A-4 truncation is shown in Figure 6.5.1, lane E, where the two closely migrating bands at 23 and 22 kDa correspond to full-length and N-terminal-deleted (A-4) protein, respectively (for further details see Wingfield et al., 1987b, and references therein; also see Langley et al., 1987a). The two forms can be resolved under native conditions by chromatofocusing, which takes advantage of the small difference in their isoelectric points; full-length and A-4 BGH have pI values of 8.2 and 8.0, respectively (Wingfield et al., 1987b). Protein purified by the method of Wood et al. (1989) has a single and correct N-terminus. Recombinant hormones have been expressed that contain the full BGH sequence (e.g., Langley et al., 1987a), as well as others with the ∆A-1 deletion (e.g., Wingfield et al., 1987a; Bogosian et al., 1989) and the A-9 deletion (Wingfield et al., 1987a). Purified A-1 BGH retains the initiating Met at the N-terminus (MetPhePro...), but when the initiating Met is placed in front of the Ala, the Met is removed to give the authentic sequence (AlaPhePro...). Deletion of at least the first nine N-terminal residues does not appear to effect the biological activity of BGH. Another source of chemical heterogeneity in both authentic and recombinant growth hor-



mones is the deamidation of specific Asn and Gln residues (Secchi et al., 1986). This is partly responsible for the multiple band patterns observed in isoelectric focusing gels (Wingfield et al., 1987a; 1987a). In typical authentic and recombinant BGH preparations, ∼20% of the protein is deamidated. However, because these modified forms have lower pIs than unmodified protein, they can be separated by chromatofocusing (Wingfield et al., 1987a). The cationexchange chromatography step (at pH 6.9 in 4.5 M urea) used by Wood et al. (1989) for purification of pituitary BGH also appears to separate deamidated from unmodified protein. Deamidated protein is more positively charged than unmodified protein, and will therefore bind more tightly to a negatively charged cation-exchange matrix. Authentic and recombinant BGH both exhibit self-association with the characteristics of a rapid monomer-dimer equilibrium. The dissociation constant (Kd) for authentic BGH is 6.6 × 10−6 M (Fernandez and Delfino, 1983), which is similar to that determined for both the A-1 and A-9 recombinant forms of BGH (P.T.W., unpub. observ.). At concentrations of 1.0 and 0.1 mg/ml, ∼76% and 42% of the protein will be dimeric, respectively. Under physiological conditions, the monomer appears to be the active species. Monomeric protein binds to two molecules of the extracellular domain of the cell-surface receptor, forming a 1:2 molar complex analogous to the human growth hormone (Staten et al., 1993).


Other approaches used to extract and fold recombinant BGH Langley et al. (1987a) extracted protein from inclusion bodies using 6 M guanidine⋅HCl (pH 8.0) as the denaturant. The extract was allowed to oxidize in air >72 hr at room temperature in the presence of the denaturant. Oxidized protein was fractionated by gel filtration in guanidine⋅HCl, the fractions making up the monomer peak pooled, and the protein folded by removal of denaturant using dialysis. The rationale for this method is that, in a protein with four cysteine residues that form two disulfide bonds (such as BGH), there are three possible intramolecular disulfide linkage combinations. Hence, even if the oxidation proceeds in a random manner as expected for unfolded protein, at least 33% of the protein produced will have the correct pattern. In practice, the recovery of correctly folded oxidized protein was significantly higher than 33%—suggest-

ing that even in 6 M guanidine⋅HCl, BGH may contain some elements of native-like structure. This method is geared toward large-scale protein production and gives good protein recovery. However, the long incubation period in guanidine⋅HCl appears to result in ∼50% of the protein being deamidated (Langley et al., 1987b). Bogosian et al. (1989) used 4.5 M urea at pH 10.7 to extract BGH from inclusion bodies. The extract was then stirred for 48 hr, allowing the protein to oxidize in air. This method is mentioned because it results in formation of a novel concatenated dimer, formed by the interlocking of disulfide loops (Violand et al., 1989). It is most likely that under the conditions used for extraction, the protein folding and oxidation is initiated from partially folded associated protein. Ovine growth hormone (which is closely related to BGH) was folded and purified with an overall recovery of 30% by a procedure similar in approach to Basic Protocol 1 (Wallis and Wallis, 1989). This method, however, used 2 M guanidine⋅HCl instead of urea as the cosolvent to maintain solubility during folding. It also employed air oxidation for 24 hr instead of the redox buffer system. Interleukin 2 Interleukin 2 (IL-2) is a member of the cytokine family. It is composed of a compact core bundle of four antiparallel α-helices (Bazan, 1992; McKay, 1992). IL-2 plays important roles in the proliferation and differentiation of T lymphocytes as well as in regulation of the immune system. The biological effects of IL-2 are mediated via binding to IL-2 receptors (Minami et al., 1993). The human IL-2 (hIL-2) gene has been cloned and expressed in E. coli as insoluble inclusion bodies (Devos et al., 1983). The purification of recombinant hIL-2 has been described and the protein thoroughly characterized (Liang et al., 1985; Kato et al., 1985; Weir and Sparks, 1987). Basic Protocol 2 details the solubilization, refolding, and purification of recombinant hIL-2. This procedure can be scaled up to yield hIL-2 on the scale of hundreds of milligrams. hIL-2 has three cysteines; hence, three possible disulfide-linked forms can exist. The native isomer consists primarily of one form— i.e., the one that contains a disulfide linkage between Cys-58 and Cys-105 (Robb et al., 1984; Wang et al., 1984; Liang et al., 1985).


Reversed-phase-HPLC has been used to separate these isomers (Browning et al., 1987; also see Fig. 6.5.3). HIV-1 integrase An early part of the life cycle of all retroviruses (including the human immunodeficiency virus, HIV) is the integration of a DNA copy of the viral genome into the host chromosome. This step is essential for viral replication. Retroviral DNA integration is carried out by a defined set of DNA cutting and joining reactions, catalyzed by integrase protein (Katz and Skalka, 1994). The HIV-1 integrase is a 288-residue protein (mol. wt. 32,200) which has been expressed in E. coli (Sherman and Fyfe, 1990). The protein is located in the pellet obtained by low-speed centrifugation following cell breakage, but is apparently not highly aggregated into inclusion bodies, as it can be extracted into high-ionicstrength solution (e.g., 1 M NaCl) without a denaturant. The insolubility of the protein is probably a result of nonspecific binding to E. coli nucleic acid. The salt-extraction procedure was originally used by Terry et al. (1988) for extraction of avian sarcoma-leukosis virus integrase expressed in E. coli. Dissection of the HIV-1 integrase by preparation of a series of deletion mutants (Bushman et al., 1993; see also references therein) demonstrated that a central core region of residues 50 to 212 was enzymatically active, carrying out a subset of the reactions catalyzed by the full-length enzyme. This deletion mutant is interesting for structural studies as it has better solubility than the full-length enzyme, which is notoriously difficult to handle as a result of limited solubility in the usual aqueous solvents. The HIV-1 integrase deletion mutant (IN50-212) is expressed in E. coli as a fusion protein with the N-terminal extension sequence GlySerSerGlyHisHisHisHisHisHisSerSer GlyLeuValProArgGlySerHisMet. This sequence (a His tag) contains a six-residue histidine repeat which is responsible for the selective high-affinity binding of the fusion protein to a nickel-chelate column. The boldface portion of the His tag sequence indicates the location of the specific thrombin-cleavage site between Arg-16 and Gly-17, which is exploited in removal of the tag using thrombin. In Basic Protocol 3 the His-tagged IN50-212 is expressed as an insoluble (i.e., inclusionbody-type) protein, and is purified under denaturing conditions, taking advantage of the fact that metal-chelate affinity chromatography

(MCAC) can be carried out in the presence of protein denaturants such as guanidine⋅HCl. Following purification, the protein is folded and the His tag removed by exploiting the specific thrombin-cleavage site of the tag sequence (underlined above). The cleaved protein has the N-terminal sequence GlySerHisMetHisGlyGln.... The Met at position 4 corresponds to Met 50 of the wild-type integrase and residues 1 to 3 (GlySerHis) are derived from the tag. Excess thrombin is removed by affinity chromatography using an immobilized inhibitor (p-aminobenzamidine). In the last step of the purification, the IN50-212 is subjected to gel-filtration to remove any aggregated multimers. The protein produced is 166 residues long, the molecular weight estimated from the DNA coding sequence is 18,200, and the calculated isoelectric point is ∼7. Although the folded deletion mutant has enhanced solubility as compared with the wildtype enzyme, the zwitterionic detergent CHAPS is included in buffers to maintain solubility, especially during protein folding. Following protein purification the detergent can be removed or exchanged for other buffer additives. Substitution of Phe-185 for a Lys residue appears to increase the solubility of IN50-212. This mutant protein, expressed in E. coli with the N-terminal His tag, was extracted with 1 M salt and purified by MCAC. The extraction with high-ionic-strength salt solution in the absence of denaturant is analogous to the procedure used with the full-length protein, discussed above. The crystal structure of this protein has been determined at 2.5-Å resolution. The overall topology consists of five β-sheets flanked by helical regions, showing that the integrase domain belongs to a superfamily of polynucleotidyl transferases that includes ribonuclease H (Dyda et al., 1994).

Critical Parameters Bovine growth hormone (BGH) Protein extraction. The protein should be extracted from the inclusion bodies in a monomeric and fully reduced state. This provides a defined starting point from which to develop a reproducible folding protocol. Based on pilot-scale experiments in which washed pellets (from step 4 of Basic Protocol 1) were extracted with various protein denaturants (P.T.W., unpub. observ.; also see UNIT 6.3), 8 M guanidine⋅HCl was chosen for solubilization of aggregated BGH. The most effective ones were



Supplement 4


urea and guanidine⋅HCl at 6 M to 8 M concentrations, each of which solubilized >80% of the protein. The extracts were further analyzed by gel filtration on Sephacryl S-300 in the presence of the same concentration of the particular denaturant used for the initial extraction. The results indicated that guanidine⋅HCl at concentrations >6 M yielded solubilized protein of which >80% was monomeric. Although 8 M urea also solubilized >80% of the protein, only ∼20% of this protein was monomeric, the remainder consisting of dimers and higher-order aggregates. Protein folding and oxidation. Basic Protocol 1 describes an empirical approach based on some of the principles discussed in UNITS 6.1 & 6.4. The determination of basic physicochemical/conformational properties for a particular protein will take some of the guesswork out of developing a suitable folding protocol. For BGH, purified authentic protein from bovine pituitary glands was available, and had been fairly well characterized before recombinant proteins were produced. This situation is not common, as the natural counterparts of many recombinant proteins are rare and may never have been purified. The optimal ratio of reduced to oxidized glutathione in the folding buffers used in Basic Protocol 1 was empirically determined. However, it is commonly observed that the highest folding yields are obtained when the ratio of GSH/GSSG is between 5 and 10 and the total reduced and oxidized glutathione concentration is in the range of 1 to 10 mM. Deamidation. As mentioned above, BGH is susceptible to deamidation. Although deamidated forms (charge isomers) can be resolved by ion-exchange methods (Wingfield et al., 1987a), their formation should be minimized. This is achieved by limiting the amount of time the protein spends in the fully unfolded state, especially at pH 9.5. For example, the guanidine⋅HCl extract should be processed immediately and diluted as described in step 6 of Basic Protocol 1. The dialysis steps (at pH 9.5) should be performed for the times indicated, not longer. The folded protein is stable and does not appear to deamidate spontaneously on storage. Scaleup. For larger-scale production, a Manton-Gaulin-APV homogenizer (UNIT 6.1) is used to break cells instead of the French press. In addition, a larger Sephadex G-100 column (e.g., 5.0 × 90–cm) is used in step 12 of Basic Protocol 1.

Interleukin-2 Protein purification. Recombinant hIL-2 should be extracted as a monomer (see Basic Protocol 2, step 5). The extraction has been tested with various concentrations of acetic to determine the optimal condition; 20% acetic acid has been found to extract the maximum amount of monomeric hIL-2 from inclusion bodies. However, in the gel-filtration step (see Basic Protocol 2, step 8), 20% acetic acid is too corrosive for the metal parts of the chromatography system (e.g., the stands, pump, and fraction collector); thus 10% acetic acid is recommended. Protein folding. The hIL-2 monomer that elutes from the Sephadex G-100 column (see Basic Protocol 2, step 8) yields two peaks upon analytical reversed-phased chromatography (see Support Protocol). These two peaks represent the refolded, oxidized hIL-2 and unfolded, reduced hIL-2. The proportion of the oxidized form is increased by the dialysis step (see Basic Protocol 2, step 9). The time and volume recommended in the dialysis step have been determined empirically to produce the best yield of oxidized hIL-2. Longer periods of dialysis (e.g. 48 hr) causes do not increase the yield of oxidized hIL-2. Scaleup. The parameters for larger-scale production are the same as those for BGH (see above). HIV-1 integrase MCAC. It is important to use low concentrations of 2-mercaptoethanol in the sample and column buffer (1 mM is safe) and to completely avoid dithiothreitol, as either reductant will strip Ni2+ from the MCAC matrix (see UNIT 9.4). Protein folding. Because HIV-1 integrase is not a very soluble protein and has a high tendency toward self-association and aggregation (Hickman et al., 1994), the solubility and stability of the folded protein must be maintained by including a relatively high salt concentration as well as the detergent CHAPS in the folding buffer and in all subsequent column buffers. The buffer additives indicated for HIV-1 integrase are usually not required for folding of the average protein. Thrombin cleavage. It is necessary to perform pilot-scale experiments to optimize the conditions for removing the His tag using thrombin. The usual parameters to vary are enzyme/substrate ratios and incubation time and temperature. Thrombin is a serine protease, and can be irreversibly inhibited with either PMSF or AEBSF. Removal of the tag can be

6.5.22 Supplement 4


monitored by SDS-PAGE (UNIT 10.1), in which case the thrombin should be quenched with 1 mM PMSF or AEBSF prior to addition of SDS and heating. There is an ∼2-kDa mass difference between tagged and nontagged protein (see Fig. 6.5.2). The digestion should be as complete as possible at the specific cleavage site (see Background Information); overdigestion may result in nonspecific cleavage at other sites. It is also important to avoid denaturation of the (thrombin) (e.g., by vigorous stirring or overheating), as partially denatured proteases often exhibit some loss of specificity, cleaving at unpredictable sites.

Troubleshooting Bovine growth hormone (BGH) Most precautions and guidelines for folding and oxidation are given in annotations to the individual steps of Basic Protocol 1. It should be cautioned that, when using a mixture of oxidized and reduced glutathione for oxidizing proteins that contain both free and disulfidelinked cysteines in the native state (e.g., IL-2), the potential exists for glutathionylation, in which the unpaired cysteine(s) form a mixed disulfide with glutathione. This modification will cause charge heterogeneity, and can be readily detected by electrospray ionization mass spectrometry (ESI-MS), which will indicate a mass increase of 305.3 for each GSH moiety incorporated. Interleukin-2 Low-molecular-weight impurities extracted together with monomeric hIL-2 in acetic acid (see Basic Protocol 2, step 5) are usually eluted earlier than the oxidized form of hIL-2 in RPHPLC (see Basic Protocol 2, step 12) and can thus be removed. If the RP-HPLC peak containing the oxidized form of hIL-2 is contaminated by low-molecular-weight impurities, SDSPAGE analysis of each fraction composing the peak is recommended to avoid pooling fractions containing impurities. Additional information regarding troubleshooting is given in annotations to individual steps of Basic Protocol 2. HIV-1 integrase MCAC. Protein will not bind to the column if the His tag has been nonspecifically degraded or removed by E. coli proteases. SDS-PAGE (UNIT 10.1) of the cell extracts and fractions during the purification should indicate if the tag is present. The His tag adds ∼2 kDa to the

apparent mass of the protein. The denaturant concentration should be high enough to maintain solubility of the protein during chromatography. Insoluble or aggregated protein will not bind to the column and will be located in the column flowthrough or will clog the top of the matrix. If protein aggregates on the column, a higher concentration of denaturant should be tried, although aggregation should not happen using 6 M guanidine⋅HCl in column buffer A. Further details on troubleshooting the use of the Ni-NTA resin are given in the manufacturer’s literature (Qiagen, 1992). Thrombin cleavage. If undigested (Histagged) protein still remains after digestion, the standard procedure is to reapply the mixture to an MCAC column, whereupon the matrix will specifically bind the unprocessed protein containing the His tag. The processed (cleaved) protein is not bound and elutes in the column flowthrough. This approach, although satisfactory for monomeric proteins, is complicated with multimeric proteins and those that exhibit reversible self-association—e.g., HIV-1 integrase. For example, in a partially digested dimeric protein, the subunits can be arranged in three possible configurations—TT, TM, and MM—where T is still tagged and M has had the tag cleaved. It may be possible to separate TT and TM by gradient elution from the MCAC column, as the former might be expected to bind more tightly than the latter. A simpler approach is to use solvent conditions that dissociate the protein subunits without denaturing them. For HIV-1-integrase, this can be achieved with 2 to 3 M urea.

Anticipated Results BGH The purification of ∆ A-9 BGH is summarized in Table 6.5.1. About 90 to 100 mg of protein are obtained from 50 g (wet weight) of cells with an overall yield in the range 15% to 25%. For larger and/or multidomain proteins, much lower yields (1% to 5%) may be more typical. Gel analysis of the BGH in cell extracts and in purified protein is shown in Figure 6.5.1. Interleukin-2 If the expression of hIL-2 in E. coli is ∼10%, 20 g of cell pellet will yield 4 to 5 mg of purified protein, with a specific activity of 5 × 106 U/mg. The purity of the hIL-2 should be >98% as indicated by analytical C18 column chromatography (see Support Protocol).



Table 6.5.1

Purification of a Recombinant Bovine Growth Hormone Analoga

Total Proteinc (mg)

Stage of Purificationb Cells Washed pellet (step 4) Dialysis supernatant (step 8) DEAE-Sepharose pool (step 11) Sephadex G-100 pool (step 14)

5000 375 200 131 95

Specific BGH contentd (%) 7.5 60 75 95 99

Total BGH (mg) 375 225 150 125 94

Yield (%) 100 60 40 33 25

aThe summary refers specifically to a biologically active analog of BGH (δ-9 BGH) in which the full-length sequence is

truncated at the N-terminus by eight residues and serine is substituted for glycine at the first position. Similar relative yields were obtained with a δ-1 analog in which the N-terminal Ala of the native sequence was replaced by Met, but the amount of this protein expressed in E. coli was several-fold lower than that obtained for δ-9 BGH (Wingfield et al., 1987a) bThe numbers in parentheses refer to Basic Protocol 1 steps. cDeterminations were made using the Bio-Rad Protein Assay Kit, except those for the DEAE and Sephadex pools, which

were made by UV absorbance measurements. dPercentage of total protein that is BGH. Estimates made by densitometric scanning of Coomassie blue stained

SDS-polyacrylamide gels.

a

b

c

d

In a scaled-up process (see Critical Parameters), 25 to 30 mg of purified hIL-2 may be obtained from 200 g of E. coli cell pellet.


HIV-1 integrase Figure 6.5.4 shows results of SDS-PAGE of the purified IN50-212 before (lane B) and after (lane A) removal of the His tag by thrombin digestion. The molecular weight difference of ∼2000 kDa corresponds to the removal of the

Figure 6.5.4 Results of SDS-PAGE of H1V-1 integrase50-212 on a 12.5% polyacrylamide gel stained with Coomassie blue. Lane A, recombinant HIV-1 integrase50-212; lane B, recombinant HIV-1 integrase50-212 with N-terminal His tag; lane C, extract of HIV-1 integrase–expressing E. coli cells used for purification.

16 amino acid residues comprising the N-terminal His tag. The overall recovery of purified protein is ∼30%. For example, from 1 g (wet weight) of cells (containing ∼5% IN50-212) ∼1.5 mg of purified protein are obtained (see Figure 6.5.4, lane C, for SDS-PAGE analysis of whole IN50212–expressing E. coli). Most of the steps, including protein folding, have high recoveries (80% to 95%). The yield of thrombin-digested


protein from purified tagged protein is between 60% to 70%. The enzymatic and biophysical properties of the protein are detailed in Hickman et al. (1994).

Time Considerations BGH Basic Protocol 1 for the purification of BGH takes 5 days. Protein can be stored at −80°C at the end of step 4 (as washed pellets) or step 10 (as pooled fractions from ion-exchange). Day 1: Cell breakage, preparation of washed pellets and extraction with guanidine⋅HCl (steps 1 to 5) requires 1⁄2 day of work. The protein extract is then dialyzed overnight (step 5). Day 2: After changing the dialysis buffer (step 7), dialysis is continued at least 6 hr. At this step the dialysis can be left overnight or directly processed by ion-exchange chromatography (steps 8 to 10), which takes 6 to 8 hr and can be run overnight. Day 3: Ion-exchange chromatography is run, or if ion-exchange was performed on day 2, the pooled fractions are concentrated for several hours (step 11), then applied to the gel-filtration column and chromatographed for several hours (steps 12 and 13). The gel-filtration column is run is overnight if a low-pressure matrix (e.g., Sephadex G-100) is used or on the same day if a medium-pressure matrix (e.g., Superdex 200) is used. Day 4: The protein is analyzed (e.g., by SDS-PAGE or isoelectric focusing), concentrated if required (step 14), and frozen or prepared for lypohilization. Interleukin-2 In Basic Protocol 2, preparation and lysis of the cells (steps 1 and 2) will take 1 hr, cell breakage and preparation of washed pellets (steps 3 and 4) will take 3 hr, acid extraction (steps 5 and 6) will take 1 hr, Sephadex G-100 column chromatography (step 7) will take 12 hr, dialysis of the pooled fractions will be carried out overnight, and RP-HPLC (steps 9 to 12) will require 7.5 hr. The final dialysis is run overnight. In the Support Protocol for resolution of native and misfolded forms of hIL-2, it will take ∼1 hr to run a sample after the blank runs to establish the baseline profile have been completed.

HIV-1 integrase Basic Protocol 3 is usually carried in two stages. In stage 1, purification of the unfolded protein that still contains the His tag (steps 1 to 8) is usually carried out on a relatively large scale (using 100 g of cells). If the Superdex 200 column used in step 5 and the MCAC column used in step 6 are run using the Pharmacia Biotech Biopilot system, this stage will take 3 to 4 days to complete. It will take longer if low-pressure columns are used. In stage 2, protein folding (steps 11 and 12), removal of the His tag by thrombin cleavage (steps 13 and 14), affinity chromatography (steps 15 to 17), and gel filtration (step 19) are performed. This stage is carried out repeatedly with relatively small amounts of protein (i.e., 30 mg; this represents 98% pure protein (see Support Protocol 2). The relationships between the protocols are shown in Figure 6.6.1. STRATEGIC PLANNING A variety of pGEX expression vectors are commercially available (Pharmacia Biotech) that contain a tac promoter for chemically inducible, high-level protein expression. The available pGEX vectors have an open reading frame encoding glutathione-S-transferase (GST) followed by multiple cloning sites. These are followed by termination codons in each reading frame (Figs. 6.6.2 and 6.6.3). A fragment of DNA containing the genetic sequence for the polypeptide of interest is ligated into an appropriate pGEX vector and transformed into E. coli. It should be noted that although expression in E. coli is efficient, there is no post-translational modification machinery. Successful expression of GST

Contributed by Sandra Harper and David W. Speicher Current Protocols in Protein Science (1997) 6.6.1-6.6.21 Copyright © 1997 by John Wiley & Sons, Inc.


6.6.1 Supplement 9

expression of GST fusion protein (Basic Protocol 1)

on-column affinity purification (Basic Protocol 2 or Alternate Protocol 1)

cleavage on column (Alternate Protocol 3)

elution

batchwise purification with affinity resin (Alternate Protocol 2)

elution

cleavage in solution (Basic Protocol 3)

elution

cleavage bound to resin (Alternate Protocol 4)

cleavage in solution (Basic Protocol 3)

protein purification by affinity chromatography (Support Protocol 1) and/or HPLC (Support Protocol 2)

Figure 6.6.1 Flow chart showing the relationships between the various protocols in this unit.

fusion proteins using baculovirus systems (Davies et al., 1993) and yeast (Mitchell et al., 1993) have also been reported. One factor that influences which pGEX vector to choose is whether or not the GST moiety will ultimately be cleaved away from the protein of interest. The pGEX-2T and pGEX-4T series of vectors contain a protease cleavage site for thrombin, and the pGEX-3X and pGEX-5X series of vectors contain protease cleavage sites for factor Xa. A more recently developed expression vector is the pGEX-6P series, which contains a cleavage site for PreScission protease (Pharmacia Biotech). PreScission protease has the advantage that it is effective at low temperatures (5°C). It is also a GST fusion protein, a feature that facilitates removal of the protease from the target protein after cleavage. Fusion proteins with a thrombin recognition site have the advantage that relatively small amounts of thrombin and short digestion times at 37°C will often cleave the fusion protein with high efficiency. Thrombin digestions are often the most cost effective on a per milligram of cleaved target polypeptide basis. Factor Xa is more expensive and typically requires use of much higher enzyme-to-substrate ratios for efficient cleavage. Solutions of factor Xa also have a more limited shelf life since freezing and thawing inactivates this enzyme.

Expression and Purification of GST Fusion Proteins

For preparation of cDNA inserts encoding the desired polypeptide, see Ausubel et al. (1994) or Sambrook et al. (1989). Briefly, a set of oligonucleotides is designed for polymerase chain reaction (PCR) amplification of the region of interest of a pertinent cDNA. These oligonucleotides should also contain appropriate restriction sites adjacent to the desired coding region that are compatible with a restriction site in the cloning site of the selected pGEX vector (see Figs. 6.6.2 and 6.6.3). PCR amplification is performed,

6.6.2 Supplement 9


pGEX-1λT thrombin Leu Val Pro Arg ↓ Gly Ser Pro Glu Phe Ile Val Thr Asp CTG GTT CCG CGT GGA TCC CCG GAA TTC ATC GTG ACT GAC TGA CGA

BamHI

stop codons

EcoRI Tth111I Aat II

BalI glutathione-S- transferase Ptac

r

pSj10∆Bam7Stop7

Ap

BspMI

PstI

pGEX ~4950 bp

la

NarI EcoRV BssHII ApaI Bst EII MluI

Alw NI

p4.5

q

cl

pBR322 ori

Figure 6.6.2 pGEX vectors are plasmid expression vectors that express a cloned gene as a fusion protein with glutathione-S-transferase (GST). The lac repressor gene binds to the lac promoter (ptac) and represses expression of the GST fusion protein until induction with isopropyl-1-thio-β-D-galactopyranoside (IPTG). The polypeptide of interest can be inserted immediately after the GST gene using the polylinker site shown in brackets (pGEX-1λT, shown here, is the most common; see Fig. 6.6.3 for other PGEX polylinkers). Protease cleavage sites (brackets above the polylinker sequences) are located between the GST carrier protein and the protein of interest so that the GST moiety can be removed. Restriction endonuclease sites are indicated below the sequence of the polylinker and on the plasmid. An important consideration in selecting a vector and appropriate cloning sites is to minimize the number of extraneous residues introduced into the N-terminal of the target polypeptide. Vector map courtesy of Pharmacia Biotech.

followed by digestion of the PCR product and the pGEX vector with the appropriate restriction enzymes. The PCR product is then ligated into the pGEX vector and transfected into a suitable E. coli host. Several transformants should be grown in minicultures and induced with isopropyl-1-thio-β-D-galactopyranoside (IPTG) to check for expression of the desired fusion protein. Fusion protein expression can be monitored by SDS-PAGE (UNIT 10.1) or by Western blot (UNIT 10.10) detection of the GST fusion protein using an antibody specific for either the target protein or the GST moiety. Once successful expression is achieved, the integrity of the DNA should be verified by sequencing to ensure that no errors were introduced during PCR. Before conducting a large-scale purification, it is worthwhile to perform a small pilot purification (∼10-fold less than protocol descriptions) to determine optimal conditions. The purification can then easily be scaled up. All stages of purification should be monitored using SDS-PAGE (UNIT 10.1). In most cases, GST fusion protein expression is very high and a major band at the expected molecular weight (the GST moiety contributes 26 kDa to the molecular weight) is obvious when uninduced and induced cells are compared on SDS gels (Fig. 6.6.4). This band can then be monitored at each step of the purification. However, as noted above, if the level of protein expression obtained is low or band identification is ambiguous, the fusion protein can be monitored by Western blot



Supplement 9

pGEX-2T thrombin Leu Val Pro Arg ↓ Gly Ser Pro Gly Ile His Arg Asp CTG GTT CCG CGT GGA TCC CCG GGA ATT CAT CGT GAC TGACTG ACG

stop codons

BamHI SmaI EcoRI pGEX-2TK kinase

thrombin

Leu Val Pro Arg ↓ Gly Ser Arg Arg Ala Ser Val CTG GTT CCG CGT GGA TCT CGT CGT GCA TCT GTT GGA TCC CCG GGAATT CATCGT GAC TGA

stop codon

BamHI SmaI EcoRI pGEX-4T-1

thrombin

Leu Val Pro Arg ↓ Gly Ser Pro Glu Phe Pro Gly Arg Leu Glu Arg Pro His Arg Asp CTG GTT CCG CGT GGA TCCCCG GAATTC CCG GGT CGA CTC GAG CGGCCG CAT CGT GAC TGA

BamHI

SalI

EcoRI SmaI

stop codons

NotI

XhoI

pGEX-4T-2

thrombin

Leu Val Pro Arg ↓Gly Ser Pro Gly Ile Pro Gly Ser Thr Arg Ala Ala Ala Ser CTG GTTCCG CGT GGA TCC CCA GGA ATT CCC GGG TCGACT CGA GCG GCC GCA TCG TGA

BamHI

EcoRI SmaI

SalI

stop codons

NotI

XhoI

pGEX-4T-3

thrombin

Leu Val Pro Arg ↓ Gly Ser Pro Asn Ser Arg Val Asp Ser Ser Gly Arg Ile Val Thr Asp CTG GTT CCG CGT GGA TCC CCG AAT TCC CGG GTC GAC TCG AGC GGC CGC ATC GTG ACT GAC TGA

BamHI

factor Xa

EcoRI SmaI

SalI

stop codons

NotI

XhoI

pGEX-3X

Ile Glu Gly Arg ↓Gly Ile Pro Gly Asn Ser Ser ATC GAA GGT CGT GGG ATC CCC GGG AAT TCA TCG TGA CTG ACT GAC

stop codons

BamHI SmaI EcoRl pGEX-5X-1

factor Xa

Ile Glu Gly Arg ↓ Gly Ile Pro Glu Phe Pro Gly Arg Leu Glu Arg Pro His Arg Asp ATC GAA GGT CGT GGG ATC CCC GAATTC CCG GGTCGA CTC GAG CGG CCG CAT CGT GAC TGA

BamHI

EcoRl SmaI

Sall

XhoI

Not l

stop codons

pGEX-5X-2

factor Xa

Ile Glu Gly Arg ↓ Gly Ile Pro Gly Ile Pro Gly Ser Thr Arg Ala Ala Ala Ser ATC GAA GGT CGT GGG ATC CCC GGA ATTCCC GGG TCG ACT CGA GCGGCC GCATCG TGA

BamHI

EcoRl SmaI

Sall

XhoI

Not l

stop codons

pGEX-5X-3

factor Xa

Ile Glu Gly Arg ↓ Gly Ile Pro Arg Asn Ser Arg Val Asp Ser Ser Gly Arg lle Val Thr Asp ATC GAA GGT CGT GGG ATC CCC AGG AAT TCC CGGGTC GAC TCG AGC GGC CGC ATC GTG ACT GAC TGA

BamHI

EcoRI SmaI

SalI

XhoI

NotI

Stop codons

Figure 6.6.3 Variations on the polylinker site shown in Figure 6.6.2 (courtesy of Pharmacia Biotech). Expression and Purification of GST Fusion Proteins

6.6.4 Supplement 9


95 66 43 36

66

25 17

29

6 18

1

2

L

S 30 C

Figure 6.6.4 SDS gel stained with Coomassie brilliant blue showing the expression of a glutathione-S-transferase (GST) fusion protein. Lane 1, total cell lysate before induction with isopropyl1-thio-β-D-galactopyranoside (IPTG). Lane 2, total cell lysate after 3-hr induction with IPTG. The position of the fusion protein is indicated by the arrow.

P

L

S 25 C

P

L

S

P

18 C

Figure 6.6.5 SDS gel stained with Coomassie brilliant blue showing the effects of different growth temperatures on solubility of an expressed glutathione-S-transferase (GST) fusion protein. E. coli transfected with a recombinant pGEX-2T vector were grown at different temperatures as indicated. Proportional aliquots of whole-cell lysate after sonication (L), the supernatant after centrifuging the lysate (S), and the remaining pellet (P) are shown. This recombinant was primarily in inclusion bodies (P) at 37°C (not shown), 30°C, and 25°C. At 18°C, however, nearly all of the expressed protein was in a soluble native form.

analysis (UNIT 10.10) using a GST-specific antibody (Pharmacia Biotech). It is recommended that the lysed cell extract, extracted pellet, and all other collected fractions from the purification be saved on ice until after careful analysis of the entire purification by SDS-PAGE and/or immunoblotting to ensure that fractions containing the fusion protein are not mistakenly discarded. Basic Protocol 1 describes protein production in cells grown at 37°C; however, at this temperature some fusion proteins may be found in inclusion bodies in a denatured form. As an alternative to attempting to renature the protein after extraction from inclusion bodies (UNIT 6.1), expression at lower temperatures, such as 30°, 25°, 20°, or 15°C, can be evaluated to determine whether the protein can be obtained in good yield in the soluble fraction (Fig. 6.6.5). When expressing fusion proteins at lower temperatures, the initial overnight culture can be grown at 37°C followed by growth at a lower temperature prior to induction. EXPRESSION OF GLUTATHIONE-S-TRANSFERASE FUSION PROTEIN Transformed E. coli cells expressing the glutathione-S-tranferase (GST) fusion protein of interest are grown in culture in the presence of isopropyl-1-thio-β-D-galactopyranoside (IPTG) at the desired preparative scale. Since the expression level of GST fusion proteins is usually very high, adequate amounts of protein can usually be conveniently obtained by preparing a few liters of cells grown in shaker cultures. This protocol describes the preparation of 1.8 liters of transfected E. coli in three 600-ml units using 2-liter flasks in a shaker incubator. Moderate further scale-up is feasible by using more or larger flasks.

BASIC PROTOCOL 1



Supplement 9

Further scale-up can be accomplished using a fermentor (see UNITS 5.3 & 5.4). This protocol describes protein production in cells grown at 37°C. At this temperature, however, some fusion proteins may be recovered from inclusion bodies in a denatured form, and culture conditions may need to be modified to improve protein yield in the soluble fraction (see Strategic Planning). Culture growth can be monitored by reading the optical density at 550 nm (OD550) as well as by analysis of the bacterial culture using SDS-PAGE. Cells should not be allowed to grow for extended periods of time after induction as cell lysis can occur; this releases proteases that may degrade the fusion protein. Visual inspection of the cells using a microscope is a useful method for identifying cell breakage (see UNITS 5.1-5.3 & 6.1-6.5 for additional details concerning recombinant protein expression in E. coli). Materials Luria broth (LB medium; UNIT 5.2; pH adjusted to 7.2) 5 mg/ml ampicillin (see recipe) Glycerol culture of transformed E. coli cells expressing GST fusion protein of interest in a pGEX vector 100 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG; see recipe) 2-liter culture flasks 500-ml culture flasks Large centrifuge bottles (e.g., 1-liter capacity) Low-speed refrigerated centrifuge (e.g., Beckman J6-B and JS-4.2 rotor or equivalent), 4°C Additional reagents and equipment for SDS-polyacrylamide gel electrophoresis (SDS-PAGE; UNIT 10.1) Grow bacterial cells 1. Prepare LB medium and add 600 ml to each of three 2-liter flasks and 100 ml to each of two 500-ml flasks. Autoclave 20 to 30 min at a slow exhaust (liquid) setting. Flasks should be filled to only 20% to 30% of their capacity to ensure adequate aeration of the medium during cell growth. Autoclave LB medium immediately after preparing it to prevent any incidental bacterial growth from occurring. It can then be stored up to 1 month at room temperature under sterile conditions.

2. Allow medium to cool to room temperature. Add 1 ml of 5 mg/ml ampicillin to one flask containing 100 ml LB medium. It is important to allow medium to cool before proceeding, because ampicillin is inactivated at temperatures >50°C. Flame the opening of all bottles and flasks to reduce the risk of contamination.

3. Using an inoculating loop, transfer some of the glycerol culture containing the transfected E. coli expressing the GST fusion protein of interest to the flask containing 100 ml LB medium with ampicillin. Sterile flame the inoculating loop as well as the opening of all bottles. Allow the loop to cool before transferring the inoculating culture. If the loop temperature is too high, all the cells could be killed during the transfer.

4. Incubate the inoculated culture on an environmental shaker set at 250 to 300 rpm overnight at 37°C. Expression and Purification of GST Fusion Proteins

5. The next morning, remove the culture from the shaker and read the optical density at 550 nm (OD550) using a UV/visible light spectrophotometer. The OD550 of the overnight culture should be ∼1.0. Use medium from the second 500-ml flask that did not receive ampicillin as a reference to zero the spectrophotometer.

6.6.6 Supplement 9


6. Using sterile technique, add 6 ml of 5 mg/ml ampicillin to each 2-liter flask containing 600 ml LB medium (0.1 mM final concentration). 7. Dilute the overnight culture 1:20 by adding 30 ml to each of the three 2-liter flasks containing 600 ml LB medium. 8. Incubate the 600-ml cultures on a shaker at 37°C at 250 to 300 rpm until the OD550 is 0.5 to 0.7. It should take ∼2 hr for the culture to reach this early log stage of growth at 37°C. If cells are grown at lower temperatures to shift the expressed protein from inclusion bodies into the soluble fraction, this time must be increased, since the cells will grow more slowly at lower temperatures.

Induce expression of fusion protein 9. Remove a 1-ml sample from each flask and save for gel analysis. Induce cells by adding 6 ml of 100 mM IPTG per flask (1.0 mM final). Incubate at 37°C for an additional 2.5 to 3 hr. Optimal growth conditions (OD550 at time of induction, growth temperature, and growth time after induction) for each recombinant protein should be empirically determined. The conditions given here will generally be suitable for growth at either 37° or 30°C. At lower temperatures (≤25°C), longer growth times will definitely be needed, and the best protein yields will usually be obtained if the OD550 prior to induction is 0.65 to 0.85.

10. Remove cultures from the shaker 2.5 to 3 hr after induction. Remove 1 ml from each culture and save for gel analysis. Check final OD550. Culture growth can be monitored at OD550. When the cells reach saturation, they will stop dividing. A typical SDS-PAGE gel of an uninduced control culture and an induced culture after 3 hr growth is shown in Figure 6.6.4.

Recover bacterial cells 11. Pour each culture into a large centrifuge bottle and centrifuge 20 min at 4000 × g, 4°C. Medium from the second 500-ml flask without ampicillin can be used to balance the centrifuge bottles.

12. Carefully decant the supernatant, leaving 15 to 50 ml in the centrifuge bottle. Resuspend the pelleted cells in the remaining supernatant. Transfer the cell suspension to a 50-ml centrifuge tube. 13. Centrifuge 20 min at 4000 × g, 4°C. Decant the supernatant. 14. Freeze cell pellet by placing in a −80°C freezer. 15. Analyze saved 1-ml samples from before and after induction using SDS-PAGE (UNIT 10.1) to check for protein expression. As noted above, a 1-ml aliquot of culture is removed prior to induction and a second 1-ml sample is removed after the 3-hr induction. These aliquots can be centrifuged for 2 min and the supernatants carefully removed with a pipet. These samples can be stored at 0° to 4°C overnight prior to running a gel; for longer-term storage prior to gel analysis, store frozen (−20°C) to prevent proteolysis. The cell pellet can be directly resuspended in SDS sample buffer (200 ìl) and heated 3 to 5 min at 90°C. Typical results are shown in Figure 6.6.4. Purification of Recombinant Proteins


Supplement 9

BASIC PROTOCOL 2

AFFINITY CHROMATOGRAPHY PURIFICATION OF A SOLUBLE GST FUSION PROTEIN Soluble glutathione-S-transferase (GST) fusion proteins can easily be purified from cell lysate supernatants by affinity chromatography on glutathione–Sepharose 4B using either batch or column loading of the sample. Protease inhibitors should be added to the lysis buffer to minimize potential proteolysis. As an aid for cell lysis, 1% (v/v) Triton X-100 may be added to the lysis buffer. After cells are lysed, centrifugation is used to pellet any unlysed cells and inclusion bodies. Location of the fusion protein can usually easily be determined by analyzing a small aliquot of both the supernatant and the pellet by SDS-PAGE (UNIT 10.1). The glutathione–Sepharose 4B can be regenerated and reused for subsequent purifications. To avoid potential cross-contamination of different proteins or mutant forms of a single protein, however, it is recommended that a given column or batch of resin be reserved for use with a single protein. Preequilibration of the glutathione column (steps 1 to 4) can be performed either prior to or in parallel with cell lysis. All steps except SDS-PAGE (room temperature) should be performed in a cold room at 4°C, unless otherwise noted. Materials Glutathione–Sepharose 4B resin (Pharmacia Biotech) PBS (APPENDIX 2E) Glutathione buffer (see recipe) PBS/EDTA/PMSF buffer (see recipe) Pelleted E. coli culture expressing fusion protein (see Basic Protocol 1, step 14) Lysis buffer (see recipe), ice cold Wash buffer (see recipe), ice cold PBS/EDTA (see recipe) 2.5 × 8–cm glutathione–Sepharose 4B column (e.g., Bio-Rad Econo) Peristaltic pump Sonicator equipped with microtip probe (e.g., Branson) Dounce homogenizer 60-ml centrifuge bottles (capable of handling force of 48,000 × g) High-speed refrigerated centrifuge (e.g., Beckman JZ-21M centrifuge and JA-18 rotor or equivalent), 4°C Additional reagents and equipment for pouring chromatographic columns (UNIT 8.3) and for SDS-PAGE (UNIT 10.1) Preequilibrate glutathione column 1. Pour 20 ml glutathione–Sepharose 4B resin into 2.5 × 8–cm column (glutathione column; see UNIT 8.3 for details of pouring procedure). This amount of resin should be adequate for purification of fusion protein from three 600-ml cultures that contain ∼20 to 40 mg fusion protein per culture. Although glutathione– Sepharose 4B has an advertised minimum binding capacity of 8 mg/ml resin, the actual capacity may be substantially different and should be empirically determined. The actual amount of resin used and column size can be increased or decreased depending on the amount of fusion protein to be purified.

2. Wash the glutathione column with 5 to 10 bed volumes PBS at a flow rate of 1.5 ml/min to remove the ethanol storage solution. Expression and Purification of GST Fusion Proteins

A bed volume is one-half the amount of glutathione–Sepharose 4B that was added to the column as a 50% slurry.

6.6.8 Supplement 9


New columns should be prepared as described in steps 1 to 4. Previously used columns should be preequilibrated using steps 3 to 4 only, to ensure that the column is completely reduced and has maximal binding capacity. Use of a peristaltic pump is recommended for convenient control of flow rates. Compression of the resin bed indicates that the column pressure is too high and that the flow rate should be lowered.

3. Wash the glutathione column with 3 to 5 bed volumes glutathione buffer at 1.5 ml/min. Previously used columns may become partially oxidized on storage and should be preequilibrated (steps 3 to 4) within 24 hr before use.

4. Wash the glutathione column with 10 bed volumes PBS/EDTA/PMSF at 1.5 ml/min. Lyse cells 5. Resuspend each pelleted 600-ml culture in 15 ml ice-cold lysis buffer. Pellets should be resuspended in 25 to 50 ìl buffer per milliliter of culture.

6. Sonicate the suspension using a probe-tip sonicator ten times for 10 sec each, with 1-min rests between sonications to lyse the cells. Save a sample (∼100 µl) of the lysate for gel analysis and transfer remainder to a 60-ml centrifuge tube. The cells are usually adequately lysed at the point when the suspension turns a slightly darker color and becomes clearer. To minimize proteolysis in the sample, it is essential to keep the cells on ice throughout the sonication procedure, and sonication should be performed in short bursts to minimize sample heating. Avoid excessive sonication, as this can lead to co-purification of E. coli host proteins along with the fusion protein of interest. Avoid frothing during sonication, which can denature the fusion protein.

7. Centrifuge the lysate 20 min at 48,000 × g, 4°C. Unbroken cells, large cellular debris, and inclusion-body protein will be pelleted.

8. Decant the supernatant containing the soluble fusion protein into a clean 50-ml centrifuge tube. 9. Add a volume of ice-cold wash buffer equal to the volume of lysis buffer used in step 5 to the pellet. Use a dounce homogenizer to resuspend the pellet. Pellets should be resuspended in 25 to 50 ìl buffer per milliliter of culture.

10. Analyze the lysate, supernatant, and resuspended pellet using SDS-PAGE (UNIT 10.1) to verify that the fusion protein is in the supernatant. If the fusion protein is in the supernatant, proceed to the next step. If the protein is in the pellet, it is necessary to purify the GST fusion proteins from inclusion bodies (see Alternate Protocol 2) or to start over, shifting the protein into the supernatant by growing the cultures at a lower temperature (see Troubleshooting and Fig. 6.6.5).

Load the column 11. Load the supernatant onto a preequilibrated glutathione column (see step 4). Collect fractions and analyze multiple fractions across the unbound peak by SDS-PAGE (UNIT 10.1) to verify that the fusion protein has bound and that column capacity was not exceeded. For a 2.5-cm diameter column, a sample-loading flow rate of ≤0.1 ml/min is recommended to permit complete binding of the fusion protein to the resin. Faster flow rates may decrease the yield of bound fusion protein due to slow kinetics of association at 4°C, unless a large excess of resin is used. This step is most conveniently performed overnight; the flow rate



Supplement 9

can be adjusted so that most of the supernatant will be loaded by the next morning. Do not allow the column to run dry. SDS-PAGE analysis of fractions collected during sample loading will reveal whether the fusion protein is bound to the column or is present in the unbound fractions. Absence of fusion protein in early unbound fractions combined with its appearance in late unbound fractions indicates that column capacity has been exceeded. If this condition is observed, reduce the protein load or increase the column size.

12. Wash the column with 5 to 10 bed volumes PBS/EDTA/PMSF at 1.5 ml/min. 13. Wash the column with 10 bed volumes PBS/EDTA at 1.5 ml/min to remove the PMSF. If samples are to be cleaved with thrombin or factor Xa, any serine protease inhibitor (e.g., PMSF) must be removed from the sample before cleavage.

Elute the fusion protein 14. Elute the fusion protein from the column by washing the column with 5 bed volumes glutathione buffer. A flow rate of 0.3 ml/min for a 2.5-cm column is recommended to elute the fusion protein in a minimal volume. Elution of the fusion protein can be monitored at A280 either with an online UV monitor or by reading the absorption of individual fractions.

15. Analyze the fractions by SDS-PAGE (UNIT 10.1) and pool fractions containing the GST fusion protein. Store at 0° to 4°C. Fusion protein should typically be >90% pure at this point. ALTERNATE PROTOCOL 1

AFFINITY CHROMATOGRAPHY PURIFICATION OF GST FUSION PROTEIN FROM INCLUSION BODIES In some cases, fusion proteins are entirely or primarily located in a denatured, aggregated form in inclusion bodies. Glutathione-S-transferase (GST) fusion proteins can often be purified from inclusion bodies after solubilization in urea or another denaturant followed by renaturation by dialysis (UNIT 6.3). Other methods of solubilization include addition of detergents such as Sarkosyl (N-laurylsarosine; Grieco et al., 1992; Frangioni, 1992). After denaturation and renaturation, it is important to ensure that the protein has regained its native conformation and function (UNITS 6.4 & 6.5). Although denatured GST will not bind to the glutathione column and hence will not be recovered by this method, the possibility that the GST moiety has refolded when the fusion partner has not folded properly should also be considered. As an alternative to purifying proteins from inclusion bodies, growing the cells at a lower temperature will often shift the fusion protein into the supernatant while still producing ≥10 mg of fusion protein per liter of culture when pGEX vectors are used (see Strategic Planning and see Troubleshooting). All steps should be performed in a cold room at 4°C unless otherwise noted. Additional Materials (also see Basic Protocol 2) U buffer (see recipe) Triton X-100 PBS/glycerol buffer (see recipe) Low-speed refrigerated centrifuge (e.g., Beckman J6-B and JS-4.2 rotor or equivalent), 4°C Additional reagents and equipment for dialysis (APPENDIX 3B)


1. Preequilibrate the glutathione column, lyse the cells, and separate the lysate pellet, which includes the inclusion bodies, and supernatant (see Basic Protocol 2, steps 1 to 9).

6.6.10 Supplement 9


2. Centrifuge washed pellet (see Basic Protocol 2, step 9) 20 min at 48,000 × g, 4°C. Decant the supernatant and resuspend the pellet in 12 ml freshly prepared U buffer per 600 ml original culture. Incubate 2 hr on ice. Pellets should be resuspended in 20 ìl U buffer per ml of culture.

3. Centrifuge 20 min at 48,000 × g, 4°C. Carefully transfer the supernatant to a clean 50-ml centrifuge tube. The extracted fusion protein should now be in the supernatant.

4. Add Triton X-100 to the supernatant to give a final concentration of 1% (v/v). 5. Dialyze the sample (APPENDIX 3B) 2 to 3 hr in PBS/glycerol buffer. Dialysis buffer volume should be 20 times the sample volume.

6. Dialyze sample overnight in PBS/EDTA/PMSF. Dialysis buffer volume should be >100 times the sample volume.

7. Remove the sample from the dialysis bag and centrifuge 20 min at 4000 × g, 4°C. 8. Column purify the fusion protein (see Basic Protocol 2, steps 11 to 15). BATCH PURIFICATION OF GST FUSION PROTEIN Soluble glutathione-S-transferase (GST) fusion proteins in cell lysate supernatants or renatured proteins extracted from inclusion bodies can be batch purified on glutathione– Sepharose 4B as an alternative to column-based purification (see Basic Protocol 2 and see Alternate Protocol 1). Batch purification requires less equipment and is relatively quick and easy, but resulting protein yield and sample purity are lower than in a chromatographic separation. In addition, the room temperature incubations recommended by the resin manufacturer (Pharmacia Biotech), especially the batch incubation of E. coli lysate supernatant or inclusion body extract with glutathione-Sepharose, increase the risk of proteolytic degradation of the fusion protein.

ALTERNATE PROTOCOL 2

Prepare fusion protein 1. Extract soluble fusion proteins (see Basic Protocol 2, steps 5 to 10). If the fusion protein is in inclusion bodies, lyse the cells and collect the pellet (Basic Protocol 2, steps 5 to 9, followed by Alternate Protocol 1, steps 2 to 7).

Prepare resin slurry 2. Prepare a 50% slurry of glutathione–Sepharose 4B in PBS: For each milliliter of bed volume, centrifuge 1.33 ml of a 75% slurry of glutathione–Sepharose 4B for 5 min at 500 × g, room temperature. Discard the supernatant. Wash with 10 bed volumes PBS. Invert tube containing resin several times to mix, then centrifuge 5 min at 500 × g, room temperature, and remove supernatant. Add 1 ml PBS for each 1.33 ml of original slurry. Mix well before using. Bind fusion protein to resin 3. Add 2 ml of 50% slurry of equilibrated glutathione-Sepharose to each 100 ml bacterial lysate supernatant. Incubate 30 min at room temperature with gentle agitation on a platform or orbital shaker. 4. Centrifuge the suspension 5 min at 500 × g, room temperature. Remove supernatant and save at 0° to 4°C for later analysis by SDS-PAGE (UNIT 10.1) to determine the efficiency of binding of the fusion protein to the resin.



Supplement 9

5. Wash the pellet with 10 bed vol PBS. A bed volume is one-half the volume of the 50% slurry that was added to the column.

6. Centrifuge the suspension 5 min at 500 × g, room temperature. Discard the supernatant. 7. Repeat the wash and centrifugation steps for a total of three washes with 10 bed volumes each of PBS. Elute fusion protein 8. Elute the bound fusion protein by gently resuspending the sedimented resin in 1.0 ml glutathione buffer per milliliter resin bed volume. Incubate 10 min at room temperature with gentle agitation. 9. Centrifuge the suspension 5 min at 500 × g, room temperature. Transfer supernatant to a separate tube. 10. Repeat the elution and centrifugation (steps 8 to 9) a total of three times. Store at 0° and 4°C. The supernatants may be pooled into one tube or analyzed separately by SDS-PAGE (UNIT 10.1) to monitor for fusion protein content. The yield of fusion protein can be monitored by measuring the absorbance at 280 nm (A280). The extinction coefficient will partially depend on the absorbance characteristics of the experimental component of the fusion protein. For the GST moiety alone, the concentration can be estimated using 1 A280 = ∼0.6 mg/ml protein. As noted above, batch purification at room temperature increases the risk of proteolytic digestion of the target protein. To minimize such degradation, this procedure can alternatively be performed in a cold room at 4°C with incubation times in steps 3 and 8 increased 2- to 4-fold. BASIC PROTOCOL 3

PROTEASE CLEAVAGE OF FUSION PROTEIN IN SOLUTION TO REMOVE GST AFFINITY TAG The glutathione-S-transferase (GST) affinity tag is removed by cleaving with thrombin (pGEX-T vectors) or factor Xa (pGEX-X vectors). Conditions for optimal cleavage of each recombinant fusion protein must be empirically determined. Some of the parameters that can be varied include temperature, enzyme-to-substrate ratio, length of incubation, and buffer conditions. Proteolysis can usually be performed in the glutathione buffer used to elute the fusion protein from the column, either with or without addition of NaCl or Ca2+. After proteolysis, the glutathione must be removed by dialysis prior to rechromatography on the glutathione column to separate the cleaved target protein from the GST moiety and uncleaved fusion protein. Materials Solution of affinity-purified fusion protein (see Basic Protocol 2, step 15, or Alternate Protocol 2, step 10) Thrombin (Sigma) reconstituted in water at 0.5 U/µl or 1 µg/µl factor Xa (Boehringer Mannheim) 0.15 M PMSF in isopropanol PBS/EDTA/PMSF buffer (see recipe) Beckman J6-B centrifuge and JS-4.2 rotor (or equivalent), 4°C


Additional reagents and equipment for dialysis (APPENDIX 3B) and SDS-PAGE (UNIT 10.1)

6.6.12 Supplement 9


1. To the solution of affinity-purified fusion protein, add an appropriate amount of thrombin (when using pGEX-T vectors) or factor Xa (when using pGEX-X vectors) per microgram of purified fusion protein and digest protein 2 to 8 hr in a shaking water bath at 37°C (thrombin) or 25°C (factor Xa). The appropriate amount of enzyme must first be empirically determined for each fusion protein by surveying a range of conditions in a pilot analytical proteolysis experiment. A convenient method is to digest 100 ìg fusion protein per condition over a range of enzyme-to-substrate ratios and for differing incubation times. Typical digestion times range from 2 to 8 hr and typical enzyme-to-substrate-ratios are 1:100, 1:350, 1:1000, and 1:3000 (in units of enzyme per microgram fusion protein) at 37°C for bovine plasma thrombin, or 1:10, 1:25, 1:50, 1:100, 1:300 (microgram enzyme per microgram fusion protein) at 25°C for bovine plasma factor Xa. At the desired times, proteolysis is stopped by adding an aliquot of the sample to boiling SDS sample buffer, and samples are analyzed on an SDS minigel (UNIT 10.1; 2 ìg/lane) to determine the best digestion conditions. If digestion is incomplete at the highest enzyme-to-substrate ratio tested with digestion times of 6 to 8 hr, consider reengineering the protease cleavage site by introducing a linker between the GST moiety and target polypeptide to decrease steric hindrance. In some cases, >10-fold improvement in cleavage efficiency can be achieved by adding as few as two glycines next to the thrombin site. The SDS gel in Figure 6.6.6 shows a typical thrombin digestion optimization experiment where the amount of enzyme has been varied.

2. Stop preparative digestion by adding a 1:500 dilution of 0.15 M PMSF stock solution. Incubate sample an additional 15 min at 37°C for thrombin or 30 min at 25°C for factor Xa to covalently inhibit the enzyme with the PMSF. 3. Dialyze the sample (APPENDIX 3B) twice versus 2 liters PBS/EDTA/PMSF for a minimum of 4 hr per buffer change at 4°C. Complete removal of glutathione is important if samples are to be rechromatographed on glutathione-Sepharose to remove the GST moiety and uncleaved fusion protein. Larger volumes of dialysate may be necessary depending on sample volume: e.g., if the sample volume is >40 ml, increase the dialysis buffer volume to 4 liters per change or use three changes of buffer. In addition, since glutathione equilibrates slowly during dialysis, when

66 36

18 14 6 1

2

3

4

5

6

7

8

9

10

Figure 6.6.6 SDS gel stained with Coomassie brilliant blue showing pilot thrombin digestions of two recombinant glutathione-S-transferase (GST) fusion proteins. Samples were digested 3 hr at 37°C in buffer with varying enzyme-to-substrate ratios. Lane 1, a 78-kDa fusion protein. Lanes 2 to 5, thrombin digestion of the 78-kDa fusion protein using enzyme-to-substrate ratios of 1:3000, 1:1000, 1:350 and 1:100, respectively. Lane 6, a 45-kDa fusion protein. Lanes 7 to 10, thrombin digestion of the 45-kDa fusion protein using substrate ratios of 1:3000, 1:1000, 1:350, and 1:100, respectively. In each case, an enzyme-to-substrate ratio of 1:1000 was chosen as the optimal digestion condition



Supplement 9

dialysis tubing with a MWCO 3 bed volumes glutathione buffer at 1.5 ml/min. A bed volume is one-half the amount of glutathione-Sepharose that was added to the column. The same 2.5 × 8–cm column used for initial purification of a given fusion protein can be used for repurification of the same target polypeptide after protease cleavage. GlutathioneSepharose must be fully reduced in order for the GST moiety to bind efficiently. If the column was washed with glutathione buffer 10 to 50 mg/liter of E. coli culture grown on an environmental shaker) and the facile purification methods for both initial isolation and subsequent separation of cleaved polypeptide and GST moiety. Since nondenaturing purification conditions are employed, polypeptides that do not normally contain posttranslational modifications usually retain their functional and antigenic properties. Other advantages of this system include availability of several alternative protease cleavage sites and the large number of bacterial hosts that can be used. Purification of most soluble GST fusion proteins is straightforward, and success in the

purification of insoluble products is determined largely by the ability to refold the fusion protein after extracting it from inclusion bodies. Since the GST moiety refolds fairly readily after urea solubilization, the presence of this moiety may facilitate refolding of the adjacent polypeptide. However, the high protein expression level of the vector, even at lower growth temperatures means that the preferred strategy for solubilizing proteins that are initially expressed in inclusion bodies is to attempt to grow the transfected cells at lower temperatures, where the protein often remains in the soluble fraction (see Fig. 6.6.5 and Troubleshooting). Difficulties with vector construction, protein expression, and protease cleavage may be greater when the molecular weight of the desired polypeptide is >100 kDa. The GST moiety can usually be easily removed from the protein of interest by cleavage with thrombin or factor Xa. For some studies, such as some immunological or functional assays, removal of the GST moiety may not be required. An important consideration is that the GST moiety is a dimer; hence, steric hindrance effects and nonequivalence of functional sites on the associated fusion partner within this dimeric structure must be considered.

Troubleshooting Contamination of the glutathione-S-transferase (GST) fusion protein after affinity purification with E. coli host cell proteins is usually an indication that sonication has been too severe. Other contaminants may represent de-



Supplement 9


graded fragments of the fusion protein that might be eliminated by the addition of protease inhibitors and/or carefully keeping all samples, buffers, and sample tubes as cold as possible (i.e., keeping the materials at 0°C, on ice, often results in noticeably lower proteolysis than occurs where samples are kept at 4°C in a refrigerator). Degradation of fusion proteins during expression can sometimes be minimized by adding isopropyl-1-thio-β-D-galactopyranoside (IPTG) later in the bacterial culture growth with cells at higher OD at time of induction and by decreasing the duration of the induction period. If the fusion protein does not completely bind to the glutathione-Sepharose resin, several options may be tried: increasing the quantity of glutathione-Sepharose, decreasing the protein loading flow rate, ensuring that the column is completely reduced by pretreating it with a freshly prepared glutathione buffer within 24 hr prior to using the column for purification, and ensuring that any denaturants or reducing reagents used to treat the sample have been completely removed through exhaustive dialysis. Expression of GST fusion protein in inclusion bodies can be addressed in several ways. The preferred method is to lower the growth temperature from 37° to 30°, 25°, 20°, or 15°C. The optimal growth temperature is that temperature where >50% of the expressed protein is in the supernatant. This strategy is efficient even if the expressed fusion protein yield decreases somewhat at the lower growth temperatures. It is usually more time and cost effective to grow a larger quantity of cells than to attempt to renature fusion protein extracted from inclusion bodies since the latter method nearly doubles the time required to purify the fusion protein. In one study, when the authors’ laboratory expressed eleven GST fusion proteins at 37°C, four proteins were primarily in inclusion bodies and the remaining seven proteins were partially in the supernatant, although a substantial amount of fusion protein was also in inclusion bodies. When these eleven fusion proteins were expressed at 30°C, ten were primarily (>80%) in the soluble fraction. The remaining recombinant protein was primarily in inclusion bodies at either 30° or 25°C, but could be switched to the soluble fraction and produced in high yield by growing the cells at 18°C, as shown in Figure 6.6.5. If lowering the growth temperature does not result in expression of a soluble protein, renaturing the protein after extracting inclusion bodies with a chaotropic reagent such as urea

can be attempted (see Alternate Protocol 1). Finally, even when most of a protein is primarily in inclusion bodies, there may be enough protein in the soluble fraction that it can be purified in low milligram yields per liter of culture. In some cases, using high enzyme-to-substrate ratios and prolonged incubation times is not sufficient to cleave the fusion protein in good yield. To minimize steric hindrance at the thrombin cleavage site, modified pGEX vectors that contain a glycine-rich, nine-amino-acid “kinker” region immediately before or after the thrombin cleavage site (Guan and Dixon, 1991; Hakes and Dixon, 1991) can be tried. Sometimes introduction of as few as one or two extra glycines immediately after to the thrombin site may substantially improve the rate of protease cleavage. The probability that extraneous residues on the N-termini of cleaved polypeptides could significantly affect structural and functional properties of expressed polypeptides is very likely to increase as the length of the extraneous sequence increases. Hence, the number of extra residues remaining on the target polypeptide after protease cleavage should be minimized. Although thrombin cleavage is usually more cost effective than factor Xa cleavage, a minimum of two extra residues from the thrombin recognition site (GS) are left on the cleaved peptide. Although no residues from the factor Xa recognition site are left on the cleaved polypeptide, at least several extraneous amino acids are introduced onto the N-terminal of all pGEX-X vectors from the sequence remaining at the cloning site (see Figs. 6.6.2 and 6.6.3). Depending upon the pGEX vector and cloning site utilized, as few as two and as many as thirteen extra amino acids may remain on the cleaved N-terminal end of the target protein. The yield of GST fusion proteins is typically quite high, i.e., 10 to 50 mg of fusion protein per liter of culture when cells are grown on an environmental shaker. When low protein expression does occur, it can often be improved by optimizing the bacterial culture growth conditions: for example, growing the cells to a somewhat higher OD550 at the time of induction, increasing the length of the induction period, using another strain of E. coli such as a protease-negative strain, or growing the cells at higher temperatures (although proteins are more likely to be in inclusion bodies at higher temperatures; see Background Information). Caution should be taken when using higher cell ODs and longer growth times after induction,

6.6.20 Supplement 9


since the cells may start to lyse under these conditions. For additional information concerning GST fusion protein expression, Pharmacia Biotech publishes a reference list, bulletin number 181117-27, that classifies related publications into several useful categories to assist in troubleshooting a variety of parameters.

Anticipated Results Yields of fusion protein can vary widely. Typical yields are 10 to 50 mg/liter, but can occasionally be much lower, especially if the fusion protein is toxic to the cells or is unstable. In some cases, >50 mg/liter can be obtained when expression conditions have been well optimized. A single-step affinity purification should yield fusion protein that is >90% pure in most cases. The relationship between yield of fusion protein and yield of cleaved, repurified recombinant target polypeptide is in part due to the mass of the target polypeptide. A good final yield of a cleaved protein after repurification on a glutathione column followed by HPLC gel filtration might be ∼2 mg/liter for a 10-kDa protein and ∼10 mg/liter for a 50-kDa protein.

Time Considerations Protein expression takes 1.5 days of intermittent work requiring approximately 3 to 4 hr of operator time. Longer induction periods may be required at temperatures lower than 30°C, but total operator time remains the same. A small-scale batch purification can be completed in 1 day. A large-scale column purification, cleavage of fusion protein, and repurification of the cleaved peptide by affinity and gel filtration chromatography will take ∼5 to 8 days of intermittent work requiring several hours of operator time per day. The purification should be completed in as short a time as practical to minimize proteolysis, aggregation, and precipitation of impure fractions.

Literature Cited Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, K. (eds.). 1994. Current Protocols in Molecular Biology. John Wiley & Sons, New York. Davies, A.H., Jowett, J.B.M., Jones, I.M. 1993. Recombinant baculovirus vectors expressing glutathione-S-transferase fusion proteins. Bio/Technology 11:933-936. Frangioni, J.V. 1992. Solubilization and purification of enzymatically active glutathione-S-transferase (pGEX) fusion proteins. Anal. Biochem. 210:179-187. Gearing, D.P., Nicola, N.A., Metcalf, D., Foote, S., Willson, T.A., Gough, N.M., and Williams, R.L. 1989. Production of leukemia inhibitory factor in Escherichia coli by a novel procedure and its use in maintaining embryonic stem cells in culture. Bio/Technology 7:1157-1161. Grieco, F., Hull, J., and Hull, R. 1992. An improved procedure for the purification of protein fused with glutathione-S-transferase. Biotechniques 13:856-857. Guan, K.L. and Dixon, J.E. 1991. Eukaryotic proteins expressed in Escherichia coli: An improved thrombin cleavage and purification procedure of fusion proteins with glutathione-S-transferase. Anal. Biochem. 192:262-267. Hakes, D.J. and Dixon, J.E. 1991. New vectors for high level expression of recombinant proteins in bacteria. Anal. Biochem. 202:293-298. Mitchell, D.A., Marshall, T.K., and Deschenes, R.J. 1993. Vectors for the overexpression of glutathione-S-transferase fusion proteins in yeast. Yeast 9:715-722. Sambrook, J., Fritsch, E.F., and Maniatis, T. 1989. Molecular Cloning: A Laboratory Manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Smith, D.B. and Johnson, K.S. 1988. Single-step purification of polypeptides expressed in Escherichia coli as fusions with glutathione-Stransferase. Gene 67:31-40.

Key Reference Smith and Johnson, 1988. See above. Original description of the pGEX system.

Contributed by Sandra Harper and David W. Speicher The Wistar Institute Philadelphia, Pennsylvania



Supplement 9

Expression and Purification of Thioredoxin Fusion Proteins

UNIT 6.7

This unit describes a gene fusion expression system that uses thioredoxin, the product of the Escherichia coli trxA gene, as the fusion partner. The system is particularly useful for high-level production of soluble fusion proteins in the E. coli cytoplasm; in many cases heterologous proteins produced as thioredoxin fusion proteins are correctly folded and display full biological activity. Although the thioredoxin gene fusion system is routinely used for protein production, high-level production of peptides—i.e., for use as antigens— is also possible because the prominent thioredoxin active-site loop is a very permissive site for the introduction of short amino acid sequences (10 to 30 residues in length). The inherent thermal stability of thioredoxin and its susceptibility to quantitative release from the E. coli cytoplasm by osmotic shock can also be exploited as useful tools for thioredoxin fusion protein purification. In addition, a more generic method for purification of any soluble thioredoxin fusion employs a modified form of thioredoxin (called “His-patch Trx”), which has been designed to bind to metal chelate resins. Protein fusions to His-patch Trx can usually be purified in a single step from cell lysates (see Strategic Planning). The first step is construction of a fusion of trxA to any desired gene and expression of the fusion protein in an appropriate host strain at 37°C (see Basic Protocol). Additional protocols describe E. coli cell lysis using a French pressure cell and fractionation (see Support Protocol 1), osmotic release of thioredoxin fusion proteins from the E. coli cytoplasm (see Support Protocol 2), and heat treatment to purify some thioredoxin fusion proteins (see Support Protocol 3). STRATEGIC PLANNING The thioredoxin gene fusion expression vectors pTRXFUS and hpTRXFUS, both of which carry the E. coli trxA gene (Fig. 6.7.1), are used for high-level production of C-terminal fusions to thioredoxin. The vector hpTRXFUS differs from pTRXFUS in that it contains a modified E. coli trxA gene which produces a mutant protein (“His-patch” thioredoxin) that can specifically bind to metal chelate matrices charged with nickel or cobalt, otherwise known as native metal-chelate affinity chromatography (MCAC; UNIT 9.4). The trxA translation-termination codon has been replaced in both vectors by DNA encoding a ten-residue peptide linker sequence that includes an enterokinase (enteropeptidase; LaVallie et al., 1993a) cleavage site. This highly specific site can be cleaved with enterokinase following purification of the fusion protein to release the protein of interest from its thioredoxin fusion partner. Immediately downstream of the DNA encoding the enterokinase site in pTRXFUS and hpTRXFUS lies a DNA polylinker sequence containing a number of unique restriction endonuclease sites that can be used for forming in-frame translational fusions of any desired gene to trxA. Downstream of the DNA polylinker lies the E. coli aspA transcription terminator. Replication of these vectors is controlled by a modified colE1 replication origin similar to that found in pUC vectors (Norrander et al., 1983). Plasmid selection and maintenance is ensured by the presence of the β-lactamase gene on the vector. The vector pALtrxA-781 (Fig. 6.7.1) is very similar to pTRXFUS. However in this plasmid the trxA gene is followed by a translation termination codon, and the sequences encoding the enterokinase-site peptide linker are absent. A unique RsrII site, present in both pALtrxA-781 and pTRXFUS, allows for the easy insertion of short peptide-encoding DNA sequences into trxA within the region that encodes the active-site loop. Contributed by John McCoy and Edward LaVallie Current Protocols in Protein Science (1997) 6.7.1-6.7.14 Copyright © 1997 by John Wiley & Sons, Inc.


6.7.1 Supplement 10

BLA

ori

pALtrxA–781 pTRXFUS p L hpTRXFUS

trxA aspA Rsr II

TGGTGCGGTCCGTGCAAA W C G33 P34 C K

Sfi I

pALtrxA–781:

Xba I

Sal I

Pst I

AACCTGGCCTAGCTGGCCATCTAGAGTCGACCTGCAG N L A *

aspA terminator

thioredoxin

Kpn I Bam HI Xba I

pTRXFUS:

Sal I

Pst I

AACCTGGCCGGTTCTGGTTCTGGTGATGACGATGACAAGGTACCCGGGGATCCTCTAGAGTCGACCTGCAG

N

L

thioredoxin

A

G

S

G

S

G

D

D

D

D

K

fusion point

aspA terminator

linker peptide enterokinase site

Figure 6.7.1 Thioredoxin gene fusion expression vectors pTRXFUS, hpTRXFUS, and pALtrxA-781. pALtrxA-781 contains a polylinker sequence at the 3′ end of the trxA gene. pTRXFUS and hpTRXFUS contain a linker region encoding a peptide that includes the enterokinase cleavage site between the trxA gene and the polylinker. The sequence surrounding the active site loop of thioredoxin has a single RsrII site that can be used to insert peptide coding sequence. The asterisk indicates a translational stop codon. Abbreviations: trxA, E. coli thioredoxin gene; BLA, β-lactamase gene; ori, colE1 replication origin; pL, bacteriophage λ major leftward promoter; aspA terminator, E. coli aspartate amino-transferase transcription terminator.


pTRXFUS, hpTRXFUS, and pALtrxA-781 carry the strong bacteriophage λ promoter pL (Shimatake and Rosenberg, 1981) positioned upstream of the trxA gene. Transcription initiation at the pL promoter is controlled by the intracellular concentration of λ repressor protein (cI). cI857-containing strains (Shatzman et al., 1990) can be used for heat inductions of pL at 42°C; alternatively, in the strains carrying the wild-type repressor, pL can be induced by a prior induction of the E. coli SOS stress response. However, it is often desirable to express heterologous genes in E. coli at temperatures considerably lower than 42°C, or under conditions where cells are not undergoing a physiological stress. Strains GI698, GI724 and GI723 were designed to allow the growth and induction of pL expression vectors, including pTRXFUS, hpTRXFUS, and pALtrxA-781, under mild conditions over a wide range of temperatures (see Table 6.7.1; Mieschendahl et al., 1986). Each of these strains carries a wild-type allele of cI stably integrated into the E. coli chromosome at the nonessential ampC locus. A synthetic trp promoter integrated into ampC upstream of the cI gene in each strain directs the synthesis of cI repressor only when intracellular tryptophan levels are low. When tryptophan levels are high, synthesis of cI is switched off; therefore, the presence of tryptophan in the growth medium of GI698, GI723, or GI724 will block expression of λ repressor and thus will turn on pL. Because the three strains carry ribosome-binding sequences of different strengths at the 5′-end of

6.7.2 Supplement 10


Table 6.7.1 E. coli Strains for Production of Thioredoxin Fusion Proteins at Varying Temperatures

Strain

Desired production temperature (°C)

Pre-induction growth temperature (°C)

Induction period (hr)

GI698

15

25

20

GI698 GI698

20 25

25 25

18 10

GI724 GI724

30 37

30 30

6 4

GI723

37

37

5

their respective cI genes, they maintain intracellular concentrations of λ repressor that increase in the order GI698 < GI724 < GI723. The choice of which strain to use for a particular application is dependent on the desired culture conditions as described below. Although some thioredoxin fusion proteins produced at 37°C are insoluble, expression at lower temperatures can often result in the fusion protein being produced in a soluble form. Each of the three pL host strains GI698, GI723, and GI724 is suitable for the production of thioredoxin fusion proteins over a particular temperature range. Table 6.7.1 indicates the correct strain for expression of thioredoxin fusion proteins at any temperature between 15°C and 37°C. The induction protocol at any of these temperatures is the same as that described for induction of GI724 at 37°C (see Basic Protocol), except the preinduction growth temperature and the length of the induction period vary according to the strain used and the temperature chosen. Cultures should be grown at the indicated preinduction growth temperature until they reach a density of 0.4 to 0.6 OD550/ml. They should then be moved to the desired induction temperature and induced by the addition of 100 µg/ml tryptophan. Low-temperature inductions are best performed in strain GI698. However, this strain makes only enough cI repressor protein to maintain the vectors in an uninduced state at temperatures below 25°C. GI698 should therefore never be grown above 25°C when it carries a pL plasmid. A nonrefrigerated water bath can be maintained below room temperature by placing it in a 4°C room and setting the thermostat to the desired temperature. It is often a good idea to collect time points during the course of a long induction period and to fractionate cells from these time points (see Support Protocol 1, steps 9 to 13). Although a particular fusion protein may be soluble during the early part of an induction, during the later phases of induction it may become unstable or its concentration inside the cell may exceed a critical threshold above which it will precipitate and appear in the insoluble fraction.



Supplement 10

BASIC PROTOCOL

CONSTRUCTION AND EXPRESSION OF A THIOREDOXIN FUSION PROTEIN This protocol describes construction and subsequent expression of a gene fusion between trxA (encoding thioredoxin) and a gene encoding a particular protein or peptide. After a clone carrying the correct fusion sequence is constructed, analyzed, and isolated, cultures are grown and expression is induced. The protocol is described in terms of the E. coli host strain GI724 with expression at 30°C; it may also be applied to strains GI698 and GI723 (also available from Genetics Institute) for expression at other temperatures by using the parameters specified in Table 6.7.1 (see Strategic Planning). Materials DNA fragment encoding desired sequence Thioredoxin expression vectors (Fig. 6.7.1): pTRXFUS or pALtrxA-781 (Genetics Institute or Invitrogen) or hpTRXFUS (Genetics Institute) E. coli strain GI724 (Genetics Institute or Invitrogen), grown in LB medium and made competent LB medium (UNIT 5.2) IMC plates (see recipe) containing 100 µg/ml ampicillin CAA/glycerol/ampicillin 100 medium (see recipe) IMC medium (see recipe) containing 100 µg/ml ampicillin 10 mg/ml tryptophan (see recipe) SDS-PAGE sample buffer (see recipe) 30°C convection incubator 18 × 50–mm culture tubes Roller drum (New Brunswick Scientific) 250-ml culture flask 70°C water bath Microcentrifuge, 4°C Additional reagents and equipment for SDS-PAGE (UNIT 10.1), and Coomassie brilliant blue staining (UNIT 10.5) Construct the trxA gene fusion 1. Use DNA fragment encoding the desired sequence to construct either an in-frame fusion to the 3′-end of the trxA gene in pTRXFUS or hpTRXFUS, or a short peptide insertion into the unique RsrII site of pALtrxA-781. A precise fusion of the desired gene to the enterokinase linker sequence in pTRXFUS or hpTRXFUS can be made by using the unique KpnI site trimmed to a blunt end with the Klenow fragment of E. coli DNA polymerase. The desired gene can usually be adapted to this blunt-end construct by using a synthetic oligonucleotide duplex ligated between it and any convenient downstream restriction site close to the 5′ end of the gene. When designing the fusion junction, note that enterokinase is able to cleave —DDDDK↓X—, where X is any amino acid residue except proline. Synthetic oligonucleotides encoding short peptides for insertion into the thioredoxin active-site loop at the RsrII site will insert only in the desired orientation, because the RsrII sticky end consists of three bases.

2. Transform the ligation mixture containing the new thioredoxin fusion plasmid into competent GI724 cells. Plate transformed cells onto IMC plates containing 100 µg/ml ampicillin to select transformants. Incubate plates in a 30°C convection incubator until colonies appear. Expression and Purification of Thioredoxin Fusion Proteins

Strains GI698, GI723, and GI724 are all healthy prototrophs that can grow under a wide variety of growth conditions, including rich and minimal media and a broad range of growth temperatures (see Table 6.7.1). These strains can be prepared for transformation with pL-containing vectors by growing them in LB medium at 37°C. LB medium may also

6.7.4 Supplement 10


be used for these strains during the short period of outgrowth immediately following transformation. This growth period of 30 min to 1 hr is often used to express drug-resistance phenotypes before plating out plasmid transformations onto solid medium. Subsequently, however, these strains should be grown only on minimal or tryptophan-free rich media, such as IMC medium containing 100 ìg/ml ampicillin (for expression of the fusion protein) or CAA/glycerol/ampicillin 100 medium (for plasmid DNA preparations). Except during transformation, LB medium should never be used with these three strains when they carry pL plasmids because LB contains tryptophan. The pL promoter is extremely strong and should be maintained in an uninduced state until needed so that expression of the protein will not lead to selection of mutant or variant cells with lower expression due to undesirable genetic selections or rearrangements in the expression strain.

3. Grow candidate colonies in 5 ml CAA/glycerol/ampicillin 100 medium overnight at 30°C. Prepare minipreps of plasmid DNA and check for correct gene insertion into pTRXFUS by restriction mapping. 4. Sequence plasmid DNA of candidate clones to verify the junction region between thioredoxin and the gene or sequence of interest. Induce expression 5. Streak out frozen stock culture of GI724 containing thioredoxin expression plasmid to single colonies on IMC plates containing 100 µg/ml ampicillin. Grow 20 hr at 30°C. Occasionally there is induction of pL plasmids grown in GI698 and GI724 at 37°C, even in medium containing no tryptophan. Such induction appears to be a temperature-dependent phenomenon. If growth at 37°C prior to pL induction is essential, then GI723 should be used as the host strain because GI723 produces higher levels of cI repressor than both GI698 and to GI724. Otherwise, plasmid-containing GI698 should be grown at 25°C and plasmid-containing GI724 should be grown at 30°C prior to induction (see Table 6.7.1).

6. Pick a single fresh, well-isolated, colony from the plate and use it to inoculate 5 ml IMC medium containing 100 mg/ml ampicillin in an 18 × 150–mm culture tube. Incubate overnight at 30°C on a roller drum. 7. Add 0.5 ml overnight culture to 50 ml fresh IMC medium containing 100 µg/ml ampicillin in a 250-ml culture flask (1:100 dilution). Grow at 30°C with vigorous aeration until absorbance at 550 nm reaches 0.4 to 0.6 OD/ml (∼3.5 hr). 8. Remove a 1-ml aliquot of the culture (uninduced cells). Measure the optical density at 550 nm and harvest the cells by microcentrifuging 1 min at maximum speed, room temperature. Carefully remove all the spent medium with a pipet and store the cell pellet at −80°C. 9. Induce pL by adding 0.5 ml of 10 mg/ml tryptophan (100 µg/ml final) to remaining cells immediately. 10. Incubate 4 hr at 37°C. At hourly intervals during this incubation, remove 1-ml aliquots of the culture and harvest cells as in step 8. 11. Harvest the remaining cells from the culture 4 hr post-induction by centrifuging 10 min at 3000 rpm (e.g., in a Beckman J6 rotor), 4°C. Store the cell pellet at −80°C. Procedures for further analysis of these cells are outlined in the support protocols.

Verify induction 12. Resuspend the pellets from the induction intervals (steps 8 and 10) in 200 µl of SDS-PAGE sample buffer/OD550 cells. Heat 5 min at 70°C to completely lyse the cells and denature the proteins. Run the equivalent of 0.15 OD550 cells per lane (30 µl) on an SDS-polyacrylamide gel (UNIT 10.1).



Supplement 10

13. Stain the gel 1 hr with Coomassie brilliant blue (UNIT 10.5). Destain the gel and check for expression. Most thioredoxin fusion proteins are produced at levels that vary from 5% to 20% of the total cell protein. The desired fusion protein should exhibit the following characteristics: it should run on the gel at the mobility expected for its molecular weight; it should be absent prior to induction; and it should gradually accumulate during induction, with maximum accumulation usually occurring 3 hr post-induction at 37°C. SUPPORT PROTOCOL 1

E. COLI LYSIS USING A FRENCH PRESSURE CELL A small, 3.5-ml French pressure cell can be used as a convenient way to lyse E. coli cells. The whole-cell lysate can be fractionated into soluble and insoluble fractions by microcentrifugation. Other lysis procedures may be used—for example, sonication (UNIT 6.6) or treatment with lysozyme-EDTA (UNIT 6.5). For use of the larger 40-ml French pressure cell, see UNIT 6.2. Materials Cell pellet from 4-hr post-induction culture (see Basic Protocol) 20 mM Tris⋅Cl, pH 8.0 (APPENDIX 2E), 4°C Lysis buffer: 20 mM Tris⋅Cl (pH 8.0) with protease inhibitors (optional)—0.5 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM p-aminobenzamidine (PABA), and 5 mM EDTA SDS-PAGE sample buffer (see recipe) French press and 3.5-ml mini-cell (SLM Instruments), 4°C Additional reagents and equipment for SDS-PAGE (UNIT 10.1) Lyse the cells 1. Resuspend cell pellet from 4-hr post-induction culture in 20 mM Tris⋅Cl, pH 8.0, to a concentration of 5 OD550/ml. Protease inhibitors can be included in the resuspension if desired. Cells can also be resuspended at densities of 100 OD550/ml or greater; however, at high densities cell lysis may be less efficient.

2. Place 1.5 ml resuspended cell pellet in the 3.5-ml French pressure cell. Hold the cell upside down with the base removed, the piston fully extended downwards, and the outlet valve handle that holds the nylon ball seal in the open position (loose). Before filling the pressure cell, check that the nylon ball, which seals the outlet port and sits on the end of the outlet valve handle, is not deformed. If it is, replace it with a new one. Both the condition of the nylon ball and its seat in the pressure-cell body are critical for the success of the procedure.

3. Bring the liquid in the pressure cell to the level of the outlet port by raising the piston slowly to expel excess air from the cell. With the outlet valve open and at the same time maintaining the piston in position, install the pressure-cell base. Gently close the outlet valve. CAUTION: Do not over-tighten the valve as this will deform the nylon ball and may irreparably damage its seat on the pressure-cell body.

4. Turn the sealed cell right-side-up and place it in the hydraulic press. Expression and Purification of Thioredoxin Fusion Proteins

5. Turn the pressure regulator on the press fully counter-clockwise to reset it to zero pressure. Set the ratio selector to medium. Turn on the press. CAUTION: The larger (50-ml) pressure cell is usually used with the selector set on high. The small (3.5-ml) cell is only used on medium ratio.

6.7.6 Supplement 10


6. Slowly turn the pressure regulator clockwise until the press just begins to move. Allow the press to compress the piston. The press will stop moving after a few seconds.

7. Position a collection tube under the pressure-cell outlet. Slowly increase the pressure in the cell by turning the pressure regulator clockwise. Monitor the reading on the gauge and increase the pressure to 1000 on the dial, corresponding to an internal cell pressure of 20,000 lb/in2. 8. While continuously monitoring the gauge, very slowly open the outlet valve until lysate begins to trickle from the outlet. The lysate should flow slowly and smoothly, and the cell pressure should not drop more than 100 divisions on the dial. At 20,000 lb/in2 and 5 OD550/ml, cell lysis will be complete after one passage through the press. Lower pressures and/or higher cell densities may require a second passage.

Fractionate the lysate 9. Remove a 100-µl aliquot of the lysate and freeze at −80°C (whole-cell lysate). 10. Fractionate the remainder of the lysate by microcentrifuging 10 min at maximum speed, 4°C. 11. Remove a 100-µl aliquot of the supernatant and freeze at −80°C (soluble fraction). Discard the remainder of the supernatant. Because this is a pilot experiment, it would not produce enough material to warrant saving any remaining supernatant.

12. Resuspend the pellet in an equivalent volume of lysis buffer. Remove a 100-µl aliquot and freeze at −80°C (insoluble fraction). 13. Lyophilize the 100-µl aliquots to dryness in a Speedvac evaporator. Solubilize in 100 µl SDS-PAGE sample buffer. Analyze 30-µl samples by SDS-PAGE (UNIT 10.1). This crude fractionation provides a fairly reliable indication of whether a protein has folded correctly. Usually proteins in the soluble fraction have adopted a correct conformation and proteins in the insoluble fraction have not. However, occasionally proteins found in the soluble fraction are not truly soluble; instead they form aggregates that do not pellet in the microcentrifuge. Conversely, sometimes a protein found in the insoluble fraction may be there because it has an affinity for cell-wall components and cell membranes, and it may not be intrinsically insoluble. Occasionally proteins can be recovered from these insoluble fractions by extracting with agents such as mild detergents.

OSMOTIC RELEASE OF THIOREDOXIN FUSION PROTEINS Thioredoxin and some thioredoxin fusion proteins can be released with good yield from the E. coli cytoplasm by a simple osmotic shock procedure.

SUPPORT PROTOCOL 2

Materials Cell pellet from 4-hr post-induction cultures (see Basic Protocol) 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA/20% (w/v) sucrose, ice-cold 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA, ice-cold Additional reagents and equipment for SDS-PAGE (UNIT 10.1) 1. Resuspend cell pellet from 4-hr post-induction cultures at a concentration of 5 OD550/ml in ice-cold 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA/20% sucrose. Incubate 10 min on ice. 2. Microcentrifuge 30 sec at maximum speed, 4°C, to pellet the cells.



Supplement 10

3. Discard the supernatant and gently resuspend the cells in an equivalent volume of ice-cold 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA. Incubate 10 min on ice and mix occasionally by inverting the tube. Osmotic release from the cytoplasm occurs at this stage.

4. Microcentrifuge 30 sec at maximum speed, 4°C. Save the supernatant (osmotic shockate). Resuspend the cell pellet in an equivalent volume 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA (retentate). 5. Lyophilize 100-µl aliquots of osmotic shockate and retentate to dryness in a Speedvac evaporator. 6. Solubilize each in 100 µl SDS-PAGE sample buffer. Analyze 30-µl aliquots by SDS-PAGE (UNIT 10.1). The osmotic-shock procedure provides a substantial purification step for some thioredoxin fusion proteins. This procedure will remove most of the contaminating cytoplasmic proteins as well as almost all of the nucleic acids. However the shockate will contain as contaminants about half of the cellular elongation factor-Tu (EF-Tu) and most of the E. coli periplasmic proteins. SUPPORT PROTOCOL 3

PURIFICATION OF THIOREDOXIN FUSION PROTEINS BY HEAT TREATMENT Wild-type thioredoxin is resistant to prolonged incubations at 80°C. A subset of thioredoxin fusion proteins also exhibit corresponding thermal stability, and heat treatment at 80°C can sometimes be used as an initial purification step. Under these conditions the majority of contaminating E. coli proteins are denatured and precipitated. Materials Cell pellet from 4-hr post-induction cultures (See Basic Protocol) 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA SDS-PAGE sample buffer (see recipe) 80°C water bath 10-ml glass-walled tube Additional reagents and equipment for lysis using a French pressure cell (Support Protocol 1) and SDS-PAGE (UNIT 10.1) 1. Resuspend cell pellet from 4-hr post-induction cultures at a concentration of 100 OD550/ml in 20 mM Tris⋅Cl (pH 8.0)/2.5 mM EDTA. It is important to start off with a high protein concentration in the lysate to ensure efficient precipitation of denatured proteins.

2. Lyse the cells at 20,000 lb/in2 in a French pressure cell (see Support Protocol 1, steps 2 to 8). Collect whole-cell lysate in a 10-ml glass-walled tube. 3. Incubate whole-cell lysate 10 min at 80°C. Remove 100-µl aliquots after 30 sec, 1 min, 2 min and 5 min and plunge immediately into ice. At 10 min, plunge the remaining heated lysate into ice.


A glass-walled tube (not plastic) provides good thermal conductivity to provide a rapid rise in temperature to 80°C and then a rapid drop in temperature to 4°C. A suitable volume to use in a 10-ml glass tube is 1.5 ml lysate. For large-scale work, a glass-walled vessel should be used and the lysate should be mixed well during both heat treatment and cooling.

4. Microcentrifuge the aliquots 10 min at maximum speed, 4°C to pellet heat-denatured, precipitated proteins.

6.7.8 Supplement 10


5. Remove 2-µl aliquots of the supernatants and add 28 µl SDS-PAGE sample buffer. Analyze the samples by SDS-PAGE (UNIT 10.1) to determine the heat stability of the fusion protein and the minimum time of heat treatment required to obtain a good purification. REAGENTS AND SOLUTIONS Use deionized, distilled water in all recipes and protocol steps. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Casamino Acids (CAA), 2% (w/v) 20 g Casamino Acids (Difco-certified) H2O to 1 liter Autoclave or filter sterilize through a 0.45-µm filter Store ≤2 months at room temperature Do not use technical-grade Casamino Acids because it has a higher NaCl content.

CAA/glycerol/ampicillin 100 medium 800 ml 2% (w/v) Casamino Acids (see recipe; 1.6% final) 100 ml 10× M9 salts (see recipe; 1× final) 100 ml 10% (v/v) glycerol (sterile; 1% final) 1 ml 1 M MgSO4 (sterile; 1 mM final) 0.1 ml 1 M CaCl2 (sterile; 0.1 mM final) 1 ml 2% (w/v) vitamin B1 (sterile; 0.002% final) 10 ml 10 mg/ml ampicillin (sterile; 100 µg/ml final) Prepare fresh IMC medium 200 ml 2% (w/v) Casamino Acids (see recipe; 0.4% final) 100 ml 10× M9 salts (see recipe; 1× final) 40 ml 20% (w/v) glucose (sterile; 0.5% final) 1 ml 1 M MgSO4 (sterile; 1 mM final) 0.1 ml 1 M CaCl2 (sterile; 0.1 mM final) 1 ml 2% (w/v) vitamin B1 (sterile; 0.002% final) 658 ml glass-distilled H2O (sterile) 10 ml 10 mg/ml ampicillin (sterile; optional; 100 µg/ml final) Use fresh IMC plates 15 g agar [Difco; 1.5% (w/v)] 4 g Casamino Acids [Difco-certified; 0.4% (w/v)] 858 ml glass-distilled H2O (sterile) Autoclave 30 min Cool in a 50°C water bath 100 ml 10× M9 salts (see recipe; 1× final) 40 ml 20% (w/v) glucose (sterile; 0.5% final) 1 ml 1 M MgSO4 (sterile; 1 mM final) 0.1 ml 1 M CaCl2 (sterile; 0.1 mM final) 1 ml 2% (w/v) vitamin B1 (sterile; 0.002% final) 10 ml 10 mg/ml ampicillin (sterile; optional; 100 µg/ml final) Mix well and pour into petri plates Store ≤1 month at 4°C Purification of Recombinant Proteins


Supplement 17

M9 salts, 10× 60 g Na2HPO4 (0.42 M) 30 g KH2PO4 (0.24 M) 5 g NaCl (0.09 M) 10 g NH4Cl (0.19 M) H2O to 1 liter Adjust pH to 7.4 with NaOH Autoclave or filter sterilize through a 0.45-µm filter Store ≤6 months at room temperature SDS-PAGE sample buffer 15% (v/v) glycerol 0.125 M Tris⋅Cl, pH 6.8 (APPENDIX 2E) 5 mM Na2EDTA 2% (w/v) SDS 0.1% (w/v) bromphenol blue 1% (v/v) 2-mercaptoethanol (2-ME; add immediately before use) Store indefinitely at room temperature Tryptophan, 10 mg/ml Heat 500 ml glass-distilled H2O to 80°C. Stir in 5 g L-tryptophan until dissolved. Filter sterilize the solution through a 0.45 µm filter and store ≤6 months in the dark at 4°C. COMMENTARY Background Information


Two significant problems plague researchers who hope to express heterologous proteins in Escherichia coli: inefficient initiation of translation of many eukaryotic mRNA sequences on bacterial ribosomes (Stormo et al., 1982), and proteins that often form insoluble aggregates, called inclusion bodies, that are composed of misfolded or denatured proteins (Mitraki and King, 1989). Although successful protocols for refolding eukaryotic proteins from inclusion bodies can be developed, the process is always uncertain and usually time-consuming; in most instances it is preferable to prevent inclusion-body formation in the first place. The use of trxA fusions provides a solution to both problems. Inefficient initiation of translation of eukaryotic messages in E. coli can often be improved by modifying sequences at the 5′ end of the gene. A more reliable technique that avoids the problem entirely is to use a gene-fusion strategy in which the gene of interest is linked in-frame to the 3′ end of a highly translated partner gene. In this case protein synthesis always initiates on the same efficiently translated fusion-partner mRNA, thus ensuring high-level expression. Some earlier gene-fusion expression systems, for example the trpE and lacZ systems, offer very reliable ways of

producing large quantities of any desired eukaryotic protein. However, these gene-fusion systems still suffer from the pervasive inclusion-body problem. They are thus mainly useful for the production of antigens, rather than correctly folded, biologically active proteins. More recently the maltose binding protein (MBP; Riggs, 1994; UNIT 5.1) and glutathioneS-transferase (GST) gene fusion expression systems (see UNIT 6.6) have proven more successful in producing soluble fusion proteins; these systems retain the translation advantage of the earlier fusion systems. Apart from the obvious advantages in making a correctly folded product, the synthesis of soluble fusion proteins also allows for the development of generic purification schemes based on some unique property of the fusion partner. Why would any particular eukaryotic protein produced in the E. coli cytoplasm be more soluble when it is linked to a fusion partner than it would be by itself? It is likely that physical properties of the fusion-partner protein are important, with efficient self-folding and high solubility being useful in this role. It is possible that some good fusion partners (proteins that fold efficiently and are highly soluble), by virtue of their desirable physical qualities, are able to keep folding intermediates of linked heterologous proteins in solution long enough for



them to adopt their correct final conformations. In this respect the fusion partner may serve as a covalently joined chaperone protein, in many ways fulfilling the role of authentic chaperone proteins (McCoy, 1992), analogous to the covalent chaperone role proposed for the N-terminal pro regions of a number of protein precursors (Silen et al., 1989; Shinde et al., 1993). Many of the known properties of E. coli thioredoxin (Holmgren, 1985) suggested that it would make a particularly effective fusion partner in an expression system. First, thioredoxin, when overproduced from plasmid vectors, can accumulate to 40% of the total cellular protein, yet even at these expression levels all of the protein remains soluble. Second, the molecule is small (11,675 Mr) and would contribute a relatively modest amount to the total mass of any fusion protein, in contrast to other systems such as the lacZ system. Third, the tertiary structure of thioredoxin (Katti et al., 1990) reveals that both the N- and C-termini of the molecule are accessible on the surface and in good position to link to other proteins. The structure also shows that the molecule has a very tight fold, with >90% of its primary sequence involved in strong elements of secondary structure. This provides an explanation for thioredoxin’s observed high thermal stability (Tm = 85°C), and suggests that the molecule might possess the robust folding characteristics that could make it a good fusion-partner protein. In support of this view, complete thioredoxin domains are found in a number of naturally occurring multidomain proteins, including E. coli DsbA (Bardwell et al., 1991), the mammalian endoplasmic reticulum proteins ERp72 (Mazzarella et al., 1990), and protein disulfide isomerase (PDI; Edman et al., 1985). These proteins can all be considered as natural precedents for thioredoxin fusion proteins. The synthesis of small peptides in E. coli is often difficult, with the products frequently being extensively degraded or insoluble. The thioredoxin tertiary structure revealed that the characteristic active site, —CGPC—, protrudes from the body of the protein as a surface loop, with few interactions with the rest of the molecule. The loop does not seem to contribute to the overall stability of thioredoxin, so the production of peptides as insertions at this site was an attractive possibility. In this location they would be protected from host-cell aminoand carboxypeptidases, and thioredoxin’s high solubility should help keep them in solution. In addition, the conformation of peptides inserted at this position would be constrained, which

could be an advantage for applications in which it is desirable for the peptide to adopt a particular form. Thioredoxin has indeed proven to be an excellent partner for the production of soluble fusion proteins in the E. coli cytoplasm (LaVallie et al., 1993b). Figure 6.7.2 demonstrates the production of soluble fusion proteins between thioredoxin and eleven human and murine cytokines and growth factors using the trxA vectors. All of these mammalian proteins had been previously produced in E. coli only as insoluble inclusion bodies. As thioredoxin fusions, the growth factors are not only made in a soluble form, but in most cases they are also biologically active in in vitro assays. Experience gained while working with these and a number of other trxA fusion proteins shows that two further characteristics of thioredoxin can be exploited as purification tools. The first is the inherent thermal stability of the molecule, a property that is retained by some thioredoxin fusion proteins. This enables heat treatment to be used as an effective purification step. The second additional property relates to thioredoxin’s cellular location. Although E. coli thioredoxin is a cytoplasmic protein, it has been shown to occupy a special position within the cell—it is primarily located on the cytoplasmic face of the adhesion zones that exist between the inner and outer membranes of the E. coli cell envelope (Lunn and Pigiet, 1982). From this location thioredoxin is quantitatively released to the exterior of the cell by simple osmotic shock or freeze/thaw treatments, a remarkable property that is retained by some thioredoxin fusion proteins, thus providing a simple purification step. A more generic method for purification of any soluble thioredoxin fusion employs a modified form of thioredoxin (called “Hispatch Trx”), which has been designed to bind to metal chelate resins (Lu et al., 1996; UNIT 9.4). If the fusion protein is soluble after lysis or osmotic release (see Support Protocols 1 and 2), then any of the conventional purification methods described in Chapter 8 can be used. Alternatively, an affinity-based purification method can be used where an immobilized arsenical compound forms an adduct with the redox-sensitive vicinal dithiols present at the active site of thioredoxin. This affinity matrix was originally prepared by covalently linking 4-aminophenylarsenine oxide to cyanogen bromide–activated Sepharose 4B (Hannestad et al., 1982). Hoffman and Lane (1992) coupled the same ligand to Affi-Gel 10 (Bio-Rad); the



Supplement 10

1

2

3

4

5

6

7

8

9

10

11

12

MW(kDa) 97.4 66.2

45.0

31.0

21.5

14.4

Figure 6.7.2 Expression of thioredoxin gene fusions. The gel shows proteins found in the soluble fractions derived from E. coli cells expressing eleven different thioredoxin gene fusions. Lane 1, host E. coli strain GI724 (negative control, 37°C); lane 2, murine interleukin-2 (IL-2; 15°C); lane 3, human IL-3 (15°C); lane 4, murine IL-4 (15°C); lane 5, murine IL-5 (15°C); lane 6, human IL-6 (25°C); lane 7, human MIP-1a (37°C); lane 8, human IL-11 (37°C); lane 9, human macrophage colony-stimulating factor (M-CSF; 37°C); lane 10, murine leukemia inhibitory factor (LIF; 25°C); lane 11, murine steel factor (SF; 37°C); and lane 12, human bone morphogenetic protein-2 (BMP-2; 25°C). Temperatures in parentheses are the production temperature chosen for expressing each fusion. This is a 10% SDS-polyacrylamide gel, stained with Coomassie brilliant blue.


resultant affinity matrix has a 10-atom spacer which makes it more useful for affinity chromatography. The resin is commercially available from Invitrogen (ThioBond) and Sigma. Protocols detailing the affinity-based purification approach can be obtained on-line from the I nvitr og en web site (http://www.invitrogen.com/manuals.html), and a recent example from the literature is the expression of human glutamate decarboxylase (Papouchado et al., 1997). Other specific purification approaches include immunoaffinity chromatography using antibodies against either the target protein or against the thioredoxin moiety (monoclonal antibodies against thioredoxin are commercially available). Thioredoxin fusions with additional histidine tags or thioredoxin fusions based on mutant thioredoxins with metalchelating affinity (Lu et al., 1996) can be purified using metal-chelate chromatography (see UNIT 9.4). Finally, if the fusion protein is insoluble, then attempts to solubilize and fold the fusion protein can be made prior to purification,

in a technique analogous to that used with insoluble GST fusion proteins (see UNIT 6.6). Thioredoxin from E. coli is a very soluble protein that can be readily denatured and refolded (Kelley et al., 1987).

Critical Parameters Lack of protein solubility leading to inclusion-body formation in E. coli is a complex phenomenon with many contributing factors: simple insolubility as a result of high-level expression, insolubility of protein-folding intermediates, lack of appropriate bacterial chaperone proteins, and lack of glycosylation mechanisms in the bacterial cytoplasm. Fusion of heterologous proteins to thioredoxin or to other fusion partners can help address most of these solubility issues. However, another important factor contributing to inclusion body formation is the inability to form essential disulfide bonds in the reducing environment of the bacterial cytoplasm, which leads to incorrect folding. Thermal lability of even correctly



folded heterologous proteins in the absence of these stabilizing disulfide cross-links is a significant problem, so the expression of fusion genes should be attempted over a wide range of temperatures, even as low as 15°C (the limit for E. coli growth is ∼8°C). Thermal denaturation is a time-dependent process, so it is also prudent to monitor the solubility of the expressed fusion protein over the time course of induction. A great many proteins contain distinct structural domains. For example, hormone receptor proteins usually have an extracellular ligandbinding domain, a transmembrane region, and an intracellular effector domain. Sometimes, expressing these domains individually as fusion proteins can yield better results than expressing the entire protein. The exact positions chosen for boundaries of the domains to be expressed in the fusion protein are important and can be determined from a knowledge of the tertiary structure of the protein of interest, by homology comparisons with similar proteins, by limited proteolysis or other domain-mapping experiments, or empirically by generating multiple fusions that test different boundary positions. It is important to be consistent in treating samples for loading on gels. For example, using different heating conditions from one experiment to the next can result in a mobility shift for the protein of interest.

Anticipated Results Thioredoxin fusion protein yields are usually in the range of 5% to 20% of total cell protein. At these expression levels, a 1-liter induction culture in a shaker flask will yield ∼3 g (wet weight) of cells, 300 mg total protein, and 15 to 60 mg of thioredoxin fusion protein. The final recovered yield will depend on factors such as solubility of the fusion protein and the efficiency of downstream purification procedures.

Time Considerations From a single colony on a plate, the basic induction protocol requires an overnight growth to prepare a liquid inoculum and a 3.5-hr preinduction growth at 30°C the next day, followed by a 4-hr 37°C induction period. These times are significantly longer if lower induction temperatures are required (see Table 6.7.1). Lysis of a sample in the French pressure cell should require ≤ 5 min, and both the heattreatment and osmotic-shock procedures require exc dT

Tm

Figure 7.9.10

Tm always differs from the temperature of maximum partial heat capacity, Tmax, due to the difference in the heat capacities between the native and unfolded states, ∆Cp = Cp,U − Cp,N. Tmax − Tm decreases with an increase in the relative ratio of the area under the excess heat capacity profile to ∆Cp. Calorimetric enthalpy, ∆Hexp (Tm). Calorimetric enthalpy is defined as the area under the excess heat capacity function, i.e.,

I

∞

∆Hexp (Tm ) = Qtot = < C p (T ) > exc dT 0

Equation 7.9.11 Measuring Protein Thermostability by Differential Scanning Calorimetry

Equation 7.9.12

The enthalpy of ionization arises from two different processes, ionization of protein groups and ionization of buffer; i.e.: ion ion ∆H ion ( Tm ) = ∆H pr ( Tm ) + ∆v ⋅ ∆H buf ( Tm )

Equation 7.9.13

0

Tm

∆Hexp ( Tm ) = ∆H cnf ( Tm ) + ∆H ion ( Tm )

Qtot

T

=

release protons that will be absorbed or released by the buffer. This process will be accompanied by heat effects and will be included in the experimentally measured calorimetric enthalpy, ∆Hcal(Tm). Thus ∆Hcal(Tm) consists of two parts, the conformational enthalpy of protein unfolding, ∆Hcnf(Tm), and the ionization effects, ∆Hion (Tm):

and is referred to as the enthalpy at the transition temperature. Enthalpy of the conformational transition, ∆Hcnf(Tm). When a protein unfolds, internal groups buried in the native state can bind or

where Hion pr (Tm) is the enthalpy of ionization of protein, Hion buf (Tm) is the enthalpy of ionization of buffer, and ν is the change in the protonation state of the protein (see below). When the enthalpies of ionization of protein groups and of the buffer are close, the calorimetric and conformational enthalpies are also close to each other, i.e, ∆Hcal (Tm) ≅ ∆Hcnf(Tm). Such a situation occurs in glycine or acetate buffer systems with an acidic pH. In the acidic pH range, the most probable titratable groups of the protein are the carboxylic acids (aspartic and glutamic acid). Their heats of ionization are very close to the heats of ionization for carboxylic groups of glycine or acetic acid. However, at neutral or alkaline pH, titratable groups (histidine, lysine, arginine, tyrosine, and cysteine) have heats of ionization very different from those of the buffer systems used (e.g., phosphate and cacodylate). In these cases, differences between calorimetric and conformational enthalpies can be significant (see, for example, Yu et al., 1994). Changes in the protonation state of the protein upon unfolding, ∆ν. Under certain solvent conditions protein unfolding is accompanied by a change in the protonation state of the protein, ∆ν. This difference in the number of protons bound to the unfolded, νU, and native, νN, states can be determined from the dependence of the transition temperature on the pH values of the solution dTm/dpH as: ∆ν = νU − ν N =

∆H exp ( Tm ) 2.303 ⋅ R ⋅ Tm2

⋅

dTm dpH

Equation 7.9.14



dTm =0 dpH

60 55

T m (oC)

50

dTm >0 dpH

dTm >0 dpH

45 40 35 30 25

2

3

4

5

6

7

8

9

10

pH

Figure 7.9.2 Dependence of the transition temperature on solution pH for a hypothetical protein.

Typically, the curve representing the dependence of the transition temperature on the pH of the solution has a bell shape (Fig. 7.9.2). On the left shoulder, dTm/dpH > 0 and thus ∆ν < 0—i.e., the protein releases protons upon unfolding. On the right shoulder of Tm dependence on pH, dTm/dpH < 0 and thus the protein absorbs protons upon unfolding (∆ν > 0). At the maximum of the Tm(pH) function, dTm/dpH = 0 and thus the protonation states of the native and unfolded states of a protein are identical (∆ν = 0). Van’t Hoff enthalpy, ∆HνH (Tm), and modes of unfolding transition. The so-called van’t Hoff enthalpy for a two-state unfolding process is determined by the temperature dependence of the equilibrium constant, Keq: ∆H vH (Tm ) = − R ⋅

d ln K eq d (1 / T )

= R⋅T2 ⋅

Equation 7.9.15

where Keq is defined as K eq ( T ) =

FN ( T ) FN ( T ) + FU ( T )

Equation 7.9.16

d ln K eq dT

From this definition, it follows that van’t Hoff enthalpy can be estimated from Equation 7.9.8 or Equation 7.9.9. However, for the calorimetric curves ∆HνH (Tm) can be estimated as: ∆H vH ( Tm ) =

4 ⋅ R ⋅ Tm2 ⋅ < C p ( Tm ) > max ∆H exp ( Tm )

Equation 7.9.17

where max is the maximum value of the excess heat capacity function. A direct estimate of two enthalpies—experimental calorimetric enthalpy, ∆Hexp (Tm), and van’t Hoff enthalpy, ∆HνH(Tm)—from a single DSC experiment provides a unique way to determine the modes of protein unfolding (Biltonen and Freire, 1978; Privalov and Potekhin, 1986; Kidokoro and Wada, 1987; Marky and Breslauer, 1987; Ghosaini et al., 1988). When the ratio β = ∆HvH(Tm)/∆Hcal(Tm) equals 1 (in practice, ∼0.95 to 1.05), this indicates that the observed transition is two-state, proceeding from the native to the unfolded state without a significant population of intermediates: i.e., the unfolding reaction can be represented by N ↔ U. Deviations from unity indicate that the transition is more complicated. There are two possible explanations of β deviating from unity for

Characterization of Recombinant Proteins


Supplement 12

Measuring Protein Thermostability by Differential Scanning Calorimetry

the reversible unfolding of the protein showing a single-isotherm DSC profile. 1. The reaction is not monomolecular. Information on whether or not the reaction is monomolecular can be obtained from the dependence of the transition temperature for the reversible reaction on the protein concentration. The possibility of oligomerization of the protein in the native or/and unfolded states should be examined by performing experiments under exactly the same solvent conditions using protein concentrations that differ by at least a factor of 5. The dependence of the transition temperature, Tm, on protein concentration, w, will be an indication of the mode of unfolding reaction:

of the heat capacity functions of the unfolded and native states at the transition temperature: ∆Cp = Cp,U(Tm) − Cp,N(Tm). An alternative, indirect method for the determination of ∆Cp uses a basic thermodynamic equation, ∆Cp = (∂∆H/∂T)p, which shows that the slope of the dependence of the conformational enthalpy of unfolding, ∆Hcnf(Tm), on the transition temperature, Tm, is equal to ∆Cp. Heat capacity change upon unfolding, ∆Cp, defines the temperature dependencies of the enthalpy

a. ∂Tm/∂w = 0: N ↔ U, unfolding reaction is monomolecular; b. ∂Tm/∂w > 0: N ↔ nU, unfolding reaction is of nth order and the native protein exists in an oligomeric state; c. ∂Tm/∂w < 0: N ↔ Un, unfolding reaction is of nth order and the unfolded protein exists in an oligomeric state.

Equation 7.9.18

Situations (a) and (b) are very common and the thermodynamic formalism for their analysis is well developed. Situation (c) is rare for reversible reactions, but does occur for irreversible reactions when irreversibility is caused by aggregation (usually nonspecific) in the unfolded state. 2. Intermediate states. An alternative explanation for β being >1 is the existence of intermediate states during equilibrium reversible unfolding: i.e., N ↔ I1 ↔ I2, ... In ↔ U. Consider the most common example of such an unfolding reaction, which is the unfolding of a two-domain protein. Both domains have similar intrinsic stability, ∆G1(T) ≈ ∆G2(T), and their interactions with one another are described by Gibbs free energy ∆G12(T). The ratio of calorimetric to van’t Hoff enthalpy, β, will depend on the strength of the interdomain interactions. If ∆G12(T) = 0, i.e., there is no interaction between domains, each domain will unfold independently of the other, and β = 2. If there is a strong interaction between domains, i.e., neither domain can exist in a folded state if the other domain is unfolded, unfolding will be very cooperative and β = 1. Heat capacity change upon unfolding, ∆Cp, and thermodynamic functions for protein unfolding: enthalpy, ∆H(T), entropy, ∆S(T), and Gibbs free energy, ∆G(T). One way of estimating the heat capacity change upon protein unfolding is directly from the difference

I

T

∆H (T ) = ∆H (T0 ) +

∆C p ⋅ dT

T0

and entropy functions,

I

T

∆S (T ) = ∆S(T0 ) +

∆C p ⋅ d ln T

T0

Equation 7.9.19

and thus the Gibbs free energy function: ∆G(T ) = − R ⋅ T ⋅ ln Keq = ∆H (T ) − T ⋅ ∆S( T ) Equation 7.9.20

The integration constants ∆H(To) and ∆H(To) in Equation 7.9.19 and Equation 7.9.20 have to be determined experimentally. The integration constant the enthalpy function can be set as ∆H(To) = ∆Hexp(Tm). The integration constant for the entropy function (Equation 7.9.19) for a two-state transition is determined as follows. If the transition is two-state—i.e., ∆Hexp(Tm) ≈ ∆HνH(Tm)—then at the transition temperature (Tm) the population of native and unfolded states is equal. This means that at Tm the equilibrium constant of the reaction Keq = FN/FU is equal to 1, and thus the Gibbs free energy difference between the native and unfolded states at Tm is equal to zero and ∆G(Tm) = ∆H(Tm) − Tm⋅∆S(Tm) = 0. Thus the entropy change upon protein unfolding at the transition temperature, ∆S(Tm), is equal to: ∆S ( Tm ) =

∆H ( Tm ) Tm

Equation 7.9.21



This equation provides a constant of integration for Equation 7.9.20, ∆S(Tm) = ∆S(To). A complete thermodynamic description of the system at any other temperature can then be calculated as follows:

I

T

∆H (T ) = ∆H (Tm ) +

∆C p ⋅ dT

Tm

I I

T

∆Cp T

Tm

=

∆H(Tm ) Tm

⋅ dT

T

+

∆Cp ⋅ d ln T

Tm

Equation 7.9.23

∆G(T ) = ∆H (Tm ) − T ⋅

I

T

+

I

∆H (Tm ) Tm

T

∆C p ⋅ dT − T ⋅

Tm

Gill, S.C. and von Hippel, P.H. 1989. Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182:319-326. Izatt, R.M. and Christensen, J.J. 1976. Heats of proton ionization, pK, and related thermodynamic quantities. In CRC Handbook of Biochemistry and Molecular Biology, Vol. 1: Physical and Chemical Data (G.D. Fasman, ed.) pp. 151-269. CRC Press, Cleveland.

Equation 7.9.22

∆S(T ) = ∆S(Tm ) +

Ghosaini, L.R., Brown, A.M., and Sturtevant, J.M. 1988. Scanning calorimetric study of the thermal unfolding of catabolite activator protein from Escherichia coli in the absence and presence of cyclic mononucleotides. Biochemistry 27:52575261.

∆C p ⋅ d ln T

Tm

Equation 7.9.24

where the heat capacity change upon protein unfolding, ∆Cp, is a function of temperature itself. In practice, however, if no extrapolations over a wide temperature range are necessary, as a first approximation a temperature-independent ∆Cp is a reasonable assumption.

Time Considerations Sample preparation (of which dialysis is the most time-consuming step) usually requires 12 to 16 hr. Consult UNIT 4.4 for possible ways of reducing dialysis time. Baseline data is usually collected in overnight experiments; a calorimetric experiment with a protein sample requires several hours, depending on the heating rate. However, optimization of experimental conditions (reversibility and equilibrium) when first working with an unfamiliar sample may take several days. Time requirements for data analysis vary in individual cases.

Literature Cited Biltonen, R.L. and Freire, E. 1978. Thermodynamic characterization of conformational states of biological macromolecules using differential scanning calorimetry. CRC Crit. Rev. Biochem. 5:85124.

Jaenicke, L. 1974. A rapid micromethod for the determination of nitrogen and phosphate in biological material. Anal. Biochem. 61:623-627. Kidokoro, S. and Wada, A. 1987. Determination of thermodynamic functions from scanning calorimetry data. Biopolymers 26:213-229. Krishinan, K.S. and Brandts, J.F. 1978. Scanning calorimetry. Methods Enzymol. 49:3-14. Mabrey, S. and Sturtevant, J.M. 1976. Investigation of phase transitions of lipids and lipid mixtures by sensitivity differential scanning calorimetry. Proc. Natl. Acad. Sci. U.S.A. 73:3862-3866. Makhatadze, G.I. 1998. Heat capacities of amino acids, peptides and proteins. Biophys. Chem. 71:1-26. Makhatadze, G.I., Medvedkin, V.N., and Privalov, P.L. 1990. Partial molar volumes of polypeptides and their constituent groups in aqueous solution over a broad temperature range. Biopolymers 30:1001-1010. Marky, L.A. and Breslauer, K.J. 1987. Calculating thermodynamic data for transitions of any molecularity from equilibrium melting curves. Biopolymers 26:1601-1620. Pace, C.N., Vajdos, F., Fee, L., Grimsley, G., and Gray, T. 1995. How to measure and predict the molar absorption coefficient of a protein. Protein Sci. 4:2411-2423. Plotnikov, V.V., Brandts, J.M., Lin, L.N., and Brandts, J.F. 1997. A new ultrasensitive scanning calorimeter. Anal. Biochem. 250:237-244. Privalov, G., Kavina, V., Freire, E., and Privalov, P.L. 1995. Precise scanning calorimeter for studying thermal properties of biological macromolecules in dilute solution. Anal. Biochem. 232:79-85. Privalov, P.L. and Potekhin, S.A. 1986. Scanning microcalorimetry in studying temperature-induced changes in proteins. Methods Enzymol. 131:4-51. Privalov, P.L. and Khechinashvili, N.N. 1974. A thermodynamic approach to the problem of stabilization of globular protein structure: A calorimetric study. J. Mol. Biol. 86:665-684. Sanchez-Ruiz, J.M. 1992. Theoretical analysis of Lumry-Eyring models in differential scanning calorimetry. Biophys. J. 61:921-935.



Supplement 12

Scopes, R.K. 1974. Measurement of protein by spectrophotometry at 205 nm. Anal. Biochem. 59:277-282. Sturtevant, J.M. 1972. Calorimetry. Methods Enzymol. 26:227-253. Winder, A.F. and Gent, W.L. 1971. Correction of light-scattering errors in spectrophotometric protein determinations. Biopolymers 10:12431251. Wintrode, P.L., Makhatadze, G.I., and Privalov, P.L. 1994. Thermodynamics of ubiquitin unfolding. Proteins 18:246-253. Yu, Y., Makhatadze, G.I., Pace, C.N., and Privalov, P.L. 1994. Energetics of ribonuclease T1 structure. Biochemistry 33:3312-3319.

Key References

Pace et al., 1995. See above. Detailed discussion on the calculation of protein extinction coefficients from amino acid compositions. Privalov, P.L. 1979. Stability of proteins: Small globular proteins. Adv. Protein Chem. 33:167241. Privalov, P.L. 1982. Stability of proteins: Proteins which do not present a single cooperative system. Adv. Protein. Chem. 35:1-104. Makhatadze, G.I. and Privalov, P.L. 1995. Energetics of protein structure. Adv. Protein. Chem. 47:307-425. Three reviews describing and assessing the determination and interpretation of protein stability by DSC.

Biltonen and Freire, 1978. See above.

Sanchez-Ruiz, 1992. See above.

Privalov and Potekhin, 1986. See above.

Detailed discussion of the analysis of irreversible denaturation processes.

Kidokoro and Wada, 1987. See above. Three references that discuss in detail the theoretical background for deconvolution of the heat capacity profile, each describing a separate approach to data analysis. Izatt and Christensen, 1976. See above.

Contributed by George I. Makhatadze Texas Tech University Lubbock, Texas

Basic reference containing tables of heats of ionizations and other thermodynamic information for different buffer systems.

Measuring Protein Thermostability by Differential Scanning Calorimetry



Characterizing Recombinant Proteins Using HPLC Gel Filtration and Mass Spectrometry

UNIT 7.10

It is essential to determine the identity, sequence integrity, purity, oligomeric state, and conformational integrity of a protein before structural or functional studies are carried out. This is true for proteins isolated from any source, but is particularly important for recombinant proteins, which are subject to several problems associated with their artificial overexpression in a heterologous host. These problems include expression in a denatured state, aggregation, mutations, incorrect translation (e.g., stop codon readthrough), incorrect proteolytic processing, and unwanted post-translational modifications. Often these problems cannot be detected using SDS-PAGE. Fortunately, most sources of heterogeneity in recombinant proteins can be identified by a preliminary analysis of the sample using a combination of analytical high-performance liquid chromatography/fast protein liquid chromatography (HPLC/FPLC) and matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS). Analytical gel filtration is used to detect and quantify contaminants in the sample (including protein aggregates), and can give an approximate measure of the protein size in solution (Stokes radius). Mass spectrometry is then used to determine the molecular weight of the protein to within a few daltons, which enables the detection and identification of incorrect sequence length and most chemical modifications of amino acid side chains. Apart from detecting unwanted modifications to the protein of interest, HPLC and mass spectrometry can be used to help characterize post-translational modifications such as glycosylation (see Chapter 12). In this unit, analytical HPLC gel filtration is used to characterize a purified recombinant protein sample in the Basic Protocol. A method for determining the Stokes radius of a protein using HPLC is described in Support Protocol 1. MALDI-MS is used to characterize a protein sample in Support Protocol 2. NOTE: Milli-Q-purified or equivalent quality water should be used for all buffers. STRATEGIC PLANNING Selecting Column Type and Size The first step in planning a size-exclusion fractionation is selecting the type of column to be used. Several factors need to be considered. Most importantly, the molecular weight of the recombinant proteins to be analyzed should fall within the fractionation range of the column, which is determined by the pore size of the particles. The fractionation range should be large enough to encompass the molecular weights of all species to be separated, and the protein of interest should be near the midpoint of the fractionation range. The fractionation range may be increased by placing two columns with different pore sizes in series. Further selection of the column depends upon several related factors: the required resolution, the sample volume, and the desired run time. For analytical gel filtration, the aim is to maximize resolution while minimizing run time. Resolution is affected by the particle size and size distribution, as well as the column length and the flow rate (for a detailed discussion, see UNITS 8.1 & 8.3, or Neue, 1997). Columns with small, rigid, well-defined particles (500 psi). Resolution may Contributed by Gillian E. Begg, Sandra L. Harper, and David W. Speicher Current Protocols in Protein Science (1999) 7.10.1-7.10.15 Copyright © 1999 by John Wiley & Sons, Inc.



Table 7.10.1

Common Prepacked Columns for HPLC/FPLC Size-Exclusion Chromatographya

Fractionation range (kDa)

Resinb

Particle size (µm)

Dimensions available (i.d. × length, mm)

Supplierc

5-150 10-500

S S

4 4

4.6 × 300 4.6 × 300

TH TH

5-150 10-500 20-10,000 5-100

S S S S

5 5 8 10

TH TH TH TH

TSK G3000SW

10-500

S

10

TSK G4000SW

20-7000

S

13

Bio-Sil SEC 125 Bio-Sil SEC 250 Bio-Sil SEC 400 Protein-Pak 60 Protein-Pak 125 Protein-Pak 200SW Protein-Pak 300SW Superose 12 HR 10/30 Superdex Peptide Superdex 75 HR Superdex 200 HR Superose 6 HR

5-100 10-300 20-1000 1-20 2-80 1-60 10-300 1-3000 0.1-7 3-70 10-600 5-5000

S S S S S S S CL A CL A/D CL A/D CL A/D CL A

5 5 5 10 10 10 10 10 13 13 13 13

7.8 × 300 7.8 × 300 7.8 × 300 7.5 × 300 7.5 × 600 7.5 × 300 7.5 × 600 7.5 × 300 7.5 × 600 7.8 × 300 7.8 × 300 7.8 × 300 7.8 × 300 7.8 × 300 8.0 × 300 8.0 × 300 10.0 × 300 10.0 × 300 10.0 × 300 10.0 × 300 10.0 × 300

Preparative TSK G2000SW

5-100

S

13

TSK G3000SW

10-500

S

13

TSK G4000SW

20-7000

S

17

TSK G2000SW

5-100

S

20

TSK G3000SW

10-500

S

20

Superdex 30 HiLoad

1-10

CL A/D

34

Superdex 75 HiLoad

3-70

CL A/D

34

Superdex 200 HiLoad

10-600

CL A/D

34

Column type Microanalytical TSK Super SW2000 TSK Super SW3000 Analytical TSK G2000SWXL TSK G3000SWXL TSK G4000SWXL TSK G2000SW

21.5 × 300 21.5 × 600 21.5 × 300 21.5 × 600 21.5 × 300 21.5 × 600 55.0 × 300 55.0 × 600 55.0 × 300 55.0 × 600 16.0 × 600 26.0 × 600 16.0 × 600 26.0 × 600 16.0 × 600 26.0 × 600

TH TH BR BR BR WA WA WA WA APB APB APB APB APB TH TH TH TH TH TH TH TH

aData reported by manufacturers. bAbbreviations: CL A, cross-linked agarose; CL A/D, cross-linked agarose/dextran; S, coated silica. cAbbreviations: APB, Amersham Pharmacia Biotech; BR, Bio-Rad; TH, TosoHaas; WA, Waters. Addresses and phone

numbers of suppliers are provided in the SUPPLIERS APPENDIX.

Characterizing Recombinant Proteins Using HPLC and MS



be further enhanced by increasing the length of the column; if a longer column is not available, two columns may be placed in series to achieve the same effect. Low-pressure gel filtration columns with softer, larger particles are generally less expensive and allow larger protein loads, but require lower flow rates, which markedly increases the run time for the separation while providing reduced resolution even at low flow rates. If it is necessary to scale up an analytical separation, the sample volume will determine the size of the column required—the larger the sample volume and amount of protein to be separated, the greater the column volume needed. One option is to increase the length of the column; for example, a 60-cm column has twice the sample volume capacity of a 30-cm column, with little loss of resolution. Increasing the diameter of the column will also increase the maximum load in proportion to the square of the increase in diameter. Table 7.10.1 lists prepacked columns for both preparative work (i.d. = 21.5 or 55 mm) and analytical work (i.d. ≤ 10 mm). Another factor to consider when selecting a column is the potential for interactions between the sample and the column. For example, silica-based columns are available in the smallest particle sizes and can be operated at higher pressures compared to agarose gels; however, some adsorption can occur due to electrostatic interactions with unmodified silanol groups. Usually these effects can be minimized by maintaining at least a physiological ionic strength. Stainless steel columns can withstand high back pressures, but also have drawbacks. For example, some samples might interact with the metal surfaces (especially stainless steel end frits) and common buffer components (e.g., halogens such as chloride) can react with incompletely passivated stainless steel surfaces such as the compression screws. For samples or buffer systems that are incompatible with stainless steel, glass columns should be used. Selecting Instrument Type An isocratic (one pump, one buffer) HPLC or FPLC system is required for the columns described in this unit. Selection of an appropriate chromatography system involves two major decisions: operating pressure range and whether or not biocompatibility is needed. FPLC, HPLC, and biocompatible systems are appropriate for different types of applications. An FPLC system is used to run columns that require moderate back pressure (typically 50 to 600 psi). An HPLC system is used to run columns with higher back pressures (typically >500 psi). Biocompatible systems are available for operation in a variety of pressure ranges and are used in cases where the protein of interest may be adversely affected by trace metal ions that could exist with a system containing stainless steel. Biocompatibility is a heterogeneously applied term, generally assumed to mean a system with few or no stainless steel components. The importance of biocompatibility for most protein applications has not been clearly established. Many companies sell integrated computer-controlled systems that consist of reliable pumps, an injector, a UV detector, a fraction collector, and a computer output for data collection and manipulation. Of course, a dual-pump HPLC or FPLC system may be used by utilizing only one of the pumps. It is most critical to have a pumping system that can deliver a constant flow rate in the desired range, which is typically 0.1 to 1.5 ml/min for analytical columns and up to 5 ml/min for preparative columns. Because of the small number of samples typically analyzed, an autosampler is not necessary. However, it is important to have a computer integrated with the injector to ensure precise starting points for sample injection. While most binary HPLC and FPLC systems are relatively expensive, there are less expensive alternatives, such as integrated isocratic systems that cover the required pressure range (e.g., 0 to 2500 psi) and cost under $10,000 (e.g., D-Star Instruments Integrated Isocratic HPLC system, available through Thomas Scientific).



Supplement 16

Selecting Appropriate Buffers The buffer selected for gel filtration should be optimized for protein stability. Highly hydrophobic or membrane proteins, for example, require a nondenaturing detergent in the buffer (see next section). Silica-based resins are not stable above pH 7.5, and stainless steel columns should typically not be used with halogens. Buffer type and strength may affect resolution and mass recovery. The ionic strength of the buffer is also important. In high–ionic strength buffers (>1 M), hydrophobic interactions between the protein and the resin may occur, while at low ionic strength (7.5. Bed compression can occur if the flow rate or back pressure is too high. To prevent particulate matter from clogging the frit or column matrix, buffers should be filtered and samples should be filtered or centrifuged prior to injection. Another method is to install a guard column in front of the fractionating column, which will trap any particles or precipitating proteins that are accidentally introduced into the system. Air bubbles can occur if the buffer has not been degassed properly, or they may be introduced into the system during sample injection if the sample loop is not completely filled with buffer and sample. A void may be filled using some of the same column matrix; however, care should be taken not to disturb the existing column bed. An increase in back pressure may be due to a clogged frit. The best remedies for this are to reverse flush the column, clean the column according to the manufacturer’s recommendations, or replace the end fitting. When cleaning the column, make sure the cleaning buffers are compatible with all other components of the

system. The column should be run at a lower flow rate and the detector should be taken out of line. The column frit should also be changed if a visible space develops between the adaptor and the filter. Resolution of the columns should be checked with a standard protein mix rather than with an experimental sample. An experimental sample may contain either a leading or trailing edge due to other components or impurities, such as aggregated or proteolyzed material. Routine column tests using a well-characterized standard mixture also facilitate comparative monitoring of a column’s performance over the lifetime of the column. Problems with mass spectrometry usually involve failure of the protein sample to ionize (no signal), or signal suppression due to internal standards or buffer components such as high salt or detergent. Ionization of a sample may be improved by experimenting with different matrices, sample concentrations, sample/matrix ratios, and laser power. Signal suppression due to buffer components can be remedied by dialyzing the sample into an MS-compatible buffer such as ammonium bicarbonate.

Anticipated Results If the recombinant protein is pure and exists in a single oligomeric state with no aggregation or degradation, it is anticipated that the HPLC chromatograph will show a single, symmetrical peak. Additional peaks that elute before the protein of interest may be oligomers or aggregates of the protein, or contaminants. Peaks eluting after the protein of interest may be proteolysis products of the protein or contaminants. To identify the components, collect fractions throughout the separation procedure and analyze by another technique such as SDSPAGE (UNIT 10.1), MALDI-MS (see Support Protocol 2), immunoblotting (UNIT 10.10), or Nterminal sequencing (UNIT 11.10). MALDI-MS analysis of a pure sample with the correct sequence will show a single peak with a mass that should be within 0.1% of that expected from the amino acid sequence. Final confirmation of identity can be obtained by N-terminal sequence analysis of the intact protein and/or tryptic peptide mapping (UNIT 7.3). Additional peaks in the spectrum may sometimes be identified by their mass (Table 7.10.3).

Time Considerations Equilibration of the gel filtration column typically takes several hours. In some cases, equilibration may have to be performed over-



night. An analytical HPLC gel filtration run can typically be performed in ∼30 min depending upon flow rate and column length.

Literature Cited Beavis, R.C. and Chait, B.T. 1996. Matrix-assisted laser desorption ionization mass-spectrometry of proteins. Methods Enzymol. 270:519-551.

Seigel, L.M. and Monty, K.J. 1966. Determination of molecular weights and frictional ratios of proteins in impure systems by use of gel filtration and density gradient centrifugation. Application to crude preparations of sulfite and hydroxylamine reductases. Biochim. Biophys. Acta 112:346-362.

Key References

Cantor, C.R. and Schimmel, P.R. 1980. Biophysical Chemistry, Part II: Techniques for the Study of Biological Structure and Function. W.H. Freeman, New York.

Beavis and Chait, 1996. See above.

Neue, U.D. 1997. HPLC Columns: Theory, Technology, and Practice. John Wiley & Sons, New York.

Neue, 1997. See above.

Overall, C.M. 1987. A microtechnique for dialysis of small volume solutions with quantitative recoveries. Anal. Biochem. 165:208-214. Schagger, H. 1994. Chromatographic techniques and basic operations in membrane protein purification. In A Practical Guide to Membrane Protein Purification (G.V. Jagow and H. Schagger, eds.) pp. 23-57. Academic Press, San Diego.

A practical guide to MALDI-MS of proteins, including details of sample preparation and data analysis.

A theoretical and practical guide to HPLC, including properties and selection of columns, methods development, and troubleshooting.

Contributed by Gillian E. Begg, Sandra L. Harper, and David W. Speicher The Wistar Institute Philadelphia, Pennsylvania



Supplement 16

Rapid Screening of E. coli Extracts by Heteronuclear NMR

UNIT 7.11

Assessing whether a protein or protein complex is amenable to structural analysis is an important component in the structural genomics effort. In particular, if complete sets of structures for entire genomes are to be obtained within a reasonable time frame, high throughput methodologies for all steps along the way have to be developed. These days, cloning and expression systems are highly optimized and a variety of commercially available vectors can be used. However, heterologous proteins or protein domains expressed in bacteria may not be soluble or correctly folded, necessitating intricate solubilization and refolding schemes prior to structural or functional studies. NMR spectroscopy is an important tool for assessing the solubility, stability, and structural integrity of a gene product. It allows efficient evaluation of many variations of polypeptide length and sequence (without time- and labor-intensive purification/refolding procedures) using 1H-15N-HSQC spectroscopy of samples of 15N-labeled proteins directly from crude E. coli extracts (Clore and Gronenborn, 1996). In addition to screening for particular properties of the expressed protein alone, it is also possible to map intermolecular interactions such as ligand binding (Huth et al., 1997). In this unit, the basic methodology for bacterial growth, isotope labeling, and spectroscopic evaluation of the protein structure is provided. STRATEGIC PLANNING Choice of Expression System A variety of efficient expression systems are available (see Chapter 5). Important for the present purpose is a tightly controlled transcription and translation system. As an example, the T7 promoter-driven system originally developed by Studier (Studier et al., 1990) has enjoyed widespread use, although other systems (λ pL, tac, trp, lac, araBAD promoters) can be employed as well. In addition, the decision regarding whether expression as an untagged protein or fusion protein is preferable should take into account any previous knowledge about solubility and stability of the protein. For novel and uncharacterized proteins or domains, it is generally advisable to express these as fusions, increasing the probability of producing large amounts of protein in a folded form. For most N-terminal fusion proteins, such as those containing domains from glutathione S-transferase, maltose binding protein, protein A, thioredoxin, or the immunoglobulin-binding domain of streptococcal protein G (GB1), yields are high and there is generally no need to optimize expression conditions. However, careful attention should be paid to the selection of host strains in each particular expression vector–host system. It must be established that growth of the bacterial host in a modified minimal medium is not impaired since labeling is achieved using 15NH4Cl and glucose as sole nitrogen and carbon sources, respectively. Unfortunately, the published genotypes of host strains may not list important amino acid requirements. A comprehensive description of methods for biosynthetic enrichment of proteins with 15N is provided in Muchmore et al. (1989). NMR Instrumentation Any high-field NMR spectrometer (11.75 to 21.15 Tesla, 500 to 900 MHz 1H frequency) equipped with a triple-resonance gradient probe can be used to obtain 1H-15N HSQC spectra. Higher field instruments obviously produce spectra with higher resolution and sensitivity, although excellent performance for screening purposes can be obtained with 500-MHz instruments and cryo-probes. For high-throughput structural genomics appliContributed by Angela M. Gronenborn Current Protocols in Protein Science (2003) 7.11.1-7.11.8 Copyright © 2003 by John Wiley & Sons, Inc.



cations, an automated sample changer may be advantageous. For an introduction to NMR of proteins, see UNIT 17.5. BASIC PROTOCOL

15

N PROTEIN LABELING AND NMR CHARACTERIZATION

This protocol describes the preparation and screening of crude extracts for a GB1-fusion system. Variations of this Basic Protocol are necessary for different expression systems. The key to success is tight repression prior to induction and robust expression after induction. In general, 50 to 250 ml of bacterial culture should suffice for several NMR samples. Materials Bacteria (E. coli strains BL21(DE3) or HMS174) containing the expression plasmid (GEV1 or GEV2) carrying the coding sequence for the protein of interest Minimal medium (see recipe) 1 M IPTG PBS, pH 7.4 20 mM sodium phosphate, pH 5.4 D2O Shaker incubator (New Brunswick Scientific) French pressure cell (UNIT 6.7); alternatively the bacteria can be ruptured in a microfluidizer Concentrators (e.g., Centriprep-3, Amicon) NMR spectrometer Grow bacteria, express protein, and prepare extract 1. Transform (UNIT 5.2) or obtain a 50-ml culture of bacteria (E. coli strains BL21(DE3) or HMS174) containing an expression plasmid (GEV1 or GEV2) carrying the coding sequence for the protein of interest. It is best to start with fresh transformants or a previously tested glycerol stock.

2. At an OD600 of ∼1⁄3 of the OD600 reached at stationary phase without induction, harvest the cells by centrifuging 10 min at 1200 × g, 25°C. UNIT 5.3

describes growth monitoring of E. coli.

It is important not to form a hard cell pellet since it needs to be resuspended again for further growth.

3. Resuspend the cells in pre-warmed (37°C) minimal medium containing 15NH4Cl as sole nitrogen source. Culture for an additional 30 min. This minimal media works well for growth in shake flasks as well as for small fermentations.

4. Induce protein production by adding 1 to 2 ml of 1 M IPTG per liter of culture and incubate an additional 3 to 5 hr. 5. Harvest cells by centrifuging 10 min at 4000 × g, 4°C, decant the supernatant and resuspend the cell pellet in 20 ml PBS, pH 7.4. Lyse the cells by passing them through a French pressure cell two times (UNIT 6.7). 6. Remove the cell debris by high-speed centrifuging 20 min at 20,000 × g, 4°C. Rapid Screening of E. coli Extracts by Heteronuclear NMR

7. Concentrate the supernatant in a Centriprep-3 device to a volume of ∼1 ml. At the same time change the buffer to 20 mM sodium phosphate, pH 5.4 by adding ∼10 ml of buffer to the 1-ml concentrate and concentrating this sample again two to three



HSQC

15H

1H

Figure 7.11.1 Schematic outline of cloning, expression, extract preparation, and NMR screening.

times. Transfer 0.3 to 0.5 ml of this concentrated extract to an NMR tube and add D2O to yield a final concentration of 5% (v/v). The general outline of this approach is illustrated in Figure 7.11.1.

Perform NMR spectroscopy 8. Record 1H-15N HSQC spectra using pulse sequences previously described (Bax et al., 1990; Piotto et al., 1992) in the experimental libraries of the instrument manufacturer. In general, 128 × 512 complex points in the indirect (15N) and acquisition (1H) dimensions are recorded with total acquisition times of 64 msec in both dimensions and 16 scans per T1 increment. Figure 7.11.2 presents 1H-15N HSQC spectra of crude E. coli extracts containing 15N-labeled GB1 domain (56 amino acids) or Il-1β (153 amino acids).

ENRICHMENT OF 15N-LABELED PROTEIN For cases in which a fusion construct is employed for expression, partial purification or enrichment of the 15N-labeled fusion protein can be carried out in one step by passing the extract over an affinity column. For highly expressing constructs (>0.1 mg/ml of culture), this is not necessary, but it could be useful for poorer expressers. In the case of the GB1 fusion protein vector (Huth et al., 1997), a His-tag resides at the C-terminus in addition to the GB1 domain at the N-terminus and both domains can be used for affinity purification (e.g., employing a Ni2+ affinity column to bind the His tag; UNIT 9.4).

ALTERNATE PROTOCOL

Additional Materials (also see Basic Protocol) 50 mM Tris⋅Cl, pH 7.5/5 mM EDTA/5 mM benzamidine solution 20-ml IgG Sepharose fast-flow column 50 mM Tris⋅Cl, pH 7.5/150 mM NaCl 5 mM sodium acetate buffer, pH 5.0 5 M sodium acetate buffer, pH 3.5 10 to 20 mM sodium phosphate, pH 5 to 6 Additional reagents and equipment for SDS-PAGE (UNIT 10.1) 1. Grow cells and induce protein expression as described in Basic Protocol, steps 1 to 4.



Supplement 31

IL-1b

GB-1

Figure 7.11.2 SDS/PAGE of crude extract of IL-1β and GB1 and the corresponding 1H-15N HSQC spectra. Samples prepared as described in the Basic Protocol were separated on SDS/PAGE and Coomassie stained. The identical samples were used for NMR spectroscopy. Crosspeaks are labeled with the residue name and number.

2. Harvest cells by centrifuging 10 min at 4000 × g, 4°C. Decant the supernatant and resuspend the pelleted cells in 20 ml 50 mM Tris, pH 7.5/5 mM EDTA/5 mM benzamidine solution. Disrupt the cells using a French pressure cell as described in Basic Protocol, step 5. 3. Remove the cell debris by high-speed centrifuging 20 min at 20,000 × g, 4°C, and apply the supernatant containing the fusion protein onto a 20-ml IgG Sepharose fast-flow column equilibrated in 50 mM Tris⋅Cl, pH 7.5/150 mM NaCl either at room temperature or 4°C depending on stability of protein. 4. Wash the column with one to two column volumes of 5 mM sodium acetate buffer, pH 5.0 and then elute the GB1 fusion protein with 5 M sodium acetate buffer, pH 3.5. Collect 1-ml column fractions and check for protein by SDS-PAGE (UNIT 10.1). 5. Concentrate the protein-containing fractions in a Centriprep-3 device to a volume of ∼1 ml. At the same time, change the buffer to 10 to 20 mM sodium phosphate, pH 5 to 6 (see Basic Protocol, step 7). The final buffer is selected based on known properties of the expressed protein. If not information is available, a low molarity phosphate buffer, slightly on the acidic side, is recommended.

6. Transfer 0.3 to 0.5 ml of this concentrated solution to an NMR tube, and add D2O to a final concentration of 5% (v/v). 7. Acquire NMR spectra as described in Basic Protocol, step 8. Figure 7.11.3 presents the 1H-15N HSQC spectra of two different GB1- fusion proteins after one-step purification on IgG Sepharose. The spectrum in panel A illustrates the appearance of a fusion partner that is basically a random coil (resonances reside between 7.8 and 8.4 ppm for 1H) while the resonances in panel B are dispersed over the entire spectral widths, indicative of a folded fusion partner. Rapid Screening of E. coli Extracts by Heteronuclear NMR



A

B

Figure 7.11.3 1H-15N HSQC spectra of GB1-fusion proteins. (A) The fusion partner exhibits a random coil structure (GB1-GCN4) and (B) a folded fusion partner (GB1-Fas) is present. Resonances arising from the GB1 portion are labeled by residue name and number. For GB1-GCN4 (A) all additional cross-peaks are located in the narrow region of 7.8-8.4 ppm, indicative of a random coil conformation for the GCN-4 portion. For GB1-Fas (B), cross-peaks are distributed over the entire region of the spectrum, indicative of folded Fas protein.

REAGENTS AND SOLUTIONS Use Milli-Q-purified water or its equivalent for the preparation of all buffers. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Minimal medium 2 ml 20% (w/v) glucose 500 µl trace elements (see recipe) 500 µl 1 M MgCl2⋅6 H2O 300 µl 5 mg/ml thiamine-vitamin B1 100 µl 10% yeast extract (optional) 50 µl 50 mg/ml antibiotic Add sterile water to 50 ml and mix Add to 950 ml of minimal salts (see recipe) Make fresh for use within 1 or 2 days Filter sterilize all stock solutions through a 0.22-µm filter Minimal salts 13 g/liter KH2PO4 (anhydrous) 10 g/liter K2HPO4 (anhydrous) 9 g/liter Na2HPO4 (anhydrous) 2.4 g/liter K2SO4 1.1 g/liter NH4Cl (either 14N or 15N) Dissolve in 950 ml H2O and autoclave Store up to several months at room temperature



Supplement 31

Trace elements 6.0 g CaCl2⋅2H2O 6.0 g FeSO4⋅7H2O 1.15 g MnCl2⋅4H2O 0.8 g CoCl2⋅6H2O 0.7 g ZnSO4⋅7H2O 0.3 g CuCl2⋅2H2O 0.02 g H3BO3 0.25 g (NH4)6Mo7O24⋅4H2O 5.0 g EDTA Add reagents one at a time to 1 liter of water with vigorous stirring. Wait 5 to 10 min between each addition. Each reagent must be completely dissolved before adding the next one. After addition of EDTA, stir until the solution is golden-brown. This may take 2 to 3 days, including a transition from green to a yellowish-brown color. When the color is golden yellow, sterilize by filtering through a 0.22-µm filter. Store up to several months in the dark at room temperature. COMMENTARY Background Information

Rapid Screening of E. coli Extracts by Heteronuclear NMR

A prerequisite for any structural investigation by NMR is the availability of a “well-behaved” sample. For proteins, it is necessary to have an over-producing organism, fast and effective purification methods, and the knowledge of conditions that allow the sample to survive days, or sometimes weeks, at or above room temperature. Thus, it is imperative to establish conditions (buffer, salt, pH) under which the protein is biochemically stable, folded, and amenable to spectroscopy. Usually, these conditions are determined by trial and error, using large amounts of protein, costly purification procedures, and considerable time and manpower. As such, they are ill-suited for any high-throughput strategies. For many proteins, sub-cloning of truncated versions or domains as well as point mutants is a common approach and frequently a large number of protein variants is evaluated before one with the desired biochemical and biophysical properties emerges. The methodology outlined in this unit addresses these issues and is aimed at overcoming the shortcomings traditionally encountered. Some specific advantages of this system are as follows: (1) The coupling between the highly efficient T7 polymerase expression system with the N-terminal GB1 domain in the GEV vectors ensures a high level of protein production. Depending on the particular case, protein expression can reach 25% of total cellular protein. (2) The addition of the highly soluble GB1 domain enhances the solubility of the fusion

partner. Indeed it has been shown that using this approach, the structure determination of a protein-protein complex that was insufficiently soluble and stable became possible (Zhou et al., 2001). (3) Sample conditioning can be carried out without having to cleave the fusion protein. This entails evaluating different buffer compositions and pH values for obtaining the best spectral resolution. The particular advantage of GB1 compared to other fusion proteins is its small size (6.2 kD). Most other fusion partners are fairly large (GST, 27.5 kD, and MBP, 38 kD), and as a result, their properties can easily mask those of the protein of interest. For NMR spectroscopy, this implies that a larger number of signals will arise from the protein tag, obscuring the observation of resonances from the interesting partner. For the GB1 fusions, this implies that all screening and even a structural characterization can be carried out on the fused protein (Huth et al., 1997). (4) There appears to be no interaction between the GB1 and its fusion partner for all cases studied to date (Huth et al., 1997; Zhou et al., 2001). Thus, the likelihood of inducing an artificial conformational change in the protein of interest by coupling to the GB1 domain is minimal. (5) The use of the GB1 fusion vector in combination with screening of E. coli extracts by 1H-15N HSQC spectroscopy is highly amenable to automation. In particular, the NMR screening part can be carried out in highthroughput mode using spectrometers with automated sample changers.



(6) Using 1H-15N HSQC spectra for screening purposes exploits the intrinsically higher sensitivity of this experiment compared to other 2-D experiments, overcoming the inherent low sensitivity of NMR. Since chemical shifts are extremely sensitive to the local environment, a qualitative assessment of structural identity and similarity can be carried out by NMR, even without resonance assignments. The pattern of cross-peaks in a 1H-15N HSQC spectrum of uniformly 15N-labeled protein can be used as a fingerprint of the structure. Thus, closely similar resonance frequencies of amide groups provide a reliable measure for similarities in overall structure for related proteins or mutants and this property can be exploited for fast and efficient screening of mutant libraries by NMR (Gronenborn et al., 1996). The high quality of the 1H-15N HSQC spectra is remarkable and a direct result of the high expression levels reached in the cell. When combined with the sensitivity of a cryoprobe, these spectra can be recorded directly without further purification or concentration on 90% isotopic enrichment, the target gene has to be completely repressed before induction and this repression has to be completely lifted upon induction. Not all E. coli expression systems exhibit these properties. In other organisms, such as the yeast Pichia pastoris (UNIT 5.7) and the insect cell/baculovirus system (UNITS 5.4 & 5.5), large amounts of protein (25% to 50%) can be produced and media for labeling over-expressed proteins with NMR-active isotopes are available (Creemers et al., 1999; de Lamotte et al., 2001). Methodologies for these systems are, however, not as advanced as those for protein production in E. coli. More developmental work is needed in alternate systems and the associated costs need to come down substantially to allow for routine use. Complete media for labeling in E. coli are commercially available from Cambridge Isotopes (www.isotope.com), Isotec (www.isotec.com), Spectra Stable Isotopes (www.spectrastableisotopes.com) and Silantes GmbH (www.silan-

tes.com); media for labeling in pichia or SF9 cells are in development. Higher-field NMR instruments will obviously have higher sensitivity and therefore allow screening at lower concentrations (expression levels). The availability of cryoprobes for 500- and 600-MHz spectrometers that provide significant improvement in signal-to-noise ratios for protein samples and allow much shorter data collection times has changed the requirement for high field instruments. It is possible to record 1H-15N HSQC spectra on 15N-labeled proteins on a 600-MHz spectrometer equipped with a cryoprobe on 10 µM or 100 µM samples in 5 to 10 hr or 30 min, respectively.

Troubleshooting A frequent, but rarely reported problem with expression systems is their leakiness. If repression of the desired protein is not efficient before introduction of isotopically labeled media, substantial amounts of unlabeled material may accumulate and dilute the concentration of the labeled material. If a new expression system is investigated for purposes of screening with the methodology described in this unit, it is advised to monitor production of the expressed protein along the growth curve in medium containing 14NH Cl by SDS/PAGE prior to growth in 4 15NH Cl-containing medium. 4 Slow, insufficient growth can be caused by lesions in amino acid metabolism genes in the host strains. Commonly used host strains that grow well in the minimal medium described in this unit are BL21, DH5α, and HMS174. Strains that grow poorly or not at all are TOP10 and HB101. In addition, optimization of expression for the system at hand should be done before setting up a screening program for a large number of samples. For the GB1 fusions discussed in this unit, high yields are invariably observed, but this may not translate to other systems to the same extent.

Anticipated Results Several examples of 1H-15N HSQC spectra recorded on crude extracts and enriched samples are provided in Figures 7.11.2 and 7.11.3. With current instrumentation, excellent signalto-noise ratios can be achieved, provided the parameters for sample preparation are optimized (using >20 µM labeled protein) and the parameters for the NMR experiment are set correctly. The most likely reason for low signal intensity is aggregation of the proteins, caused by instability or incorrect folding of the ex-



Supplement 31

pressed protein. Adding small amounts of detergent (CHAPS) or salt may alleviate problems associated with aggregation. Figure 7.11.2 displays 1H-15N HSQC spectra of GB1 and IL-1β recorded on crude extracts and Figure 7.11.3 those of the partially enriched GB1 fusion proteins of GB1-HMG-I with and without the DNA ligand.

Time Considerations After optimization of the system, all steps, from expression to the completed NMR spectrum can be completed in 1 Kg) amounts of target protein will be needed. This is often referred to as the scale of operation. When the source material is very Contributed by Alan Williams Current Protocols in Protein Science (1995) 8.1.1-8.1.9 Copyright © 2000 by John Wiley & Sons, Inc.

UNIT 8.1

expensive, unstable, or difficult to obtain, it is necessary to develop and optimize the purification strategy on a small scale first, then scale up. Conversely, if multiple samples are to be processed, as for analytical applications, it may be desirable to develop the purification process at a larger scale, then scale down. The amount of target protein to be purified may also be determined by limited availability of source material.

Source Material Once the scale of operation has been determined, the amount of source material needed to begin the procedure must be clearly defined. This obviously depends on the concentration of target protein in the source material, which can vary greatly. Certain proteins and peptides are present in natural sources only in very minute quantities. Purification of even a microgram of pure material may require many kilograms of tissue, or many liters of fermentation broth if microorganisms are the source. On the other hand, some recombinant proteins are expressed at such high levels in a host that the protein may be present in milligram per milliliter or higher concentrations. Source materials containing very low or very high concentrations of target protein each present unique problems that must be addressed in selecting a purification strategy.

Yield of Purification Procedure The next major consideration in deciding how much protein to purify is what yield is necessary at each step of the purification procedure to ensure that there will be enough pure target protein at the end, taking into account the availability of source material. If a 75% yield of target protein is obtained at each of four steps in a purification procedure, less than one-third of the desired protein will remain at the end of the procedure (Fig. 8.1.1). It is obviously advantageous to use as few steps as possible in a purification process while maintaining as high a yield as possible at each step. The yield from any purification step may be reported as total protein, percentage of total activity, or change in specific activity. The yield information from progressive purification steps is used to construct a purification table (e.g., see Table 8.1.1). In Table 8.1.1, the yield is reported as percent of total activity remaining from the crude

Conventional Chromatographic Separations

8.1.1 CPPS

95%/step 90%/step

85%/step 80%/step 75%/step

Target protein recovered (%)

100 90 80 70 60 50 40 30 20 10 0 0

1

2

3

4

5

6

7

8

9

10

Number of steps in procedure

Figure 8.1.1 Theoretical yields from multistep protein purifications. Each curve represents a multistep process with a given percentage yield per step. The percentage of the original quantity of target protein remaining at the end of purification is plotted against the number of purification steps.

homogenate; the step yield is the percent activity relative to the previous step; and the purification factor reflects the change in specific activity relative to the crude homogenate. It is not unusual to have yields >100% based on activity (e.g., the yield after anion-exchange chromatography in Table 8.1.1). This is most often due to the removal of inhibitors or contaminants that interfere with the activity assay.

Assessment of Protein Purity

Overview of Conventional Chromatography

Assessing protein purity is often the most time-consuming and labor-intensive step in the purification process. It is generally desirable to limit the number of steps in a purification process simply to reduce the number of assays that must be performed to assess purity at each step. Protein purity is judged from a structural and/or functional perspective, depending on the nature of the target protein and its intended use. If the intention is to examine the function of the target molecule, the activity of the target protein must be maintained during the purification process. A functional assessment of the purity of the target protein will thus be required at each step. If the intention is to examine the structure of the target molecule, structure must be maintained, but function may or may not have to be preserved. A structural assessment of purity will, however, be required in this case. If struc-

tural/functional relationships involving the target molecule are to be examined, assessment of both structural and functional purity will be required. Methods for assessing protein purity must quantitate the amount of target protein relative to the amount of contaminant(s) in the sample. This requires two separate analytical methods: one for the target protein and one for the total protein including contaminants. For example, with enzymatic proteins, assessment of functional purity typically relies on kinetic assays in which the amount of substrate consumed or product produced per unit time is proportional to the amount of enzyme present. The total activity of the sample is calculated and compared to the total amount of protein to give the specific activity of the sample. Methods for determining total protein are discussed in Chapter 3. An increase in specific activity at a purification step reflects a loss of contaminant proteins. Functional purity of structural proteins, denatured proteins (i.e., proteins that have lost their function during the purification procedure), and proteins for which a functional assay does not exist must usually be assessed by measuring structural attribute(s)—e.g., molecular weight, pI, presence of a metal ion or cofactor, or presence of an antibody binding


Table 8.1.1

Purification Table for an Arbitrary Enzymea

Purification step

Total activity (U)

Crude homogenate 12,000 × g supernatant (NH4)2SO4 fraction (20-50%) HIC AEX Gel filtration

200 180 150 125 208 148

Total protein Specific Yield (% total Step Purification (mg) activity (U/mg) activity) yieldb factorc 50,000 35,000 25,000 2500 29.3 5.9

0.004 0.005 0.006 0.05 7.1 25.1

100 90 75

— 90 83

1.00 1.25 1.50

62.5 104 74

83 166 71

12.50 1775 6275

aAbbreviations: HIC, hydrophobic-interaction chromatography; AEX, anion-exchange chromatography. bPercentage activity relative to previous step. cIncrease in specific activity.

site. As in the case of functional purity, an increase in the amount of target protein relative to total protein reflects an increase in purity.

STEPS IN A PURIFICATION STRATEGY A purification strategy for any protein at any scale of operation can be broadly divided into three sequential stages—capture, intermediate purification, and polishing. Each stage represents a set of specific problems that may be encountered during a purification process. The nature of the sample and the scale of operation will dictate what equipment and methodology are appropriate to solve the problem.

Capture Stage The capture stage is the initial purification of the target protein from the source material. The goal of capture is to concentrate the target protein while removing as much of the major contaminant(s) as possible; to this end, an adsorptive chromatographic technique should be employed. Ion-exchange chromatography (UNIT 8.2) and hydrophobic-interaction chromatography (HIC; UNIT 8.4) are generally the best chromatographic techniques for capture at any scale of operation. Certain affinity techniques are also effective, including lectin affinity chromatography (UNIT 9.1), dye affinity chromatography (UNIT 9.2), immunoaffinity chromatography, and metal-chelate affinity chromatography (MCAC). Affinity chromatography is very useful for capture on a small scale (e.g., 1.5. At this value, the peaks are completely separated from each other (i.e., the purity of each peak is >99.9%).


efficiency, and selectivity of the system. Each of these factors must be considered and controlled to achieve success. The theoretical expression for resolution is Equation 8.1.2, where k is the average capacity factor for the two peaks, N is the efficiency factor for the system, and α is the selectivity factor of the medium. Rs =

k=

VR2 − Vm Vm

Equation 8.1.3

In Equation 8.1.2 for Rs, presented in the discussion of Resolution, k is the average of k1 (capacity factor for peak 1) and k2 (capacity factor for peak 2). Adsorption techniques such as ion exchange, HIC, chromatofocusing, RPC, and affinity chromatography can have high capacity factors because experimental conditions can be manipulated so that the elution volume for a peak can exceed the total bed volume (Vm as is the case with peaks 2 and 3 in Fig. 8.1.4.) However, in gel filtration, which is a nonadsorptive technique, all peaks must elute within the volume Vm−Vo as is the case with peak 1 in Fig. 8.1.4. The impact of experimental conditions on capacity at various stages of protein purification are discussed for different purification techniques in subsequent units of Chapter 8.

α−1 k 1 ) ( N−2 ) ( ) ( 4 1+k α Equation 8.1.2

The impact of various experimental factors on resolution in various stages of protein purification is discussed for specific chromatographic techniques in subsequent units of Chapter 8.

Capacity The capacity or retention factor (k) is a measure of retention of a sample component. It should not be confused with the loading capacity of a column, which is expressed as milligrams of sample bound per milliliter of gel and represented by the area under a peak. The capacity factor may be calculated for any individual peak in a chromatogram. For example, the capacity factor for peak 2 in Figure 8.1.4 is derived from Equation 8.1.3, where VR2 is the elution volume of peak 2 and Vm is the volume of the mobile phase (i.e., the total bed volume).

Efficiency The efficiency factor (N) is a measure of zone broadening (peak width) occurring on a column. It can be calculated for any given peak from Equation 8.1.4, where VR is the elution volume (i.e., the total volume of eluant that has

gel filtration

absorptive techniques 2

3

A280

1

V0

Wb1 VR1 VM VR2 VR3 Elution volume

Figure 8.1.4 Hypothetical chromatogram. Vo = void volume; VR1 = elution volume for peak 1; VR2 = elution volume for peak 2; VR3 = elution volume for peak 3; VM = volume of mobile phase; Wb1 = peak width for peak 1; Wb2 = peak width for peak 2.



A

A280

high efficiency

low efficiency

B

A280

high efficiency

low efficiency

Elution volume

Figure 8.1.5 Effect of selectivity and efficiency on resolution. (A) Chromatogram obtained using experimental technique that provides good selectivity. Where conditions are chosen to provide high efficiency, two narrow and distinct peaks are obtained (high resolution). When the conditions provide only low efficiency, the peaks are broadened and overlap more (lower resolution). (B) Chromatogram obtained using experimental technique that provides bad selectivity. Even when high-efficiency conditions are used, there is lower resolution than in (A), and the resolution deteriorates further with low-efficiency conditions.

passed through the column at peak maximum) and Wh is the peak width at half the peak height.  VR  N = 5.54   W  h

2

Equation 8.1.4

Overview of Conventional Chromatography

Efficiency (N) may be expressed as the number of theoretical plates for the column under specific experimental conditions (see UNIT 8.3).

Efficiency is also frequently defined as the number of plates per meter of chromatographic bed, or in terms of H, the height equivalent to a theoretical plate. H is simply the column length (L) divided by the efficiency factor (N); i.e., H = L/N. The main cause of zone broadening (i.e., loss of efficiency) in a chromatographic bed is diffusion. Diffusion perpendicular to the flow is restricted by the walls of the column. Therefore, longitudinal diffusion is the primary fac-


tor contributing to zone broadening. While a protein is adsorbed to the medium, little or no diffusion takes place, but once the protein is unbound, diffusion begins. The amount of diffusion that occurs is proportional to the time required for the material to emerge from the system. Loss of efficiency resulting from diffusion is minimized if the distances available for diffusion in the mobile phase, gel beads, and system are minimized. In practice, efficiency increases with increasing uniformity in particle size and with decreasing bead size. Good experimental technique is required for high efficiency. Unevenly packed chromatography beds and trapped air will lead to channeling, zone broadening, and consequent loss of resolution. Loss of efficiency also stems from system effects such as dead volumes in the system, poor mixing during gradient formation, and pulsations in flow.

Selectivity is more important than efficiency (N) in determining resolution because Rs is directly proportional to selectivity, but is proportional only to the square root of efficiency (see Fig. 8.1.5 along with Equation 8.1.1 and Equation 8.1.2 for Rs in the discussion of Resolution). Hence, a four-fold increase in efficiency is required to double resolution, as compared with a two-fold increase in selectivity. In practice, selectivity depends partly on the chromatographic technique employed but can usually be controlled by manipulating experimental conditions, such as the pH and ionic strength of the mobile phase. Because this can be done easily and predictably, selectivity is the factor that is exploited to achieve maximum resolution in column chromatography rather than efficiency, which is fixed by the particle size and uniformity of the medium selected.

LITERATURE CITED Selectivity Selectivity of a chromatographic medium defines the ability of that medium to separate peaks (i.e., it is a measure of the distance between two peaks in a chromatogram; see Fig. 8.1.5). The selectivity factor (α) can be calculated from a chromatogram using Equation 8.1.5, where k2 is the capacity factor for the second peak (see discussion of Capacity), k1 is the capacity factor for the first peak, VR2 is the elution volume of the second peak, VR1 is the elution volume of the first peak (see Fig. 8.1.2), and Vm is the volume of the mobile phase.

α=

k2 VR2 − Vm VR2 = ≈ k1 VR1 − Vm VR1

Corran, P.H. 1989. Reversed-phase chromatography of proteins. In HPLC of Macromolecules: A Practical Approach (R.W.A. Oliver, ed.) pp. 127156. IRL Press, Oxford. Giddings, J.C. and Keller, R.A. (eds.) 1965. Dynamics of Chromatography, Part 1: Principles and Theory. Marcel Dekker, New York. Janson, J.-C. and Ryden, L. (eds.) 1989. Protein Purification: Principles, High Resolution Methods and Applications. VCH Publishers, New York.

Contributed by Alan Williams Pharmacia Biotech Piscataway, New Jersey

Equation 8.1.5



Ion-Exchange Chromatography

UNIT 8.2

Ion-exchange chromatography separates biomolecules on the basis of charge characteristics. Charged groups on the surface of a protein interact with oppositely charged groups immobilized on the ion-exchange medium. As illustrated in Figure 8.2.1, the charge of a protein depends on the pH of its environment (the operating pH). The pH at which the net charge of a protein is zero (i.e., where the number of positive charges equals the number of negative charges) is known as the isoelectric point (pI). When the operating pH is greater than the pI, the protein will have a net negative charge, and should bind to anion-exchange media, which are positively charged. When the operating pH is less than the pI, the protein will have a net positive charge, and should bind to cation-exchange media, which are negatively charged. The Strategic Planning section outlines the basic steps in planning and carrying out ion-exchange chromatography to separate proteins. Basic Protocol 1 describes batch adsorption of protein to an ion-exchange medium followed by elution using a step gradient of increasing salt concentration. This technique accommodates a wide range of sample volumes and is most often used in the initial capture stage of protein purification (UNIT 8.1). Batch techniques have minimal system requirements. The Alternate Protocol describes use of a buffer of a different pH to elute via step gradient. Basic Protocol 2 describes adsorption of protein to an ion-exchange medium in a column, followed by elution with a linear gradient. Basic Protocol 2 provides a higher resolution than Basic Protocol 1, and is therefore used in the intermediate purification and final polishing stages of protein separation (UNIT 8.1). Support Protocol 1 describes a pilot experiment to determine initial conditions for batch or column chromatography (i.e., pH required for binding, change in pH or salt concentration required for elution, and available capacity of a medium). Support Protocol 2 describes a means of calculating the dynamic capacity (UNIT 8.1) of an ion-exchange column, Support Protocol 3 describes methods for producing continuous gradients of pH and salt concentration to elute proteins from ion-exchange

+ Net positive charge of protein isoelectric point (pl)

Will bind to cation exchanger

select system pH above pI for anion exchange

0

system pH 4

6

8

10

select system pH below pI for cation exchange Will bind to anion exchanger

– Net negative charge of protein

Figure 8.2.1 Net charge of a protein as a function of pH, showing the pH ranges in which protein is bound to anion or cation exchangers. The pH range over which the protein is stable may be only a small fraction of the binding range; this must also be taken into consideration when choosing an ion-exchange medium. Contributed by Alan Williams and Verna Frasca Current Protocols in Protein Science (1999) 8.2.1-8.2.30 Copyright © 1999 by John Wiley & Sons, Inc.


8.2.1 Supplement 15

columns, Support Protocol 4 describes regeneration of used ion-exchange media, and Support Protocol 5 details storage of ion-exchange media. STRATEGIC PLANNING There are a number of basic steps in developing an ion-exchange method for protein purification. First, an appropriate ion-exchange medium must be selected, as well as the optimal operating pH and buffer system for the medium and sample (see following discussions on Selecting an Ion-Exchange Medium and Selecting a Buffer System). It is then necessary to decide whether batch or column chromatography is appropriate given the purity, protein concentration, and physical characteristics of the sample, the sample volume to be used, and the availability of suitable equipment (see following discussion on Selecting Batch Versus Column Purification). Next, pilot experiments are conducted to determine conditions for binding and eluting the protein (see Support Protocol 1). It is also necessary to make sure that the capacity of the medium is sufficient to isolate the desired quantity of protein (see Support Protocol 1 and Support Protocol 2). When these initial conditions have been determined, the medium and sample are prepared, the sample is bound to the medium, unbound sample components are washed away, and bound sample components are selectively eluted and collected (see Basic Protocol 1, Basic Protocol 2, or Alternate Protocol). The medium may then be cleaned and regenerated (see Support Protocol 5). Finally, results are evaluated and conditions optimized if necessary (see Critical Parameters). Selecting an Ion-Exchange Medium A wide variety of ion-exchange media are commercially available, but no miracle medium exists that is best for every protein purification. Criteria for selecting an ion-exchange medium include the specific requirements of the application, the pI and molecular size of the sample components (i.e., target protein and contaminants), and the available equipment (e.g., pumps and columns). It is necessary to begin by selecting either anion or cation exchange. This requires knowledge of the pI and pH stability of the target protein. If the pI of the target protein is known, an anion-exchange medium with an operating pH above the pI of the target protein or a cation-exchange medium with an operating pH below the pI of the target protein should be selected. If the pI of the target protein is unknown, it is desirable to determine it before beginning. The optimal operating pH can be determined empirically (see Support Protocol 1). Because the pI for most proteins is below pH 7 (Gianazza and Righetti, 1980), it is reasonable to select an anion-exchange medium and an operating pH of 8.5 to start, then evaluate the results and optimize conditions as necessary. It is also useful to know the pI and binding characteristics of the contaminants present in the protein solution. For example, if the pH of the binding buffer is higher than the pI of a major contaminant, that contaminant will not bind to an anion-exchange medium. If the starting material is crude cell lysate, DNA will be present and generally needs to be removed from the protein. Since DNA is anionic, it binds tightly to anion-exchange media and typically is not eluted with the salt and pH conditions used during protein purification. Selecting a Buffer System A buffer system must be selected for the desired pH range. As with selection of an ion-exchange medium, there are several factors that must be considered in selecting a buffer system, including the type of ion exchange to be performed, the pH stability of the sample and the pH range to be used, the required buffering capacity, and, finally, the cost. Ion-Exchange Chromatography

8.2.2 Supplement 15


Table 8.2.1

Buffers for Ion-Exchange Chromatographya

pKa (25°C)

pH range

Bufferb

Anion exchange 4.75 5.68 5.96 6.46 6.80 7.76 8.06 8.52 8.88 8.64 9.50 9.73 10.47 11.12

4.5-5.0 5.0-6.0 5.5-6.0 5.8-6.4 6.4-7.3 7.3-7.7 7.6-8.5 8.0-8.5 8.4-8.8 8.5-9.0 9.0-9.5 9.5-9.8 9.8-10.3 10.6-11.6

N-Methylpiperazine Piperazine L-Histidine bis-Tris bis-Tris propane Triethanolamine Tris⋅Cl N-Methyldiethanolamine Diethanolamine 1,3-Diaminopropane Ethanolamine Piperazine 1,3-Diaminopropane Piperidine

Cation exchange 2.00 2.88 3.13 3.81 3.75 4.21 4.76 5.68 7.20 7.55 8.35

1.5-2.5 2.4-3.4 2.6-3.6 3.6-4.3 3.8-4.3 4.3-4.8 4.8-5.2 5.0-6.0 6.7-7.6 7.6-8.2 8.2-8.7

Maleic acid Malonic acid Citric acid Lactic acid Formic acid Butanedioic acid Acetic acid Malonic acid Phosphate HEPES BICINE

Working Temperature concentration (mM) factorc 20 20 20 20 20 20 20 50 20 (pH 8.4) 20 20 20 20 20 20 20 20 50 50 50 50 50 50 50 50

−0.015 −0.015 −0.017 −0.02 −0.028 −0.028 −0.025 −0.031 −0.029 −0.026 −0.026 −0.031

−0.0024 +0.0002 −0.0018 +0.0002 −0.0028 −0.0140 −0.0180

aInformation from Pharmacia Biotech (1995). bAbbreviations: BICINE, N,N-bis[2-hydroxyethylglycine; bis-Tris, bis[2-hydroxyethyl]iminotris[hydroxymethyl]methane; bis-Tris propane, 1,3-bis[tris(hydroxymethyl)methylamino]-propane; HEPES, N-[2-hydroxyethyl]piperazine-N′-[2ethanesulfonic acid]. cChange in pK per °C (i.e., ∂pK /∂T). a a

Anionic buffers (e.g., acetate and phosphate) are preferred for cation exchange and cationic buffers (e.g., Tris⋅Cl, ethanolamine, and piperazine) are preferred for anion exchange. It is important to ensure that the buffering ion will have the same charge as the ion exchanger, and hence will not be bound. A constant buffering capacity and pH will thus be maintained during the ion-exchange experiment. Table 8.2.1 lists a variety of buffers useful for anion and cation exchange. Additives to be included in the buffers (e.g., detergents or protease inhibitors) should also carry the same charge as the ion-exchange medium, to preclude binding. Selecting Batch Versus Column Purification If the sample volume is large in relation to the size of the pumps and columns available in the laboratory, batch adsorption techniques (see Basic Protocol 1) are appropriate for the capture stage of purification. Batch methods, in which the sample and medium are directly mixed without use of a column, are employed to reduce the sample volume and



Supplement 15

Table 8.2.2

Ion-Exchange Media for Batch Adsorption

Medium

Type

Suppliera

Dextran bead matrix QAE Sephadex A-25 QAE Sephadex A-50 SP Sephadex C-25 SP Sephadex C-50

Strong anion Strong anion Strong cation Strong cation

Amersham Pharmacia Biotech Amersham Pharmacia Biotech Amersham Pharmacia Biotech Amersham Pharmacia Biotech

Microgranular cellulose matrix QA-52 Strong anion QA-53 Strong anion SE-52 Strong cation SE-53 Strong cation

Whatman Whatman Whatman Whatman

Polymer-coated silica matrix Accell Plus QMA Strong anion Accell Plus CM Weak cation

Waters Waters

Agarose bead matrix DEAE Bio-Gel A CM Bio-Gel A

Bio-Rad Bio-Rad

Weak anion Weak cation

aFor addresses and telephone numbers of suppliers, see SUPPLIERS APPENDIX.

total protein content prior to column chromatography. Batch techniques are also very useful for capture of the target protein from extremely crude samples (e.g., cell lysates and samples containing particulates and/or aggregates). In batch methods the flow and packing characteristics of the medium are of minor importance, but economy and capacity are of primary importance. Ion-exchange media suitable for batch purification at the capture stage of protein purification are presented in Table 8.2.2. High resolution is required in the intermediate purification and polishing stages of protein purification, and this can only be achieved using column chromatography. However, higher-than-necessary resolution is frequently traded off for greater throughput (i.e., amount of material processed in a defined time). The packing and flow characteristics of the medium are of primary importance when using column chromatography. Columns packed with smaller beads usually require higher operating pressures than columns packed with larger beads. However, smaller beads offer higher resolution, resulting from the increased efficiency that comes with decreasing bead size. Larger beads are desirable for applications requiring higher throughput. A medium with the smallest bead size that will provide the necessary throughput (and whose operating pressure can be accommodated by available equipment) should be selected. Table 8.2.3 lists a variety of media suitable for column purification at various stages in the purification process. See SUPPLIERS APPENDIX for contact information of suppliers of these media, all of whom provide technical support for selection and use of specific media. Capacity of Ion-Exchange Media


The amount of ion-exchange medium required for a particular application will be determined by the protein capacity of the medium (i.e., the amount of protein that can be bound to the matrix) at the chosen pH. The amount of medium required will in turn determine the size of column required. The capacity of an ion-exchange medium is dependent on a variety of factors, including charge and molecular size of the components in the sample and experimental conditions employed.

8.2.4 Supplement 15



8.2.5


Supplement 15

34 90 90 200 200

PS/DVB

Agarose Agarose Agarose Agarose

Agarose

Q Sepharose High Performance SP Sepharose High Performance Q Sepharose Fast Flow

SP Sepharose Fast Flow

SP Sepharose Big Beads Agarose Agarose

SOURCE 15S

STREAMLINE DEAE

STREAMLINE SP

Hi-Trap Qc HiTrap SPd

34

PS/DVB

SOURCE 15Q

200

15

15

10

PS/DVB

Mono S

Avg. bead diameter (µm)

10

Matrix

B/C

B/C

C

C/I

C/I

C/I/P

C/I/P

I/P

I/P

I/P

I/P

Use

BU

BU

BU

PP/BU

PP/BU

PP/BU

PP/BU

PP/BU

PP/BU

PP

PP

Form available

Ion-Exchange Media for Column Purificationa

Amersham Pharmacia Biotech Mono Q PS/DVB

Mediumb

Table 8.2.3

170

205

210

215

215

170

170

ND

ND

160

320

Ionic capacity (meq/ml)

—

—

—

—

—

—

—

—

—

—

—

Available capacity (mg/ml)

70

55

80

50

120

90

108

85

54

75

65

Dynamic capacity (mg/ml)

continued

4.5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.0; 100 cm/hr; 0.5 × 5–cm column Human IgG; 100 mM sodium acetate, pH 5.0; 300 cm/hr; 0.5 × 5–cm column 4.5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.0; 100 cm/hr; 0.5 × 5–cm column 55 mg/ml lysozyme; 20 mM phosphate, pH 6.8; 1 ml/min; 0.63 × 3–cm column 4.5 mg/ml BSA; 5 mM Tris⋅Cl, pH 8.0; 100 cm/hr; 0.5 × 5–cm column RNase; 100 mM sodium acetate, pH 5.0; 150 cm/hr; 0.5 × 5–cm column Human serum albumin; 50 mM Tris⋅Cl, pH 8.3 Human IgG; 100 mM sodium acetate, pH 5.0; 300 cm/hr; 0.5 × 5–cm column 2 mg/ml BSA; 100 mM sodium acetate, pH 5.0; 300 cm/hr BSA; 50 mM Tris⋅Cl, pH 7.5; 300 cm/hr; 5-cm-i.d. expanded-bed column Lysozyme; 50 mM phosphate, pH 7.5, 300 cm/hr; 5-cm-i.d. expanded-bed column

Conditions for determining dynamic capacity


8.2.6

Supplement 15


PM

PM PM

Macro-Prep high S

Macro-Prep high Q

Macro-Prep high S

65 35 100 65 35

PM PM PM PM PM PM PM PM PM PM PM

Waters Protein-Pak Q HR Protein-Pak SP HR Protein-Pak Q HR Protein-Pak SP HR Protein-Pak Q HR Protein-Pak SP HR 8 8 15 15 40 40

100 100 100

PM PM PM

50

50

10

10


TosoHaas Toyopearl QAE-550C Toyopearl SP-550C Toyopearl SUPER Q-650C Toyopearl SUPER Q-650M Toyopearl SUPER Q-650S Toyopearl SP-650C Toyopearl SP-650M Toyopearl SP-650S

Econo-Pac high Qe Econo-Pac high Sf

PM

Matrix

P P I/P I/P B/C/I B/C/I

C/I C/I/P I/P

I/P

C/I/P

C/I C/I C/I

B/C/I

B/C/I

I/P

I/P

Use

PP PP PP/BU PP/BU BU BU

PP/BU PP/BU PP/BU

PP/BU

PP/BU

PP/BU PP/BU PP/BU

P/B

PP/BU

PP/BU

PP/BU

Form available

Ion-Exchange Media for Column Purificationa, continued

Bio-Rad Macro-Prep high Q

Mediumb

Table 8.2.3

200 225 200 225 200 225

150 150 150

240

240

330 160 280

160

400

127

115


ND ND ND ND

126g 45h 50h 50h

65 40 75 40 60 20

ND

143g

— — — — — —

ND ND ND

60

40

50

20


70g 110h 129g

—

—

—

—


continued

BSA; 20 mM Tris⋅Cl, pH 8.2 Cytochrome c; 25 mM MES, pH 5.0 BSA; 20 mM Tris⋅Cl, pH 8.2 Cytochrome c; 25 mM MES, pH 5.0 BSA; 20 mM Tris⋅Cl, pH 8.2 Cytochrome c; 25 mM MES, pH 5.0

— — —

—

—

— — —

5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.3; 100 cm/hr; 1.5 × 10–cm column 5 mg/ml human IgG; 20 mM sodium acetate, pH 5.0; 100 cm/hr; 1.5 × 10–cm column 5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.3; 100 cm/hr; 1.5 × 10–cm column 5 mg/ml human IgG; 20 mM sodium acetate, pH 5.0; 100 cm/hr; 1.5 × 10–cm column



8.2.7


Supplement 15

CP

CP

CP

CP

CP

CP

CP

CP

BioSepra Q-Hyper D

S-Hyper D

Q-Hyper D

S-Hyper D

Q-Hyper D

S-Hyper D

Q-Hyper D

S-Hyper D

60

60

35

35

20

20

10

10


C/I

C/I

I/P

I/P

I/P

I/P

P

P

Use

BU

BU

PP/BU

PP/BU

PP/BU

PP/BU

PP

PP

Form available

100

200

100

250

100

250

110

250


—

—

—

—

—

—

—

—


100

100

100

40

105

90

105

90


5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.6; 150 cm/hr; determined at 10% breakthrough 5 mg/ml lysozyme; 50 mM acetate, pH 4.5; 150 cm/hr; determined at 10% breakthrough 5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.6; 150 cm/hr; determined at 10% breakthrough 5 mg/ml lysozyme; 50 mM acetate, pH 4.5; 150 cm/hr; determined at 10% breakthrough 5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.6; 150 cm/hr; determined at 10% breakthrough 5 mg/ml lysozyme; 50 mM acetate, pH 4.5; 150 cm/hr; determined at 10% breakthrough 5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.6; 150 cm/hr; determined at 10% breakthrough 5 mg/ml lysozyme; 50 mM acetate, pH 4.5; 150 cm/hr; determined at 10% breakthrough


hConditions for determining available capacity: 5 mg/ml lysozyme; 20 mM phosphate, pH 6.0.

gConditions for determining available capacity: 5 mg/ml BSA; 50 mM Tris⋅Cl, pH 8.7.

f Prepacked column suitable for use with syringe. For medium specification, see Macro-Prep high S.

e Prepacked column suitable for use with syringe. For medium specification, see Macro-Prep high Q.

d Prepacked column suitable for use with syringe. For medium specification, see SP Sepharose High Performance.

c Prepacked column suitable for use with syringe. For medium specification, see Q Sepharose High Performance.

bFor addresses and telephone numbers of suppliers, see SUPPLIERS APPENDIX.

aAbbreviations: B, batch; BSA, bovine serum albumin; BU, bulk; C, capture; CP, composite polymer, I, intermediate purification; i.d., inner diameter; MES, 2-[N-morpholino]ethanesulfonic acid; ND, not determined; P, polishing; PM, polymethacrylate; PP, prepacked; PS/DVB, polystyrene/divinylbenzene.

Matrix

Ion-Exchange Media for Column Purificationa, continued

Mediumb

Table 8.2.3

The number of charged groups on a specified amount of matrix that can participate in ion exchange is termed ionic or total capacity. However, ion-exchange media are porous and the ion-exchange groups are usually distributed throughout the bead. The amount of protein that will bind is dependent on the ability of the sample components to diffuse into the bead, and this ability is related to the molecular weight and shape of the components and the pore structure of the matrix (see UNIT 8.3). In addition, large molecules such as proteins and nucleic acids can bind to multiple ion-exchange groups (or sterically hinder other molecules from binding) depending on the shape and surface-charge distribution of the protein, as well as the spatial orientation of the exchange ligands on the matrix. These factors are taken into account in the available and dynamic capacity of a matrix or column. Available capacity is defined as the amount of a specific protein that will bind to an ion-exchange medium at a defined pH, salt concentration, and sample concentration in a batch purification process. When medium is packed in a column, the amount of a specific protein that will bind is dependent on many of these factors, as well as on the column dimensions and flow rate. The capacity of a column operating under defined conditions is termed the dynamic capacity. The dynamic capacity is typically less than the available capacity for the same sample under the same operating conditions. Ion-exchange media in which the charge state of the ligand is sensitive to the operating pH are termed weak ion exchangers. The usable pH range of weak ion exchangers is narrower than that of strong ion-exchange media, in which the charge on the ligand used for exchange is constant (i.e., insensitive to pH). Ion-exchange media with DEAE (diethylaminoethyl) or CM (carboxymethyl) groups as the ligand for exchange are considered weak ion exchangers. Because the ionic capacity of weak ion-exchange media is variable, strong ion-exchange media are recommended for pilot experiments (see Support Protocol 1). When selecting ion-exchange media for batch or column purifications, the amount of medium should be estimated from the manufacturer-supplied capacity data for a protein similar in size to the target protein to be purified. Be aware of what type of capacity (i.e., ionic, available, or dynamic) is being reported and the experimental conditions under which the reported capacity was determined (see Table 8.2.3). A general recommendation is to begin pilot experiments at ∼10% of the published capacity data, then determine the capacity for your system. Preparing Solutions All aqueous solutions for use in ion-exchange chromatography should be prepared with distilled or deionized water (Milli-Q purified or equivalent). Solutions and samples for column chromatography should be filtered, degassed, and equilibrated to the appropriate temperature prior to use. Special attention must be given to any additives to be included in the buffer to be sure that they will not bind to the column (see preceding discussion on Selecting a Buffer System). If the chosen additives increase the viscosity of the mobile phase, the flow rate should be decreased. Chromatography Systems


Column techniques typically result in higher resolution than batch techniques, and should be used during the intermediate purification and polishing stages (UNIT 8.1) of a purification procedure. However, to achieve high resolution, an appropriate chromatography system is required. A chromatography system may be described simply by the functional operations required to perform a chromatographic separation. The minimum functions required (Fig. 8.2.2) include a buffer delivery system capable of gradient formation, a mechanism for introducing sample into the system, the column itself (properly packed with the appro-

8.2.8 Supplement 15


Table 8.2.4 Scale-Up Calculations for Ion-Exchange Column Chromatography

Scale

1×

10×

100×

Sample Sample concentration (mg/ml) Sample load (mg) Sample volume (ml)

1 10 10

1 100 100

1 1000 1000

Column Column diameter (cm) Cross-sectional area (cm2) Bed height (cm) Bed volume (ml)

0.5 0.2 5 1

1.6 2 5 10

5 20 5 100

Flow rate Volumetric flow (ml/min)a Linear velocity (cm/hr)

2 300

20 300

200 300

Gradient Gradient volume (ml)

20

200

2000

aLinear flow (cm/hr) multipled by the cross-sectional area of the column (cm2) equals the volumetric flow rate (ml/hr); cm/hr × cm2 = cm3/hr = ml/hr.

priate ion exchanger), an in-line UV monitor suitable for use at 280 nm, a chart recorder, and a fraction collector to preserve the separation. Additional unit operations may include system controllers for automated operation as well as data-acquisition systems. The equipment required to perform each of these unit operations is ultimately determined by the chromatography medium and column selected. Specifically, the operating pressure and flow characteristics of the packed column will be the most important factors for selecting equipment to fit the column, or for selecting columns to fit existing equipment. It is highly recommended that prepacked columns be used and that the manufacturer be consulted prior to purchase as to equipment requirements for proper operation. This will save time and money and increase the probability of success. If columns are to be packed in the laboratory, the manufacturer’s packing instructions for the medium should be followed and column efficiency and function should be tested with representative standards prior to use with the actual sample. There is no single packing methodology that is optimal for all types of media. Table 8.2.3 lists a variety of ion-exchange media available in prepacked columns and/or in bulk for packing in the laboratory. Several manufacturers offer inexpensive disposable ion-exchange columns, suitable for many applications, that can be operated using only a syringe. Scale-Up Conditions Ion-exchange separations are typically developed and optimized on a small scale before the entire sample is committed. It is thus important to choose an ion exchanger that will allow simple and convenient scale-up, so that methods established on a small column can be applied more or less directly to a larger one. Scaling up in ion-exchange chromatography can be approached in two different ways— either the same ion-exchange medium can be used in a larger-diameter column with the same bed height, or a different medium having the same charged group immobilized on a matrix with a higher throughput can be substituted for the medium used in the small-scale separation.



Supplement 15

The former approach, using the same medium throughout, is the simplest. Scaling up a separation with the same medium and bed height used in the small-scale optimization is achieved by employing simple scale factors to adapt the flow, gradient volume, and sample load to the increased volume of a larger-diameter column. The scale factor for the required increase in flow rate is the ratio of the cross-sectional area of the larger column to that of the smaller column (i.e., the factor by which the total flow of solution through the column must be increased so that the same linear velocity used in the small-scale experiment is maintained). The increase in gradient volume and sample loading are directly proportional to the increase in column volume. Table 8.2.4 shows an example of changes that must be made to various parameters for a 10- and 100-fold increase in scale. BASIC PROTOCOL 1

BATCH ADSORPTION AND STEP-GRADIENT ELUTION WITH INCREASING SALT CONCENTRATION This protocol describes adsorption of a protein to an ion-exchange medium by direct mixing of the protein-containing sample and the medium (in this case, an anion-exchange gel). This batch purification is in contrast to column purification (see Basic Protocol 2), in which the gel is packed in a column and the sample is passed through it. Batch adsorption is particularly useful where large volumes of sample are involved, which is usually the case in the capture stage of protein purification (UNIT 8.1). The equipment used here is sufficient for sample volumes up to 1.5 liter. Larger samples may be divided and processed with parallel apparatuses, or processed sequentially on a single apparatus. Different-sized funnels and side-arm flasks may be used when appropriate. Once the starting conditions have been determined (see Support Protocol 1), the range of pH and salt concentration where binding and elution occur is well defined. Conditions for batch adsorption and elution used in this procedure are based on the results, presented in Figure 8.2.3, of a test tube pilot experiment performed as described in Support Protocol 1. Buffer volumes and equipment were chosen on the basis of a sample volume of 1 liter containing 50 to 1000 mg of total protein. Elution of sample components adsorbed to an ion-exchange medium can be achieved either by increasing the salt concentration of the mobile phase, as in this protocol, or by changing its pH (see Alternate Protocol). In batch techniques, discontinuous (step) gradients are used for elution. Column chromatography (see Basic Protocol 2) allows the use of either step or continuous gradients. Formation of continuous gradients requires special equipment (see Basic Protocol 2 and Support Protocol 3). The number of incremental steps to include in a step gradient must be judged empirically based on the resolution achieved. Ideally (as in the procedure described here), three steps should be used in which all weakly bound materials are desorbed and eluted in the first gradient step using wash buffer, the target protein is eluted in the second gradient step using elution buffer, and all strongly adsorbed materials are eluted in the third gradient step using regeneration buffer. A minimum increment of 50 mM NaCl is recommended for experiments using step-gradient elution with increasing salt concentration. Steps in the gradient may be added, omitted, or modified as determined empirically.


Materials QAE Sephadex A-25 (Amersham Pharmacia Biotech) or equivalent anion-exchange gel Binding buffer: 20 mM Tris⋅Cl, pH 7.5 (or other buffer as determined empirically; see Support Protocol 1) Protein sample to be purified Wash buffer: 20 mM Tris⋅Cl (pH 7.5)/100 mM NaCl (or other buffer/salt solution as determined empirically; see Support Protocol 1)



Elution buffer: 20 mM Tris⋅Cl (pH 7.5)/350 mM NaCl (or other buffer/salt solution as determined empirically; see Support Protocol 1) Regeneration buffer: 20 mM Tris⋅Cl (pH 7.5)/2 M NaCl (also see Support Protocol 5) Boiling water bath (optional) 500-ml sintered-glass filter funnel, medium porosity Three 2000-ml side-arm flasks Conductivity meter Swell and equilibrate ion-exchange gel 1. Add 10 g QAE Sephadex A-25 to 1 liter binding buffer and swell 2 days at room temperature or 2 hr in a boiling water bath. The volume of swollen gel will be ∼70 ml.

2. Mount 500-ml sintered-glass filter funnel on 2000-ml side-arm flask and apply suction. Centrifugation (∼5000 × g, 1 min) may be used instead of filtration to collect the gel, and may be preferred for processing small sample volumes. It is also possible to simply allow the gel to settle under gravity, then decant or aspirate the supernatant; this may be preferable for very large-scale operations.

3. Gently swirl swollen gel to resuspend. Pour gel slurry into funnel as fast as the fluid level in funnel permits, allowing buffer to collect in flask. Continue until all gel has been collected. Release suction after all buffer has been removed. IMPORTANT NOTE: Never use a magnetic stir-bar to resuspend chromatography medium. This can damage beads by grinding them against the bottom of the vessel, resulting in fine particles that can slow the filtration process.

4. Add 200 ml binding buffer to funnel and resuspend gel using a stirring rod. Allow resuspended gel to stand for 5 min, then apply suction to remove buffer. 5. Repeat step 4 at least three times to ensure equilibration. The gel is considered equilibrated when the pH and salt concentration of the eluant are the same as those of the binding buffer. If the procedures here are followed, there is little need to measure pH and salt concentration of the eluant because the gel is swollen and washed in binding buffer and the chance of error is very small. The worst-case scenario is that the target protein will not bind (see Troubleshooting). The repeated washings are more for the removal of fines than for adjustment of pH or salt concentration.

6. Add enough binding buffer (∼100 ml) to produce an ∼50% (v/v) slurry and allow gel to stand in funnel until sample has been prepared. Adsorb sample to gel 7. Adjust pH and salt concentration of protein sample to initial optimal values. For a pilot experiment to determine optimal initial values for pH and salt concentration, see Support Protocol 1. Samples prepared by salt precipitation may require desalting prior to ion exchange (see UNIT 8.3). The sample can be dialyzed against the binding buffer. Also, the salt concentration of a sample can be reduced by adding distilled water until a desired salt concentration is achieved, as measured by a conductivity meter.

8. Combine gel and sample in 2000-ml beaker or wide-mouth flask. The ∼50% gel slurry is swirled and poured from the funnel; transfer may be aided with a rubber policeman, spoon, stirring rod, or wash bottle containing binding buffer.



Supplement 15

9. Gently mix sample and gel by swirling every 15 min or shake on a platform shaker at sufficient rate to maintain gel in suspension. Allow 1 to 2 hr for binding at room temperature or 3 to 4 hr at 4°C. IMPORTANT NOTE: Excessive shaking may result in foaming and possible denaturation of the target protein.

10. Collect gel by filtration as in step 3. Save filtrate for assay. This filtrate is saved in case the target protein did not bind. All filtrates should be kept until the fraction containing the target protein has been identified. Storage conditions depend on the stability of the target protein; as a general rule they should be refrigerated to minimize microbial growth. Bacteriostatic agents may be added for prolonged storage.

11. Add 100 ml binding buffer to funnel and resuspend gel using a stirring rod. Allow resuspended gel to stand for 5 min, then apply suction to remove buffer. Repeat three or four times, pooling all washings with the filtrate from step 10. The gel containing the bound sample may be packed in a column for subsequent elution steps (see Basic Protocol 2).

Elute with step gradient of increasing salt concentration 12. Transfer the sintered-glass funnel to the top of a clean 2000-ml side-arm flask. 13. Add an equal volume of wash buffer to gel in funnel and resuspend using a stirring rod. Allow resuspended gel to stand for 5 min, then apply suction to remove buffer. Save filtrate. The wash buffer is the first step of the gradient. The filtrate will contain sample components that were weakly adsorbed to the gel (see annotation to step 10).

14. Measure absorbance of filtrate at 280 nm using wash buffer as blank. 15. Repeat steps 13 and 14 until the last filtrate shows no significant absorbance at 280 nm. Pool and save the filtrates. 16. Transfer the sintered-glass funnel to the top of a clean 2000-ml side-arm flask. 17. Add an equal volume of elution buffer to gel in funnel and resuspend using a stirring rod. Allow resuspended gel to stand 5 min, then apply suction to remove buffer. Save filtrate. The elution buffer is the second step of the gradient. The filtrate will contain the sample components that were more strongly adsorbed to the gel and should include the target protein.

18. Measure the absorbance of the filtrate at 280 nm using elution buffer as a blank. 19. Repeat steps 17 and 18 until the last filtrate shows no significant absorbance at 280 nm. Pool and save the filtrates for assay and subsequent purification of the target protein. If gel is to be reused, continue with remaining steps. Otherwise, discard the gel.

Regenerate ion-exchange medium for reuse 20. Transfer the sintered-glass funnel to the top of a clean 2000-ml side-arm flask. 21. Add an equal volume of regeneration buffer to gel in funnel and resuspend using a stirring rod. Allow resuspended gel to stand for 5 min, then apply suction to remove buffer. Repeat five times and discard filtrates. Ion-Exchange Chromatography

The regeneration buffer removes strongly bound materials from the gel. See Support Protocol 5 for additional considerations in regenerating ion-exchange media.



22. Repeat step 11 for a total of five washes to reequilibrate the gel with binding buffer. 23. Check the pH and conductivity of the last filtrate to ensure that the gel is properly equilibrated. See Support Protocol 5 for information on storing ion-exchange media.

pH-BASED STEP-GRADIENT ELUTION The net charge of a protein is pH dependent. Therefore, altering the pH of the mobile phase to make it closer to the pI of a protein can change its net charge, causing it to desorb and elute from an ion exchanger (see Figure 8.2.1). Continuous pH gradients are very difficult to produce at constant salt concentration because charge characteristics of sample components, buffering ions, and ion-exchange media depend on the pH of the system. pH-based step gradients are much easier to produce and more reproducible. For anionexchange media, elution occurs when the pH is decreased. For cation exchange, the pH is raised for elution.

ALTERNATE PROTOCOL

The first basic protocol can be modified for pH-gradient elution by selecting a buffer at a pH suitable for elution, based on results obtained in a pilot experiment (see Support Protocol 1). This elution buffer is then used in steps 17 to 19 of the first basic protocol. Incubation times and buffer volumes should be increased by 20% when using pH elution. When scouting methods for ion exchange, it is best to start with changes in salt concentration (see Basic Protocol 1) as this technique is simpler and more reproducible. Elution with pH change is only recommended when the ion-exchange behavior of the sample is well known and the resolution required cannot be attained with a change in salt concentration. pH elution may also be appropriate in cases where it provides a superior ionic/pH environment for loading in the next chromatographic step. A combination of increasing salt concentration and pH change may also be used for elution. COLUMN CHROMATOGRAPHY WITH LINEAR GRADIENT ELUTION Column chromatography is generally preferred to batch chromatography (see Basic Protocol 1), especially where high resolution is required, although it accommodates smaller sample volumes. Column chromatography requires specialized equipment to achieve high resolution, including a column containing the medium and a fluid-delivery system appropriate for the pressure and flow rate at which the column is to be operated. It is the technique preferred for the intermediate purification and polishing stages of protein purification (UNIT 8.1). Also in contrast to the first basic protocol, this technique uses a linear gradient for elution. Although more difficult to produce than step gradients (see Basic Protocol 1), linear gradients lead to better resolution (see Support Protocol 3).

BASIC PROTOCOL 2

The example presented here uses a RESOURCE Q anion-exchange column from Amersham Pharmacia Biotech, which may be used with either FPLC (fast protein liquid chromatography) or HPLC (high-performance liquid chromatography) systems. Optimal conditions for other columns should be obtained from the manufacturer. Procedures mentioned in the protocol steps should be carried out according to the manufacturer’s instructions for the particular chromatography system being used. NOTE: All buffers, media, and other system components should be filtered, degassed, and equilibrated to the same temperature before use. Conventional Chromatographic Separations


Supplement 15

Materials Liquid chromatography system (FPLC or HPLC) Elution buffer: binding buffer (see Basic Protocol 1 and Support Protocol 1) containing 1 M NaCl Binding buffer (see Basic Protocol 1 and Support Protocol 1) RESOURCE Q chromatography column (1-ml packed bed volume; Amersham Pharmacia Biotech) Protein sample to be purified Conductivity meter 0.22-µm filter Prepare chromatography system 1. Set up the liquid chromatography system according to manufacturer’s instructions, without the column in-line. 2. Test system performance by running a blank gradient ranging from 0% to 100% elution buffer in 20 ml at a constant flow rate of 5 ml/min. See Figure 8.2.2 for a diagram of a typical column chromatography system. The gradient is formed by placing binding and elution buffer in the appropriate buffer reservoirs (see Support Protocol 3 for detailed discussion of continuous gradient formation).

pump

injector

mixer

buffer reservoirs column gradient maker

detector

recorder

fraction collector Ion-Exchange Chromatography

Figure 8.2.2 Liquid column chromatography system with gradient maker.



Stability of the column at extremes of pH must be considered when choosing binding and elution buffers. The working stability of the RESOURCE Q column is pH 2 to 12, and the stability for cleaning is pH 1 to 14. The flow rate for the test should be appropriate for the column to be connected. For the RESOURCE Q column, the recommended flow rate is 1 to 10 ml/min and the maximum operating pressure is 4 bar.

3. Test for system leaks and ascertain monitor stability according to system documentation. Test accuracy of gradient composition using conductivity meter. Adding 0.1% (v/v) acetone to the elution buffer used in step 2 will allow observation of gradient composition at 280 nm using the UV monitor. Do not include acetone in the elution buffer used for the chromatographic run.

Prepare column 4. Purge system with binding buffer to remove any air, then install column in system. New columns may require precycling prior to use; consult product documentation. Used columns may require cleaning or removal of storage solution prior to use (see Support Protocol 4).

5. Wash column with 5 Vc (5 ml for RESOURCE Q column) of elution buffer at a flow rate of 5 ml/min and check for leakage. Vc refers to the packed bed volume of a column and is calculated according to the equation Vc = πr2 × L, where r is the radius of the column and L is the bed height (i.e., the height of the packed medium in the column). As the RESOURCE Q column used here has a packed bed volume of 1 ml, the column is washed with 5 ml of elution buffer. Consult product literature to find Vc for other columns.

6. Equilibrate column with 5 to 10 Vc of binding buffer at 5 ml/min. Collect one fraction at the end of the equilibration stage and measure pH and conductivity to ensure the pH and salt concentration of the fraction are the same as that of binding buffer. If the pH and salt concentration of the fraction are not the same as that of the binding buffer, continue passing binding buffer through the column until equilibration is complete.

Purify protein sample and regenerate column 7. Adjust pH and salt concentration of sample to binding conditions determined in Support Protocol 1. Filter sample prior to injection using a 0.22-µm filter to remove particulates that may clog system or column. For initial experiments, a sample containing 25 mg total protein is appropriate. For columns packed with beads of average diameter ≥90 ìm, a 1-ìm filter may be used; for columns packed with beads of average diameter 34 to 90 ìm, a 0.45-ìm filter may be used. The 0.22-ìm filter is appropriate for columns packed with beads of average diameter 0.3 M; and the available capacity will be >50 mg/ml.



Supplement 15

3. Add 2 ml of 50% slurry of Q Sepharose Fast Flow to each tube and mix. SP Sepharose Fast Flow (Amersham Pharmacia Biotech) or equivalent may be used for cation exchange.

4. Allow gel to settle to bottom of tube or centrifuge 1 min at ∼5000 × g, room temperature. 5. Decant supernatant from each tube and replace with 5 ml of the same buffer added in step 1. Gently shake or vortex to resuspend gel, then let tubes stand 2 min. 6. Repeat steps 4 and 5 three times to equilibrate gels in each tube. 7. Decant supernatants and resuspend each equilibrated gel in 1 ml of the same buffer added in step 1. 8. Set up eight identical aliquots (≥1 ml) of the protein to be purified in tubes numbered 1 to 8, each containing 0.1 to 1 mg total protein. Adjust the pH of each to obtain eight different pH values corresponding to those of tubes 1 to 8 in step 1. Quantity of protein to be used depends on availability of sample and extinction coefficient of target protein. pH may be adjusted by titrating with acid or base using pH paper or by diluting 1:1 with buffer of desired pH.

9. Add the corresponding pH-adjusted protein aliquot to each of the 8 tubes from step 7. Mix gently for 10 min by periodic swirling, then allow gel to settle. If the salt concentration of the sample is too high, no binding will occur. It is therefore recommended that the sample be desalted into the appropriate buffer as in UNIT 8.3.

10. Assay supernatant from each tube for the target protein. Measuring the absorbance of the supernatant at 280 nm is usually sufficient. Figure 8.2.3 shows a set of possible results. A decrease or absence of target protein activity (or absorbance) in the supernatant indicates binding. Specific assays for the target protein may also be used.

11. Select the lowest pH at which the highest quantity of target protein is bound to use as the binding pH for further work. Select the highest pH where no binding occurs as the elution pH. For anion exchange, it is best to select a binding pH no more than 0.5 to 1 pH unit (pH 7.5 to 8.0 in Fig. 8.2.3) above the highest pH at which no binding occurs. If too high a pH is chosen, elution becomes difficult and high salt concentrations may be required. For cation exchange, select a binding pH 0.5 to 1 unit below the lowest pH where no binding occurs. Select the lowest pH where no binding occurs as the elution pH.

Determine salt concentration for binding and elution 12. Set up a series of test tubes, numbered 1 to 10. Using the binding buffer selected in step 10 in place of the different buffers in step 1, perform steps 1 to 7 to equilibrate the Q Sepharose Fast Flow with buffer in all ten tubes. 13. To each tube, add the same amount of the protein to be purified that was added to the tubes in step 8, equilibrated in the binding buffer selected in step 10. 14. Mix contents of tubes gently for 10 min, then allow gel to settle. 15. Decant supernatants from each tube and discard. Wash gel two times with 5 ml of the binding buffer, allowing the gel to settle and decanting the supernatants after each addition of buffer. Ion-Exchange Chromatography



16. Add 2 ml of the binding buffer to each tube, then sequentially add water and 4 M NaCl to each tube as follows: tube 1: 1.90 ml H2O and 0.10 ml 4 M NaCl (0.10 M NaCl final) tube 2: 1.80 ml H2O and 0.20 ml 4 M NaCl (0.20 M NaCl final) tube 3: 1.70 ml H2O and 0.30 ml 4 M NaCl (0.30 M NaCl final) tube 4: 1.60 ml H2O and 0.40 ml 4 M NaCl (0.40 M NaCl final) tube 5: 1.50 ml H2O and 0.50 ml 4 M NaCl (0.50 M NaCl final) tube 6: 1.40 ml H2O and 0.60 ml 4 M NaCl (0.60 M NaCl final) tube 7: 1.30 ml H2O and 0.70 ml 4 M NaCl (0.70 M NaCl final) tube 8: 1.20 ml H2O and 0.80 ml 4 M NaCl (0.80 M NaCl final) tube 9: 1.10 ml H2O and 0.90 ml 4 M NaCl (0.90 M NaCl final) tube 10: 1.00 ml H2O and 1.00 ml 4 M NaCl (1.00 M NaCl final). 17. Mix contents of tubes gently for 10 min, then allow gel to settle. 18. Assay supernatant from each tube for the target protein. Figure 8.2.3 shows a set of possible results. Presence of target protein in the supernatant indicates elution of the target protein.

19. Select the highest salt concentration at which no elution occurs as the maximum salt concentration allowable for binding target protein and for washing away unbound sample components. Select a salt concentration at least 0.05 M above the concentration where no protein binds to the gel as the salt concentration for elution. Determine available capacity 20. Prepare a buffer/salt solution with the binding pH selected in step 10 and the binding salt concentration selected in step 19. 21. Set up a series of test tubes numbered 1 to 10. Using the buffer/salt solution prepared in step 20 in place of the different buffers in step 1, perform steps 1 to 7 to equilibrate the Q Sepharose Fast Flow with buffer/salt solution in all ten tubes. 22. Add aliquots of sample to each tube to obtain the following amount of target protein: tube 1: 10 mg tube 2: 20 mg tube 3: 30 mg tube 4: 40 mg tube 5: 50 mg tube 6: 60 mg tube 7: 70 mg tube 8: 80 mg tube 9: 90 mg tube 10: 100 mg. 23. Mix gently for 10 min, then allow gel to settle. 24. Assay supernatant from each tube for the target protein. Figure 8.2.3 shows a set of possible results. Presence of target protein in supernatant indicates that the available capacity has been exceeded.

25. Select the highest protein concentration at which the target protein does not appear in the supernatant as the available capacity. Calculate the amount of ion-exchange medium needed for batch purification based on 50% of the available capacity (defined in UNIT 8.1). If a column system is used in pilot experiments, however, it is best to determine the dynamic capacity of the medium (see Support Protocol 2). However, 20% of the available capacity is a safe and reasonable starting point for pilot experiments using column chromatography.



Supplement 15

SUPPORT PROTOCOL 2

MEASUREMENT OF DYNAMIC (COLUMN) CAPACITY AND BREAKTHROUGH CAPACITY OF ION-EXCHANGE COLUMNS For chromatographic separations that will be performed on a routine basis as well as separations that will be scaled up, it is necessary first to optimize conditions with respect to resolution and to determine the capacity of the medium with respect to the target protein (see Support Protocol 1). Knowledge of the capacity allows optimal use of the gel medium in terms of cost and yield, and also allows working limits to be set on the sample load to ensure robustness of a particular purification step. After capacity has been determined, the separation can be optimized with respect to time. There are two capacity terms that commonly appear in the literature (in addition to available capacity; also see Support Protocol 1 and UNIT 8.1): dynamic capacity and breakthrough capacity. The dynamic (or column) capacity is the capacity of an ion-exchange column under defined conditions of flow rate, pH, salt concentration and sample concentration. The breakthrough capacity (QB) for a system is obtained by calculating the amount of protein that has been absorbed by the column when the sample is first detected in the effluent, or when the recorder signal reaches some arbitrarily defined percent of full-scale deflection. Breakthrough capacity values at 50% full-scale deflection (QB50) are commonly reported in the literature. However, it is generally recommended to use only 10% to 20% of the published QB as the practical capacity of the column to ensure high recovery of the target protein. Additional Materials (also see Basic Protocol 2) Ion-exchange gel of unknown capacity, and column Elution buffer capable of eluting target protein in single step of salt concentration or pH (e.g., 2 M NaCl; see Support Protocol 1) 1. Pack a defined amount of the ion-exchange gel into the column. See UNIT 8.3 for guidelines on column packing. Normally, a quantity of gel sufficient to give a packed bed volume of ∼1 ml is sufficient. In the case of a prepacked column, the amount of gel is predetermined. The packed bed volume (Vc) is calculated according to the equation Vc = π r2 × L, where r is the radius of the column and L is the bed height (i.e., the height of the packed medium in the column).

2. Prepare chromatography system and column (see Basic Protocol 2, steps 1 to 6) using the optimal binding and elution conditions determined in Support Protocol 1. Do not exceed 75% of the flow rate used for packing. The effect of flow rate on capacity should be determined in a true optimization.

3. Prepare protein solution and inject into column (see Basic Protocol 2, steps 7 and 8). Apply solution continually until the recorder shows >50% full-scale deflection, then stop. The absorbance of the binding buffer corresponds to 0% and the absorbance of the sample corresponds to 100% full-scale deflection.

4. Wash column with binding buffer until 0% full-scale deflection is approached. 5. Elute proteins with a buffer providing a single-step increase in salt concentration or pH. Continue collecting eluant until recorder shows ≤2% full-scale deflection, then pool fractions. A typical chromatogram for this procedure is presented in Figure 8.2.4.

6. Assay pooled fractions from step 5 for target protein. Calculate maximum amount of protein (A) that can be bound to the column according to the equation A = Cp × V, Ion-Exchange Chromatography



Full-scale deflection (%)

sample application

washing

elution

100

50

pool x Elution volume (ml)

Figure 8.2.4 Typical chromatogram for determining the capacity of an ion-exchange column. The volume of eluant (x) at 50% full-scale deflection is used to calculate the breakthrough capacity (QB50).

where Cp is the protein concentration of the pooled fractions (mg/ml) and V is the volume of the pooled fractions (ml). 7. Calculate the column (dynamic) capacity (mg/ml) using the equation dynamic capacity = A/Vc. 8. Calculate breakthrough capacity (QB50) using the equation QB50 = (Cs × x)/Vc, where Cs is the concentration of protein in the original sample and x is the volume of sample that has been applied to the column at the point where 50% full-scale deflection has been attained. If QB50 greatly exceeds the dynamic capacity, then the target material may still be adsorbed on the column, and the elution conditions should be reexamined. See Figure 8.2.4 for graphical depiction of these variables.

GRADIENT-FORMATION TECHNIQUES

SUPPORT PROTOCOL 3

Simple Gradient Mixers Gradient mixers of the type shown in Figure 8.2.5 can generate linear pH or salt gradients in gravity-based or one-pump chromatography systems. Similarly designed gradient mixers used for casting electrophoresis gels may also be used for chromatography. The reservoir closest to the column is filled with the binding buffer and the second reservoir is filled with an equal volume of the elution buffer. The gradient begins when the valve between the two reservoirs is opened. Continuous mixing is required in the reservoir closest to the column to ensure gradient linearity and reproducibility.



Supplement 15

stirrer

gradient maker

mixing chamber

elution buffer

valve

binding buffer

Figure 8.2.5 Gradient mixer for forming gradients of pH and salt concentration used during ion-exchange chromatography. The apparatus shown is an Amersham Pharmacia Biotech Gradient Mixer GM-1, which can be used for preparing gradients up to 500 ml. A linear gradient of salt concentration is produced by filling the reservoirs with buffers at the same pH and different salt concentration. The mixing chamber contains the lower-salt-concentration (binding) buffer, and the other chamber contains the higher-salt-concentration (elution) buffer. A continuous linear or nonlinear pH gradient may be produced by filling the reservoirs with buffers at the same salt concentration and different pH. A gradient of increasing pH is used for cation-exchange separations, and a gradient of decreasing pH is used for anion-exchange separations. The buffer chambers are joined by a channel controlled by a valve, and outflow from the mixing chamber is controlled by another valve.

Gradient Mixing with Multichannel Peristaltic Pumps Peristaltic pumps that can accommodate three or more pumping channels simultaneously (Fig. 8.2.6) can be used to form linear or complex gradients (see Critical Parameters). Linear gradients are easily formed by using the same size tubing for all channels, pumping with one channel from the elution buffer reservoir to the binding buffer reservoir, and pumping with the other two channels from the binding buffer reservoir to the column. Continuous mixing is required in the binding buffer reservoir closest to the column to ensure gradient linearity and reproducibility. Switch-Valve-Based Gradient Mixing


Switch-valve-based gradient formation is useful for forming step, linear, and complex gradients (see Critical Parameters). Gradients are formed by proportioning variable amounts of binding buffer and elution buffer through a single three-way valve using a system controller. When programmed to deliver 10% elution buffer, the controller will open the valve port for binding buffer 90% of the time and the port for elution buffer 10% of the time. As higher switching rates are required for lower flow rates to ensure gradient accuracy, high-quality system controllers will vary the rate at which the valve switches (i.e., the number of switching events per unit time) as a function of flow rate. A dynamic mixer should be included in the flow path just after the proportioning valve to ensure gradient accuracy. Switch-valve-based gradients are generally not very accurate from 1% to 10% elution buffer (90% to 99% binding buffer) or from 1% to 10% binding buffer (90% to 99% elution buffer). Therefore, the ionic strength of the binding and elution buffers may be adjusted to allow a gradient of the desired composition to be formed by mixing the two buffers in the range of 20% to 80% elution buffer (80% to 20% binding buffer).



channel 1 channel 2 channel 3

resembles a label

multichannel peristaltic pump

this

is

elution buffer

text

binding buffer

mixer

channel 3

to column

Figure 8.2.6 Peristaltic pump accommodating three pumping channels for continuous gradient formation.

Gradient Formation by Autoblending Some chromatography systems allow gradient formation over a range of both pH and salt concentration. Four solvents—a high-pH buffer, a low-pH buffer, a concentrated salt solution, and water—are blended via four two-way valves. A system controller proportions the high- and low-pH buffers to achieve the desired pH gradient while simultaneously proportioning the concentrated salt solution and water to achieve the desired salt-concentration gradient. Gradients formed in this way are usually difficult to reproduce but may be useful for scouting initial binding and elution conditions. Gradient Formation with Twin Pumps Use of two positive-displacement pumps, as in HPLC and FPLC systems, provides the highest degree of accuracy, precision, and reproducibility in gradient formation. One pump delivers binding buffer while the other delivers elution buffer. A system controller allows gradient formation simply by controlling the flow rate of each pump. A dynamic mixer should be included in the flow path, after the pumps and before the column, to ensure gradient accuracy. CLEANING AND REGENERATION OF ION-EXCHANGE MEDIA Proper maintenance of chromatography media is as much an art as protein purification itself. Knowing what kind of contaminating material from the sample is bound to the column helps in selecting a regime for maintaining good column hygiene. It is best to practice preventative maintenance by regenerating the medium after every run, using a high-salt buffer or large change in pH, and periodically cleaning the column and checking the system flow characteristics. If any increase in column back-pressure is noted, the column should be cleaned as soon as possible. Once enough contaminant has bound to a column to reduce or block the flow, it is probably too late to save the medium without investing considerable time and effort.

SUPPORT PROTOCOL 4



Supplement 15

Routine Washing of Columns and Media Columns or bulk medium should be washed after every run with salt solution until an ionic strength of ≥2 M is reached. This should remove any substances bound by ionic forces. This is recommended after every run. Removal of Contaminants Alkali-soluble contaminants. Contaminants such as lipids, proteins, and nucleic acids can usually be removed by washing with 2 to 3 column volumes of 0.1 M NaOH, followed by 2 to 3 column volumes of Milli-Q-purified (or equivalent) distilled water and then 2 to 3 column volumes of binding buffer. Many media can tolerate brief exposure to 1 M NaOH, if harsher conditions are required to remove contaminants; however, manufacturer’s guidelines should be consulted regarding pH stability of the medium. Hydrophobic contaminants. Lipids and other hydrophobic materials may be removed by washing with alcohol solutions (e.g., 70% ethanol) or nonionic detergents. When using alcohols or organic solvents to remove hydrophobic materials, it is often effective to wash first with a gradient from 0% to 100% and then with a gradient from 100% to 0%. Repeat washing with the alternating gradients until no contaminants are detected in the eluant. Metallic contaminants. Contamination with metals may cause blue or gray discoloration at the top of an ion-exchange column. These metals can originate from impurities in buffer salts or water, or they can leach from metallic system components in contact with buffer solutions. Metal contaminants may usually be removed by treating the column with several column volumes of 10 mM HCl (i.e., pH 2) saturated with EDTA. Manufacturer’s guidelines should be checked for pH stability of medium and column before exposure to mild or strong acid solutions. Precipitated materials. Precipitated materials that have accumulated in a column are very difficult to remove. Unless the precipitated material can be physically removed by removing the gel at the top of the column, the material will have to be resolubilized. Detergents, urea, and guanidine⋅Cl can be introduced to help dissolve contaminants. The viscosity of concentrated solutions used for cleaning may require very low flow rates. The column may be equilibrated and incubated in 6 M urea, then washed with distilled water and buffer. Degradative enzymes (e.g., proteases and lipases) may also be introduced in an attempt to degrade precipitates on the column; however, enzymatic cleaning can be quite expensive. It may be useful to reverse the direction of flow through the column (i.e., from bottom to top) when attempting to remove particulate material from the top of the bed. SUPPORT PROTOCOL 5


STORAGE OF ION-EXCHANGE MEDIA Bacterial and microbial growth can seriously interfere with the chromatographic properties of any chromatography medium and endanger the sample as well. Prior to storage, all media should be cleaned and sanitized. During prolonged experiments or storage (i.e., >24 hr at room temperature or 48 hr at 4°C), a bacteriostatic or antimicrobial agent should be added to the ion exchanger. Antimicrobials chosen to be added for storage should not bind to the ion exchanger and should be easily removed when the gel is to be reused. Azide is anionic and will bind to anion-exchange media. Treatment with 1 M NaOH for 1 hr will sanitize most media by lowering the bacterial count 100-fold, and will help solubilize dead cells (Pharmacia Biotech, 1997). Storage of sanitized media in 0.1 M NaOH will not leave any toxic materials on the column—a significant consideration in pharmaceutical production. Manufacturer’s guidelines should



be checked for stability of the medium and column in NaOH. Silica-based media will not tolerate high pH. Exposure to 70% ethanol for 3 to 4 hr can be used to sanitize most media. Most ion-exchange media and columns can be stored for prolonged periods of time in 20% ethanol. Manufacturer’s guidelines should be checked for stability of the medium and column in ethanol. Effective antimicrobial agents appropriate for use with anion-exchange media include either 0.001% phenyl mercuric salts or 0.002% chlorhexidine, in weakly alkaline solution. An effective antimicrobial agent for use with cation-exchange media is 0.005% merthiolate in weakly acidic solution. An agent that is effective for use with either anion or cation exchangers is 0.05% trichlorobutanol in weakly acidic solution. COMMENTARY Background Information

Critical Parameters

The first ion exchangers were synthetic resins designed for applications such as demineralization, water treatment, and recovery of ions from waste. Early ion-exchange resins were tightly cross-linked hydrophobic polymers, highly substituted with ionic groups, and had very high capacity for small ions. The high degree of cross-linking provided mechanical strength, but the limited porosity restricted use of these media with large molecules. In addition, the high charge density resulted in very strong binding, and the hydrophobic matrix tended to adsorb and denature labile biological materials. The first ion exchangers designed for use with biomolecules were developed by Peterson and Sober (1956). These cellulose-matrix ion exchangers were very hydrophilic and had little tendency to denature proteins, but had low capacities and poor flow characteristics. Modern ion-exchange media have very high capacity, often >100 mg/ml. Modern particle technology has significantly increased the mechanical strength and flow characteristics of ion-exchange media so that separations that once required hours or days may now be completed in a few minutes. Ion-exchange can easily be coupled to other methods of protein separation in a multistep purification. Samples purified by ion exchange can typically be applied to a hydrophobic-interaction chromatography (HIC) medium simply by adding NaCl (or other salts) to a concentration sufficient for binding (UNIT 8.4). Samples to be further purified by chromatofocusing or an additional round of ion exchange will require desalting (UNIT 8.3). Samples may usually be applied directly to gel-filtration columns (UNIT 8.3) with no intervening treatment.

The most critical aspects of any ion-exchange experiment are selection of the appropriate ion-exchange medium and of the initial conditions for binding and elution of the target protein or contaminants. Once these parameters are selected, conditions can be optimized with respect to the three primary parameters of any chromatographic separation—i.e., resolution, capacity, and speed. These three parameters are mutually exclusive. Resolution will typically decrease when the sample load is increased and when the flow rate is increased (to increase speed). It is generally recommended that resolution be optimized first, then capacity, and finally speed. Optimization of resolution This parameter is most easily optimized by controlling the selectivity (UNIT 8.3) of the medium through manipulation of binding and elution conditions. Construction of chromatographic titration curves followed by optimization of the shape of the elution gradient is a common approach to achieving maximum resolution. Chromatographic titration curves can be constructed by performing a series of anionand cation-exchange experiments at different pH values as in Figure 8.2.7. For each of these experiments, the elution salt concentrations (in mM NaCl) for sample components A, B, and C are plotted on the y axis and the operating pH is plotted on the x axis. It is then simple to observe the pH ranges within which optimal resolution between the various sample components is achieved. The different sample components are usually detected by UV absorption, but may also be viewed by other means (e.g., fluorescence or bioactivity). Data for chroma-



Supplement 15

A

a. cation exchange b. cation exchange at pH 3.0 at pH 5.0

C B A

A280

C

B

B C

c. cation exchange at pH 7.0

A

A

Elution NaCl concentration d. anion exchange at pH 8.0

e. anion exchange at pH 10.0

C

BC

f. anion exchange at pH 11.0

A

B

B A

A

A280

C

Elution NaCl concentration

a cation exchange

Increasing mM NaCl

B

A

b

B c C 9

0 3

4

5

10

11

pH

6

Increasing mM NaCl

B C d

e A f anion exchange

Figure 8.2.7 Construction of a chromatographic titration curve. (A) Six chromatographic separations (a to f) of a sample containing proteins A, B, and C are carried out at six different operating pHs using a linear gradient of increasing salt concentration. (B) The elution ionic strengths for components A, B, and C (y axis) are plotted against the operating pH (x axis) for experiments a to f. Note that the salt concentration in mM NaCl (y axis) is ascending in both directions. The optimal resolution with respect to component A occurs below pH 5 with cation exchange and above pH 9 with anion exchange.




tographic titration curves can often be generated by fast screening with low-volume gradients (e.g., 5 to 10 Vc) and low sample mass (e.g., 0.5

—

BR

Bio-Gel HTP (DNA-grade)

Dry

10

>0.8

—

BR

Small particle size to increase capacity and selectivity for dsDNA; suitable for batch use or short columns only

Hydroxylapatite Fast Flow

Dry

>12

>0.7

—

CB

Hydroxylapatite High Resolution Hydroxyapatite Type I

Dry

>15

>1.2

—

CB

Recommended for applications requiring high flow rate and binding capacity Recommended for analytical applications

Hydrated

—

—

—

S

Hydroxyapatite Type III

Hydrated

—

—

—

S

HA-Ultrogel

Hydrated

—

—

—

S

Macro-Prep Ceramic Type I

Dry

—

—

>25

BR

Macro-Prep Ceramic Type II

Dry

—

—

>12.5

BR

Hydroxylapatite HPLC grade TSK gel HA-1000

Dry

>15

>1.2

—

CB

Suitable for HPLC

Prepacked column

>20

—

—

TH

Suitable for HPLC; available as prepacked glass and stainless steel columns; use with guard column kit HA-1000 recommended

Contains silica gel as crystal initiator Microcrystalline; requires no precycling; good flow rates even in large columns Ceramic, macroporous; high protein-binding capacity; best for acidic proteins; excellent flow rates and mechanical stability Macro-Prep Physical properties as Type I; recommended for purification of nucleic acids; low affinity for albamin, therefore recommended for immunoglobulin work

aAbbreviations: BR, Bio-Rad; CB, Calbiochem; S, Sigma; TH, TosoHaas. For addresses and phone numbers of suppliers, see SUPPLIERS APPENDIX.

Materials Hydroxylapatite matrix (see Table 8.6.1) Loading buffer: 10 mM sodium phosphate, pH 6.8 (APPENDIX 2E) Protein sample to be purified Gradient solutions: 0.01 and 0.4 M sodium phosphate, pH 6.8 (APPENDIX 2E) Loading buffer (see above) with 0.02% sodium azide

Hydroxylapatite Chromatography

Side-arm flask and vacuum source Glass chromatography column (see Strategic Planning) with two adaptors or removable end piece and one adaptor (Fig. 8.3.1), tubing, and optional packing reservoir (e.g., Pharmacia Biotech C or XK series)

8.6.2 Supplement 8


Tubing clamps Peristaltic pump Buffer reservoirs Fraction collector UV monitor with 280-nm filter, connected to a chart recorder Gradient mixer Prepare gel 1. Calculate the amount of hydroxylapatite matrix required based on protein-binding capacity. Capacity of the matrix is usually supplied by the manufacturer

If matrix is supplied dry 2a. Degas a sufficient volume of loading buffer in a side-arm flask under vacuum. Add one part matrix to six parts loading buffer and resuspend by swirling gently. Do not use a magnetic stirrer as this will damage the hydroxylapatite crystals.

3a. Allow the slurry to settle for a minimum of 20 min, then decant the fines, which are in the cloudy solution above the settled matrix. 4a. Add a volume of loading buffer approximately equal to the volume of the settled bed and mix gently by swirling. Allow matrix to settle. 5a. If the matrix settles to give a sharp dividing line between the bed and clear buffer above, swirl gently to resuspend and proceed to packing of the column (step 6). If the supernatant buffer is still cloudy, repeat steps 3a and 4a, then proceed to step 6. If matrix is supplied hydrated 2b. Resuspend the matrix (in its own storage buffer as supplied by the manufacturer) by swirling gently. 3b. Pour a sufficient volume of the suspension into a beaker and allow to settle for a minimum of 20 min. Decant the fines. 4b. If the buffer in which the matrix was shipped is different from the loading buffer (10 mM sodium phosphate) add a volume of loading buffer approximately equal to that of the matrix, mix by swirling gently, allow to settle for a minimum of 20 min, then decant. 5b. Add a volume of loading buffer approximately equal to that of the matrix and swirl gently to resuspend. Proceed to packing of the column. Pack column 6. Examine the column to check that it is clean and undamaged and that sinters, support nets, and O rings are present and intact. 7. Insert bottom adaptor if used, or screw in the end piece. Mount column in a laboratory stand with clamps to ensure that it is vertical. If the volume of the resin slurry is greater than that of the column, fit a packing reservoir (Fig. 8.3.1). 8. Pour a small volume of loading buffer into the column and check that it flows freely through the bottom sinter. Clamp off the outlet tubing with a tubing clamp, leaving ∼2 cm of buffer above the sinter. 9. Swirl the hydroxylapatite slurry (step 5a or 5b) gently to resuspend, then pour it all into the column at once and allow 2 to 3 cm to settle under gravity. Release the tubing clamp and allow the rest of the bed to pack under gravity flow. 10. Attach a peristaltic pump to the bottom outlet and pump 5 bed volumes of loading buffer through the column to equilibrate.



Supplement 8

As a rough guide, a standard general-purpose matrix such as Bio-Gel HT (see Table 8.6.1) may be used at a linear flow rate of 75 cm/hr. Fine-particle DNA columns should be run at ∼1⁄10 this rate. Macroporous material—e.g., Macro-Prep Ceramic (Table 8.6.1) can tolerate flow rates up to 5000 cm/hr. The linear flow rate (cm/hr) is equal to the volumetric flow rate (cm3/hr) divided by the column cross-sectional area (cm2). The column should be equilibrated at a flow rate 25% to 50% higher than that anticipated for the separation to ensure that there is no further compaction of the bed during the run. If the manufacturer recommends a maximum flow rate, this should never be exceeded.

11. Insert the top adaptor, lower it onto the column bed, and tighten. Buffer will emerge from the tubing, which should remain filled to prevent air from being drawn into the column.

12. Place the tubing from the top adaptor into a reservoir of loading buffer clamped to the support stand at the same level as the top of the column. 13. Pump a further 5 bed volumes of loading buffer through the column. If the bed shrinks so that there is a gap between the top adaptor and the surface of the bed, carefully loosen the top adaptor, lower it onto the bed and retighten. The column is now ready and may be used at once or stored for several weeks at 4°C.

Prepare sample 14. If necessary, free sample of particulate matter by centrifuging or passing it through a 0.45-µm filter. 15. Adjust sample to the same pH and a similar salt concentration as the loading buffer. If only the pH of the sample is different from that of the loading buffer, this can be carefully adjusted by adding 10 mM sodium dihydrogen phosphate (to decrease pH) or 10 mM disodium hydrogen phosphate (to increase pH). If the salt concentration is too high, the sample should be dialyzed against an appropriate buffer. Buffer exchange may also be effected by use of a desalting column (UNIT 8.3) or a membrane concentration device (UNIT 4.4). The sample may also be diluted with low-ionic-strength buffer.

Chromatograph sample and regenerate column 16. If the column has been stored for some time, pump a little loading buffer through it and check the bed for air bubbles or cracks and the tubing connections for leaks. 17. Transfer the top adaptor tubing from the buffer reservoir to the sample and pump the sample onto the column, collecting fractions with a fraction collector and monitoring the A280 using a UV monitor with a 280-nm filter, connected to a chart recorder. Take care not to introduce air into the column.

18. Wash the column with loading buffer until the A280 returns to baseline, collecting the flowthrough and wash fraction, as these contain the material which does not bind to the matrix. 19. Degas appropriate volumes of 0.01 and 0.4 M sodium phosphate, pH 6.8, as in step 2a. Connect gradient mixer to column and place the two gradient solutions in the respective reservoirs. Elute the bound material using a linear gradient from 0.01 to 0.4 M sodium phosphate with a gradient volume 6 to 10 times the bed volume. Elute any material still bound by washing with 0.4 M sodium phosphate, pH 6.8. Collect 20 to 40 fractions of a size appropriate to the volume of the gradient until all bound material has been eluted (as judged by the A280 of the eluant). Hydroxylapatite Chromatography

20. Regenerate column by washing with 5 bed volumes of loading buffer, then wash with 2 bed volumes of loading buffer containing 0.02% sodium azide. Store column at 4°C.

8.6.4 Supplement 8


CAUTION: Sodium azide is toxic and potentially explosive. Avoid contact with metals and acids and do not lyophilize solutions containing azide. See APPENDIX 2A for additional guidelines on the handling and disposal of sodium azide. If the manufacturer recommends a different procedure for column regeneration, this should be followed. Figure 8.6.1 and Table 8.6.2 show results for low-pressure chromatography.

1.2

protein concentration

1.0

1.0

0.8

0.8

proteinase activity

0.6

0.6

0.4

0.4

0.2

0.2

A595 (HIV proteinase assay)

A599 (Bradford protein assay)

1.2

21 25 29 33 37 41 45 49 53 57 61 65 69 73 Fraction number

Figure 8.6.1 Chromatogram illustrating purification of HIV-2 proteinase on a hydroxylapatite column (A.V. Broadhurst and A.J. Ritchie, unpub. observ.). 139 g recombinant E. coli expressing HIV-2 proteinase were lysed by lysozyme digestion and sonication. The soluble fraction was treated with 1% streptomycin sulfate to precipitate nucleic acids. The supernatant was brought to 25% saturation with solid ammonium sulfate and the pellet was discarded. The ammonium sulfate was increased to 40% saturation and the pellet containing the proteinase was dialyzed against 10 mM sodium phosphate, pH 6.8, containing 0.0125% Tween 20. This fraction was applied to a 4.5 × 19–cm Bio-Rad HT hydroxylapatite column, bed volume 300 ml, at a linear flow rate of 10 cm/hr. The column was washed with 250 ml 10 mM sodium phosphate (pH 6.8)/0.0125% Tween followed by 250 ml 25 mM sodium phosphate (pH 6.8)/0.0125% Tween 20. Bound proteins were eluted with a 600 ml gradient of 25 to 250 mM sodium phosphate, (pH 6.8)/0.0125% Tween. The fractions were assayed for protein by the Bradford method (the dotted plot and the y axis on the left; Bradford, 1976; also see UNIT 3.4) and for proteinase activity using a colorimetric assay (the solid plot and the y axis on the right; Broadhurst et al., 1991). Table 8.6.2 Typical Purification Scheme for HIV-2 Proteinase Incorporating Hydroxylapatite Chromatographya

Purification stage Crude E.coli lysate 40% ammonium sulfate pellet Pooled peak fractions from hydroxylapatite column Pooled concentrated fractions from G-75 gel-filtration column

Total proteinase Specific activity activity (U) (U/mg) 79,120 54,590 31,995

13.9 48.2 194

18,333

561

U is arbitrarily defined as the amount of enzyme required to release 1 µmol product/min at 37° in a colorimetric assay (Broadhurst et al., 1991). a1



Supplement 8


SIMPLE COLUMN CHROMATOGRAPHY FOR PROTEIN PURIFICATION USING HYDROXYLAPATITE If only basic equipment is available, simplified methodology in which the pump is replaced by a gravity column may be applied successfully. The method described below is also useful for quick trial runs to assess the suitability of hydroxylapatite for particular applications. Additional Materials (also see Basic Protocol 1) Step gradient solutions: e.g., 0.025, 0.05, 0.10, 0.25, and 0.4 M sodium phosphate, pH 6.8 Column with integral bottom sinter and airtight lid: e.g., Kontes Flex-Column (Kontes Glass) or Bio-Rad Econo-Column Frit to place on top of resin bed (Kontes Glass) 1. Prepare matrix. (see Basic Protocol 1, steps 1 to 5a or 5b). 2. Pour matrix slurry into column and wash with 5 bed volumes of degassed loading buffer, allowing the column to drain by gravity. Do not allow column to run dry. 3. Place frit on top of bed to prevent disturbance of the surface. Avoid trapping air bubbles beneath it. 4. Prepare sample (see Basic Protocol 1, steps 14 and 15). Load sample on top of the frit and allow it run through the bed. When all the sample has been loaded, wash with 1.5 bed volumes of loading buffer. Combine the flowthrough and wash fractions, which will contain unbound material. 5. Wash with 8 bed volumes of loading buffer and discard the wash fractions. 6. Wash sequentially with 2.5 bed volumes each of step gradient solutions of increasing concentration—e.g., 0.025, 0.05, 0.10, 0.25, and 0.4 M sodium phosphate, pH 6.8 (degassed before use; see Basic Protocol 1, step 2a). Collect each wash separately.


BATCH PROCESSING FOR PROTEIN PURIFICATION USING HYDROXYLAPATITE If hydroxylapatite is to be used with very crude material early in the purification schedule, or if the volume to be processed is very large, this batch separation procedure may be appropriate. Batch processing does not require sophisticated equipment; it can be done quickly and is useful for trial runs. Materials (also see Basic Protocol 1) Step gradient solutions: e.g., 0.025, 0.05, 0.10, 0.25, and 0.4 M sodium phosphate, pH 6.8 Sintered-glass funnel Side-arm flask and vacuum source NOTE: Degassing of buffers is not essential for batch processing. 1. Prepare matrix (see Basic Protocol 1, steps 1 to 5a or 5b). 2. Carefully remove all supernatant buffer from the settled matrix by decantation. 3. Prepare sample (see Basic Protocol 1, steps 14 and 15).


If the sample volume is small, add sufficient 10 mM sodium phosphate, pH 6.8, to make it approximately twice the bed volume.

8.6.6 Supplement 8


4. Mix the sample with the hydroxylapatite matrix by gentle swirling. Leave for 15 min. 5. Pour slurry into sintered-glass funnel and apply a very gentle vacuum until the buffer just reaches the top of the matrix. IMPORTANT NOTE: Do not allow the matrix to dry out.

6. Wash matrix in funnel with a small amount of loading buffer. Collect and combine the flowthrough and wash fractions. 7. Wash well with loading buffer and discard wash. 8. Elute bound proteins by washing with a step gradient of increasing phosphate concentration (see Alternate Protocol 1, step 6). HPLC OF PROTEINS AND NUCLEIC ACIDS USING HYDROXYLAPATITE Hydroxylapatite columns can be used for the chromatography of both proteins and nucleic acids. HPLC is especially useful for analytical applications, as the technique offers high resolution, excellent reproducibility, and fast analysis times. For occasional users, readypacked columns are recommended. These are usually of good quality and are tested and certified before sale.

BASIC PROTOCOL 2

The protocol described here is a general one applicable to the separation of proteins or nucleic acids. Using a method of this type, it is possible to quantitate the amount of a particular protein or nucleic acid in a sample by comparison with a range of standard samples of known concentration. Materials 0.01 and 0.4 M sodium phosphate buffer, pH 6.8 (APPENDIX 2E; prepare using boiled H2O) Protein or nucleic acid standard solutions Protein or nucleic acid test samples 0.2 M sodium hydroxide 0.01 M sodium phosphate, pH 6.8 (APPENDIX 2E) with 0.02% (w/v) sodium azide 0.45-µm filters (for buffers) 0.22-µm filters (for samples) ∼7.5-mm × 7.5-cm HPLC column packed with suitable grade of hydroxylapatite (Table 8.6.1) Guard column recommended for HPLC column used HPLC apparatus equipped with pumps, gradient mixer, autosampler or injection loop, integrator, and UV monitor Prepare buffers and sample 1. Filter 0.01 and 0.4 M sodium phosphate buffers through a 0.45-µm filter. Degas in side-arm flasks under vacuum before use. 2. Filter sample through a 0.22-µm filter. If sample is very turbid it may be necessary to centrifuge it first to avoid blocking the filter.

Prepare column and chromatograph samples 3. Follow any instructions supplied by the manufacturer relating to preparation of column and assembly of HPLC system. 4. Equilibrate column with 0.01 M sodium phosphate buffer, pH 6.8, prepared as in step 1.



Supplement 8

4 2 3

A280

1

0

5

10

15

20

25

Elution time (min) Figure 8.6.2 Chromatogram illustrating purification of monoclonal antibody from mouse ascites fluid. A Bio-Rad Bio-Gel HPHT column with guard column was used and 0.5 ml mouse ascites fluid was applied. The column was eluted with a linear gradient of 0.01 to 0.30 M sodium phosphate, pH 6.8 at a flow rate of 1.0 ml/min. Peak 1 contains low-molecular-weight material, peak 2 contains albumin and serum protease, peak 3 contains other serum (ascites) proteins, and peak 4 contains essentially pure IgG. The linear gradient is traced via the conductance (dotted line). Adapted from Bio-Rad Technical Bulletin US/EG 1115.

5. Program the HPLC to inject 20 µl of sample and set the UV monitor to 260 nm for nucleic acids or 280 nm for proteins. Place 0.01 and 0.4 M sodium phosphate buffers, pH 6.8, in the appropriate reservoirs of the gradient mixer and run a 30-min linear gradient from 0.01 to 0.4 M sodium phosphate at a flow rate of 1 ml/min. 6. Run a standard through the column to to check its performance using the parameters in step 5. Collect 1-ml fractions. 7. Run the standard at a range of concentrations to generate a standard curve. 8. Run the test samples. 9. Use the integrated peak areas of the standards and test samples to calculate the concentration of the protein/nucleic acid of interest in the samples. See Figure 8.6.2 for representative results.

Regenerate and store column 10. At the end of each sample batch, inject 1 column volume of 0.2 M sodium hydroxide to remove any strongly bound material. Wash extensively with 0.4 M sodium phosphate. 11. Store column at room temperature in 0.01 M sodium phosphate, pH 6.8 containing 0.02% sodium azide. Hydroxylapatite Chromatography

CAUTION: Sodium azide is toxic and potentially explosive. Avoid contact with metals and acids and do not lyophilize solutions containing azide. See APPENDIX 2A for additional guidelines on the handling and disposal of sodium azide.

8.6.8 Supplement 8


COMMENTARY Background Information Calcium phosphate gels have been used for chromatography for many years. The material originally used was unsatisfactory with respect to flow rate, stability, and elution characteristics, and fell out of favor as newer matrices such as HA Ultrogel and Macro-Prep Ceramic became available. In the 1950s, Tiselius and co-workers described a process for preparing hydroxylapatite, a crystalline form of calcium phosphate, and demonstrated its usefulness for protein separation (Tiselius et al., 1956). Material prepared by this method demonstrated superior flow rates and has been commercially available for many years. Recently, some new forms of hydroxylapatite with spherical particles have been developed, and these have demonstrated even better flow rates as well as enough mechanical strength to be used in highpressure systems. These technological advances, combined with hydroxylapatite’s unique mode of separation, make this medium a useful addition to the chromatographer’s repertoire. Protein binding to hydroxylapatite is mediated through interactions between the amino and carboxyl groups of the protein and the Ca2+ and PO43– groups of the hydroxylapatite crystal lattice. Basic proteins are thought to bind via electrostatic interactions between their amino groups and the surface PO43– groups, the positive charge rather than the chemical nature of the group being the important factor (Gorbunoff, 1984b). Acidic and neutral proteins bind to the Ca2+ sites via their carboxyl groups; this interaction encompasses both electrostatic interactions and specific effects (Gorbunoff, 1984b). As a general principle, basic proteins with an isoelectric point (pI) of 8 can be eluted by increasing the concentration of the phosphate buffer, and they are also displaced by very low concentrations of Mg2+ or Ca2+ ions and moderate concentrations of NaCl (Gorbunoff, 1984a). Acidic or neutral proteins can be eluted by increasing the phosphate concentration, but not by adding Mg2+ or Ca2+ ions or NaCl. Increasing the pH of the elution buffer promotes desorption of proteins irrespective of pI (Gorbunoff, 1984a). There are many proteins whose interactions with hydroxylapatite are not as easily predicted on the basis of pI alone—e.g., β-lactoglobulin, ovomucoid, trypsinogen (Gorbunoff and Timasheff, 1984), phosphoproteins (Bernardi et al., 1972), and metalloproteins (Atkinson et al.,

1973). Tertiary structure is also important for binding, and proteins with floppy or unstructured regions such as the glycoprotein ovomucoid, or those that have been denatured, show reduced affinity for hydroxylapatite (Gorbunoff, 1984a). The distribution of surface charge may also play a part (Gorbunoff, 1984b). Nucleic acids bind to hydroxylapatite by interaction of the charged phosphate backbone with the Ca2+ ions at the crystal face. Doublestranded DNAs bind more tightly than singlestranded DNAs, possibly because their planar nature maximizes the interaction. (Martinson, 1973). Small double- and single-stranded DNAs appear to separate by size. tRNAs can also be separated, and secondary and tertiary structures may be important in this context (Kawasaki et al., 1986). Hydroxylapatite chromatography is performed using mild conditions at moderate pH and salt concentration, which favors the preservation of biological activity. The matrix has low nonspecific binding, and small molecules such as amino acids do not bind well. It is chemically stable at pHs between 6 and 11 and can be used with denaturing agents, detergents, and reducing agents such as dithiothreitol. EDTA and chelating buffers such as PIPES should be avoided as these bind the calcium. The unique mode of separation of hydroxylapatite, combined with the improved handling characteristics of the newer forms available, make it a useful adjunct to the more commonly used techniques such as ion-exchange and hydrophobic-interaction chromatography. Hydroxylapatite has proved useful in the separation of IgG idiotypes that cannot easily be separated by other techniques (Juarez-Salinas et al., 1986), as well as for the isolation of immunoglobulins from ascites fluid and tissue culture media (Smith et al., 1984). Another application may be the secondary purification step for recombinant proteins with N-terminal affinity tags such as maltose-binding protein (MBP), glutathione-S-transferase (GST), or 6His. When these proteins have been isolated by affinity chromatography, any breakdown products produced by C-terminal degradation and any premature termination fragments will also bind to the resin, as they are tagged. These contaminants can be difficult to remove by gel filtration or ion-exchange chromatography as they may differ very little in size or charge from the protein of interest; hence hydroxylapatite chromatography may assist in the separation.



Supplement 8

As an example, during the preparation of human cytomegalovirus proteinase for crystallographic studies, protein that had been purified using Mimetic Green and Q Sepharose and that appeared to be homogeneous on PAGE, separated into two peaks on a hydroxylapatite column (A.V. Broadhurst, unpub. observ.). The ability of hydroxylapatite to bind and separate nucleic acids can be advantageous during chromatography of proteins derived from recombinant E. coli. Crude extracts contain large quantities of DNA, which can bind to proteins or ion-exchange resins and contaminate the product. Hydroxylapatite can be used to remove nucleic acids during the purification.

Critical Parameters Selecting format Careful consideration must be given to choosing the most appropriate chromatographic format for the application. If hydroxylapatite is to be used with very crude material early in the purification schedule, or if the volume to be processed is very large, a batch separation (see Alternate Protocol 2) may be suitable. Batch processing does not require sophisticated equipment; it can be done quickly and is useful for trial runs. Low-pressure chromatography (see Basic Protocol 1 and Alternate Protocol 1) can be used to prepare milligram to gram quantities of protein. Semiautomated or fully automated systems are available, but satisfactory results can be obtained with quite basic equipment. This type of chromatography is used primarily for preparative rather than analytical purposes. HPLC (see Basic Protocol 2), although it can be scaled up for preparative use, is particularly suitable for analytical applications. Its main advantages are high resolution, reproducibility, and rapid throughput.


Selecting matrix Various grades and types of hydroxylapatite are commercially available (see Table 8.6.1). In general, hydroxylapatite with a small particle size will be suitable for analytical applications, giving excellent resolution but operating at low flow rates. Material with a larger particle size can be used at higher flow rates but with some loss of resolution. Hydroxylapatite can be difficult to work with as its crystalline structure is fragile and easily damaged, but some of the newer types (TSKgel HA-1000 from TosoHaas and MacroPrep ceramic hydroxylapatite from Bio-Rad) are purported to be more robust and

are suitable for use with-high pressure systems. There are several types of ready-prepared columns available commercially. These are designed for particular applications such as HPLC or low-pressure chromatography; some are of particular utility for the separation of proteins and others for separation of nucleic acids. Selecting buffers Sodium or potassium phosphate buffers at near-neutral pH are used for the majority of hydroxylapatite applications. Low phosphate concentration and low pH increase protein binding; high phosphate concentration and high pH promote desorption. In this context, low and high pH refer to the range 6 to 8, over which phosphate is an effective buffer. Potassium phosphate is more soluble than sodium phosphate at lower temperatures and is recommended for cold-room use. The protein-binding capacity of hydroxylapatite is maximal at pH 6.6 to 7 and decreases sharply above pH 7 (Atkinson et al., 1973). The matrix itself is chemically stable over a pH range of 6 to 11, but dissolves below pH 5.5. Buffers other than phosphate have been employed—e.g., MOPS with a phosphate gradient (de Gunzberg et al., 1984)—but should be used with caution as they may damage the matrix. 10 mM Tris⋅Cl can be used, but may cause some deterioration over time. Chelating buffers such as PIPES must be avoided. Buffer pH should always be within the range over which the sample protein is known to be stable. Various buffer additives can be employed. Glycerol, added to stabilize proteins, can be used at a concentration of 10% to 15%; it increases buffer viscosity so that a reduction in flow rate may be necessary. Reducing agents such as dithiothreitol and 2-mercaptoethanol, as well as nonionic detergents such as Triton X-100 and Tween 20, are compatible with hydroxylapatite. SDS has been used, but can reduce nonspecific binding and alter elution characteristics; it is also rather insoluble at 4°C and therefore is probably best avoided. Chelating agents such as EDTA are contraindicated as they bind to the calcium ions of hydroxylapatite and will rapidly destroy the column.

Troubleshooting Many of the problems encountered during hydroxylapatite chromatography are common to all types of column chromatography, but a few are unique owing to the crystalline mineral nature of the matrix.

8.6.10 Supplement 8


Column bed drying out or becoming filled with bubbles Drying out of the bed is usually caused by the system not being airtight. Ensure that all O rings and seals are present and in good condition. It may help to have the buffer reservoir slightly higher than the top of the column. Drying out can also occur if the top sinter or the top of the bed becomes clogged, causing the pump to create a vacuum in the bed. This might be alleviated by reducing the flow rate; if this does not work, remove the top sinter, wash it, and replace. If the problem persists it may be necessary to modify the sample preparation. Bubbles that appear in the gel are usually caused by inadequate degassing of the buffers. Dissolved gases will be released under reduced pressure or if the buffers/column are moved from a colder to a warmer environment. Degas all buffers (unless they contain volatile solvents) under vacuum before use. Bubbles may also appear if high concentrations of detergent are included in the buffers; this can be minimized by avoiding shaking or vigorous stirring of the buffers and using a moderate flow rate. Inadequate or decreasing flow rate This problem may be caused by packing the bed at too high a flow rate, causing it to become compressed. Manufacturers’ recommended flow rates should never be exceeded; this is particularly important with hydroxylapatite because of its inherently fragile nature. The presence of fines (resulting from inadequate removal before packing the bed) or degraded material can also cause poor flow rates. Ensure that the matrix is prepared as recommended; hydroxylapatite is easily damaged by rough treatment such as stirring with magnetic stirring bars and should be resuspended only by gentle swirling. If hydroxylapatite columns are stored for long periods the top of the bed may become hard and crusted, causing reduced flow rate. This results from the reaction of the calcium ions with carbonate from water used to prepare the buffers. Boiled, degassed water should be used. The hard, crusted material can be removed and replaced with a little fresh hydroxylapatite slurry. If the flow rate achievable is not high enough to allow the experiment to be performed in a reasonable time, consider whether a matrix with a larger particle size would be more suitable, or whether a batch process might be preferable.

If the flow rate is constant during the preliminary washing and equilibration of the column but decreases during sample loading, the problem may lie with the sample itself. It is important that particulate matter be removed from the sample before loading or it will block the sinter or the top of the gel bed. The sinter can be removed, washed, and replaced as an interim measure, but improved sample preparation is the cure. Crude extracts of recombinant E. coli, which are increasingly used as the starting point for protein purification, can be problematic. Even if the extract is clarified by centrifugation, it may be extremely viscous due to the high DNA content. It is often possible to reduce the viscosity enough to achieve a reasonable flow rate by diluting one part sample with three parts buffer before loading. If this fails, try precipitating the nucleic acids with 0.01% aqueous polyethyleneimine or 1% streptomycin sulfate, or treat the sample with DNase. UNIT 4.5 contains additional guidelines on clarification of samples and nucleic acid removal. Protein not bound to column If the protein of interest is only partially bound, with significant amounts appearing in the flowthrough, the binding capacity of the column may have been exceeded. Try increasing the ratio of matrix to total protein. If the protein is not bound using 10 mM sodium phosphate, pH 6.8, try reducing the molarity of the sodium phosphate. Concentrations as low as 1 mM can be used, although the buffering capacity will be small at this concentration. If the protein is stable at lower pH, the pH of the loading buffer can be decreased to 6 to facilitate binding. Do not reduce the pH below 5.8, as the hydroxylapatite matrix will dissolve at lower pH. Adding 0.005 mM CaCl2 to the buffer will increase binding of acidic proteins. CaCl2 should only be used with low molarities of phosphate (0.4 M are required for complete elution and the chromatography is to be done at 4C, use potassium phosphate buffer instead of sodium phosphate, as it is more soluble. Desorption of proteins can also be achieved by increasing the pH of the buffer. Basic proteins can be eluted with Ca2+ at concentrations of ∼0.003 M. This cannot be done in phosphate buffer as calcium phosphate will precipitate out. Wash the column with 1 mM NaCl, then with 0.003 M CaCl2 in 1 mM NaCl.

Anticipated Results High recoveries can be expected with hydroxylapatite; near-quantitative yields have been reported with 90% to 95% being common. Retention of activity is usually excellent because of the mild conditions employed (Kato et al., 1987). Depending on the nature of the sample to be chromatographed, >20-fold purification is achievable.

Time Considerations Small columns can be prepared and a chromatographic run carried out in a working day. Larger columns take longer to run and therefore will need to be prepared in advance.

Literature Cited

protein utilising the principle of protein-dye binding. Anal. Biochem. 72:248-254. Broadhurst, A.V., Roberts, N.A., Ritchie, A.J., Handa, B.K., and Kay, C. 1991. Assay of HIV-1 proteinase: A colorimetric method using small peptide substrates. Anal. Biochem. 193:280-286. de Gunzberg, J., Part, D., Guiso, N., and Veron, G. 1984. An unusual adenosine 3′, 5′-phosphate dependent protein kinase from Dictyostelium discoideum. Biochemistry 23:3805-3812. Gorbunoff, M.J. 1984a. The interaction of proteins with hydroxyapatite. 1. Role of protein charge and structure. Anal. Biochem. 136:425-432. Gorbunoff, M.J. 1984b. The interaction of proteins with hydroxyapatite. 2. Role of acidic and basic groups. Anal. Biochem. 136:433-439. Gorbunoff, M.J. and Timasheff, S.N. 1984. The interaction of proteins with hydroxyapatite. 3. Mechanism. Anal. Biochem. 136:440-445. Juarez-Salinas, H., Ott, G.S., Chen, J.-C., Brooks, T.L., and Stanker, L.H. 1986. Separation of IgG idiotypes by high performance hydroxylapatite chromatography. Methods Enzymol. 131:615622. Kato, Y., Nakamura, K., and Hashimoto, T. 1987. High performance hydroxyapatite chromatography of proteins. J. Chromatogr. 398:340-346. Kawasaki, T., Ikeda, K., Takahashi, S., and Kuboki, Y. 1986. Further study of hydroxylapatite highperformance liquid chromatography using both proteins and nucleic acids, and a new technique to improve chromatographic efficiency. Eur. J. Biochem. 155:249-257. Martinson, H.G. 1973. The basis of fractionation of single-stranded nucleic acids on hydroxylapatite. Biochemistry 12:2731-2736. Smith, G.J., McFarland, R.D., Reisner, H.M., and Hudson, G.S. 1984. Lymphoblastoid cell–produced immunoglobulins: Preparative purification from culture medium by hydroxylapatite chromatography. Anal. Biochem. 141:432-436. Tiselius, A., Hjerten, S., and Levin, O. 1956. Protein chromatography on calcium phosphate columns. Arch. Biochem. Biophys. 65:132-155.

Atkinson, A., Bradford, P.A., and Selmes, I.P. 1973. The large scale preparation of chromatographic grade hydroxylapatite and its application in protein separation procedures. J. Appl. Chem. Biotechnol. 23:517-529.

Key References

Bernardi, G., Giro, M.-G., and Gaillard, C. 1972. Chromatography of polypeptides and proteins on hydroxyapatite columns: Some new developments. Biochim. Biophys. Acta 278:409-420.

A series of three consecutive papers describing the basis of interaction of proteins with hydroxylapatite. Practical aspects are described, and there is extensive discussion of theoretical considerations.

Bio-Rad Technical Bulletin 1115 US/EG. Bio-Gel HPHT for protein and nucleic acid HPLC: New high performance hydroxyapatite column. BioRad Laboratories, Hercules, Calif. Bradford, M.M. 1976. A rapid and sensitive method for the quantitation of microgram quantities of

Gorbunoff, 1984a. See above. Gorbunoff, 1984b. See above. Gorbunoff and Timasheff, 1984. See above.

Anne V. Broadhurst Roche Products Ltd. Welwyn Garden City, Hertfordshire United Kingdom


8.6.12 Supplement 8


HPLC of Peptides and Proteins High-performance liquid chromatography (HPLC) is an essential tool for the purification and characterization of biomacromolecules. The choice of the chromatographic method and the type of high-performance equipment are determined by the molecular nature of the investigated molecules and the aim of the research. There are eight basic modes of HPLC currently in use for peptide and protein analysis and purification, namely size-exclusion chromatography (HP-SEC), ion exchange chromatography (HP-IEX), normal phase chromatography (HP-NPC), hydrophobic interaction chromatography (HP-HIC), reversed-phase chromatography (RP-HPLC), hydrophilic interaction chromatography (HP-HILIC), immobilized metal ion affinity chromatography (HPIMAC), and biospecific/biomimetic affinity chromatography (HP-BAC), and a number of subsets of these chromatographic modes, e.g., mixed mode chromatography (HP-MMC), charge transfer chromatography (HP-CTC), or ligand-exchange chromatography (HP-LEC). In terms of usage, versatility, and flexibility, RP-HPLC techniques dominate the application world with peptides and proteins at the analytical- and laboratory-scale preparative levels. All of these various chromatographic modes can be operated under isocratic or gradient elution conditions, and in analytical or preparative situations. They all have common start-up procedures, which are outlined in the initial sections of this unit, and specific standard conditions that are detailed for the major modes and which represent starting points for further method development. As an example for appropriate method development, the RP-HPLC mode is described in greater detail and discussed according to four possible intended purposes, i.e.: (1) the purification of one component out of a natural or synthesized sample; (2) the simultaneous purification of several components; (3) the desalting of proteins or polypeptides obtained from other purification procedures; and (4) the characterization of the physicochemical properties of peptides or proteins. They all require a good understanding of the underlying common principles of polypeptide-ligand interaction. The basics of these principles are touched upon with references to further reading. Finally, a short section of this unit is dedicated to troubleshooting; however, many of the “check-back” confirmatory procedures im-

plicit to sound operational practices and the identification of suitable alternatives for the separation strategy are included in the section on method development.

THE PROPERTIES OF PEPTIDES AND PROTEINS AND THEIR IMPLICATIONS FOR HPLC METHOD DEVELOPMENT Peptides and proteins are a class of molecules containing amino acids as the fundamental units. The chemical organization (i.e., the primary structure or amino acid sequence) and the folded structure (i.e., the secondary, tertiary and quaternary structure) are the essential features of a polypeptide or protein, around which a chromatographic separation can be designed. Two sets of factors must be considered. The first relates to the structural properties of the amino acid entities themselves; the second relates to the chemical and physical attributes of the separation system per se.

Biophysical Properties of Peptides and Proteins The 20 naturally occurring L-α-amino acids found in peptides and proteins vary dramatically in terms of the properties of the side chain or R-groups. Table 8.7.1 lists some of the fundamental properties of the common L-α-amino acids found in peptides and proteins. This chemical diversity becomes even greater in circumstances where some of these side chains have been post-translationally modified with carbohydrates or lipid moieties. The sidechains are generally classified according to their polarity (e.g., non-polar or hydrophobic versus polar or hydrophilic). The polar side chains are divided into three groups: uncharged polar, positively charged or basic, and negatively charged or acidic side chains. Peptides and proteins generally contain several ionizable basic and acidic functionalities. They therefore typically exhibit characteristic isoelectric points with the overall net charge and polarity in aqueous solution varying with pH, solvent composition and temperature. Cyclic peptides without ionizable side chains will have zero net charge, and they represent an exceptional subgroup. The number and distribution of charged groups will influence the polarizability and ionization status of a peptide or protein, as well

Contributed by Reinhard I. Boysen and Milton T.W. Hearn Current Protocols in Protein Science (2001) 8.7.1-8.7.40 Copyright © 2001 by John Wiley & Sons, Inc.

UNIT 8.7


8.7.1 Supplement 23

Table 8.7.1

Properties of the Common L-α-Amino Acid Residues

3-Letter code

1-Letter code

Ala Arg Asn Asp Cys Gln Glu Gly His Ile Leu Lys Met Phe Pro Ser Thr Trp Tyr Val α-aminoe α-carboxylf

A R N D C Q E G H I L K M F P S T W Y V — —

Mass (Da) 71.08 156.19 114.10 115.09 103.14 128.13 129.12 57.05 137.14 113.16 113.16 128.17 131.20 147.18 97.12 87.08 101.11 186.21 163.18 99.13 — —

Accessible pKa of Partial spec. Rel. surface area a c hydrophobicityd 3 vol. (Å ) side-chain (Å2)b 88.6 173.4 117.7 111.1 108.5 143.9 138.4 60.1 153.2 166.7 166.7 168.6 162.9 189.9 122.7 89 116.1 227.8 193.6 140 — —

115 225 160 150 135 180 190 75 195 175 170 200 185 210 145 115 140 255 230 155 — —

— 12.48 — 3.9 8.37 — 4.07 — 6.04 — — 10.54 — — — — — — 10.46 — 7.7-9.2g 2.75-3.2g

0.06 −0.85 0.25 −0.20 0.49 0.31 −0.10 0.21 −2.24 3.48 3.50 −1.62 0.21 4.8 0.71 −0.62 0.65 2.29 1.89 1.59 — —

aFrom Zamyatnin, 1972. bFrom Chothia, 1974. cFrom Dawson et al., 1986. dFrom Wilce et al., 1995. eα-amino group present on all primary amino acids. fα-carboxyl group present on all primary amino acids. gFrom Rickard et al., 1991.

as the microscopic and global hydrophobicity. These important factors ultimately determine the selection of the optimal separation conditions for the resolution of peptide and protein mixtures. Table 8.7.1 can be used to evaluate the impact of amino acid composition on retention behavior. For example, this information can be used to direct the choice of eluent composition or the gradient range in RP-HPLC; to assess the impact on retention of amino acid substitution or deletion with small peptides; or alternatively to guide the identification of peptide fragments derived from tryptic digestion of proteins for further sequence analysis.

HPLC of Peptides and Proteins

Parameters of the Mobile Phase/Stationary Phase These parameters directly impact on the molecular properties of the polypeptide or protein during liquid chromatographic separations, and are listed in Table 8.7.2. In solution, a polypeptide or protein can, in principle, explore a relatively large array of conformational space. For small peptides (up to ∼15 amino acid residues) a defined secondary structure (α-helical, β-sheet or β-turn motif) is generally absent. With increasing polypeptide chain length, depending on the nature of the amino acid sequence, specific regions/domains of a polypeptide or protein can adopt preferred secondary, tertiary or quaternary structures. In aqueous solutions this folding, which internalizes the hydrophobic residues and thus stabilizes the

8.7.2 Supplement 23


Table 8.7.2 Chemical and Physical Factors of the Chromatographic System that Contribute to the Variation in the Resolution and Recovery of Polypeptides, Proteins and Other Biomacromolecules in HPLC Systems (Hearn, 2000a)

Mobile phase contributions

Stationary phase contributions

Organic solvents pH Metal ions Chaotropic reagents Oxidizing or reducing reagents Temperature Buffer composition Ionic strength Loading concentration and volume

Ligand composition Ligand density Surface heterogeneity Surface area Pore diameter Pore diameter distribution Particle size Particle size distribution Particle compressibility

Table 8.7.3 Relevant Absorption Bands and Extinction Coefficients in Proteins (Campbell and Dwek, 1984)

Group Peptide bond His Cys Trp Tyr Phe

Wavelength (nm)

Log εmax

190-210 211 250 280 274 257

2.0-3.8 3.8 2.5 3.7 3.1 2.3

polypeptide structure, becomes a significant feature of peptides and proteins for chromatographic separations. A critical factor in the selection of an HPLC procedure is that the choice of experimental conditions will inevitably cause perturbations of the conformational status of these biomacromolecules. Although polypeptide and protein conformational stability can be manipulated in a number of ways (e.g., mobile and stationary phase composition, temperature) in HPLC, in most cases an integrated biophysical experimental strategy—including 1H 2-dimensional NMR (UNIT 17.5), Fourier transformed infrared (FTIR), ESI-MS (UNIT 16.1), or circular dichroism–optical rotatory dispersion (CD-ORD) spectroscopy—is required in order to determine the secondary and higher-order structure of a polypeptide or protein in solution or in the presence of specific ligands. Availability of such instrumentation is not mandatory, but the quality of the interpretation of the experimental results will become more substantial when additional results are

independently obtained with such spectrometric procedures to confirm the participation of conformational or self-self aggregation effects with peptides or proteins under HPLC conditions.

DETECTION OF PEPTIDES AND PROTEINS IN HPLC The peptide bond absorbs strongly in the far-ultraviolet (UV) region of the spectrum (∼λ = 205 to 215 nm). Hence UV detection is the most widely used method for detection of peptides and proteins in HPLC (Table 8.7.3). Besides absorbing in the far-UV range, the aromatic amino acid residues (and to some extent cysteine) also absorb light above 250 nm. Knowledge of the UV spectra, in particular the extinction coefficients of the non-overlapping absorption minima of these amino acids, allows, in conjunction with UV-diode array detection (DAD) and second derivative or difference UV-spectroscopy, verification of peak purity and determination of the aromatic amino



Supplement 23

Table 8.7.4 UV Cutoff Values of Different Organic Solvents in RP-HPLC

Eluent Methanol Acetonitrile Isopropanol

UV cutoff (nm)a 205 188 205

aWavelength at which the absorbance of a 1-cm-

long cell filled with the solvent was 1.0, measured against water as reference.

acid content of peptides and proteins. Moreover, the knowledge of the relative UV/VIS absorbancy of a peptide or protein is therefore crucial, since the choice of detection wavelength of peptides and proteins in RP-HPLC (and in the other HPLC modes) depends on the different UV cutoffs of the eluents used (Table 8.7.4). The common use of λ = 215 nm as the preferred detection wavelength for most analytical reversed-phase applications (and for those of other HPLC modes) with peptides and proteins is a good compromise between detection sensitivity and potential detection interference due to buffer absorption. However, wavelengths between 230 and 280 nm are frequently employed in preparative applications, where the use of more sensitive detection wavelengths could result in overloading of the detector response (usually above an absorbance value of 2.0 to 2.5 AU).

START-UP PROCEDURES Correct selection of these first important steps may take more time than the ultimate experimental procedure if a high-performance separation of high resolution, robustness, and reproducibility is to be achieved. They require good planning and thorough work. The following details are representative of the types of equipment, materials, chemicals, and experimental protocols that can be routinely required for isocratic or gradient elution HPLC. Sample Peptide or protein sample (kept at 4°C if not used)


Apparatus Pump module Mixing chamber Spectrophotometer with analytical or preparative flow cell Injection valve

Analytical (10 to 100 µl) or preparative (500 to 1000 µl) sample loop Column oven or thermostated column coolant– jacket coupled to recirculating cooler Autosampler (optional) Computer, printer, and software, e.g., Beckman, System Gold; Hewlett-Packard HP1090A liquid chromatograph, or Waters 600/486 HPLC system with attendant data management systems and system automation controllers Chemicals Acetonitrile (HPLC grade) Methanol (HPLC grade) Acetone (HPLC grade) Thiourea or sodium nitrate Milli-Q water Glassware Two 1-liter eluent bottles Two 1-liter measuring cylinders 10-ml measuring cylinder Waste bottle All glassware coming into contact with sample before and during analysis should be rinsed three times with Milli-Q water Mobile phase filtration facility Vacuum pump 1-liter reservoir Support base with glass frit and integral vacuum connection Funnel Clamp 47-mm membrane filter (0.2 µm PTFE) Gases Helium Nitrogen (for autosampler) Columns See respective sections below

8.7.4 Supplement 23


HPLC peptide standards See respective sections below Tools Screwdrivers: 1⁄ -in. and 1⁄ -in. flat-head screwdrivers 8 4 Phillips screwdriver no. 2 Wrenches: 12-in. adjustable wrench (for compressed gas tank regulator) Three open-end wrenches (two 1⁄4-in. × 1⁄16-in.; one 1⁄2-in. × 1⁄16-in.) for fittings columns and valves Two long-jaw needle-nose pliers Two bastard files Two hex key (Allen) wrench sets (metric and nonmetric) Tweezers Pump seal insertion tool (if required) Inner reamer Teflon tape Flashlight Magnetic pick-up tool Logbook All steps must be documented to facilitate troubleshooting and reproduction Spare parts Zero dead volume union Ferrules (steel, rheodyne) 1⁄16-in. short and long (for rheodyne valves) Bushing nuts One-piece PEEK fitting Tubing (steel, PEEK) Column frits In-line filter Inlet filter Fuses Miscellaneous Graduated 25-µl Hamilton glass syringe 1-ml syringe with truncated needle Conical vials for autosampler Laboratory coat Gloves Safety glasses Stopwatch

PREPARATION OF THE SAMPLE The following considerations are relevant to the preparation of samples containing peptides and proteins. 1. Dissolve the sample with half the target volume of eluent A (weak mobile phase). If the sample is not soluble, a small amount of eluent B (strong mobile phase) may be added (typi-

cally 100 µl. The same consideration is valid if the sample loop is exchanged from an analytical separation (e.g., a sample loop of 50 µl) to a semipreparative separation (e.g., a sample loop of ≥500 µl) on the same column. 4. Determine the gradient delay and present the results graphically in a format similar to that shown as Figure 8.7.1.

8.7.6 Supplement 23


D V = [6.1 min – (10/2)] × 2 ml/min = 2.2 ml

0.3

Absorbance (254 nm)

90%

0.2

50% 0.1 6.1 min

10%

0.0 0

5

10

15

Time (min) Figure 8.7.1 Graphical illustration of the approach employed to determine the gradient delay volume, Dv. In this figure the gradient profile is illustrated for a defined flow rate (2 ml/min), with the gradient profile recorded from 10% to 90% buffer B at a specified wavelength such as 254 nm, as described in the text with acetonitrile-water-acetone mixtures.

The dwell volume, VD, is the volume of eluent from the pump heads to the column inlet (including the mixing chamber volume). The dwell volume values range from 2 to 7 ml; autosamplers in particular make a large contribution to the delay volume. It should be determined with an accuracy of ±0.5 ml. The profile can be used for diagnostic purposes, since the volume accuracy of the pump delivery is also monitored. Knowledge of the gradient delay is essential for method development, since it allows the accurate calculation of the S and k0 values (Ghrist and Snyder, 1988a; Hearn, 1991a). Its determination is particularly important when establishing segmented gradients (since various errors can accumulate here), and when an established HPLC method is transferred from one instrument to another instrument.

CONNECTING THE COLUMN The following considerations are relevant to the preparation of the HPLC column for use with samples containing peptides and proteins. 1. Flush the column with eluent B with the inlet connected to the injector and the outlet facing the waste collector (5 min at 1 ml/min for analytical columns). This procedure re-

moves air that may have been trapped and replaces the storage buffer. 2. Connect the column outlet to the detector. Start the flow with 0.5 ml/min and slowly increase the flow rate to 1 ml/min. 3. Equilibrate the column first with eluent B until a stable baseline is reached or alternatively with 10 column volumes (∼15 min for analytical column at 1 ml/min) and then with eluent A again with 10 column volumes. The pressure should be monitored and documented for each eluent since it can be used for diagnostic purposes.

PROGRAMMING THE HPLC INSTRUMENT The following considerations are relevant to the programming of the HPLC instrument for use with samples containing peptides and proteins. 1. Program, according to the manufacturer’s handbook, the pump, the detector, the integration module, and the autosampler. Test the method with a test run before leaving the instrument alone. 2. Program a shutdown method for overnight runs, which will switch off the lamp and pump. This approach prolongs lamp life and saves eluents.



Supplement 23

INJECTING THE SAMPLE The following considerations are relevant to the injection of the samples containing peptides and proteins onto the HPLC column. 1. Switch the sample loop to the load position and rinse with eluent A. This procedure removes eluent B (which may be present from rinsing the loop or previous runs). Failure to do so can cause pre-elution of the peptide or protein sample, particularly in conjunction with partially filled sample loops. 2. Load the sample slowly into the loop avoiding air bubbles. Do not squirt the sample into the loop too fast as it will end up in the waste. 3. Inject the sample by switching the valve swiftly into the inject position. If the switching is done too slowly, the pumps might shut down because the pressure limit is exceeded, as the valve is blocked in the intermediate switching position.

TESTING THE FUNCTIONAL HPLC SYSTEM The following considerations are relevant to the testing of the HPLC system for use with samples containing peptides and proteins. 1. Produce a blank run (inject eluent A) and run a gradient from 100% eluent A to 100% eluent B under the same conditions as intended for the peptide or protein sample. Repeat if peaks occur. This procedure cleans the column of peptides and proteins from previous separations, which have not been removed by the flushing process. 2. Measure the dead volume of the column with thiourea or sodium nitrate (or any other noninteractive solute). 3. Test the column performance with a gradient run and an appropriate test mixture (see below for details). First, this test allows the evaluation of the column bed integrity (low integrity will be associated with split, fronting or tailing peaks) and column performance (in terms of plate numbers). Second, this test allows, if repeated at regular intervals, the monitoring of the performance during the lifetime of a column, and the assessment of batch-tobatch differences of column fillings.

LOGBOOKS


Record keeping is essential for liquid chromatography system maintenance and confirmation/substantiation of the experimental results. Three types of logbooks are recommended, the system logbook, the column

logbook, and the experiment/assay logbook (Dolan and Snyder, 1989).

System Logbook This logbook should contain information on: 1. The module identification (brand, model, serial number, purchase data, warranty information) for the entire liquid chromatography (LC) system: injector, autosampler, pump(s), detector, software, column. 2. Module replacements, maintenance records and upgrades. 3. Reference chromatograms and operating parameters. 4. Maintenance: activity, time, operator, and name/reliability of service engineer. 5. Column replacement (cross-reference to column logbook).

Column Logbook This logbook should contain information including a summary of use, column life in months and number and type of samples, cause of failure, and suggestions for extending life. For each column this would be tabulated as follows. 1. Date column first used. 2. Specification: mobile phase flow rate, plate number, peak shape, dead volume over the lifetime of the column. 3. Performance of new column (validated with a test mixture). 4. Record of use (instrument, operator, number and type of samples). 5. Storage information. 6. Maintenance performed (type of backflush protocols, frit replacement, etc.). 7. Revaluation of column performance. 8. Cause of failure.

Experiment/Assay Logbook This logbook should contain information on: 1. Equipment configuration. 2. Operating condition(s). 3. Mobile phase recipe(s) (literature references if available). 4. Sample pre-treatment method(s) (literature references if available). 5. Assay procedure(s) (literature references if available). 6. Data analysis procedure(s). In the following sections, illustrative examples of standard operating protocols for the major modes of HPLC are described based on

8.7.8 Supplement 23


Sample loop size: 20 to 200 µl Isocratic elution Eluent A: 50 mM KH2PO4, pH 6.5, 0.1 M KCl Flow rate: 0.5 ml/min Detection: 214 nm Temperature: room temperature Peptide standards for column testing as described, e.g., in Mant and Hodges (1991a) Method development in the HP-SEC of peptides and proteins can be performed via the following steps: 1. Select sorbent of the most appropriate average pore size, packed into a column of suitable length. 2. Check for “ideal” and “non-ideal” retention effects. 3. Optimize plate number (adjust the flow rate or change to a column of different length). An example of HP-SEC for the separation of peptides and proteins is illustrated in Figure 8.7.2. In this example, the resolution of the 50S ribosomal proteins from Thermus aquaticus was achieved on a tandem TSK-250 column (75-mm length × 7.5-mm i.d. and 300-mm length × 7.5-mm i.d.) at a flow rate of 0.5 ml/min with a 50 mM (NH4)2SO4/20 mM NaH2PO4 buffer system, pH 5.0.

the instrumental system validation procedures described above.

STANDARD OPERATING CONDITIONS FOR HP-SEC The separation of peptides and proteins by high-performance size-exclusion chromatography (HP-SEC) is based on the concept that molecules of different sizes (hydrodynamic volume, Stoke’s radius) permeate to different extents into porous SEC separation media and thus exhibit different permeation coefficients according to differences in their molecular weights (Regnier, 1983). However, many SEC materials are slightly hydrophobic or can weakly act as ion exchangers. These properties lead to nonideal behavior (specifically electrostatic or hydrophobic interactions between the peptide or protein and the matrix). This feature is not necessarily a disadvantage, since mixedmode selectivities can be achieved, but can be suppressed by the addition of a salt at a reasonably high ionic strength, i.e., ≥100 mM, to the mobile phase (Mant et al., 1987).

Chromatographic Conditions

Column: e.g., TSK-250 (10 µm, 300 Å, 300mm length × 7.5-mm i.d.) Sample size: 25 mm, it may be necessary to remove air bubbles trapped under the column’s bottom net; this is done by drawing air out of the net using tubing connected to a water aspirator.

8. Make sure that the resin/start buffer slurry is uniform and, without letting the resin settle, quickly pour the slurry into the column. Use additional start buffer to rinse the sides of the container from which the slurry was poured, and pour this rinse liquid into the column as well. Use a wash bottle filled with start buffer to dislodge any resin stuck to the sides of the column. 9. After the resin has been poured into the column, add start buffer until the column is full to capacity. At this point, a tray or a tote can be placed under the column to catch any buffer that might spill out in subsequent steps.



Supplement 40

Figure 8.8.3 Schematic diagram showing the tubing connections, valve placement, pump configuration, and monitoring options for expanded-bed adsorption chromatography. Valves that control column operation are labeled V1, V2, V3, and V4. This diagram shows a hydraulic column; adaptor movement is controlled by pump 2. In a manual column, pump 2 and V3 are not present, and the column lid does not have a connector to the hydraulic line, but all other connections are the same. Buffer 1 and buffer 2 are typically equilibration buffer and elution buffer, respectively. P1 and P2 are fraction collection lines. CIP, cleaning-in-place solution (see Support Protocol 2); UV, flowthrough UV detector; cond., flowthrough conductivity meter; pH, flow-through pH meter. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources).

10. Tip the adaptor such that its bottom makes at least a 45◦ angle with respect to the surface of the buffer in the column, and then insert it into the column under the buffer, taking care not to introduce air bubbles. As soon as the entire adaptor O-ring is just inside the column, straighten the adaptor so that its bottom is parallel to the surface of the buffer, and then put the lid on the column and secure it with the screws provided by the manufacturer. IMPORTANT NOTE: No resistance should be felt when inserting the adaptor into the column. If resistance is felt, remove the adaptor from the column, make sure that the O-ring is wet and is seated properly in its groove, and then try again. Several attempts may be required to achieve proper insertion of the adaptor, as it is critical that no air bubbles be trapped in the adaptor net in the assembled column.

11a. If using a manual column: Position the adaptor so that its distance from the bottom of the column is ∼4× the height of the sedimented bed.

Expanded Bed Adsorption Chromatography

11b. If using a hydraulic column: Prime the hydraulic line with water. Open the flow path from the hydraulic pump to the column (through valve 3; Fig. 8.8.3), and set valve 2 open to waste. Use the hydraulic pump to pump water into the hydraulic chamber, and then stop the pump once the adaptor has been displaced downward by a few centimeters. Remove the column lid and aspirate the buffer/water mixture



from the hydraulic chamber, and then refill the chamber to the top with water. Once the chamber has been refilled, put the lid back on the column and secure it as in step 10. The position of the adaptor is set to allow for expansion of the resin.

Fill the system 12. Set the bottom valve (valve 4) to the bypass (waste) position and run pump 1 to fill all tubing with the appropriate start buffer. 13. Put the column on-line by setting the bottom valve (valve 4) to allow direct flow into the column.

Equilibrate the column 14. Check the orientation of the column using a spirit level, and adjust the positioning of the column as necessary to ensure that it is mounted completely vertically. For optimal results, it is important that the column be completely vertical. If a tray was placed under the column to catch liquid spillage during setup, it can now be removed.

15. Mark the height of the sedimented bed on the column using label tape. This sedimented bed height should be measured and recorded so that the degree of bed expansion upon equilibration can be calculated. The author recommends marking the column with narrow label tape, although any laboratory tape is suitable for this purpose.

16. Set each valve in the appropriate position for column equilibration (Fig. 8.8.4). Using pump 1, begin pumping equilibration buffer upward through the column at a flow velocity of 200 to 500 cm/hr. Record the actual flow velocity. 17. Allow the bed to stabilize at the flow velocity being used. Assess bed stability either visually (step 19) or using a residence time distribution test (steps 20 to 23), and once the bed has stabilized, use a second piece of label tape to mark the height of the expanded bed on the column. Continue pumping buffer upward through the column. The bed is stable when no further expansion is evident. Bed stabilization should take ∼45 min.

18. As the equilibration buffer flows through the column, monitor the pH of the outlet stream. As soon as the pH of this stream is the same as that of the equilibration buffer, begin recirculating the equilibration buffer by removing the column outlet tube from the waste container and putting it into the flask containing the equilibration buffer.

Evaluate bed stability by visual inspection 19. If using a column made of glass (e.g., a STREAMLINE 25 column), visually inspect the bed to assess bed stability. If the bed is stable at a certain bed height, it is safe to assume that the entire chromatographic bed is stable. If constant liquid densities, viscosities, and temperatures are maintained, bed expansion should be predictable from run to run. A significant decrease in bed expansion after several runs could be indicative of air trapped under the bottom distributor of the column, blockage in the bottom of the column, fouled resin, or the column’s not being absolutely vertical. See the troubleshooting guide in Expanded Bed Adsorption: Principles and Methods (Amersham Biosciences; see Internet Resources) for more information on these topics. Bed stability can be evaluated more rigorously by performing a residence time distribution test (RTD; steps 20 to 23), which assesses the degree of mixing in a column by determining the number of theoretical plates.



Supplement 40

Figure 8.8.4 Schematic diagram of the appropriate valve positions for equilibrating and expanding the resin bed in an expanded-bed adsorption chromatography column. Valves that control column operation are labeled V1, V2, V3, and V4. Note that the adaptor is at an elevated position within the column. This is necessary only as long as the degree of expansion of the resin is not well characterized. Once bed expansion is well characterized, the adaptor should be kept as close as possible to the top of the expanded bed to minimize mixing. CIP, cleaning-in-place solution (see Support Protocol 2); UV, flowthrough UV detector; cond., flowthrough conductivity meter; pH, flow-through pH meter; P1 and P2, fraction collection lines. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources).

Evaluate bed stability using a residence time distribution (RTD) test (optional) NOTE: The RTD test is required for evaluating bed stability in steel columns, as the contents of a closed steel column cannot be assessed visually. 20. Lower the adaptor to within 1 cm of the top of the equilibrated expanded bed. 21. Set the UV detector to 280 nm, and start the recorder used to plot UV detector readings. When the baseline is stable, change the feed from equilibration buffer to 0.25% (v/v) acetone in equilibration buffer. Acetone absorbs strongly at 280 nm.

22. Look for a positive step signal to arise on the recorder plot (Fig. 8.8.5). Once this UV step signal is stable (at its maximum level, referred to as the 100% signal level), switch the feed back to equilibration buffer alone. On the recorder plot, label the point at which this switch was made as time 0.


23. Look for a negative step signal to appear on the recorder plot, and then wait for the UV signal to stabilize at its baseline level (0%). Calculate the number of theoretical plates (N) using the following formula: N = t2 /σ 2 , where t is the mean residence time, calculated as the time (from time 0) needed for the UV signal to fall to the 50% level, and σ is the standard deviation, taken to be half of the time needed for the UV signal to drop from the 84.15% level to the 15.85% level (Fig. 8.8.5).



Figure 8.8.5 Chart paper output for a residence time distribution test using acetone as the tracer. A mark is placed on the paper to indicate time 0, the point at which a stable maximum signal is present and the feed is switched from tracer-containing equilibration buffer to tracer-free equilibration buffer. The quantity 2σ is calculated as the time (from time 0) needed for the UV signal (measured at 280 nm) to drop from the 84.15% level to the 15.85% level on the washout (negative-step) side of the tracer pulse. t, mean residence time; σ , standard deviation about the mean residence time. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources).

Note that 2σ , and not σ itself, is the quantity that the investigator measures directly from the recorder plot. Thus, it is necessary to divide this measurement by 2 to obtain the quantity that is entered into the formula for calculating N. Given a sedimented bed height of 15 cm and a flow velocity of 300 cm/hr, a value of N between 25 and 30 is desirable. A difference of ≥20% in N values from two runs performed under identical test conditions is indicative of air trapped under the bottom distributor of the column, blockage in the bottom of the column, fouled resin, or the column’s not being entirely vertical.

Apply sample to column 24. Raise the adaptor to a position very near the top of the column. Due to the higher viscosity of the sample mixture, bed expansion is greater when the sample, rather than buffer alone, is applied to the column. Thus, the adaptor is raised in this step (before application of the sample) to accommodate any additional bed expansion that might occur.

25. Set each valve in the appropriate position for applying the feedstock to the column (Fig. 8.8.6). Using pump 1, begin pumping the feedstock upward through the column at the same flow velocity used in step 16, continuously stirring the feedstock mixture that is fed into the pump to ensure uniformity throughout the course of sample application. Once any additional bed expansion resulting from feedstock application is complete, set the adapter ∼5 cm above the top of the expanded bed. 26. Continue pumping feedstock through the column, monitoring its application using the UV detector. Look for a large flowthrough peak, and direct this peak to waste. If a hydraulic column is being used, valve 3 should also be set open to waste during this step to allow for slight adaptor movement if pressure builds up in the bed.

Wash column 27. Once all of the feedstock has been applied to the column, change the flow to wash buffer. Using pump 1, pump the wash buffer upward through the column at the same flow velocity used for sample application. Look for some amount of bed contraction as the sample mixture is replaced with wash buffer. When this contraction is



Supplement 40

Figure 8.8.6 Schematic diagram of the appropriate valve positions for applying feedstock to an expanded-bed adsorption chromatography column. Valves that control column operation are labeled V1, V2, V3, and V4. Expect the bed to expand to a higher level during feed application (as compared with equilibration). Although it is not shown in this diagram, it is critical that the feedstock being fed to pump 1 be stirred throughout the course of application. CIP, cleaning-in-place solution (see Support Protocol 2); UV, flowthrough UV detector; cond., flowthrough conductivity meter; pH, flow-through pH meter; P1 and P2, fraction collection lines. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources).

complete, adjust the position of the adaptor so that it sits just above the top of the bed (Fig. 8.8.7). Record the position of the adaptor. Adjusting the position of the adaptor in this way will speed up the wash cycle and reduce buffer consumption by eliminating the large mixing chamber between the top of the bed and the bottom of the adaptor.

28. Continue pumping wash buffer through the column until the UV signal returns to its baseline level.

Elute product 29. Turn off pump 1 and allow the resin bed to sediment. It should be noted that the elution method presented here is known as elution in “packedbed” mode. The alternative to eluting in packed-bed mode is to immediately switch from wash buffer to elution buffer and elute the product while the bed is expanded; this is easier than eluting in packed-bed mode, but it dilutes the collected product by ∼40%.

30a. If using a manual column: When sedimentation is complete, lower the adaptor so that it sits just above the bed. Expanded Bed Adsorption Chromatography

30b. If using a hydraulic column: When sedimentation is complete, use the hydraulic pump to pump water into the hydraulic chamber (above the adaptor), and stop the pump once the adaptor sits just above the bed.



Figure 8.8.7 Schematic diagram of the appropriate valve positions for washing an expandedbed adsorption chromatography column. Valves that control column operation are labeled V1, V2, V3, and V4. The bed height decreases as the switch is made from feedstock to wash buffer, and so the adaptor is lowered accordingly. CIP, cleaning-in-place solution (see Support Protocol 2); UV, flowthrough UV detector; cond., flow-through conductivity meter; pH, flowthrough pH meter; P1 and P2, fraction collection lines. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources).

31. Set each valve in the appropriate position for eluting product from the column (Fig. 8.8.8). Using pump 1, pump 1 to 2 bed volumes of elution buffer downward through the column at a flow velocity of 100 cm/hr, and collect the eluted fraction as it comes off the column (as indicated by a peak on the UV monitor).

Perform CIP and prepare column for storage 32a. If using a manual column: Immediately after collection of the product, set each valve in the appropriate position for column equilibration (Fig. 8.8.4). Raise the adaptor to the same height at which it sat at the end of step 27. 32b. If using a hydraulic column: Immediately after collection of the product, set valve 1 and valve 4 in the appropriate positions for column equilibration (Fig. 8.8.4); set valve 2 so that it forms a dead end in the outlet line coming from the top of the column (so that flow cannot exit the top of the column), and set valve 3 open to waste. Using pump 1, pump elution buffer upward through the column to lift the adaptor to the same height at which it sat at the end of step 27. Once the adaptor is at the appropriate height, set valves 2 and 3 as shown in Figure 8.8.4. With valve 2 creating a dead end in the outlet line and valve 3 open to waste, the only way for pressure to be released from the column as the elution buffer passes through is for the fluid in the hydraulic chamber to go to waste. This will cause the adaptor to move upward.



Supplement 40

Figure 8.8.8 Schematic diagram of the appropriate valve positions for product elution (in the downward direction) from an expanded-bed adsorption chromatography column. Valves that control column operation are labeled V1, V2, V3, and V4. Note that the product is collected as a fraction through the P1 line. CIP, cleaning-in-place solution (see Support Protocol 2); UV, flowthrough UV detector; cond., flowthrough conductivity meter; pH, flow-through pH meter; P1 and P2, fraction collection lines. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources).

33. Switch the feed from elution buffer to CIP solution, and carry out the CIP process developed in Support Protocol 2. CIP solution should always be pumped upward through the column. The resin bed may initially act as a plug during this process, but the plug will fall apart as CIP proceeds.

34a. If the column is to be reused immediately after CIP: Switch the feed from CIP solution to equilibration buffer. Using pump 1, pump equilibration buffer upward through the column as in step 16, and then proceed from step 17 onwards. If cleaning was adequate, the height of the expanded bed will be approximately the same as it was in the initial run. The sedimented bed height will not change unless more resin is added to the column.


34b. If the column is to be stored after CIP: Switch the feed from CIP solution to storage solution. Using pump 1, pump the storage solution upward through the column. When storage solution appears in the outlet line, stop all pumps and allow the bed to settle. Once the bed has settled, remove the lid from the column, and then remove all liquid from the hydraulic chamber using a hose attached to a water aspirator. Refill the chamber with 20% ethanol, put the lid back on the chamber, and secure the lid using the screws provided by the manufacturer. Store the column in accordance with the manufacturer’s instructions.



ESTIMATION OF THE EQUILIBRIUM RATIO OF TARGET PROTEIN TO RESIN

SUPPORT PROTOCOL 1

While a good approximation of the resin’s capacity for binding the desired product can be obtained by working with clarified feed (i.e., the target protein isolated from the crude feedstock), a more practical option is to estimate the resin capacity using unclarified feedstock. The following is a method for estimating how much resin is needed to adsorb a given amount of product in unclarified feedstock.

Materials Unclarified feedstock containing 0.5 mg/ml target protein Chromatography resin (see Strategic Planning) End-over-end test tube mixer Additional reagents and equipment for protein assays (UNIT 3.4) 1. In the following way, prepare a series of three test tubes containing unclarified feedstock combined in different ratios with the chromatography resin to be used, making sure to maintain a constant volume of feedstock across all three test tubes.

Tube 1 10 ml unclarified feedstock containing 0.5 mg/ml target protein 0.125 g resin Ratio: 40 mg target protein per gram of resin Tube 2 10 ml unclarified feedstock containing 0.5 mg/ml target protein 0.250 g resin Ratio: 20 mg target protein per gram of resin Tube 3 10 ml unclarified feedstock containing 0.5 mg/ml target protein 0.5 g resin Ratio: 10 mg target protein per gram of resin 2. Cap tubes and place in an end-over-end mixer for 4 hr at room temperature (i.e., the temperature at which the EBA column will be run). 3. Clarify samples by performing low-speed centrifugation (1000 × g is more than adequate) at room temperature for 15 min. Assay the supernatant from each of the three tubes for target protein concentration (UNIT 3.4). Centrifugation time and speed may have to be adjusted empirically. All feedstock and resin debris must be in the pellet (leaving a clarified sample as the supernatant), because particulates will interfere with protein assays. Any of the colorimetric methods described in Unit 3 will be appropriate for assaying target protein concentration.

4. Identify the equilibrium concentration, Ceq , defined as the condition (i.e., the ratio of target protein to resin) under which the concentration of target protein in the supernatant is equal to half of the initial concentration of target protein in the feedstock. It may be necessary to extrapolate (by solving a proportionality equation based on the data points obtained in step 3) to determine Ceq . This extrapolation method assumes that Ceq falls somewhere between the highest and lowest tested ratios of target protein to resin.



Supplement 40

5. Calculate the target equilibrium capacity (Qeq ) using the following equation: Qeq = [(C0 − Ceq ) × Vl ]/Mad , where C0 is the initial concentration of target protein in the feedstock, Ceq is the equilibrium concentration identified in step 4, Vl is the volume of liquid in the test tube, and Mad is the amount (i.e., mass) of resin in the test tube. For a given condition, an EBA column should be loaded to 30% of its equilibrium capacity. SUPPORT PROTOCOL 2

DEVELOPMENT OF A CLEANING-IN-PLACE PROCESS FOR EXPANDED-BED ADSORPTION CHROMATOGRAPHY Expanded-bed adsorption (EBA) chromatography is unique among chromatographic methods in that the feed is not clarified before application to the column. As a result, EBA resins are exposed to cells, cell debris, and the products of cell lysis, as well as to culture media and their additives. In order to prolong the life of the EBA resin, a cleaning-in-place (CIP) protocol must be developed. The development of a CIP protocol for EBA resins is challenging. The appropriate CIP protocol depends on the feedstock used, but some general advice can be given. This protocol provides the basic steps for generating a CIP protocol and gives some suggestions for CIP solutions. In addition, a reference is provided in which the reader can find details on CIP methods developed for several different types of feedstock.

Additional Materials (also see Basic Protocol) CIP solution: 1.0 M NaOH, 6 M urea, or 6 M guanidine·HCl Additional components for inclusion in CIP solution (if necessary; e.g., isopropyl alcohol, acetic acid, benzonase) 1. Perform repeated expanded-bed chromatography runs (see Basic Protocol) on the same column (without performing CIP between consecutive runs) until it is evident that the degree of expansion or the chromatographic properties of the resin are changing. Changes in the chromatographic properties of the resin are signaled by, for example, the inability or decreased ability of the resin to bind the target protein, bleed-through of the target protein along with the feedstock, and the binding of contaminants that flowed through the resin in earlier runs.

2. Set all valves in the chromatography apparatus and adjust the height of the column adaptor as in Basic Protocol, step 32a (if using a manual column), or Basic Protocol, step 32b (if using a hydraulic column). Using pump 1, pump CIP solution (6 M urea or 6 M guanidine·HCl for protein A–containing media, and 1.0 M NaOH for all other types of media) upward through the column at a flow velocity of 100 to 300 cm/hr for a contact time of 8 hr or overnight. IMPORTANT NOTE: Do not start CIP with the CIP solution flowing downward through the column.

3. Switch the flow from CIP solution to equilibration buffer, adjust valve 1 so that pump 1 is connected directly to the top of the column, and then begin pumping equilibration buffer downward through the column.


4. Once two column volumes of equilibration buffer have passed through the column, adjust valve 1 so that pump 1 is connected directly to the bottom of the column, and then begin pumping equilibration buffer upward through the column. When the bed stabilizes, measure its height and use this measurement to calculate the degree of expansion of the resin. The resin is equilibrated when the pH of the outlet stream is the same as the pH of the equilibration buffer.



5a. If the chromatographic bed is successfully regenerated (i.e., if the degree of expansion and the chromatographic properties of the resin return to normal): Repeat the CIP testing process (steps 1 to 4) multiple times, each time using a shorter contact time in step 2. Identify the minimum contact time needed for effective CIP. 5b. If the chromatographic bed is not successfully regenerated: Repeat the CIP testing process (steps 1 to 4) multiple times, each time adding new components (e.g., 30% v/v isopropyl alcohol, 25% v/v acetic acid, 5 µl/l benzonase) to the CIP solution used in step 2. Determine which of these components need to be added to the CIP solution for effective CIP. Once a suitable cleaning protocol has been identified, it can be used in all EBA runs involving the feedstock in question.

ION-EXCHANGE EXPANDED-BED ADSORPTION CHROMATOGRAPHY Resins for ion-exchange EBA chromatography are chosen in the same way that resins are chosen for other types of ion-exchange column chromatography—that is, based on what is known about the pI values of the target protein and all contaminants (UNIT 8.2).


To determine the appropriate binding conditions for the target molecule in an ionexchange application, perform the basic test tube experiments described in UNIT 8.2, Support Protocol 1. Capacity determinations can be made as in UNIT 8.2, Support Protocol 2, but the results of these determinations should be treated only as approximations, since the capacity of the resin is impacted by the presence of debris. Support Protocol 1 in the current unit provides a method for estimating resin capacity in the presence of debris from unclarified feedstock.

HYDROPHOBIC-INTERACTION EXPANDED-BED ADSORPTION CHROMATOGRAPHY


Resins for hydrophobic-interaction EBA chromatography are chosen in the same way that resins are chosen for other forms of hydrophobic-interaction column chromatography— that is, based on what is known about the polarities of the target protein and all contaminants (UNIT 8.4). To determine the appropriate binding conditions for the target molecule in a hydrophobicinteraction application, perform the basic experiments described in UNIT 8.4, Support Protocol 1. Capacity determinations made using clarified feed should be treated only as approximations, since the capacity of the resin is impacted by the presence of debris. Support Protocol 1 in the current unit provides a method for estimating resin capacity in the presence of debris from unclarified feedstock.

AFFINITY EXPANDED-BED ADSORPTION CHROMATOGRAPHY Resins for affinity EBA chromatography are chosen in the same way that resins are chosen for other forms of affinity column chromatography—that is, based on what is known about the potential for specific interaction between the target protein and the ligand immobilized on the resin. There are only a few commercially available affinity ligands for affinity EBA chromatography. The protocols described in Chapter 9 also apply to initial methods development for affinity EBA processes. Support Protocol 1 in the current unit provides a method for estimating resin capacity in the presence of debris from unclarified feedstock.




Supplement 40

COMMENTARY Background Information Principles of expanded-bed operation The expanded-bed adsorption cycle is a five-step operation (Fig. 8.8.9). The first step is the creation of a sedimented bed, which must have a height of ∼15 cm to allow sufficient contact time for efficient binding of the target molecule to the resin. The column is then equilibrated, with this equilibration serving two purposes. First, it allows formation of a stable, fluidized bed when the suspended adsorbent beads attain an equilibrium between particle sedimentation velocity and upward liquid flow velocity. (At equilibrium, the bed is expanded to 2.5 to 3 times its sedimented height.) Second, it allows equilibration of the adsorbent with the sample buffer. After equilibration, crude, unclarified feedstock is applied to the expanded bed in the same upward direction; target proteins in the feedstock bind to the resin, while cells, cell debris, particulates, and contaminants pass through unhindered. Next, weakly bound material, such as residual cells, cell debris, and other types of particulate material, is washed out of the expanded bed, and once all weakly retained material has been removed, the liquid flow is stopped, allowing the adsorbent beads to quickly settle in the column due to their high density. The column adapter is then lowered to the surface


of the sedimented bed, the flow is reversed, the buffer is changed, and the captured proteins are eluted. These eluted proteins are concentrated, clarified, partially purified, and ready for further purification by packed-bed chromatography. After elution, the sedimented resin bed is regenerated by washing with downward flow using buffers specific for the type of chromatographic principle in use. This process removes the more strongly bound proteins not removed during the elution phase. Finally, a cleaning-inplace procedure is performed to remove nonspecifically bound, precipitated, or denatured substances from the bed and to restore its original performance. During this phase, a moderate upward flow is used, with the column adaptor positioned at approximately twice the sedimented bed height. The purpose of cleaning in place is to clean the adsorbent between purification runs. A review of five cleaning-in-place protocols has been published (Amersham Biosciences; see Internet Resources). Four of the five case studies in this review deal with E. coli expression systems—namely, E. coli homogenate on a DEAE resin, E. coli homogenate on an SP resin, E. coli lysate on an SP resin, and E. coli inclusion bodies on an SP resin. The fifth case study deals with S. cerevisiae homogenate on an SP resin. These case

Figure 8.8.9 Resin bed status throughout the steps of an expanded-bed adsorption chromatography experiment. Arrows indicate direction of feed flow. Cleaning in place (not shown) is the last step and is done in expanded-bed mode, similar to expansion and equilibration of the adsorbent. Reproduced with permission from Amersham Biosciences (Expanded-Bed Adsorption: Principles and Methods; see Internet Resources) and Academic Press (Shiloach and Kennedy, 2000).



studies describe media reuse for 10 to 50 cycles, which is typical. In principle, an expanded bed is similar to a fluidized bed; the major difference is that during operation, there is minimal mixing in an expanded bed, while a fluidized bed is turbulent. This means that an expanded bed permits a more efficient adsorption process than does a fluidized bed. The most efficient expanded beds are stable, classified fluidized beds. (A classified fluidized bed is one in which particles of similar sizes and densities are found together, as a class, at the same position in the column.) A review comparing the properties of expanded beds and fluidized beds can be found in Thommes (1997). The pressure drop over an expanded bed is small, and so the distribution system must build up a pressure drop to produce a plug flow (i.e., the condition in which axial mixing is minimal compared with convective flow).

In addition to generating this pressure drop, the distributor also directs flow in a strictly vertical path. This eliminates any radial flow that might propagate turbulence through the column. A specially designed liquid distribution system is needed because of the small pressure drop over the expanded bed as compared with a packed bed. During operation, the pressure drop in an expanded bed is typically 5 is very selective. The most successful applications have employed an SP resin at such pH levels. EBA chromatography with E. coli feedstocks E. coli feedstock is used in three scenarios: one in which the recombinant protein is soluble and is secreted into the fermentation broth, another in which the recombinant protein is secreted into the cell periplasm, and a third in which the recombinant protein is insoluble inside the cell and exists as an inclusion body. Recombinant proteins secreted directly into the broth can be harvested directly by EBA. In contrast, cells must be partially

broken to release product from the periplasm and completely homogenized to release inclusion bodies. When working with E. coli homogenates, at least two passes through the homogenizer (at 8000 psi) must be performed to shear all DNA present. Conditioning of the homogenate at pH α-Glc

0.1–0.5 M α-MeMan

High-Man, hybrid, and biantennary N-linked chains

LCA or LCH (Lens culinarus; lentil seeds)

Ca2+, Mn2+ α-Man > α-Glc


Bi- and triantennary N-linked chains with Fucα1-6 in core region


Similar to LCA/LCH

0.1–0.5 M GlcNAc

GlcNAc- and Sia-terminated chains, or clusters of O-GlcNAc; succinylated form selectively binds GlcNAc>Sia

0.1–0.5 M Gal or lactose

—

β-Gal β-GalNAc β-Gal

Galβ1-4 on N-linked chains; can also bind β-GalNAc β-GalNAc-terminated chains.

—

GalNAc

0.1–0.5 M GalNAc

—

GalNAc

0.1–0.5 M GalNAc

Sugar specificityb Elution conditions

PSA (Pisum sativum; Ca2+, Mn2+ α-Man peas) WGA (Triticum Ca2+, Mn2+ β-GlcNAc vulgaris; wheat germ)

RCA (Ricinus communis; castor bean) RCA-Ia RCA-IIa PHA (Phaseolus vulgaris; red kidney bean) E4PHA L4PHA GSL-1, BSL-1 [Griffonia (Bandeiraea); Simplicifolia seeds]

—

Mg2+, Ca2+ α-Gal α-GalNac

0.1–0.5 M Gal

Useful for binding

Gal-terminated N-linked chains with “bisecting” GlcNAcβ1-4 Tri- and tetrabranched N-linked chains

0.1–0.5 M melibiose

Chains terminated by α-Gal residues

Proteins with poly-N-acetyllactosaminecontaining chains; highly heterodispersed glycoproteins often contain variable lengths of this repeat Similar to DSA; prefers longer poly-N-acetyllactosamines than DSA Proteins with poly-N-acetyllactosamines

DSA (Datura stramonium; Jimsonweed seeds)

—

Galβ1-4GlcNAc

0.1–0.5 M GlcNAc or 20 mM N,N′,N′′chitotriose

LEL (Lycopersicon esculentum; tomato meat) STL (Solanum tuberosum; potato tubers) MAL (Maackia amurensis; seeds)

—

β-GlcNAc

—

β-GlcNAc

—

Siaα2-3Galβ1-4 GlcNAc

0.1–0.5 M β-GlcNAc or 20 mM chitobiose or chitotriose 0.1–0.5 M GlcNAc or 20 mM chitobiose or chitotriose 0.5 M lactose

α2-3-linked Sia to Gal in N-linked sugar chains; underlying sugars and peptide contribute to binding continued

Lectin Affinity Chromatography


Table 9.1.1

Lectins Useful for Glycoprotein Purification, continued

Acronym (organism and source)

Metal ions required

Sugar specificityb Elution conditions

Useful for binding

EBL or SNA (Sambucus nigra; elderberry bark) LFA (Limax flavus; garden slug) GNL (Galanthus nivalis; snowdrop bulbs)

—

Siaα2-6Gal or GalNAc

0.1–0.5 lactose

α2-6-linked Sia in N and O sugar chains

—

Neu5Ac

10 mM Sia

—

α1-3Man

0.5 M α-MeMan

Sia-terminated chains irrespective of linkage High-Man-type chains, but not to Glc

UEA-I (Ulex europaeus; furze gorse seeds)

—

α-L-Fuc

0.1–0.5 M L-Fuc or methyl-α-L-Fuc

Sugar chains with terminal α-Fuc, especially in α1-2 linkage, but much less with α1-3 or α1-6 linkages

AAA (Anguilla anguilla; freshwater eel)

—

α-L-Fuc

0.1–0.5 M L-Fuc

Similar to UEA-I but more broadly binds any fucosylated oligosaccharide

Lotus (Lotus tetragonolobus; asparagus pea)

—

α-L-Fuc

0.1–0.5 M L-Fuc

Outer-branch α-Fuc residues; does not bind when Fuc is linked to chitobiose core of N-linked chains

HPA (Helix promatia; albumin gland of edible snail)

—

α-GalNAc

0.1–0.5 M GalNAc

Proteins with terminal α-GalNAc or GalNAcα-O-Ser/Thr (Tn antigen)

Jackalin (Artocarpus integrifolia; jackfruit seeds)

—

α-Gal

0.1–0.2 M melibiose

VVL or VVA (Vicia villosa; hairy vetch seeds) SBA (Glycine max; soybean)

—

GalNAc

0.1–0.5 M GalNAc

Proteins with only a single O-linked chain with T antigen Galβ1-3GalNAcα-O-Ser/Thr can bind; not inhibited by lactose; Sia in chain does not interfere with binding Proteins with Tn antigen, GalNAcα-O-Ser/Thr

—

GalNAc or Gal

0.1–0.5 M GalNAc

α- or β-linked GalNAc

PNA (Arachis hypogaea; peanut)

—

Galβ1-3GalNAc

0.1–0.5 M lactose

Proteins with T antigen, but does not bind if sugar chain is sialylated

DBA (Dolicholos biflorus; horse gram seeds)

—

Terminal α-GalNAc

0.1–0.5 M GalNAc

Terminal α-GalNAc (specific for blood group A), not those in the core of O-linked chains

0.1–0.5 M GalNAc

Proteins with blood group A structure GalNAcα1-3(Fucα12)Gal–

LBA (Phaseolus lunatus; lima bean)

Mn2+, Ca2+ Terminal αGalNAc

aCAUTION: Both RCA-I and RCA-II are extremely toxic, with lethal doses of 1 molecule/cell. bAbbreviations: Fuc, L-fucos; Gal, D-galactose; GalNAc, N-acetyl-D-galactosamine; Glc, D-glucose; GlcNAc, N-acetyl-D-glucosamine; Man, Dmannose; Neu5Ac, N-acetyl-D-neuraminic acid; Sia, sialic acid.

Affinity Purification


In general, these immobilized lectins can be used in two ways. First is as an affinity step for routine purification. If a novel protein may be glycosylated, using several lectins that bind very different types of sugar structures (N-linked vs. O-linked) will probably give reasonable purification. It will also give some indication of the type of sugar chains on the protein. Both Vector and E-Y Labs sell kits with six or seven different lectins for this purpose. The second and more systematic approach is useful for better-characterized glycoproteins. Here, the lectins are chosen to resolve potentially different glycoforms that may have different biological activities. An excellent example of this second type of analysis is given by Li and Jourdian (1991). The Basic and Alternate Protocols describe the use of lectins for preparative glycoprotein purification. Con A–Sepharose and WGA–agarose were chosen for convenience and availability. The Support Protocol describes a small-scale pilot procedure to test for lectin binding and to determine elution conditions. There are many variations on the basic procedure in the literature, but all use the same principles: bind the protein to immobilized lectin through its sugar chain, wash away unbound protein, and elute bound protein with a simple sugar that resembles the sugar ligand of the bound protein. Because many proteins have sugar chains that can bind to a specific lectin, this procedure seldom yields a pure protein. BASIC PROTOCOL

CON A–SEPHAROSE AFFINITY CHROMATOGRAPHY Con A–Sepharose chromatography is used to partially purify glycoproteins that contain terminal mannose or glucose residues. The steps presented below give typical conditions for this type of chromatography, although a variety of approaches have been utilized. In this protocol, bound glycoproteins are eluted with α-methyl-D-mannoside (αMM) after the column is first washed to remove unbound and weakly bound proteins. Before proceeding, it is advisable to conduct a pilot study to test the protein of interest for lectin binding and elution conditions (see Support Protocol). In order to purify the protein effectively, a specific assay for detecting the protein of interest (i.e., activity or antibodybinding assay) must be available. Materials 10 mg/ml Con A–Sepharose (Pharmacia Biotech or Sigma) Column buffer (see recipe) 0.5 M αMM in column buffer Protein sample in column buffer 1.5 × 30–cm glass or disposable chromatographic column Glass wool NOTE: Carry out this procedure at room temperature if the protein to be isolated will tolerate this condition. If not, carry it out in a cold room, and prechill all solutions to maintain temperature. 1. Gently resuspend 50 ml settled Con A–Sepharose (10 mg lectin/ml packed resin) in 50 ml column buffer to make a slurry. Degas the slurry.

Lectin Affinity Chromatography

This volume of lectin beads should be sufficient to bind ∼100 mg glycoprotein. If a 1.5 × 30–cm column is not available, either the dimensions of the column or the total amount of resin can be modified. A 1.0 × 30–cm column will bind ∼50 mg total glycoprotein. If the amount of glycoprotein in the sample is substantially less than 50 to 100 mg, decrease the volume of the column accordingly. The ratio of input protein to lectin does not seem to matter as long as the column is not overloaded.


2. Pack a glass wool plug over a sintered glass or polypropylene frit at the bottom of the column and pour the degassed slurry into the column. The glass wool prevents the beads from clogging the frit and slowing the flow rate. Alternatively, BioGel P2 or Sephadex G10 can be used in place of the glass wool.

3. Continue packing the column until the desired level is reached (for a 50-ml volume this is ∼28 cm). Wash the gel with 2 to 3 column volumes of column buffer to remove any loosely bound or degraded Con A. 4. Wash with 2 to 3 column volumes of 0.5 M αMM in column buffer or the highest concentration of αMM that will be used. 5. Wash the column with >5 column volumes of column buffer without αMM to reequilibrate. It is important to prewash the column with the eluting sugar and then to reequilibrate the column to remove any materials that may have previously bound to the column. Check completeness of washing by phenol–sulfuric acid assay (Manzi, 1993). Estimate the amount of residual sugar using αMM as a standard. An acceptable level is 100-fold purifications of glycoproteins and nearly quantitative yields, depending upon the source of the sample and the elution conditions. If the protein of interest is reasonably stable at high salt concentrations and room temperature, specific elution conditions can likely be found that will produce good yields.

Time Considerations Initial determination of the binding and elution conditions can easily be done in 1 day, assuming the detection assay for the target protein is simple and fast. The amount of time needed to conduct a single run once the elution conditions are known can vary depending upon the size of the sample and the column, and whether elution requires long-term incubation in the presence of the sugar. Even assuming the extremes of sample size and elution conditions, a single run should be completed in 2 to 3 days.

Literature Cited Cummings, R.D. 1994. Use of lectins in analysis of glycoconjugates. Methods Enzymol. 230:66-86. Ketcham, C.M. and Kornfeld, S. 1992. Purification of UDP-N-acetylglucosamine: Glycoprotein Nacetylglucosamine-1-phosphotransferase from Acanthamoeba castellanii and identification of a subunit of the enzyme. J. Biol. Chem. 267:11645-11653.

Li, M. and Jourdian, G.W. 1991. Isolation and characterization of the two glycosylation isoforms of low molecular weight mannose-6-phosphate receptor from bovine testis. J. Biol. Chem. 266:17621-17630. Manzi, A.E. 1993. Phenol–sulfuric acid assay for hexoses and pentoses. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 17.9.1-17.9.2. John Wiley & Sons, New York.

Key References Beeley, J.G. 1984. Glycoprotein and proteoglycan techniques. In Laboratory Techniques in Biochemistry and Molecular Biology, pp. 29-99. Elsevier Science Publishing, New York. Good discussions about general properties of glycoconjugates. Cummings, R.D. 1994. See above. Most current description of using lectins to characterize sugar chain structures. Lots of details on procedures to analyze free sugar chains rather than glycoproteins. Dulaney, J.T. 1979. Binding interactions of glycoproteins with lectins. Mol. Cell. Biochem. 21:4362. Lists many references and conditions used for lectins in protein purification. Li, M. and Jourdian, G.W. 1991. See above. Excellent example of separation of various glycoforms using lectin–affinity chromatography. Montreuil, J., Bouquelet, S., Debary, H., Fournat, B., Spik, G., and Strecker, G. 1986. Glycoproteins. In Carbohydrate Analysis: A Practical Approach (M.F. Chaplin and J.F. Kennedy, eds.) pp. 166-173. IRL Press, Washington, D.C. Lists several conditions for selected lectin-affinity purifications of proteins.

Contributed by Hudson H. Freeze La Jolla Cancer Research Foundation La Jolla, California



Dye Affinity Chromatography

UNIT 9.2

Dye affinity chromatography is a protein purification procedure based on the high affinity of immobilized dyes for the binding sites on many proteins. It is a rapid, inexpensive, and versatile method that is applicable to the purification of crude cellular extracts. There are three types of dye affinity chromatography—negative chromatography, positive chromatography, and tandem chromatography. Negative chromatography (Basic Protocol 2) is the simplest of the three procedures. Some of the undesired proteins are retained by the immobilized dye while the remaining proteins, including the desired protein, flow through the column. This results in a modest-fold purification. Negative chromatography is particularly convenient for the rapid removal of degradative enzymes such as proteases and nucleases, and for the removal of very abundant proteins, such as serum albumin. A different immobilized dye is used for positive chromatography (Basic Protocol 3 and Alternate Protocol 2). This immobilized dye retains the desired protein as well as some undesired proteins. The desired protein is then selectively eluted using either a nonspecific reagent, such as a neutral salt, or a specific reagent, such as a substrate, cofactor, or competitive inhibitor. Although positive chromatography requires more investigator involvement, it results in a greater-fold purification than obtained using negative chromatography, and is more commonly used. In some cases, a desired protein can be purified to homogeneity from a crude cellular extract using positive chromatography. In tandem chromatography (Basic Protocol 4), the most elaborate of the three procedures, negative chromatography and positive chromatography are coupled sequentially. The protein mixture first passes through a column containing the immobilized dye chosen for negative chromatography. The desired protein flows through this column and immediately enters another column containing the immobilized dye chosen for positive chromatography. The desired protein, which is retained by the latter column, is then eluted using either batch or gradient elution. Tandem chromatography is particularly useful for situations in which the immobilized dye used for positive chromatography binds too many undesired proteins, making selective elution of the desired protein difficult. Use of dye affinity chromatography for protein purification requires selection of one or more immobilized dyes. In addition, positive chromatography requires selection of an elution reagent and its effective concentration. These selections can be made using chromatography, as outlined in Basic Protocol 1, or using centrifugation, as outlined in Alternate Protocol 1, depending on the inclination of the investigator. Immobilized dyes can be purchased as dry powders or suspensions or already poured into miniature columns in kit form. Free dyes also can be purchased and easily immobilized by the investigator (Support Protocol). Alternate Protocol 1 facilitates selection because many samples can be processed simultaneously, but is more expensive because more immobilized dye is required. SELECTION OF COMPONENTS USING CHROMATOGRAPHY The immobilized dye for negative or positive chromatography and the eluant for positive chromatography can be selected using a chromatographic procedure. A small volume of the protein mixture to be purified is applied to a series of miniature columns, each containing a different immobilized dye. The flowthrough and washes from each column are pooled and analyzed for total protein and for desired protein. The immobilized dye column whose pooled flowthrough and washes contain the least amount of undesired protein and the greatest amount of desired protein is appropriate for negative chromatography. Conversely, the immobilized dye column whose pooled liquid contains the greatest Contributed by Earle Stellwagen Current Protocols in Protein Science (1995) 9.2.1-9.2.16 Copyright © 2000 by John Wiley & Sons, Inc.

BASIC PROTOCOL 1


9.2.1 CPPS

amount of undesired protein and the least amount of desired protein is appropriate for positive chromatography. The immobilized dye column chosen for positive chromatography is used to select the elution reagent and its effective concentration. The column is washed with small volumes of a potential elution reagent, with the concentration of the reagent increased in small increments in each successive wash. The flowthrough for each wash is then analyzed for both total protein and desired protein. Additional potential elution reagents are examined following application of a fresh aliquot of the protein mixture to be purified. The concentration of elution reagent that provides the greatest-fold purification, recovery of desired protein, and economy of use is selected to be the elution solvent. A specific elution reagent will generally provide a greater increase in fold purification than a nonspecific elution reagent, but the specific elution reagent may cost more. Materials Protein mixture to be purified Appropriate application solvent (see step 1) Series of immobilized dyes (Table 9.2.1 and Support Protocol) Concentrated solution of an appropriate elution reagent (see step 11) Plastic disposable filtration unit with pore size ≤0.45 µm Small plastic disposable chromatography tubes with bed volume at least 1 ml

Table 9.2.1

Some Commercially Available Immobilized Dyesa

Dye

Immobilized dye

Supplierb

Reactive blue 2

Affi-Gel blue gel Blue Sepharose CL-6B Cibacron blue 3G-A-agarose Cibacron blue F3G-A DyeMatrex blue A gel Reactive blue 4-agarose Reactive blue 72-agarose Reactive brown 10-agarose Reactive green 5-agarose DyeMatrex green A gel Reactive green 19-agarose DyeMatrex red A gel Reactive red 120-agarose Red Sepharose CL-6B Reactive yellow 2-agarose Reactive yellow 3-agarose DyeMatrex orange A gel Reactive yellow 13-agarose Reactive yellow 86-agarose

Bio-Rad Pharmacia Biotech ICN Biomedicals, Sigma Pierce Amicon Sigma Sigma Sigma ICN Biomedicals, Sigma Amicon Sigma Amicon ICN Biomedicals, Sigma Pharmacia Biotech ICN Biomedicals, Sigma Sigma Amicon Sigma Sigma

Reactive blue 4 Reactive blue 72 Reactive brown 10 Reactive green 5 Reactive green 19 Reactive red 120

Reactive yellow 2 Reactive yellow 3 Reactive yellow 13 Reactive yellow 86

aScreening kits containing between 5 and 40 immobilized dyes in miniature chromatography columns can be purchased

from Affinity Chromatography, Amicon, Wako Pure Chemical, and Sigma. Alternatively, free reactive dyes such as those listed in Table 9.3.2 can be purchased, immobilized, and prepared for chromatography as described in the Support Protocol. bSupplier names and addresses are provided in SUPPLIERS APPENDIX.



Prepare protein sample for application to column 1. Prepare the crude protein mixture. If it is a solid, dissolve it in the application solvent. If the protein mixture to be purified is a suspension or a solution, dialyze it for several hours against the application solvent. A generic application solvent is 50 mM Tris⋅Cl buffer, pH 7.5, at room temperature. The identity of the buffer, its pH, and the temperature of the application solvent may be varied to maintain the integrity of the desired protein without compromising the chromatography. However, addition of reagents that increase the ionic strength of the application solvent above 50 mM may diminish retention of all proteins by the immobilized dye columns. Addition of millimolar concentrations of one or more specific reagents that bind to the desired protein to maintain its integrity may diminish the retention of the desired protein by the immobilized dye columns.

2. Clarify the protein mixture, if necessary, either by centrifugation or filtration. Filtration should be performed using a plastic disposable filtration unit having a pore size of ≤0.45 ìm.

3. Determine the amount of total protein and the amount of desired protein in 1 ml of the protein mixture. Prepare assortment of immobilized dye columns 4. Pour a sufficient volume of each immobilized dye suspension into a different chromatography tube to form a bed volume of 1 ml in each tube. Most commercial screening kits of immobilized dyes are supplied as chromatographic columns having bed volumes ranging from 0.1 to 2.5 ml; these may be used without alteration. The optimal bed volume for a particular situation depends on the total volume of the crude protein mixture available to the investigator; the greater this total volume, the more that can be used. The volume of available crude protein mixture is the principal reason the bed volumes of individual columns in the commercial kits range from 0.1 to 2.5 ml. If one or more of the immobilized dyes are available in powdered form, suspend about 0.5 g of each powdered immobilized dye in application solvent for at least 30 min with gentle stirring prior to pouring the columns.

5. Wash each immobilized dye column with 2 bed volumes of chromatographic application solvent. Let the column run dry and discard the washes. Perform chromatography and assess column performance 6. Apply the same volume of protein mixture to each immobilized dye column and collect the flowthrough in a container. The volume of protein mixture added should contain between 5 and 40 mg of total protein per milliliter of bed volume and sufficient desired protein (1% of total protein) for convenient analysis. The flowthrough can be collected in a disposable tube of sufficient volume to contain the application volume and 2 bed volumes of wash.

7. Wash each immobilized dye column with 2 bed volumes of application solvent and collect the flowthrough in the same container used in step 6. 8. Measure the concentrations of the total protein and the desired protein in the combined flowthrough from each column. 9. Compare the amount of total and desired protein in the combined flowthrough from each column. The immobilized dye whose combined flowthrough contains the least amount of total protein and the greatest amount of desired protein is most appropriate for negative chromatography (see Basic Protocol 2). The immobilized dye whose combined flowthrough Affinity Purification


contains the greatest amount of total protein and the least amount of desired protein is most appropriate for positive chromatography (see Basic Protocol 3 or Alternate Protocol 2). If positive chromatography is to be performed, continue to use only the immobilized dye column selected for positive chromatography. This column should still retain the desired protein.

Elute desired protein from positive chromatography column 10. Wash the immobilized dye column whose flowthrough contained the least amount of desired protein (i.e., the column selected for positive chromatography) with 2 bed volumes of application solvent containing the lowest concentration of the elution reagent selected for testing. Collect the flowthrough in a separate container. A typical nonspecific elution reagent is NaCl, beginning with a lowest concentration of 0.1 M. A typical specific elution reagent is a substrate, cofactor, inhibitor, effector, or ligand, beginning with a lowest concentration of 1 ìM. The specific elution reagents chosen for testing should represent a good compromise between availability, economy, and high affinity for the desired protein.

11. Measure the total protein and the total desired protein in the flowthrough. 12. Increase the concentration of the elution reagent in the application solvent by increments and repeat steps 10 and 11. A representative increment for a nonspecific elution reagent is 0.2 M and that for a specific elution reagent is an order of magnitude. The application solvent containing the lowest concentration of a reagent that elutes at least 80% of the desired protein retained by the immobilized dye column is defined as the elution solvent. If alternative elution reagents are to be investigated for reasons of economy or availability, proceed with steps 13 to 15.

Repeat elution using alternative elution solvent (optional) 13. Wash the immobilized dye column selected for positive chromatography with 2 bed volumes of 2 M NaCl followed by 2 bed volumes of application solvent and discard the flowthrough. 14. Reapply a volume of the protein mixture as described in step 6 and wash with two column volumes of application solvent. Discard the flowthrough. 15. Repeat steps 10 to 12 using alternative elution reagent. ALTERNATE PROTOCOL 1


SELECTION OF COMPONENTS USING CENTRIFUGATION The immobilized dye for negative or for positive chromatography and the eluant for positive chromatography can also be selected using a centrifugation procedure instead of the chromatography procedure described in Basic Protocol 1. A different immobilized dye is placed in a series of microcentrifuge tubes. The supernatant solution is discarded from each tube and replaced with a small volume of the protein mixture. The contents of each tube are mixed and centrifuged and the supernatants analyzed for the amount of total protein and desired protein. The immobilized dye whose supernatant contains the least amount of total protein and the greatest amount of desired protein is most appropriate for negative chromatography. Conversely, the immobilized dye whose supernatant contains the greatest amount of total protein and the least amount of desired protein is appropriate for positive chromatography. To select an elution reagent for positive chromatography, the same amount of the immobilized dye selected for positive chromatography is added to a series of microcentrifuge tubes. The protein mixture is added to each tube and the


supernatant discarded. A different concentration of a potential elution reagent is added to each tube, mixed, and centrifuged, and each supernatant analyzed for the amount of total protein and desired protein present. Any number of different potential specific and nonspecific elution reagents can be evaluated in this manner. The concentration of elution reagent that provides the greatest-fold purification, recovery of desired protein, and economy of use is selected. The centrifugation procedure allows an investigator to select an elution solvent more expeditiously than does the chromatography procedure because many samples can be processed simultaneously. However, a larger amount of each immobilized dye is required. For a list of materials, also see Basic Protocol 1. Prepare the protein sample and set up immobilized dye series 1. Prepare protein sample for application to column (Basic Protocol 1, steps 1 to 3). 2. Add a sufficient volume of an immobilized dye suspension to a capped graduated 1.5-ml disposable plastic microcentrifuge tube to bring the solid/liquid boundary to the 0.4-ml mark. Repeat for each available suspension of immobilized dye. Cap each tube and centrifuge at full speed for 1 to 2 min and discard the supernatant. If one or more of the immobilized dyes are available in powdered form, suspend about 0.5 g of each powdered dye in application solvent for at least 30 min with gentle stirring prior to adding to the microcentrifuge tubes.

3. Add 1 ml application solvent to each microcentrifuge tube. Cap each microcentrifuge tube and mix its contents by inversion, shaking, or vortexing. Let stand for 30 min to swell the immobilized dye matrix. 4. Centrifuge each tube 1 to 2 min at full speed and discard the supernatant. 5. Add 1 ml application solvent to each tube. Cap each microcentrifuge tube and mix its contents by inversion, shaking, or vortexing. Centrifuge each tube 1 to 2 min at full speed and discard the supernatant. Apply protein sample to immobilized dye tubes 6. Add the same volume of the protein mixture to each tube. The volume of protein mixture should contain between 2.5 and 20 mg of total protein and sufficient desired protein for convenient measurement.

7. Cap each microcentrifuge tube and mix its contents by inversion, shaking, or vortexing. Centrifuge each tube 1 to 2 min at full speed. 8. Measure the amount of total protein and desired protein in each supernatant. The immobilized dye whose supernatant contains the least amount of total protein and the greatest amount of desired protein is most appropriate for negative chromatography (see Basic Protocol 2). The immobilized dye whose supernatant contains the greatest amount of total protein and the least amount of desired protein is most appropriate for positive chromatography (see Basic Protocol 3 or Alternate Protocol 2).

Test application solvents for positive chromatography 9. If positive chromatography is to be performed, add a sufficient volume of the immobilized dye suspension selected for positive chromatography to a series of graduated microcentrifuge tubes to bring the solid/liquid boundary in each tube to the 0.4-ml mark. 10. Repeat steps 3 to 6. 11. Construct a series of application solvents each containing concentration of the elution reagent to be tested.



A typical nonspecific elution reagent is NaCl, beginning with a lowest concentration of 0.1 M and increasing in 0.2 M increments. A typical specific elution reagent is a substrate, cofactor, inhibitor, effector, or ligand having a lowest concentration of 1 ìM and increasing in order of magnitude increments.

12. Add 1 ml of each application solvent containing the elution reagent to a different centrifuge tube containing immobilized dye. Cap the microcentrifuge tube and mix the contents by inversion, shaking, or vortexing. Centrifuge each tube 1 to 2 min at full speed. 13. Measure the concentration of total protein and of desired protein in each supernatant. The application solvent containing the lowest concentration of a reagent that elutes at least 80% of the desired protein retained by the immobilized dye column and that represents a good choice with regard to cost and availability is defined as the elution solvent.

14. Repeat steps 11 to 13 to test additional elution reagents or to refine the concentration of the selected elution reagent, if desired. BASIC PROTOCOL 2

NEGATIVE CHROMATOGRAPHY A protein mixture is applied to an immobilized dye column selected for negative chromatography and the column is washed with application solvent. Many of the undesired proteins are retained by the column while the desired protein as well as some of the undesired proteins flow through the column. The concentrations of the total protein and total desired protein in the combined flowthrough are measured. These values should reflect a yield of >80% of desired protein and a modest-fold purification. Negative chromatography is the simplest to execute but, because the desired protein as well as some undesired proteins are not retained and appear in the column flowthrough, it is the least likely to dramatically increase the fold purification of a desired protein. However, it is particularly convenient for the rapid removal of troublesome contaminating proteins such as proteases, nucleases, or serum albumin. Materials Protein mixture to be purified Application solvent, with and without 2 M NaCl Immobilized dye selected for negative chromatography (see Basic Protocol 1 or Alternate Protocol 1) 0.02% (w/v) sodium azide Empty chromatographic column with height/diameter ratio ≤5 (sintered glass funnel can suffice) 1. Prepare the protein mixture for chromatography (see Basic Protocol 1, steps 1 to 3). 2. Divide the total protein concentration (in milligrams) to be purified by 5 to estimate the bed volume (in milliliters) of the immobilized dye needed. Add suspension of the immobilized dye to the chromatographic column until that bed volume is acquired. Division is by 5 because it is conservatively estimated that each milliliter of column bed volume can retain 5 mg of protein. Immobilized dyes may be purchased either as suspensions or as dry powders from commercial sources (see Table 9.2.1). The suspensions are easier to use because they are already swollen; therefore they are the form of choice. If the powdered form of the immobilized dye is used, it should be suspended in an application solvent for about an hour with gentle stirring prior to pouring the column.



3. Wash the column with application solvent until no further dye is eluted. The volume of application solution required for washing depends upon the storage of the column. The column matrix slowly but steadily hydrolyzes, generating soluble matrix fragments still containing covalently attached dye molecules. The longer the storage time and the higher the temperature, the greater the amount of solubilized dye. The volume of wash is determined by the investigator; when colored dye stops appearing in the wash, the column is ready for chromatography. This may require 1 or 20 column volumes. The presence of the colored dye can be detected in the washing either by visual inspection or by spectrophotometric measurement at an appropriate wavelength.

4. Apply the entire protein mixture to be purified to the immobilized dye column using a flow rate of at least 1 ml/min. Collect the flowthrough liquid in a single container. The flow rate is dictated by the size of the column and the density of its packing. A fast flow rate reduces experimental time. Equilibration of protein with immobilized dye is rapid; there is little advantage in using flow rates less than 1 ml/min.

5. Wash the column with 1 column volume of application solvent. Collect the flowthrough liquid in the same container as in step 4. 6. Measure the concentration of total protein and total desired protein in the combined flowthrough liquids. These values should reflect a good yield, >80% of desired protein, and a modest-fold purification. If the fold purification is less than expected from the results observed using Basic Protocol 1 or Alternate Protocol 1, add more immobilized dye to the column and reapply the combined flowthrough liquids.

7. Wash the column with 1 column volume of application solvent containing 2 M NaCl to displace the retained protein. Protein displacement can be ascertained by monitoring the amount of protein in the wash effluent.

8. Prepare the column for storage by washing with 2 column volumes of 0.02% sodium azide and storing the poured column at 4°C. POSITIVE CHROMATOGRAPHY USING BATCH ELUTION A protein mixture is passed through an immobilized dye column selected for positive chromatography. This immobilized dye retains the desired protein and some of the undesired proteins. The desired protein is then selectively eluted using either a nonspecific or a specific elution reagent (a higher-fold purification will likely be obtained using the specific eluant), and the concentrations of total protein and desired protein in each fraction are measured. This protocol for positive chromatography utilizes batch elution, which is simpler to execute than Alternate Protocol 2 utilizing gradient elution (the latter may yield a higher-fold purification than batch elution but requires more effort to set up).

BASIC PROTOCOL 3

Materials Protein mixture to be purified Appropriate application solvent Immobilized dye and elution solvent selected for positive chromatography (see Basic Protocol 1 or Alternate Protocol 1) Empty chromatographic column with height/diameter ratio >5 Fraction collector 1. Prepare protein sample and immobilized dye column (see Basic Protocol 2, steps 1 to 5) using the immobilized dye selected for positive chromatography.



2. Measure the concentration of the total protein and the desired protein in the flowthrough liquids. If the majority of the desired protein is retained by the column, discard the flowthrough liquid. If a significant amount of desired protein is not retained, add more immobilized dye to the column and reapply the flowthrough liquid. Typically, at least 70% and, if possible, 90% or more of the desired protein is retained.

3. Wash the column with the elution solvent selected for positive chromatography using a flow rate of at least 1 ml/min. Collect the flowthrough in a fraction collector in constant-volume fractions, typically 1 ml. 4. Measure the concentration of total protein and desired protein in each fraction. 5. Pool fractions containing the desired protein. The specific activity of the desired protein in the fractions can be used to determine which fractions to include in the pool. Typically pool all tubes containing at least 10% of the total amount of desired protein applied to the column.

6. Wash and store the column (see Basic Protocol 2, steps 7 and 8). ALTERNATE PROTOCOL 2

POSITIVE CHROMATOGRAPHY USING GRADIENT ELUTION A protein mixture is passed through an immobilized dye column selected for positive chromatography. This immobilized dye retains the desired protein and some of the undesired proteins. The desired protein is selectively eluted using a linear gradient constructed from the elution solvent selected for positive chromatography, and the concentrations of total protein and desired protein in each fraction are measured. This protocol is more elaborate to set up and operate, but sometimes generates a greater-fold purification than obtained using the batch elution system of Basic Protocol 3. Additional Materials (also see Basic Protocol 3) Empty chromatographic column having a height/diameter ratio of >5 A coated magnet and magnetic stirrer Linear gradient maker (commercial or constructed from two beakers of the same size connected by a siphon) 1. Prepare protein sample and immobilized dye column (see Basic Protocol 3, steps 1 and 2). All solutions should be carefully introduced so that the column packing is minimally disturbed.

2. Place a volume of the elution solvent selected for positive chromatography in the nonmixing reservoir of the linear gradient maker. This volume is typically equal to the column volume. The nonmixing reservoir is the reservoir more distant from the column. Its contents flow into the mixing reservoir prior to introduction onto the column.

3. Place 10% of the volume of elution solvent used in step 2 in the mixing reservoir of the gradient maker. Add application solvent to make up the remaining 90% of the volume. This procedure will generate an order-of-magnitude gradient terminating in the eluant concentration employed in batch elution.



4. Place a magnetic stir bar in the mixing reservoir. Place the mixing reservoir on a magnetic stirrer. Position the nonmixing reservoir so that the bottom of this vessel is adjacent to and coplanar with the bottom of the mixing reservoir. 5. Provide for liquid flow between the two reservoirs and between the mixing reservoir and the column. 6. Gently mix the liquid in the mixing reservoir. Initiate liquid flow between the two reservoirs and between the mixing reservoir and the column. Use a flow rate of 1 ml/min and collect 1-ml fractions. Resolution of desired and undesired proteins may be improved by narrowing the concentration range of the specific or nonspecific elution reagent, by increasing the total volume of the elution solutions, or by decreasing the flow rate of the elution solutions.

7. Measure the protein concentrations, pool the appropriate fractions, and wash the column (see Basic Protocol 3, steps 4 to 6). TANDEM CHROMATOGRAPHY In tandem chromatography, columns for negative and positive immobilized dye chromatography are sequentially linked. The protein mixture is first passed through the column containing the immobilized dye chosen for negative chromatography, which preferentially removes undesired protein. The effluent from this column is immediately passed over a second column containing the immobilized dye chosen for positive chromatography, which retains the desired protein. The desired protein is then selectively eluted from the second column using either batch or gradient elution and the elution solvent selected for positive chromatography.

BASIC PROTOCOL 4

Materials Protein mixture to be purified Appropriate application solvent Immobilized dyes selected for negative and for positive chromatography (Basic Protocol 1 or Alternate Protocol 1) Selected elution solvent 2 M NaCl 0.02% (w/v) sodium azide Empty chromatographic column with height/diameter ratio >5 (two columns are required if positive chromatography is to be performed using batch elution) Empty chromatographic column with height/diameter ratio ≥5 (if positive chromatography is to be performed using gradient elution) Linear gradient maker (for gradient elution) Fraction collector Pipettors and disposable tips Prepare the protein sample and columns 1. Prepare the protein mixture for chromatography (see Basic Protocol 1, steps 1 to 3). 2. Prepare the negative and positive columns (see Basic Protocol 2, steps 2 and 3). The immobilized dye selected for negative chromatography is placed in a column with a height/diameter ratio ≤5. The immobilized dye selected for positive chromatography is placed in another identical column (if batch elution is to be employed) or in a column with a height/diameter ratio ≥5 (if gradient elution is to be employed). Affinity Purification


3. Position the negative column immediately above the positive column. Connect the two columns so that the effluent of the upper column flows directly on top of the lower column. Pass sample through columns 4. Apply the protein mixture to the upper negative column. Collect the flowthrough liquid from the lower positive column in a single container. 5. Wash the tandem columns with 1 collective column volume of application solvent. Add the flowthrough from the washing to the container in step 4. 6. Measure the concentration of desired protein in the collective flowthrough to ascertain that the desired protein has been retained. 7. Uncouple the upper negative column and set it aside. Recover the protein 8. Wash the positive column with 2 column volumes of application solvent. 9. Subject the positive column to either batch or gradient elution using the elution solvent selected for positive chromatography. Perform and process batch elution (see Basic Protocol 3, steps 3 to 5) or gradient elution (see Alternate Protocol 2, steps 2 to 6). 10. Wash and store each column (see Basic Protocol 2, steps 7 and 8). SUPPORT PROTOCOL

IMMOBILIZATION OF REACTIVE DYES Reactive dyes must be covalently attached to an insoluble porous matrix (“immobilized”) to be used in a chromatographic protocol for protein purification. A reactive dye is a visible chromophore that has a good chemical leaving group, as illustrated in Figure 9.2.1. Immobilization of a reactive dye is a simple procedure that can be performed in the laboratory to generate immobilized dyes that cannot be purchased or to save the cost of commercial immobilization. Table 9.2.2 lists a selection of dyes that can be purchased at modest cost. Each of these dyes can be immobilized by reacting it with a chromatographic matrix in aqueous solvents using moderate conditions. The nature of the covalent attachment is illustrated in Figure 9.2.1B. This protocol is scaled for preparation of an immobilized dye column having a bed volume of 100 ml. Materials Reactive dye (Table 9.2.2) Concentrated solution of KCl Reactive chromatographic matrix: e.g., Sepharose CL-4B or CL-6B (Pharmacia Biotech) or other cross-linked agarose 4 M and 1 M NaCl 10 M NaOH 2 M NH4Cl Large sintered glass funnel Plastic disposable filter with pore size of 0.45 µm


1. Remove any salt, buffer, or surfactant from the reactive dye sample by precipitation of the dye as the potassium salt from aqueous solution and filtration of the precipitate using a sintered glass funnel mounted on a filter flask. Place several grams of the dye in a beaker and add water to make a concentrated solution of the dye. Add concentrated solution of KCl dropwise to precipitate the dye from the solution. Pass the dye


SO3–

A NH

N O

NH

N

NH SO3–

N Cl

SO3–

O

NH2

SO3–

B

NH

N O

NH

N

NH SO3–

N O

SO3–

O

MATRIX

NH2

N

C

CH2O

N

NH

SO3–

N Cl

N N

D

N (CH3)2N+H

Cl

N N

Cl N

NH N

CH3

Cl

Figure 9.2.1 Structures of some reactive dyes. The dyes are arranged so that their triazine rings, the rightmost rings in structures C and D, have a common position. (A) Reactive blue 2, color index 61211, a reactive monochlorotriazine dye, also designated as Cibacron blue 3G-A and Procion blue H-B. (B) Immobilized reactive blue 2, also designated as Affi-Gel blue gel, Blue Sepharose CL-6B, Cibacron blue 3G-A-agarose, and DyeMatrex blue A gel. (C) Reactive red 8, color index 17908, a reactive dichlorotriazine dye, also designated as Procion scarlet MX-G. (D) A positively charged reactive dye.

suspension through a plastic disposable filter with a pore size of 0.45 µm, wash the filtered precipitate with ∼100 ml of water, and air dry the washed precipitate. The most common reactive dyes—e.g., Procion (ICI) and Cibacron (Ciba-Geigy) dyes— have a sulfonated chromophore linked to a chlorotriazine group by an aminoether bridge.

2. Suspend 80 g chromatographic matrix in 280 ml water. 3. Dissolve 1.2 g reactive dye in 80 ml water and add it to the matrix suspension.



Table 9.2.2

Some Commercially Available Reactive Dyesa

Generic name

Color index numberb

Reactive black 5 Reactive blue 2

61211

Reactive blue 4 Reactive blue 5

61205 61210

Reactive blue 15

74459

Reactive blue 19 Reactive blue 114 Reactive blue 160 Reactive brown 10 Reactive green 5 Reactive green 19 Reactive orange 14 Reactive orange 16 Reactive red 4

17757 18105

Reactive red 120 Reactive violet 5 Reactive yellow 2

18097 18972

Reactive yellow 3 Reactive yellow 81 Reactive yellow 86

13245

Some commercial names

Suppliers

Remazol black B Cibacron blue 3G-A Procion blue H-B Procion blue MX-R Cibacron brilliant blue BR-P Procion blue H-GR Cibacron turquoise blue GF-P Procion turquoise H-GF Remazol brilliant blue R Drimarene brilliant blue K-BL Procion blue HE-RD Procion brown MX-5BR Cibacron brilliant green 4G-A Procion green H-4G Procion green HE-4BD Procion yellow MX-4R Remazol brilliant orange 3R Cibacron brilliant red 3B-A Procion red H-7B Cibacron brilliant red 4G-E Procion red HE-3B Remazol brilliant violet 5R Cibacron brilliant yellow 3G-P Procion yellow H-5G Procion yellow H-A Procion yellow HE-3G Procion yellow M-8G

Aldrich, ICN Biomedicals, Sigma Aldrich, ICN Biomedicals Sigma, Spectrum Aldrich, ICN Biomedicals, Sigma ICN Biomedicals, Sigma Aldrich, ICN Biomedicals, Sigma ICN Biomedicals Sigma Sigma ICN Biomedicals, Sigma ICN Biomedicals, Sigma ICN Biomedicals, Sigma ICN Biomedicals, Sigma Aldrich Aldrich, ICN Biomedicals Sigma, Spectrum ICN Biomedicals, Sigma ICN Biomedicals, Sigma Aldrich, ICN Biomedicals Sigma, Spectrum Aldrich ICN Biomedicals, Sigma

aDichlorotriazine reactive dyes are denoted as Procion MX dyes; all the remaining dyes are monochlorotriazine dyes. bNumbers are taken from the Colour Index, The Society of Dyers and Colourists and the American Association of Textile Chemists and Colorists (1971).

4. Add 40 ml of 4 M NaCl to the matrix suspension. 5. If a monochlorotriazine reactive dye is used, add 4 ml of 10 M NaOH to the matrix suspension and gently stir it with a magnetic stirrer and stir bar for 72 hr at ambient temperature or 16 hr at 55° to 60°C. If a dichlorotriazine reactive dye is used, add 0.5 ml of 10 M NaOH and gently stir the matrix suspension with a magnetic stirrer and stir bar for 4 hr at ambient temperature. 6. Filter the suspension using the sintered glass funnel and wash the solid material on the filter with copious quantities of water, then 1 M NaCl, and then water again until the filtrate is clear. 7. Suspend the material on the filter in 2 M NH4Cl, pH 8.5, and gently stir with a magnetic stirrer and stir bar for 4 hr at ambient temperature. Dye Affinity Chromatography

This step eliminates any remaining chloro groups.


8. Filter and wash the suspension as described in step 6. 9. Either store the immobilized dye as a suspension in water at 4°C or dry the immobilized dye on a sintered glass filter and store as a dry powder. COMMENTARY Background Information The surfaces of all proteins bind at least one biochemical with high affinity. Accordingly, immobilization of a biochemical on a chromatographic support should result in the retention of all proteins that have a binding site for that biochemical. The biochemical can exist in two forms: the mobile biochemical, which moves about freely in solution, and the immobilized biochemical, which is covalently attached to a chromatographic support and cannot move about freely. In chromatography, a competition exists between mobile and immobilized forms of the biochemical for the binding sites on the proteins. If the immobilized form is the dominant concentration of the biochemical, as in application to a column, the protein exists principally as a protein/immobilized dye complex and is not free to move in solution. If the mobile form is the dominant concentration of the biochemical, as in elution, the protein exists principally as a protein/mobile dye complex and moves freely in solution. Elution solutions containing a concentration gradient of the mobile form of the biochemical can often be used to distinguish among those proteins that bind the immobilized biochemical. Affinity chromatography, however, can present problems. The biochemical must be immobilized in such a manner that the protein can bind it with high affinity, often a substantial challenge. Each immobilized biochemical can bind only a limited number of different proteins, requiring construction of a large number of different affinity columns for purification of a series of proteins. Further, the degradative enzymes present in crude cellular extracts often cleave immobilized biochemicals into useless remnants, necessitating either the frequent replacement of affinity columns or the postponement of their use until the end of a purification scheme. Such postponement compromises the selective power of affinity chromatography, which should, in principle, purify a protein to homogeneity in a single step. Fortunately, immobilized dye affinity chromatography circumvents these problems. A single reactive dye can effectively bind a large number of different biochemicals. A reactive

dye is sufficiently complex that immobilization does not reduce its affinity. An immobilized reactive dye is not degraded by enzymes commonly found in crude protein mixtures, as the chemistry of the dye is quite distinct from that of typical biochemicals. An immobilized dye functions as a stable general affinity molecule that can be made selective by the nature and concentration of the mobile biochemical chosen as the elution reagent. A single immobilized dye, variously known as Cibacron blue (F)3G-A, Procion blue H-B, reactive blue 2, blue dextran, Affi-Gel blue, or DyeMatrex blue A, has contributed significantly to the purification of over 60 different proteins. Some of these proteins have been purified to homogeneity from a crude bacterial extract in a single step providing a several-thousand-fold purification with >80% yield. Immobilized dyes as a group have contributed significantly to the purification of several hundred different proteins.

Critical Parameters and Troubleshooting Several aspects of dye affinity chromatography are critical. Foremost is the ionic strength of the application solvent. Because application solvents of ionic strength >50 mM weaken the electrostatic interactions that contribute to the binding of desired proteins by immobilized dyes, the concentrations of ionic components in the application solvent should be kept to a minimum. Second is the presence of nonionic detergents in the application solvent. Nonionic detergents are frequently used to solubilize intrinsic protein located in subcellular organelles such as membranes. These nonionic detergents form micelles that encapsulate immobilized dyes, making them inaccessible to proteins. Addition of a modest concentration of a negatively charged detergent, such as SDS or deoxycholate, to an application solvent containing a nonionic detergent results in the formation of mixed micelles. The negatively charged immobilized dyes repel the negatively charged mixed micelles in solvents of modest ionic strength, preserving the accessibility of immo-



Supplement 4


bilized dyes to desired proteins (Robinson et al., 1980). The nonionic and ionic detergents should have a concentration ratio of ∼5:1. Third is the sign of the electrostatic charges on immobilized dyes. Most reactive dyes are negatively charged; this is normally an advantage, as most protein binding sites contain positively charged amino acid residues designed to bind negatively charged biochemicals, such as cAMP, glucose 6-phosphate, or oligo(dAT). However, some biochemicals have a positive charge; the proteins to which they bind must then contain negatively charged residues. These binding proteins are electrostatically repelled by negatively charged immobilized dyes in solvents of modest ionic strength. The immobilized form of a positively charged dye (Fig. 9.2.1D) retains proteins having binding sites for positively charged biochemicals, such as trypsin, thrombin, and carboxypeptidase (Clonis et al., 1987). This immobilized dye is particularly useful in removing contaminating proteases from cellular extracts by negative chromatography. The two primary problems encountered in immobilized dye affinity chromatography are the inability to retain a desired protein and the inability to elute a desired protein that has been retained. The inability to retain a desired protein may result from a number of situations. If a fresh immobilized dye column is being used, the following may pertain: The ionic strength of the protein mixture may be too high. Either dilute the protein mixture with distilled water or dialyze it against a low ionic strength application solvent, such as 50 mM Tris⋅Cl devoid of neutral salts. The concentration in the protein mixture of the biochemical that binds to the desired protein may be too high. Either dilute the mixture with distilled water or, if the biochemical is sufficiently small, dialyze the mixture against a low ionic strength application solvent. Retention of the desired protein may necessitate the presence of a metallic cation, such as Mg2+ or Zn2+. Add a minimal concentration of one or more likely metallic cations to the application solvent. A nonionic detergent may be present in the protein mixture. Add a sufficient amount of a negatively charged detergent, such as SDS or deoxycholate, to the crude protein mixture to maintain the accessibility of the immobilized dye. However, note that SDS, even in low concentrations, denatures many proteins. The immobilized dye may have a relatively weak affinity for the desired protein. Add more

immobilized dye to the column or screen a series of alternative immobilized dyes for one having a stronger affinity for the desired protein. The desired protein is designed to bind only positively charged biochemicals. Purchase or synthesize an immobilized dye having a net positive charge. The desired protein may be among a small group of proteins that is not retained by any immobilized dye. Negative chromatography can then be used to great advantage, particularly a series of negative columns in tandem. If an old immobilized dye column is being used, the following may pertain: Insufficient immobilized dye may be accessible. After repeated use of an immobilized dye column the dye may become bound with proteins, detergents, and other components in cellular extracts that are not removed by washing with 2 M NaCl. An immobilized dye column should be periodically washed with several column volumes of one or more of the following solutions to displace these dye-bound components: a strong denaturant such as 8 M urea, 6 M guanidine⋅HCl, or 1% SDS; a potent lyotropic reagent such as 3 M KSCN; a strong base such as 0.5 M NaOH; or a mixed solvent such as chloroform/methanol. Insufficient immobilized dye may be present. If an immobilized dye column has been used extensively or stored in solution, the glycosidic bonds linking the sugars in the agarose matrix will be hydrolyzed. Such hydrolysis results in loss of the immobilized dye. A new immobilized dye column is strongly colored. A depleted immobilized dye column is lightly colored and should be replaced. The inability to elute a desired protein that has been retained on an immobilized dye column presents a different problem, in that the affinity is too strong rather than too weak. Suggestions for weakening the affinity include the following: Performing the chromatography at a different pH. Most proteins exhibit a pH optimum for binding a biochemical. This pH optimum results in part from changes in the ionic character of some of the amino acid side chains in the protein that are directly involved in binding. Changes in pH that weaken the affinity of biochemicals are likely to weaken the affinity for the immobilized dye as well. Use Basic Protocol 1 to determine the effect of pH on the elution of the desired protein. Note, however, that pH values outside the range 4 to 11 may result in irreversible inactivation.

9.2.14 Supplement 4


Performing the chromatography at a different temperature. The binding of a biochemical by a protein involves a combination of weak noncovalent bonds including hydrogen bonding, hydrophobic interactions, and electrostatic interactions. Changes in temperature weaken some of these bonds and strengthen others. Accordingly, changing the temperature may weaken the affinity of the immobilized dye for the desired protein. Note, however, that the temperature should not be increased above 50°C to prevent thermal inactivation of the desired protein. Increasing the potency of the elution solvent. Nonspecific elution solvents promote elution by weakening the electrostatic and hydrophobic contributions to affinity. The electrostatic contributions will usually be minimized in a 0.2 M solution of any neutral salt. How-ever, minimization of the hydrophobic contributions requires higher concentrations and judicious choice of the neutral salt. The lyotropic or Hofmeister series reflects the relative potency of neutral salts. A 3 M solution of the neutral salt KSCN is among the most potent for weakening hydrophobic interactions (Robinson et al., 1981). Specific elution solvents, by contrast, weaken affinity by competing with the immobilized dye for the desired protein. The higher the concentration of the biochemical and the stronger its affinity for the desired protein, the greater its effectiveness as an eluant. Accordingly, the concentration of a specific elution solvent should be increased in an effort to elute the desired protein. Clearly, a limiting concentration will be set by the solubility, availability, and cost of the specific eluant. Alternatively, other biochemicals known to bind to the desired protein as substrates, coenzymes, cofactors, competitive inhibitors, and allosteric effectors may be surveyed for their efficacy as elution reagents. Adding a chelation agent. Metallic cations often function as obligatory cofactors in the binding of biochemicals to proteins. Accordingly, the affinity of the desired protein for the immobilized dye may be weakened by the presence of a chelation agent such as 0.1 M EDTA or EGTA. Combining specific elution reagents. Immobilized dyes are sufficiently large that a single dye molecule may occupy adjacent biochemical binding sites on a protein: e.g., a dye may occupy the substrate site and the coenzyme site on an enzyme surface. Alternatively, two different immobilized dye molecules may simul-

taneously bind to two remote sites on the same protein molecule: e.g., different immobilized dye molecules may simultaneously bind to the substrate site and to the allosteric site on an enzyme. In each of these situations, the elution of a desired protein can be accomplished using a combination of specific elution reagents. Adding a protein denaturant such as 6 M guanidine⋅HCl. Most proteins can be eluted from an immobilized dye column using a denaturant that unfolds protein binding sites. Because most retained proteins have already been eluted, addition of a denaturant should result in a substantial-fold purification, provided the desired protein is denatured reversibly. Such reversibility is commonly achieved by dilution or dialysis of the column eluate with application solvent. Lowering the density of the immobilized dye. Many desired proteins are oligomers containing two or more identical binding sites. If the chromatographic matrix contains a high density of accessible immobilized dye molecules, it is likely that two or more immobilized dye molecules will simultaneously bind to a single desired oligomeric protein (Hogg and Winzor, 1985). Such multiple binding makes elution more difficult. Diluting an immobilized dye matrix with an unmodified matrix will not decrease the density of the immobilized dyes on each chromatographic bead. The only remedy is to obtain the same immobilized dye from an alternative supplier, hoping to obtain a less dense sample, or to immobilize the dye in the laboratory using the Support Protocol. Selecting an alternative immobilized dye that has a weaker affinity for the desired protein. Such weakening should facilitate improved results in positive chromatography.

Anticipated Results The yield of the desired protein following immobilized dye chromatography is commonly >80%. Random differences in yield occur depending upon how completely the desired protein is retained (positive chromatography) or not retained (negative chromatography) and how completely the retained protein is eluted. Lesser yields are usually the result of proteolysis either before or during chromatography. Proteolysis may generate remnants of the desired protein that have weaker affinities for an immobilized dye. Such remnants elute prior to the elution of the dominant unproteolyzed desired protein and would not be included in the pooled fractions. Good yields of desired proteins result in part from the speed of affinity



chromatography, the temperature, the stabilization provided to the desired protein from binding an immobilized dye, and the respectful treatment of a crude protein mixture by an experienced investigator. The fold purification achieved by immobilized dye chromatography is largely controlled by the relative population of the desired protein in a protein mixture. If 10% of the mixture is the desired protein, then a 10-fold purification step is maximal. If 1000-fold purification step is feasible and has been observed. However, in most instances 10- to 100-fold purifications are observed.

Time Considerations

The Society of Dyers and Colourists and the American Association of Textile Chemists and Colorists. 1971. Colour Index, 3rd edition. The Society of Dyers and Colourists, Bradford, England, and the American Association of Textile Chemists and Colorists, Research Triangle Park, NC.

Key References Clonis, Y.D., Atkinson, T., Bruton, C.J., and Lowe, C.R. 1987. Reactive Dyes in Protein and Enzyme Technology. Macmillan, New York. A comprehensive review of the preparation, immobilization, and utilization of reactive dyes for protein purification. Lowe, C.R., Burton, S.J., Burton, N., Stewart, D.J., Purvis, D.R., Pitfield, I., and Eapen, S. 1990. New developments in affinity chromatography. J. Mol. Recognit. 3:117-122.

The preliminary survey can be completed in a few hours or a day, depending on its breadth. Once a protocol is selected, negative chromatography should be completed in 1 hr and positive chromatography in a few hours. These time estimates greatly depend on the volume of the solutions to be processed and the flow rates employed.

A lively discussion of emerging developments.

Literature Cited

An alternative discussion of the topic with more emphasis on strategy and less on detail.

Clonis, Y.D., Stead, C.V., and Lowe, C.R. 1987. Novel cationic triazine dyes in protein purification. Biotechnol. Bioeng. 30:621-627. Hogg, P.J. and Winzor, D.J. 1985. Effects of solute multivalency in quantitative affinity chromatography: Evidence for cooperative binding of horse liver alcohol dehydrogenase to blue Sepharose. Arch. Biochem. Biophys. 240:70-76. Robinson, J.B., Jr., Strottmann, J.M., Wick, D.G., and Stellwagen, E. 1980. Affinity chromatography in nonionic detergent solutions. Proc. Natl. Acad. Sci. U.S.A. 77:5847-5851. Robinson, J.B., Jr., Strottmann, J.M., and Stellwagen, E. 1981. Prediction of neutral salt elution profiles for affinity chromatography, Proc. Natl. Acad. Sci. U.S.A. 78:2287-2291.

Scopes, R.K. 1994. Protein Purification, 3rd ed. Springer-Verlag, New York. A useful discussion of immobilized dye affinity chromatography by the advocate of tandem chromatography. Stellwagen, E. 1990. Chromatography on immobilized reactive dyes. Methods Enzymol. 182:343357.

Stellwagen, E. 1993. Affinity chromatography with immobilized dyes. In Molecular Interactions in Bioseparations (T.T. Ngo, ed.) pp. 247-255. Plenum Press, New York. A detailed review of the interaction of reactive blue 2 with a typical oligomeric protein, alcohol dehydrogenase.

Contributed by Earle Stellwagen University of Iowa Iowa City, Iowa



Affinity Purification of Natural Ligands

UNIT 9.3

Immobilization of proteins, nucleic acids, and other “bioligands” is not simple. The proper bioligand needs to be selected for the specific application of the product. This decision influences all others: the matrix is chosen and activated by the method that is most appropriate for this specific application and the ligand is coupled under conditions dictated by the activation method and the nature of the ligand (e.g., is it a labile protein or a sturdy enzyme cofactor?). There are many matrices and activation and coupling methods, and new ones are constantly being developed. The most frequently used are hydrophilic matrices, which exhibit the necessary low nonspecific binding, and coupling systems that react primarily with nucleophilic residues, such as lysyl amine groups in proteins. Orientation of the resulting biomolecule is extremely important, although these effects have been only superficially investigated. Immobilized IgG is only twofold more efficient when immobilized solely by the Fc portion (Murayama et al., 1978; Doman et al., 1990; Hermanson et al., 1992), whereas the interactions of large proteins that bind macromolecular ligands (lysozyme is a good example) are changed by several orders of magnitude depending on the orientation of the molecule (Fig. 9.3.1). For this reason, it is fortunate that both cationic (e.g., fluoromethyl pyridinium) and hydrophobic (e.g., Emphaze, tosyl agarose) activation methods are available. The choice of matrix and the stability of the bound ligand determine the choice of activation and coupling methods. Preactivated matrices are available but are very costly. Therefore they are best used when small amounts of ligand are to be immobilized for “test” columns and methods development. Larger projects and large-scale applications require ligand activation in the laboratory. Because of their frequent use in the past, agarose gel is often the matrix of choice and cyanogen bromide is often the activation and coupling agent of choice. However, depending on the procedure, neither may be optimal and before beginning a project it is best to explore the range of matrices and methods carefully, particularly if later industrial level scale-up with regulatory implications is anticipated. An overview of the major activation chemistries is given in Tables 9.3.1 and 9.3.2. Of the many activation systems available, three of the most frequently used are

A

B

S

L

L

S

Figure 9.3.1 Binding of small (S) versus large (L) substrates to an oriented (A) and randomly (B) immobilized enzyme. Arrows signify the movement of the enzyme to the ligand and its subsequent binding to the ligand. Reproduced from Voivodov et al. (1993) with permission of Hüthig & Wepf Publishers, Zug, Switzerland.


Contributed by William H. Scouten

9.3.1

Current Protocols in Protein Science (1995) 9.3.1-9.3.15 Copyright © 2000 by John Wiley & Sons, Inc.

CPPS

Table 9.3.1

Selected Reagents to Activate Matrix Hydroxyl Functionsa

Group that reacts

Reagent toxicity

Activation Type of bond time (hr)

Tresyl chloride, sulfonyl Excellent chlorides

Thiols, amines

Low

0.1-1.0

Secondary amine

Cyanogen bromide

Poor

Amines

High

0.1-0.4

Bisoxiranes (epoxides)

Excellent

Thiols

Moderate

5-18

Epichlorohydrin

Excellent

Moderate

2-24

High

0.5-2

Isourea or imidocarbonate Secondary amine Secondary amine Triazine ether

Activation method

Stability of bond

References Lawson et al. (1983); Gribnau (1977) Axen et al. (1967) Axen et al. (1967)

Low High

1-2 0.5-2

Anilinyl Secondary amine

Axen et al. (1967) Axen et al. (1967)

Glutaraldehyde

Thiols, amines Good Amines, thiols, hydroxyls Good Amines Poor in base, Amines fair at pH 1 day, wash with 10 bed volumes of 20% ethanol and add 1 bed volume of 20% ethanol prior to storage. Keep column sealed to prevent evaporation.

Prepare the extract IMPORTANT NOTE: Beginning with this step, all procedures should be performed on ice or in a cold room unless otherwise indicated. 9. Thaw cell pellet (from step 4) on ice. Add 5 ml MCAC-0 buffer and 33 µl of 150× protease inhibitor cocktail and resuspend by pipetting, sonication, or homogenization. All MCAC buffers contain phenylmethylsulfonyl fluoride (PMSF) as a protease inhibitor; to reduce expense, protease inhibitor cocktail is added only to the crude extract.

10. Add 0.05 ml of 10% (v/v) Triton X-100 (0.1% final). Mix thoroughly and subject the sample to 3 cycles of freezing at −70°C and thawing on ice. Ionic detergents may interfere with binding of the protein to the resin and should not be used. Cell lysis is evidenced by a visible increase in viscosity. The pLysS plasmid in these cells encodes an endogenous lysozyme that eliminates the need for exogenous lysozyme treatment to disrupt the bacterial cell wall.

11. Add 0.05 ml of 1 M MgCl2 (final concentration 10 mM) and 0.05 ml of DNase I solution (final concentration 10 µg/ml DNase I). Mix gently and incubate 10 min at room temperature. Metal-Chelate Affinity Chromatography

The DNase I treatment reduces the viscosity of the lysate.

9.4.4 Supplement 4


12. Centrifuge 15 min at 27,000 × g (15,000 rpm in JA-20 rotor), 4°C. Decant the supernatant into a clean container on ice and discard the pellet. Set aside and freeze a 10-µl aliquot at −70°C for later analysis by SDS-PAGE (see Support Protocol 1). The supernatant can be frozen at −70°C indefinitely before continuing with the procedure.

Purify protein 13. If extract is frozen, thaw on ice. Load onto Ni2+-NTA column and allow to flow through at a rate of 10 to 15 ml/hr. Collect column flowthrough and save for SDS-PAGE (see Support Protocol 1). Charged NTA resin has a capacity of 5 to 10 mg histidine-tagged protein per milliliter of packed resin. The amount of extract that can be loaded on the column will depend on the amount of soluble histidine-tagged protein in the extract.

14. Wash column with 5 ml MCAC-0 buffer at a flow rate of 20 to 30 ml/hr. Discard flowthrough. 15. Wash column in stepwise fashion with 5 ml each of MCAC-20, MCAC-40, MCAC60, MCAC-80, MCAC-100, MCAC-200, and MCAC-1000 buffers at a flow rate of 10 to 15 ml/hr. Collect 0.5-ml fractions and save on ice for SDS-PAGE (see Support Protocol 1). Alternatively, the column can be eluted with a 5-ml linear gradient of 0 to 400 mM imidazole in MCAC buffer. The second and third fractions of each wash will contain most of the eluted proteins. Most proteins with hexahistidine tails will remain bound in 60 mM imidazole (MCAC-60) and elute with 100 to 200 mM imidazole (MCAC-100 or -200); therefore, the purified protein will elute in MCAC-100 or -200. Proteins with longer histidine tails (e.g., 10 residues) bind to Ni2+-NTA with greater affinity and require higher imidazole concentrations for elution. However, optimum washing and elution conditions must be determined for each protein. Once optimum washing and elution conditions are established, it is possible to prepare the crude extract in a buffer that contains the highest imidazole concentration in which the histidine tail remains bound to the Ni2+-NTA (e.g., MCAC-40 or -60 buffer). This decreases nonspecific binding of proteins to resin and permits use of a single buffer for extract preparation, column loading, and column washing.

16. Elute column with 1 ml MCAC-EDTA buffer at a flow rate of 10 to 15 ml/hr, collecting 0.5-ml fractions. The blue-green color of the column will disappear as nickel is removed by EDTA. More tightly bound proteins may be found in these fractions. The resin can now be recharged (repeat steps 6 to 8) and the column reused. If protein is eluted adequately with imidazole (as determined by overall yield for subsequent preparations), EDTA washing can be omitted. The column can be reequilibrated with MCAC-0 buffer and the purification repeated. The same column can be used three to five times before EDTA stripping and nickel recharging are necessary. Only one protein should be purified on any given column.

17. Analyze fractions for the presence of eluted protein. An ultraviolet (280-nm) absorbance flow monitor is helpful for following column elution but is not necessary. An alternative is to measure the OD280 of individual fractions to identify protein-containing fractions. However, imidazole will also absorb at 280 nm. A quick and easy method to determine which fractions contain eluted protein is to place 2 ìl undiluted Protein Assay Dye Reagent Concentrate (Bio-Rad) on a piece of Parafilm, add 8 ìl from fraction to be tested, and mix by pipetting up and down. Immediate appearance of blue color indicates that the fraction contains protein. This does not work in the presence



Supplement 4

of Triton X-100 because the detergent itself produces an intense blue color; for this reason, Triton X-100 is excluded from the washing and elution buffers.

18. Combine the fractions containing eluted protein and remove a 10-µl aliquot for SDS-PAGE (see Support Protocol 1). Freeze the remainder in smaller aliquots at −70°C or in liquid nitrogen. If a different buffer for the protein is desired (e.g., for proteolytic removal of the histidine tail), the protein should be dialyzed against the buffer of choice to remove the MCAC buffer prior to storage at −70°C or in liquid nitrogen.

19. If time permits, proceed immediately to analysis of fractions by SDS-PAGE and processing of protein (see Support Protocol 1). Otherwise, freeze all samples at −70°C until ready for analysis and processing. ALTERNATE PROTOCOL 1

DENATURING MCAC FOR PURIFICATION OF INSOLUBLE HISTIDINE-TAIL FUSION PROTEINS High-level expression of foreign proteins in bacteria and other cells frequently results in poor solubility of the expressed protein (see UNITS 5.1 & 5.2). These insoluble proteins form inclusion bodies in bacteria, and strong chaotropic agents such as guanidine, urea, or SDS are usually required to solubilize them. These agents denature the protein and destroy the secondary structure that is essential to other affinity purification methods (e.g., maltosebinding protein or glutathione-S-transferase fusion proteins). A significant advantage of metal-chelate affinity chromatography is that the oligohistidine tail will bind to the Ni2+-NTA resin even when the protein is denatured. In denaturing MCAC, the protein extract is solubilized with 6 M guanidine and the entire affinity purification procedure is carried out in guanidine. The purified, denatured protein is renatured during dialysis. Additional Materials (also see Basic Protocol) GuMCAC-0, GuMCAC-20, GuMCAC-40, GuMCAC-60, GuMCAC-100, and GuMCAC-500 buffers (see recipe) GuMCAC-EDTA buffer (see recipe) Appropriate final buffer for protein (e.g., for proteolytic cleavage or long-term storage) Guanidine⋅HCl Additional reagents and equipment for analysis and processing of purified proteins (see Support Protocol 1) and dialysis (APPENDIX 3B & UNIT 4.4) Express the protein 1. Prepare the pellet of E. coli expressing a histidine-tail fusion protein (see Basic Protocol, steps 1 to 4). The pellet can be stored indefinitely at −70°C before proceeding. Alternatively, extract preparation (steps 5 and 6) can be carried out immediately and the column prepared during the centrifugation at step 6.

Prepare the affinity column 2. Prepare column (see Basic Protocol, steps 5 to 7). 3. Wash column with 2 ml GuMCAC-0 buffer. During this and subsequent column washes, liquid should be allowed to drain to top of packed resin bed and resin should not be allowed to dry. Metal-Chelate Affinity Chromatography

9.4.6 Supplement 4


Prepare cell extract 4. Thaw cell pellet (from step 1) on ice. Resuspend in 5 ml GuMCAC-0 buffer by pipetting, sonication, or homogenization. 5. Freeze 10 min at −70°C and thaw at room temperature. Protease inhibitors are omitted because proteases are inactivated by guanidine. Triton X-100 is not needed at this step. Freezing is not necessary but is included because it ensures complete lysis of cells. Subsequent steps can be performed at room temperature. However, if solid-phase renaturation is used (see Alternate Protocol 2), it is better to maintain lower temperatures throughout the process.

6. Gently mix samples for 30 min using a rocker, rotating mixer, or magnetic stirrer. Centrifuge 15 min at 27,000 × g (15,000 rpm in Beckman JA-20 rotor), 4°C. Decant supernatant into a clean container and discard pellet. Set aside a 10-µl aliquot for analysis by SDS-PAGE (see Support Protocol 1). The supernatant can be frozen at −70°C indefinitely before continuing with the procedure.

Purify protein 7. If extract from step 5 is frozen, thaw at room temperature. Load onto Ni2+-NTA column and allow to flow through at a rate of 10 to 15 ml/hr. Collect flowthrough and save a 10-µl aliquot for SDS-PAGE (see Support Protocol 1). 8. Wash column with 5 ml GuMCAC-0 buffer at a rate of 20 to 30 ml/hr. Discard the flowthrough. 9. Wash column in stepwise fashion with 5 ml GuMCAC-20, -40, -60, -100, and -500 buffers at a rate of 10 to 15 ml/hr. Collect 0.5-ml fractions and save for SDS-PAGE (see Support Protocol 1). The second and third fractions from each wash will contain most of the unbound protein. The histidine tail binds slightly less avidly under denaturing conditions. Lower imidazole concentrations are therefore required for washing and elution than in the Basic Protocol.

10. Elute with 1 ml GuMCAC-EDTA buffer at a rate of 10 to 15 ml/hr, collecting 0.5-ml fractions. 11. Identify fractions containing the protein, pool together, transfer to dialysis tubing, and seal. Alternatively, fractions can be frozen at −70°C indefinitely before continuing with the procedure. Guanidine precipitates in the presence of SDS and must be removed by dialysis before SDS-PAGE. An alternative technique employs buffers that switch from 6 M guanidine to 8 M urea during affinity column washing (Stüber et al., 1990). This permits samples to be taken directly from urea fractions without dialysis and analyzed by SDS-PAGE or injected into animals for antibody production.

Renature purified protein by dialysis 12. Prepare appropriate final buffer for protein (e.g., for proteolytic cleavage or long-term storage) and add sufficient guanidine to bring final concentration to 4 M. 13. Dialyze purified protein from step 11 for ≥2 hr at 4°C against 500 ml buffer/4 M guanidine (see APPENDIX 3B). The MWCO of the dialysis membrane should be chosen to be smaller than the MW of the purified protein. In most cases an MWCO of 12 to 14 kDa is sufficient. Affinity Purification


Supplement 4

14. Remove 250 ml buffer/guanidine and add 250 ml buffer without guanidine. Continue dialysis ≥2 hr. Repeat. With some proteins, renaturation by dialysis may require longer dialysis periods and more gradual decrements in the guanidine concentration of the buffer. Conditions for each protein must be determined empirically.

15. Remove dialysis bag to a container containing 500 ml of fresh buffer without guanidine at 4°C. Continue dialysis 2 hr to overnight. 16. Remove sample from dialysis bag, divide into aliquots, and freeze at −70°C or in liquid nitrogen. If protein precipitates during dialysis, solid-phase renaturation (see Alternate Protocol 2), in which protein bound to the column is renatured before elution, should be employed.

17. Analyze fractions and process protein (see Support Protocol 1). ALTERNATE PROTOCOL 2

SOLID-PHASE RENATURATION OF MCAC-PURIFIED PROTEINS In Alternate Protocol 1, removal of denaturants by dialysis will occasionally lead to precipitation of protein, possibly due in part to entanglement or aggregation of separate protein molecules as they refold. To avoid this problem, solid-phase renaturation may be attempted. In this procedure, protein extract is prepared and bound to the column under denaturing conditions. A series of washes removes the denaturing agent before the target protein is eluted and the resulting renatured protein is eluted from the column under native conditions. Additional Materials (also see Basic Protocol) 1:1 (v/v) MCAC-20/GuMCAC-20 buffer (see recipes) 3:1 (v/v) MCAC-20/GuMCAC-20 buffer (see recipes) 7:1 (v/v) MCAC-20/GuMCAC-20 buffer (see recipes) 1. Prepare protein extract, bind to column, and wash with GuMCAC buffers (see Alternate Protocol 1, steps 1 to 9). 2. Wash column with 5 ml of 1:1 (v/v) MCAC-20/GuMCAC-20 buffer. During this and subsequent washes, liquid should be allowed to drain just to top of packed resin bed and resin should not be allowed to dry.

3. Wash column with 5 ml of 3:1 (v/v) MCAC-20/GuMCAC-20 buffer. 4. Wash column with 5 ml of 7:1 (v/v) MCAC-20/GuMCAC-20 buffer. 5. Wash column with MCAC buffers, elute proteins, and analyze (see Basic Protocol, steps 15 to 19). Slow elution (between 1 and 2 hr) with a 5-ml linear gradient from 100% GuMCAC-20 buffer to 100% MCAC-20 buffer may also yield efficient renaturation.

Metal-Chelate Affinity Chromatography

9.4.8 Supplement 4


ANALYSIS AND PROCESSING OF PURIFIED PROTEINS The success of the purification scheme (particularly during a small pilot study) should be monitored at each stage by SDS-PAGE. If the fusion protein contains a specific protease cleavage site, the histidine tail can be removed using an appropriate proteolytic procedure, if desired.

SUPPORT PROTOCOL 1

Materials Fractions from MCAC column purification (crude extract, flowthroughs, and purified protein; see Basic Protocol or Alternate Protocols 1 or 2) 2× SDS sample buffer (UNIT 10.1) MCAC-0 buffer (see recipe) Additional reagents and equipment for one-dimensional SDS-PAGE (UNIT 10.1), cleavage of proteins with factor Xa or thrombin (UNIT 6.5), and dialysis (UNIT 4.4 & APPENDIX 3B) 1. Thaw aliquots of fractions to be analyzed on ice. Mix 5 µl from crude extract and crude flowthrough fractions and 10 µl from the second and third fractions from each washing step with an equal volume of 2× SDS sample buffer. 2. Load samples onto a standard SDS-PAGE gel. Run gel and visualize to identify the fractions containing purified protein (UNIT 10.1). Guanidine must be removed by dialysis prior to addition of SDS sample buffer.

3. Thaw the remaining aliquots of fractions containing purified protein, dialyze against the appropriate proteolysis buffer, and carry out cleavage procedure if desired. If necessary, after cleavage dialyze the protein against an appropriate storage buffer and freeze in aliquots. The size of the cleaved histidine tail will generally be 50 sequence-specific proteins that have been purified using the affinity chromatography method described herein.



Supplement 11

Kadonaga, J.T. and Tjian, R. 1986. Affinity purification of sequence-specific DNA binding proteins. Proc. Natl. Acad. Sci. U.S.A. 83:58895893. First paper to describe affinity chromatography with multimerized oligonucleotides; details purification of transcription factor Sp1.

Contributed by Leslie A. Kerrigan Osiris Therapeutics Baltimore, Maryland James T. Kadonaga University of California San Diego La Jolla, California

Affinity Chromatography of DNA-Binding Proteins



Purification of DNA-Binding Proteins Using Biotin/Streptavidin Affinity Systems

UNIT 9.7

Short fragments of DNA—either natural or formed from oligonucleotides—containing a high-affinity site for a DNA-binding protein provide a powerful tool for purification. The biotin/streptavidin purification system is based on the tight and essentially irreversible complex that biotin forms with streptavidin. The experimental design of this system is illustrated in Figure 9.7.1. First, a DNA fragment is prepared that contains a high-affinity binding site for the protein of interest. A molecule of biotinylated nucleotide is incorporated into one of the ends of the DNA fragment. The protein of interest is allowed to bind to the high-affinity recognition site present in the biotinylated fragment. The tetrameric protein streptavidin is then bound to the biotinylated end of the DNA fragment. Next, the protein/biotinylated fragment/streptavidin ternary complex is efficiently removed by adsorption onto a biotin-containing resin. Since streptavidin is multivalent, it is able to serve as a bridge between the biotinylated DNA fragment and the biotin-containing resin. Proteins remaining in the supernatant are washed away under conditions that maximize the stability of the DNA-protein complex. Finally, the protein of interest is eluted from the resin with a high-salt buffer.

binding site for protein X DNA end-labeled with biotin-11-dUTP

+

protein X

+ streptavidin tetramer

+ biotin cellulose biotin-cellulose protein X

high-salt wash

Figure 9.7.1 Purification of DNA-binding proteins using the biotin/streptavidin affinity technique. The protocol involves the following steps: (1) a biotinylated, labeled DNA fragment is prepared containing a binding site for the protein to be purified; (2) the biotin-cellulose resin is prepared; (3) a binding reaction containing a crude protein fraction and the biotinylated probe is set up; (4) free streptavidin is added to the binding reaction; (5) the protein/biotinylated DNA fragment/streptavidin complex is bound to the biotin-cellulose resin; (6) unbound protein is removed by extensive resin washing; and (7) the protein is eluted from the resin with high-ionic-strength buffer. Each of these steps can be monitored and optimized in solution, using a mobility shift DNA-binding assay (see Support Protocol and Buratowski and Chodosh, 1996). Contributed by Lewis A. Chodosh and Stephen Buratowski Current Protocols in Protein Science (1998) 9.7.1-9.7.13 Copyright © 1998 by John Wiley & Sons, Inc.


9.7.1 Supplement 12

BASIC PROTOCOL

PURIFICATION USING BATCH METHOD Materials Plasmid DNA with binding site for the protein of interest Appropriate restriction endonucleases Biotin-11-dUTP (Life Technologies) Labeled and unlabeled dNTPs Klenow fragment of E. coli DNA polymerase I TE buffer (APPENDIX 2E) Biotin-cellulose (Pierce) Biotin-cellulose binding buffer (see recipe) BSA Bulk carrier DNA [e.g., poly(dI-dC)⋅poly(dI-dC), salmon sperm DNA, or E. coli DNA] Biotin-cellulose elution buffer (see recipe) Protein solution Streptavidin (Celltech; may be stored as 5 mg/ml stock for at least 2 months) DEAE membrane (Schleicher & Schuell NA45) 0.025-µm filter discs (Millipore VS) Additional reagents and equipment for mobility-shift DNA binding assay (see Support Protocol) Prepare biotinylated, labeled DNA fragment 1. Digest 50 µg of plasmid DNA in 100 µl with one or more appropriate restriction endonucleases to obtain DNA probe. 2. Add the following to the probe mixture: Biotin-11-dUTP to a final concentration of 20 µM Radioactive dNTP for incorporation into the 5′ overhang 100-fold molar excess of corresponding unlabeled dNTP Remaining two unlabeled dNTPs to a final concentration of 200 µM 5 U Klenow fragment. This reaction is identical to that used in standard DNA-binding assays (e.g., see Support Protocol), except biotinylated dUTP is incorporated into one end of the DNA fragment in place of TTP and the fragment is radiolabeled to low instead of high specific activity.

3. Precipitate the biotinylated probe. Isolate the probe by agarose gel electrophoresis using DEAE membrane. Gel purification of the biotinylated fragment removes unreacted biotin-11-dUTP. This is essential because any free biotin-dUTP binds to streptavidin and then to the biotin-cellulose column, reducing the apparent capacity of both materials to react with protein.

4. Resuspend the probe in TE buffer and measure an aliquot for Cerenkov counts. Estimate the DNA concentration by ethidium bromide dot quantitation. 5. Test the biotinylated probe to be certain it will efficiently bind to the protein of interest. Use a standard binding assay (e.g., see Support Protocol) with the biotinylated fragment as probe.


Prepare biotin-cellulose resin 6. Place 200 µl biotin-cellulose in a 1.5-ml microcentrifuge tube. Spin the resin for 30 sec in a microcentrifuge and remove the supernatant. Add to the pellet: 1.0 ml biotin-cellulose binding buffer 500 µg/ml BSA 200 µg carrier DNA

9.7.2 Supplement 12


Gently mix the tube 5 min on a rotating wheel. This step blocks nonspecific protein and nucleic acid binding sites present on the biotincellulose resin.

7. Spin the resin, remove the supernatant, and resuspend pellet in 1.0 ml of biotin-cellulose elution buffer. Gently mix 5 min on a rotating wheel. Repeat this wash. Washing removes molecules on the biotin-cellulose resin that might later be eluted from the resin by the biotin-cellulose elution buffer.

8. Spin the resin, remove the supernatant, and resuspend pellet in 1.0 ml of biotin-cellulose binding buffer. Repeat this wash. This 1:6 dilution of pretreated biotin-cellulose is ready for use and can be stored for several months.

Set up binding reaction 9. Determine the molar concentration of the protein to be purified via a mobility-shift assay (e.g., see Support Protocol). 10. Set up a standard binding reaction containing the protein to be purified, carrier DNA, and a 10-fold molar excess of biotinylated fragment relative to the protein to be purified. Allow the reaction to go to completion for ∼15 min. Use reaction conditions that optimize protein binding to its recognition site. The composition of the biotin-cellulose binding buffer should be the same as that optimized for protein-DNA binding.

11. Add a 5-fold molar excess of streptavidin relative to the biotinylated fragment. Continue the binding reaction for an additional 5 min at 30°C. Bind protein/DNA/streptavidin complex to biotin-cellulose resin 12. In a separate tube, place 2 µl pretreated biotin-cellulose (12 µl of the 1:6 dilution) for each picomole of biotinylated DNA fragment in the binding reaction. Spin the resin and remove the supernatant. One or two microliters of biotin-cellulose can easily be seen at the bottom of the microcentrifuge tube.

13. Transfer the binding reaction mix into the tube with the biotin-cellulose resin using a pipettor. Gently resuspend the resin and incubate on a rotating wheel for 30 min. This incubation can be done either at 4°C or at room temperature, depending on the stability of the protein.

14. Spin the resin and remove the supernatant. The supernatant should be measured using a mobility shift assay (e.g., see Support Protocol) to determine what percentage of the biotinylated fragment and the protein has been removed from the supernatant. It is also useful to assay the supernatant from the binding reaction for a control DNA-binding protein to determine whether the protein of interest has been specifically depleted or whether multiple DNA-binding proteins have been depleted nonspecifically. The latter observation would suggest that the matrix is acting as a nonspecific DNA-affinity column. For some applications, a protein fraction specifically depleted for a particular DNA-binding protein is a valuable reagent.

Wash the resin 15. Resuspend the biotin-cellulose pellet in 500 µl biotin-cellulose binding buffer. Mix by gently inverting the tube 1 to 2 min. Spin the resin and remove the supernatant.



Supplement 12

Repeat this procedure twice. The second time, transfer to a clean microcentrifuge tube. Transferring the reaction avoids elution of proteins that were bound nonspecifically to the walls of the tube in the first binding incubation.

Elute the protein 16. Resuspend the biotin-cellulose pellet in at least an equal volume of biotin-cellulose elution buffer. Mix gently on a rotating wheel for 20 min. The salt concentration in the biotin-cellulose elution buffer must be determined empirically. Successively higher salt concentrations may be tested until the concentration of eluted protein is maximal.

17. Spin the resin. Save the supernatant and assay for binding activity. Small volumes of protein solutions can be dialyzed effectively on 0.025-ìm filter discs. ALTERNATE PROTOCOL 1

PURIFICATION USING A MICROCOLUMN Although the batch method in the basic protocol is rapid and well-suited for analyticalscale purification, larger volumes of biotin-cellulose resin can be better handled in a microcolumn. This method is also used to elute the protein in as small a volume (i.e., as high a concentration) as possible. Additional Materials (also see Basic Protocol) Silanized glass wool 1.0-ml pipet tip Ring stand 1. Prepare the biotinylated DNA fragment and biotin-cellulose resin and set up binding reaction (see Basic Protocol, steps 1 to 11). 2. Place a small plug of silanized glass wool in the bottom of a 1.0-ml pipet tip. Firmly attach the pipet-tip microcolumn to a ring stand. Prewet the glass wool in biotin-cellulose binding buffer before insertion into the pipet tip to avoid trapping air bubbles which might denature proteins.

3. Add 500 µl binding buffer to the microcolumn. Maintain a steady flow through the glass wool plug. If the column does not flow smoothly, a pipettor can be gently inserted into the top of the microcolumn. Slightly depressing the plunger will start the column or increase the flow.

4. Add at least 40 µl of 1:1 biotin-cellulose slurry to the microcolumn. Allow the buffer to run down to the surface of the resin. 5. Equilibrate the resin with 3 column volumes of biotin-cellulose binding buffer if the resin has already been pretreated (Basic Protocol, steps 6 to 8). If the resin has not been pretreated, wash sequentially with 3 column volumes each of biotin-cellulose binding buffer, biotin-cellulose binding buffer with 500 µg/ml BSA and 200 µg/ml poly(dI-dC)⋅poly(dI-dC), and biotin-cellulose elution buffer. Finally, equilibrate with biotin-cellulose binding buffer. Purification of DNA-Binding Proteins Using Biotin/Streptavidin Affinity Systems

The biotin-cellulose microcolumn can be washed very rapidly. Each wash takes 2 or 3 min. The drop size from the pipet tip is ∼25 ìl but this will change with alterations in the ionic strength and protein concentration of the eluate.

9.7.4 Supplement 12


6. Load the binding reaction mix (Basic Protocol, step 11) onto the microcolumn and collect the flowthrough. The column can be run as fast as 6 to 10 column volumes/hr without affecting the amount of biotinylated fragment bound by the resin. If the flow rate is too slow, use a pipettor to apply pressure to the column. If the flow rate is too fast, plug the tip of the microcolumn with Parafilm in between drops.

7. Wash with 4 column volumes of biotin-cellulose binding buffer. Discard the flowthrough. 8. Wash with 3 column volumes of biotin-cellulose elution buffer. Collect 2-drop fractions and assay. Fractions as small as 5 ìl may be rapidly and effectively dialyzed on 0.025-ìm filter discs with minimal loss of volume and activity. Float a filter (shiny side up) in a petri dish on top of 20 ml dialysis buffer. Allow the filter 10 min to wet. Place the sample (5 to 100 ìl) to be dialyzed onto the surface of the filter. Surface tension will keep the sample confined in a drop unless there is a detergent in the sample. After 1 hr remove the sample. Once the protein has been eluted from the resin, the resin is effectively the same as a sequence-specific DNA-affinity microcolumn.

PURIFICATION USING STREPTAVIDIN-AGAROSE When high-quality free streptavidin is not available or cellulose is an inappropriate resin, a simple variation on the basic protocol may be employed. In this protocol, the same biotinylated DNA fragment is used but is removed from solution directly by streptavidinagarose (see Fig. 9.7.2).


Additional Materials (also see Basic Protocol) Streptavidin-agarose 1. Prepare biotinylated DNA fragment and resin, substituting streptavidin-agarose for biotin-cellulose, and set up binding reaction (see Basic Protocol, steps 1 to 11).

binding site for protein X

+

DNA end-labeled with biotin-11-dUTP

protein X

+ streptavidin agarose beads

streptavidin protein X

high-salt wash

Figure 9.7.2 Purification of DNA-binding proteins using streptavidin-agarose.



Supplement 12

2. In a separate tube, add 50 µl pretreated streptavidin-agarose (300 µl of the 1:6 dilution) for each picomole of biotinylated DNA fragment in the binding reaction. Spin the resin and remove the supernatant. 3. Transfer the binding-reaction mix into the tube with the streptavidin-agarose using a pipettor. Gently resuspend the resin and incubate on a rotating wheel for 30 min to 2 hr. 4. Wash and elute the protein (see Basic Protocol, steps 14 to 17). Like biotin-cellulose, streptavidin-agarose may also be used in a microcolumn. Follow the microcolumn alternate protocol. SUPPORT PROTOCOL

MOBILITY-SHIFT ASSAY The DNA-binding assay using nondenaturing polyacrylamide gel electrophoresis (PAGE) provides a simple, rapid, and extremely sensitive method for detecting sequence-specific DNA-binding proteins. Proteins that bind specifically to an end-labeled DNA fragment retard the mobility of the fragment during electrophoresis, resulting in discrete bands corresponding to the individual protein-DNA complexes. The assay can be used to test binding of purified proteins (see Basic Protocol) or of uncharacterized factors found in crude extracts. This assay also permits quantitative determination of the affinity, abundance, association rate constants, dissociation rate constants, and binding specificity of DNA-binding proteins. The utility of this technique is underscored by the many proteins that have been characterized using this assay. It has become clear that there is no single protocol that works best for all proteins. Rather, several variables can be changed to optimize binding. There are several options available for the design of the DNA probe, the binding reaction conditions, and the gel running conditions. The reader is referred to Buratowski and Chodosh (1996) for additional details and protocols. This protocol can be divided into four stages: (1) preparation of a radioactively labeled DNA probe containing a particular protein binding site; (2) preparation of a nondenaturing gel; (3) a binding reaction in which a protein mixture is bound to the DNA probe; and (4) electrophoresis of protein-DNA complexes through the gel, which is then dried and autoradiographed. Consult Current Protocols in Molecular Biology (Ausubel et al., 1998) or Sambrook et al. (1989) for details and protocols for basic molecular biological techniques referred to below. DNA fragments from 20 to 300 bp long may be used as probes. However, longer fragments are likely to contain binding sites for multiple proteins, which may make interpretation of the gel difficult. The DNA probe can be prepared in one of several ways, outlined below.


Materials 10× electrophoresis buffer: e.g., TAE or TBE electrophoresis buffer (APPENDIX 2E) or Tris-glycine electrophoresis buffer (UNIT 10.3) 30% (w/v) ammonium persulfate, prepared fresh N,N,N′,N′-tetramethylethylenediamine (TEMED) Nondenaturing gel mix (see recipe) Bulk carrier DNA, e.g., poly(dI-dC)⋅poly(dI-dC) BSA Protein preparation containing DNA-binding protein (crude extract or purified fraction) 10× loading buffer (see recipe)

9.7.6 Supplement 12


Constant-temperature water bath Two-head peristaltic pump 10-µl glass capillary pipet (optional) Clay-Adams screw-top loader (optional) Whatman 3MM filter paper (or equivalent) Additional reagents and equipment for digesting DNA with restriction endonucleases, DNA labeling, agarose and nondenaturing polyacrylamide gel electrophoresis, recovery of DNA from gels, oligonucleotide synthesis, PCR, ethanol precipitation, ethidium bromide dot quantitation, and autoradiography (see Ausubel et al., 1998, or Sambrook et al., 1989) Prepare the DNA probe 1a. For restriction endonuclease fragments: Isolate a small DNA fragment containing the binding site of interest from a plasmid using a standard restriction endonuclease digestion. Label the fragment by filling in a 5′ overhang with the Klenow fragment of Escherichia coli DNA polymerase and 32P-labeled nucleotide or by end labeling using polynucleotide kinase. Separate the fragment from the plasmid by gel electrophoresis. Kinased probes should be avoided in experiments using crude protein preparations that might contain phosphatase activity. Agarose gels are useful for resolving fragments as small as 50 bp. Fragments can be recovered using low-melting-temperature agarose or DEAE paper. For smaller DNA fragments, nondenaturing polyacrylamide gels can be used.

1b. For synthetic oligonucleotides: Synthesize and anneal complementary oligonucleotides to generate a double-stranded DNA fragment containing the binding site of interest. Label the probe using polynucleotide kinase. The kinase reaction can be performed before annealing to label only one strand, or after annealing to label both strands. Alternatively, the oligonucleotides can be designed to leave an overhang that can be filled in with labeled nucleotide and Klenow fragment.

1c. For PCR fragments: Generate a DNA fragment containing the binding site of interest by polymerase chain reaction (PCR). End-label one of the primers with polynucleotide kinase before the PCR reaction, or label the double-stranded PCR product after purification. 2. Following isolation of the probe, determine its concentration by ethidium bromide dot quantitation. Concentrations in the range of 2 to 50 ng/ìl are convenient.

3. Count 1 µl for Cerenkov counts in a scintillation counter to determine specific activity (cpm/µl). A typical binding reaction will contain about ∼5,000 to 20,000 cpm and ∼10 to 100 fmol probe (10 fmol DNA in a final reaction volume of 10 ìl gives a total DNA concentration of 1 nM). If desired, probes can be stored up to 4 to 6 weeks at 4°C before proceeding with the experiment.

Prepare the nondenaturing gel 4. Dilute 10× electrophoresis buffer to prepare enough 1× electrophoresis buffer to fill the tank. 5. Assemble washed glass plates and 1.5-mm spacers for casting the gel. All traces of detergent must be removed because detergent will disrupt protein-DNA interactions.



Supplement 12

6. Add 150 µl of 30% ammonium persulfate and 70 µl TEMED to 60 ml nondenaturing gel mix prepared using the same buffer as that used for electrophoresis. Swirl gently to mix. The amounts of ammonium persulfate, TEMED, and nondenaturing gel mix can be scaled up or down as necessary depending on the size and number of gels.

7. Pour the gel mix between the plates and insert a comb. Allow the gel to completely polymerize for 20 min. For optimal results, use a comb with teeth that are ≥7 mm wide.

8. Remove the comb and bottom spacer and attach the plates to the electrophoresis tank after filling the lower reservoir with 1× electrophoresis buffer. Fill the upper reservoir of the tank with 1× electrophoresis buffer. With a bent-needle syringe, remove any air bubbles trapped beneath the gel and flush out the wells. Prerun the gel 30 to 60 min at 100 V. For low-ionic-strength buffers (≤0.5×), use a pump with two heads and a flow rate of 5 to 30 ml per min to exchange buffer between the upper and lower reservoirs. Recirculation of the buffer is essential to prevent polarization due to the low buffering capacity of the buffer.

Prepare the binding reactions 9. While the gel is prerunning, assemble the binding reaction by combining the following in a 0.5-ml or 1.5-ml microcentrifuge tube: 5,000 to 20,000 cpm radiolabeled probe DNA (0.1 to 0.5 ng, ≥10 fmol) 0.1 to 2 µg nonspecific carrier DNA 300 µg/ml BSA ≥10% (v/v) glycerol Appropriate buffer and salt Water or buffer to obtain a final reaction volume of 10 to 15 µl DNA-binding protein (∼15 µg crude extract or ∼5 to 25 ng purified protein). The protein should be added last.

10. Mix gently by tapping the bottom of the tube with a finger. Avoid introducing bubbles in the mix.

11. Incubate the binding reaction mix 15 to 30 minutes in a constant-temperature water bath. Optimal incubation temperatures for different proteins can vary from room temperature to 37°C.

Run the gel 12. Load each binding reaction into the appropriate well of the prerun gel using either a 10-µl glass capillary pipet and Clay-Adams screw-top loader, or a pipettor. Load a small volume of 10× loading buffer with dyes into a separate well. The dyes are used to monitor the progress of electrophoresis. There is no stacking gel in this system, so precise loading with little mixing with the gel buffer is necessary to obtain sharp bands on the gel. Allow the sample to fall along one side of the well to prevent dilution and avoid bubbles in the well.


13. Electrophorese at ∼30 to 35 mA for the minimum time required to give good separation of free probe and the protein-DNA complexes. Stop the gel before the bromphenol blue approaches the bottom of the gel (∼1.5 to 2 hr for a 15- to 20-cm gel). Longer run times may cause a weaker signal due to partial dissociation of complexes during electrophoresis.

9.7.8 Supplement 12


Bromphenol blue migrates at approximately the same position as a 70-bp DNA probe. For probes 99% of the remaining protein can be removed with repeated column washing using the binding buffer. Moreover, ∼30% to 80% of the binding activity can be recovered from the column. The recovery efficiency depends on how effectively nonspecific binding sites have been blocked. This purification scheme may be used either with crude extracts or with partially purified protein fractions.

Time Considerations Starting with a biotinylated DNA fragment, the batch protocol (Basic Protocol) can be performed in ∼1.5 hr, the microcolumn (Alternate Protocol 1) in ∼2.5 hr, and the streptavidinagarose (Alternate Protocol 2) protocol in ∼1.5 hr. The DNA-binding assay (Support Protocol) may take 4 to 5 hr to complete from the time the gel is poured to the time it is dried onto filter paper. Only a small fraction of this time is labor intensive. Thus, several DNA-binding gels can be run simultaneously while other experiments are being performed. Furthermore, the results are often available within 2 to 3 hr after completion of the experiment.

Literature Cited Ausubel, F.A., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, K. (eds.) 1998. Current Protocols in Molecular Biology. John Wiley & Sons, New York. Buratowski, S. and Chodosh, L. 1996. Mobilityshift DNA-binding assay using gel electrophoresis. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 12.2.1-12.2.11. John Wiley & Sons, New York. Chodosh, L.A., Carthew, R.W., and Sharp, P.A. 1986. A single polypeptide possesses the binding and transcription activities of the adenovirus major late promoter. Mol. Cell. Biol. 6:4723-4733. Grabowski, P.J. and Sharp, P.A. 1986. Affinity chromatography of splicing complexes: U2, U5, and U4 + U6 small nuclear ribonucleoprotein particles in the spliceosome. Science 233:1294-1299.



Haeuptle, M.-T., Aubert, M.L., Kjiane, J., and Kraehenbuhl, J.-P. 1983. Binding sites for lactogenic and somatogenic hormones from rabbit mammary gland and liver. J. Biol. Chem. 258:305314. Kadonaga, J.T. and Tjian, R. 1986. Affinity purification of sequence-specific DNA binding proteins. Proc. Natl. Acad. Sci. U.S.A. 83:58895893. Kasher, M.S., Pintel, D., and Ward, D.C. 1986. Rapid enrichment of HeLa transcription factors IIIB and IIIC by using affinity chromatography based on avidin-biotin interactions. Mol. Cell. Biol. 6:3117-3127. Rosenfeld, P.J. and Kelly, T.J. 1986. Purification of nuclear factor I by DNA recognition site affinity chromatography. J. Biol. Chem. 261:1398-1408. Sambrook, J., Fritsch, E.F., and Maniatis, T. 1989. Molecular Cloning: A Laboratory Manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

Key References Chodosh et al., 1986. See above. Describes the biotin/streptavidin-agarose microcolumn procedure from which the second alternate protocol of this unit was drawn. Kasher et al., 1986. See above. Describes a biotin-cellulose/streptavidin procedure similar to the basic protocol of this unit.

Contributed by Lewis A. Chodosh University of Pennsylvania Philadelphia, Pennsylvania Stephen Buratowski Harvard Medical School Boston, Massachusetts



Supplement 12

Immunoprecipitation

UNIT 9.8

Immunoprecipitation is a technique in which an antigen is isolated by binding to a specific antibody attached to a sedimentable matrix. The source of antigen for immunoprecipitation can be unlabeled cells or tissues, metabolically or extrinsically labeled cells (UNIT 3.7), subcellular fractions from either unlabeled or labeled cells (see Chapter 4), or in vitro–translated proteins. Immunoprecipitation is also used to analyze protein fractions separated by other biochemical techniques such as gel filtration (UNIT 8.3) or sedimentation on density gradients (UNIT 4.2). Either polyclonal or monoclonal antibodies from various animal species can be used in immunoprecipitation protocols. Antibodies can be bound noncovalently to immunoadsorbents such as protein A– or protein G–agarose, or can be coupled covalently to a solid-phase matrix. Immunoprecipitation protocols consist of several stages (Fig. 9.8.1; see Basic Protocol 1). In stage 1, the antigen is solubilized by one of several techniques for lysing cells. Soluble and membrane-associated antigens can be released from cells grown either in suspension culture (see Basic Protocol 1) or as a monolayer on tissue culture dishes (see Alternate Protocol 1) with nondenaturing detergents. Alternatively, cells can be lysed under denaturing conditions (see Alternate Protocol 2). Soluble antigens can also be extracted by mechanical disruption of cells in the absence of detergents (see Alternate Protocol 3). All of these procedures are suitable for extracting antigens from animal cells. By contrast, yeast cells require disruption of their cell wall in order to allow extraction of the antigens (see Alternate Protocol 4). In stage 2, a specific antibody is attached, either noncovalently or covalently, to a sedimentable, solid-phase matrix to allow separation by low-speed centrifugation. In this unit, two methods for achieving this are described: the noncovalent attachment of antibody to protein A– or protein G–agarose beads (see Basic Protocol 1) and covalent coupling to Sepharose (see Alternate Protocol 5 and Support Protocol). Stage 3 is the actual immunoprecipitation, which can be achieved by incubating the solubilized antigen from stage 1 with the immobilized antibody from stage 2, followed by extensive washing to remove unbound proteins (see Basic Protocol 1). Another method is to precipitate the immune complexes using antibodies contained in an anti-immunoglobulin (anti-Ig) serum (see Alternate Protocol 6). Immunoprecipitated antigens can be dissociated from antibodies and reprecipitated by a protocol referred to as “immunoprecipitation-recapture” (see Basic Protocol 2). This procedure can be used with the same antibody for further purification of the antigen, or with a second antibody to identify components of multisubunit complexes or to study protein-protein interactions (Fig. 9.8.3). Immunoprecipitated antigens can be analyzed by one-dimensional electrophoresis (UNIT 10.1), two-dimensional electrophoresis (UNIT 10.4), or immunoblotting (UNIT 10.10). In some cases, immunoprecipitates can be used for structural or functional analyses of the isolated antigens. Immunoprecipitates can also be used as sources of immunogens for production of monoclonal or polyclonal antibodies. IMMUNOPRECIPITATION USING CELLS IN SUSPENSION LYSED WITH A NONDENATURING DETERGENT SOLUTION

BASIC PROTOCOL 1

In this protocol, cells in suspension (labeled or unlabeled) are extracted by incubation in nondenaturing lysis buffer containing the nonionic detergent Triton X-100 (steps 1 to 7). This procedure results in the release of both soluble and membrane proteins; however, many cytoskeletal and nuclear proteins, as well as a fraction of membrane proteins, are Affinity Purification Contributed by Juan S. Bonifacino, Esteban C. Dell’Angelica, and Timothy A. Springer Current Protocols in Protein Science (1999) 9.8.1-9.8.28 Copyright © 1999 by John Wiley & Sons, Inc.

9.8.1 Supplement 18

not efficiently extracted under these conditions. The procedure allows immunoprecipitation with antibodies to epitopes that are exposed in native proteins. For immunoprecipitation, a specific antibody is immobilized on a sedimentable, solidphase matrix (steps 8 to 14). Although there are many ways to attach antibodies to matrices (see Commentary), the most commonly used methods rely on the property of immunoglobulins to bind Staphylococcus aureus protein A, or protein G from group G Streptococcus (Table 9.8.1). The best results are obtained by binding antibodies to protein A or protein G that is covalently coupled to agarose beads. In this protocol, Sepharose beads are used (Sepharose is a more stable, cross-linked form of agarose). Immunoprecipitation is most often carried out using rabbit polyclonal or mouse monoclonal antibodies, which,

protein A–agarose bead

animal cell specific antigen

antibody

1

cell lysis (see Basic Protocol 1, steps 1 to 7; Alternate Protocols 1 to 4)

2

antibody binding to protein A–agarose bead (see Basic Protocol 1, steps 8 to 14)

3 antigen isolation on antibody bead (see Basic Protocol 1, steps 18 to 21)

washing (see Basic Protocol 1, steps 22 to 26) and analysis (step 27)

Immunoprecipitation

Figure 9.8.1 Schematic representation of the stages of the immunoprecipitation protocol presented in Basic Protocol 1. (1) Cell lysis: antigens are solubilized by extraction of the cells in the presence or absence of detergents. To increase specificity, the cell lysate can be precleared with protein A–agarose beads (steps 15 to 17, not shown). (2) Antibody immobilization: a specific antibody is bound to protein A–agarose beads. (3) Antigen capture: the solubilized antigen is isolated on antibody-conjugated beads.

9.8.2 Supplement 18


with some exceptions (e.g., mouse IgG1), bind well to protein A (Table 9.8.1). Antibodies that do not bind to protein A–agarose can be adsorbed to protein G–agarose (Table 9.8.1) using exactly the same protocol. For optimal time management, incubation of antibodies with protein A–agarose can be carried out either before or during lysis of the cells. The final stage in immunoprecipitation is combining the cell lysate with the antibodyconjugated beads and isolating the antigen (steps 18 to 26). This can be preceded by an optional preclearing step in which the lysate is absorbed with either “empty” protein A–agarose beads or with an irrelevant antibody bound to protein A–agarose (steps 15 to 17). The need for preclearing depends on the specific experimental system being studied and the quality of the antibody reagents. The protocol described below incorporates a preclearing step using protein A–agarose. Protein fractions separated by techniques such as gel filtration (UNIT 8.3) or sedimentation on sucrose gradients (UNIT 4.2) can be used in Table 9.8.1 Ga,b,c

Binding of Antibodies to Protein A and Protein

Protein A binding

Protein G bindingd

Monoclonal antibodiese Human IgG1 Human IgG2 Human IgG3 Human IgG4 Mouse IgG1 Mouse IgG2a Mouse IgG2b Mouse IgG3 Rat IgG1 Rat IgG2a Rat IgG2b Rat IgG2c

++ ++ − ++ + ++ ++ ++ + − − ++

++ ++ ++ ++ ++ ++ ++ ++ + ++ ++ ++

Polyclonal antibodies Chicken Donkey Goat Guinea pig Hamster Human Monkey Mouse Rabbit Rat Sheep

− − + ++ + ++ ++ ++ ++ + +

− ++ ++ + ++ ++ ++ ++ ++ + ++

Antibody

a++, moderate to strong binding; +, weak binding; −, no binding. bA hybrid protein A/G molecule that combines the features of protein A and

protein G, coupled to a solid-phase matrix, is available from Pierce. cInformation from Harlow and Lane (1999), and from Amersham Pharmacia

Biotech, Pierce, and Jackson Immunoresearch. dNative protein G binds albumin from several animal species. Recombinant variants of protein G have been engineered for better binding to rat, mouse, and guinea pig IgG, as well as for avoiding binding to serum albumin. eProtein A binds some IgM, IgA, and IgE antibodies in addition to IgG, whereas protein G binds only IgG.



Supplement 18

place of the cell lysate at this stage. After binding the antigen to the antibody-conjugated beads, the unbound proteins are removed by successive washing and sedimentation steps. Materials Unlabeled or labeled cells in suspension PBS (APPENDIX 2E), ice cold Nondenaturing lysis buffer (see recipe), ice cold 50% (v/v) protein A–Sepharose bead (Sigma, Amersham Pharmacia Biotech) slurry in PBS containing 0.1% (w/v) BSA and 0.01% (w/v) sodium azide (NaN3) Specific polyclonal antibody (antiserum or affinity-purified immunoglobulin) or monoclonal antibody (ascites, culture supernatant, or purified immunoglobulin) Control antibody of same type as specific antibody (e.g., preimmune serum or purified irrelevant immunoglobulin for specific polyclonal antibody; irrelevant ascites, hybridoma culture supernatant, or purified immunoglobulin for specific monoclonal antibody; see Critical Parameters) 10% (w/v) BSA Wash buffer (see recipe), ice cold Microcentrifuge with fixed-angle rotor (Eppendorf 5415C or equivalent) Tube rotator (capable of end-over-end inversion) Pasteur pipet attached to a vacuum trap CAUTION: When working with radioactivity, take appropriate precautions to avoid contamination of the experimenter and the surroundings. Carry out the experiment and dispose of wastes in an appropriately designated area, following the guidelines provided by the local radiation safety officer (also see APPENDIX 2B). NOTE: All solutions should be ice cold and procedures should be carried out at 4°C or on ice. Prepare cell lysate 1. Collect cells in suspension by centrifuging 5 min at 400 × g, 4°C, in a 15- or 50-ml capped conical tube. Place tube on ice. Approximately 0.5–2 × 107 cells are required to yield 1 ml lysate, which is the amount generally used for each immunoprecipitation. Labeled cells are likely to have been pelleted earlier as part of the labeling procedure. If the cells are frozen, they should be thawed on ice before solubilization.

2. Aspirate supernatant with a Pasteur pipet attached to a vacuum trap. CAUTION: Dispose of radioactive materials following applicable safety regulations (APPENDIX 2B).

3. Resuspend cells gently by tapping the bottom of the tube. Rinse cells twice with ice-cold PBS as in steps 1 and 2, using the same volume of PBS as in the initial culture. 4. Add 1 ml ice-cold nondenaturing lysis buffer per ∼0.5–2 × 107 cells and resuspend pellet by gentle agitation for 3 sec with a vortex mixer set at medium speed. Do not shake vigorously, as this could result in loss of material or protein denaturation due to foaming.

Immunoprecipitation

5. Keep suspension on ice 15 to 30 min and transfer to a 1.5-ml conical microcentrifuge tube.

9.8.4 Supplement 18


Tubes can have flip-top or screw caps. Screw-capped tubes are preferred because they are less likely to open accidentally during subsequent procedures. They are also recommended for work with radioactivity.

6. Clear the lysate by microcentrifuging 15 min at 16,000 × g (maximum speed), 4°C. Centrifugation can be carried out in a microcentrifuge placed in a cold room or in a refrigerated microcentrifuge. Take precautions to ensure that the 4°C temperature is maintained during the spin (e.g., use a fixed-angle rotor with a lid, as the aerodynamics of this type of rotor reduces generation of heat by friction). If it is necessary to reduce background, the lysate can be spun for 1 hr at 100,000 × g in an ultracentrifuge.

7. Transfer the supernatant to a fresh microcentrifuge tube using an adjustable pipet fitted with a disposable tip. Do not disturb the pellet, and leave the last 20 to 40 µl of supernatant in the centrifuge tube. Keep the cleared lysate on ice until preclearing (step 15) or addition of antibody beads (step 18). NOTE: Resuspension of even a small amount of sedimented material will result in high nonspecific background due to carryover into the immunoprecipitation steps. A cloudy layer of lipids floating on top of the supernatant will not adversely affect the results of the immunoprecipitation. When the lysate is highly radioactive—as is the case for metabolically labeled cells—the use of tips with aerosol barriers is recommended to reduce the risk of contaminating internal components of the pipet. Cell extracts can be frozen at −70°C until used for immunoprecipitation. However, it is preferable to lyse the cells immediately before immunoprecipitation in order to avoid protein degradation or dissociation of protein complexes. If possible, freeze the cell pellet from step 3 rather than the supernatant from step 7.

Prepare antibody-conjugated beads 8. In a 1.5-ml conical microcentrifuge tube, combine 30 µl of 50% protein A–Sepharose bead slurry, 0.5 ml ice-cold PBS, and the following quantity of specific antibody (select one): 1 to 5 µl polyclonal antiserum 1 µg affinity-purified polyclonal antibody 0.2 to 1 µl ascitic fluid containing monoclonal antibody 1 µg purified monoclonal antibody 20 to 100 µl culture supernatant containing monoclonal antibody. The quantities of antibody suggested are rough estimates based on the expected amount of specific antibodies in each preparation. Quantities can be increased or decreased depending on the quality of the antibody preparation (see Commentary). Substitute protein G for protein A if antibodies are of a species or subclass that does not bind to protein A (see Table 9.8.1). If the same antibody will be used to immunoprecipitate multiple samples (e.g., samples from a pulse-chase experiment; UNIT 3.7), the quantities indicated above can be increased proportionally to the number of samples and incubated in a 15-ml capped conical tube. In this case, the beads should be divided into aliquots just prior to the addition of the cleared cell lysate (step 18). Antibody-conjugated beads can be prepared prior to preparation of the cell lysate (steps 1 to 7), in order to minimize the time that the cell extract is kept on ice.

9. Set up a nonspecific immunoprecipitation control in a 1.5-ml conical microcentrifuge tube by incubating 30 µl of 50% protein A–Sepharose bead slurry, 0.5 ml ice-cold PBS, and the appropriate control antibody (select one):



Supplement 18

1 to 5 µl preimmune serum as a control for a polyclonal antiserum 1 µg purified irrelevant polyclonal antibody (an antibody to an epitope that is not present in the cell lysate) as a control for a purified polyclonal antibody 0.2 to 1 µl ascitic fluid containing irrelevant monoclonal antibody (an antibody to an epitope that is not present in the cell lysate and of the same species and immunoglobulin subclass as the specific antibody) as a control for an ascitic fluid containing specific monoclonal antibody 1 µg purified irrelevant monoclonal antibody as a control for a purified monoclonal antibody 20 to 100 µl hybridoma culture supernatant containing irrelevant monoclonal antibody as a control for a hybridoma culture supernatant containing specific monoclonal antibody. The amount of irrelevant antibody should match that of the specific antibody and the antibody should be from the same species as the specific antibody.

10. Mix suspensions thoroughly. Tumble incubation mixtures end over end ≥1 hr at 4°C in a tube rotator. Addition of 0.01% (w/v) Triton X-100 may facilitate mixing of the suspension during tumbling. Incubations can be carried out for as long as 24 hr. This allows preparation of the antibody-conjugated beads prior to immunoprecipitation.

11. Microcentrifuge 2 sec at 16,000 × g (maximum speed), 4°C. 12. Aspirate the supernatant (containing unbound antibodies) using a fine-tipped Pasteur pipet connected to a vacuum aspirator. 13. Add 1 ml nondenaturing lysis buffer and resuspend the beads by inverting the tube three or four times. For lysates prepared with detergents (this protocol and see Alternate Protocols 1 and 2), use 1 ml nondenaturing lysis buffer; for lysates prepared by mechanical disruption (see Alternate Protocol 3), use detergent-free lysis buffer (see recipe). Use of a repeat pipettor is recommended when processing multiple samples.

14. Wash by repeating steps 11 to 13, and then steps 11 and 12 once more. At this point the beads have been washed twice with lysis buffer and are ready to be used for immunoprecipitation. Antibody-bound beads can be stored up to 6 hr at 4°C until used.

Preclear lysate (optional) 15. In a microcentrifuge tube, combine 1 ml cell lysate (from step 7) and 30 µl of 50% protein A–Sepharose bead slurry. The purpose of this step is to remove from the lysate proteins that bind to protein A–Sepharose, as well as pieces of insoluble material that may have been carried over from previous steps. If the lysate was prepared from cells expressing immunoglobulins—such as spleen cells or cultured B cells—the preclearing step should be repeated at least three times to ensure complete removal of endogenous immunoglobulins. If cell lysates were frozen and thawed, they should be microcentrifuged 15 min at 16,000 × g (maximum speed), 4°C, before the preclearing step.

16. Tumble end over end 30 min at 4°C in a tube rotator. Immunoprecipitation

17. Microcentrifuge 5 min at 16,000 × g (maximum speed), 4°C.

9.8.6 Supplement 18


Immunoprecipitate 18. Add 10 µl of 10% BSA to the tube containing specific antibody bound to protein A–Sepharose beads (step 14), and transfer to this tube the entire volume of cleared lysate (from step 7 or 17). If a nonspecific immunoprecipitation control is performed, divide lysate in two ∼0.4-ml aliquots, one for the specific antibody and the other for the nonspecific control. In order to avoid carryover of beads with precleared material, leave 20 to 40 µl of supernatant on top of the pellets in the preclearing tubes. Discard beads and remaining supernatant. The BSA quenches nonspecific binding to the antibody-conjugated beads during incubation with the cell lysate.

19. Incubate 1 to 2 hr at 4°C while mixing end over end in a tube rotator. Samples can be incubated overnight, although there is an increased risk of protein degradation, dissociation of multiprotein complexes, or formation of protein aggregates.

20. Microcentrifuge 5 sec at 16,000 × g (maximum speed), 4°C. 21. Aspirate the supernatant (containing unbound proteins) using a fine-tipped Pasteur pipet connected to a vacuum aspirator. The supernatant can be kept up to 8 hr at 4°C or up to 1 month at −70°C for sequential immunoprecipitation of other antigens or for analysis of total proteins. To reutilize lysate, remove the supernatant carefully with an adjustable pipet fitted with a disposable tip. Before reprecipitation, preabsorb the lysate with protein A–Sepharose (as in steps 15 to 17) to remove antibodies that may have dissociated during the first immunoprecipitation. CAUTION: Dispose of radioactive materials following applicable safety regulations.

22. Add 1 ml ice-cold wash buffer, cap the tubes, and resuspend the beads by inverting the tube 3 or 4 times. Use of a repeat pipettor is recommended when processing multiple samples.

23. Microcentrifuge 2 sec at 16,000 × g (maximum speed), 4°C. 24. Aspirate the supernatant, leaving ∼20 µl supernatant on top of the beads. 25. Wash beads three more times (steps 22 to 24). Total wash time (steps 22 to 26) should be ∼30 min, keeping the samples on ice for 3 to 5 min between washes if necessary (see Critical Parameters).

26. Wash beads once more using 1 ml ice-cold PBS and aspirate supernatant completely with a drawn-out Pasteur pipet or an adjustable pipet fitted with a disposable tip. The final product should be 15 µl of settled beads containing bound antigen. Immunoprecipitates can either be processed immediately or frozen at −20°C for later analysis. For subsequent analysis of the isolated proteins prior to electrophoresis (e.g., comparison of the electrophoretic mobility of the antigen with or without treatment with glycosidases), samples can be divided into two or more aliquots after addition of PBS. Transfer aliquots of the bead suspension to fresh tubes, centrifuge and aspirate as in the previous steps.

27. Analyze immunoprecipitates by one-dimensional electrophoresis (UNIT 10.1), two-dimensional electrophoresis (UNIT 10.4), or immunoblotting (UNIT 10.10).



Supplement 18


IMMUNOPRECIPITATION USING ADHERENT CELLS LYSED WITH A NONDENATURING DETERGENT SOLUTION Immunoprecipitation using adherent cells can be performed in the same manner as with nonadherent cells (see Basic Protocol 1). This protocol is essentially similar to steps 1 to 5 of Basic Protocol 1, but describes modifications necessary for using the same nondenaturing detergent solution to lyse cells attached to tissue culture plates. It is preferable to use cells grown on plates rather than in flasks, because the cell monolayer is more easily accessible. Additional Materials (also see Basic Protocol 1) Unlabeled or labeled cells grown as a monolayer on a tissue culture plate (UNIT 3.7) Rubber policeman NOTE: All solutions should be ice cold and procedures should be carried out at 4°C or on ice. 1. Rinse cells attached to a tissue culture plate twice with ice-cold PBS. Remove the PBS by aspiration with a Pasteur pipet attached to a vacuum trap. CAUTION: Dispose of radioactive materials following applicable safety regulations.

2. Place the tissue culture plate on ice. 3. Add ice-cold nondenaturing lysis buffer to the tissue culture plate. Use 1 ml lysis buffer for an 80% to 90% confluent 100-mm-diameter tissue culture plate. Depending on the cell type, a confluent 100-mm dish will contain 0.5–2 × 107 cells. For other plate sizes, adjust volume of lysis buffer according to the surface area of the plate.

4. Scrape the cells off the plate with a rubber policeman, and transfer the suspension to a 1.5-ml conical microcentrifuge tube using an adjustable pipettor fitted with a disposable tip. Vortex gently for 3 sec and keep tubes on ice for 15 to 30 min. Tubes can have flip-top or screw caps. Screw-capped tubes are preferred because they are less likely to open accidentally during subsequent procedures. They are also recommended for work with radioactivity.

5. Clear the lysate and perform immunoprecipitation (see Basic Protocol 1, steps 6 to 27). ALTERNATE PROTOCOL 2

IMMUNOPRECIPITATION USING CELLS LYSED WITH DETERGENT UNDER DENATURING CONDITIONS If epitopes of native proteins are not accessible to antibodies, or if the antigen cannot be extracted from the cell with nonionic detergents, cells should be solubilized under denaturing conditions. This protocol is based on that for nondenaturing conditions (see Basic Protocol 1, steps 1 to 7), with the following modifications. Denaturation is achieved by heating the cells in a denaturing lysis buffer that contains an ionic detergent such as SDS or Sarkosyl (N-lauroylsarcosine). The denaturing lysis buffer also contains DNase I to digest DNA released from the nucleus. Prior to immunoprecipitation, the denatured protein extract is diluted 10-fold with nondenaturing lysis buffer, which contains Triton X-100; this step protects the antigen-antibody interaction from interference by the ionic detergent. Immunoprecipitation is performed as described (see Basic Protocol 1).

Immunoprecipitation

The following protocol is described for cells in suspension culture, although it can be adapted for adherent cells (see Alternate Protocol 1). Only antibodies that react with denatured proteins can be used to immunoprecipitate proteins solubilized by this protocol.

9.8.8 Supplement 18


Additional Materials (also see Basic Protocol 1) Denaturing lysis buffer (see recipe) Heating block set at 95°C (Eppendorf Thermomixer 5436 or equivalent) 25-G needle attached to 1-ml syringe 1. Collect cells in suspension culture (see Basic Protocol 1, steps 1 to 3). Place tubes on ice. 2. Add 100 µl denaturing lysis buffer per ∼0.5–2 × 107 cells in the pellet. 3. Resuspend the cells by vortexing vigorously 2 to 3 sec at maximum speed. Transfer suspension to a 1.5-ml conical microcentrifuge tube. The suspension may be very viscous due to release of nuclear DNA. Tubes can have flip-top or screw caps. Screw-capped tubes are preferred because they are less likely to open accidentally during subsequent procedures. They are also recommended for work with radioactivity.

4. Heat samples 5 min at 95°C in a heating block. 5. Dilute the suspension with 0.9 ml nondenaturing lysis buffer. Mix gently. The excess 1% Triton X-100 in the nondenaturing lysis buffer sequesters SDS into Triton X-100 micelles.

6. Shear DNA by passing the suspension five to ten times through a 25-G needle attached to a 1-ml syringe. If the DNA is not digested by DNase I in the denaturing lysis buffer or thoroughly sheared mechanically, it will interfere with the separation of pellet and supernatant after centrifugation. Repeat mechanical disruption until the viscosity is reduced to manageable levels.

7. Incubate 5 min on ice. 8. Clear the lysate and perform immunoprecipitation (see Basic Protocol 1, steps 6 to 27). IMMUNOPRECIPITATION USING CELLS LYSED WITHOUT DETERGENT Immunoprecipitation of proteins that are already soluble within cells (e.g., cytosolic or luminal organellar proteins) may not require the use of detergents. Instead, cells can be mechanically disrupted by repeated passage through a needle, and soluble proteins can be separated from insoluble material by centrifugation. The following protocol describes lysis of cells in a PBS-based detergent-free lysis buffer. Other buffer formulations may be used for specific proteins.


Additional Materials (also see Basic Protocol 1) Detergent-free lysis buffer (see recipe) 25-G needle attached to 3-ml syringe NOTE: All solutions should be ice-cold and procedures should be carried out at 4°C or on ice. 1. Collect and wash cells in suspension (see Basic Protocol 1, steps 1 to 3). 2. Add 1 ml of ice-cold detergent-free lysis buffer per ∼0.5–2 × 107 cells in a pellet. 3. Resuspend the cells by gentle agitation for 3 sec with a vortex mixer set at medium speed.



Supplement 18

4. Break cells by passing the suspension 15 to 20 times through a 25-G needle attached to a 3-ml syringe. Extrusion of the cell suspension from the syringe should be rapid, although care should be exercised to prevent splashing and excessive foaming. Cell breakage can be checked under a bright-field or phase-contrast microscope. Repeat procedure until >90% cells are broken. It is helpful to check ahead of time whether the cells can be broken in this way. If the cells are particularly resistant to mechanical breakage, they can be swollen for 10 min at 4°C with a hypotonic solution containing 10 mM Tris⋅Cl, pH 7.4 (APPENDIX 2E) before mechanical disruption.

5. Clear the lysate and perform immunoprecipitation (Basic Protocol 1, steps 6 to 27). ALTERNATE PROTOCOL 4

IMMUNOPRECIPITATION USING YEAST CELLS DISRUPTED WITH GLASS BEADS Unlike animal cells, yeast cells have an extremely resistant, detergent-insoluble cell wall. To allow extraction of cellular antigens, the cell wall needs to be broken by mechanical, enzymatic, or chemical means. The most commonly used procedure consists of vigorous vortexing of the yeast suspension with glass beads. The breakage can be done in the presence or absence of detergent, as previously described for animal cells (see Basic Protocol 1, Alternate Protocol 2, and Alternate Protocol 3). The protocol described below is suitable for mechanical disruption of most yeast species, including Saccharomyces cerevisiae and Schizosaccharomyces pombe. A protocol for metabolic labeling for yeast has been described by Franzusoff et al. (1991). Additional Materials (also see Basic Protocol 1) Unlabeled or radiolabeled yeast cells Lysis buffer, ice cold: nondenaturing, denaturing, or detergent-free lysis buffer (see recipes) Glass beads (acid-washed, 425- to 600-µm diameter; Sigma) NOTE: All solutions should be ice-cold and procedures should be carried out at 4°C or on ice. 1. Collect 10 ml of yeast culture at 1 OD600 per immunoprecipitation sample, and centrifuge 5 min at 4000 × g, 4°C. Place tube on ice. 2. Remove supernatant by aspiration with a Pasteur pipet attached to a vacuum trap. CAUTION: Dispose of radioactive materials following applicable safety regulations.

3. Loosen pellet by vortexing vigorously for 10 sec. Rinse cells twice with ice-cold distilled water as in steps 1 and 2. Radiolabeled yeast cells are likely to have been pelleted earlier as part of the labeling procedure. If the pellets are frozen, they should be thawed on ice prior to cell disruption.

4. Add 3 vol ice-cold lysis buffer and 3 vol glass beads per volume of pelleted yeast cells. Use nondenaturing lysis buffer or detergent-free lysis buffer as required for the antigen under study. If the experiment requires denaturation of the antigen, the procedure can be adapted to include this (see Alternate Protocol 2 for higher eukaryotic cells); however, the yeast cells must be broken with glass beads before heating the sample at 95°C.

Immunoprecipitation

5. Shake cells by vortexing vigorously at maximum speed for four 30-sec periods, keeping the cells on ice for 30 sec between the periods.



Check cell breakage under a bright-field or phase-contrast microscope. It is helpful to check ahead of time if the cells can be broken in this way.

6. Remove the yeast cell lysate from the beads using a pipettor with a disposable tip. Transfer to a fresh tube. 7. Add 4 vol (see step 4) lysis buffer to the glass beads, vortex for 2 sec, and combine this supernatant with the lysate from step 6. 8. Clear the lysate and perform immunoprecipitation (see Basic Protocol 1, steps 6 to 27). IMMUNOPRECIPITATION WITH ANTIBODY-SEPHAROSE This protocol, which follows the steps presented in Figure 9.8.2, relies on the formation of an insoluble immune complex between a protein antigen and an antigen-specific monoclonal (or polyclonal) antibody covalently bound to Sepharose.


Materials Unlabeled cells, surface-labeled cells (e.g., with 125I or biotin; UNIT 3.6) or biosynthetically 35S-, 3H-, or 14C-labeled cells (UNIT 3.7) Triton X-100 lysis buffer (see recipe) Dilution buffer (see recipe) Antibody (Ab)-Sepharose (see Support Protocol) Activated, quenched (control) Sepharose, prepared as for Ab-Sepharose (see Support Protocol) but eliminating Ab or substituting irrelevant Ab during coupling Tris/saline/azide (TSA) solution (see recipe) 0.05 M Tris⋅Cl, pH 6.8 2× SDS sample buffer (UNIT 10.1) NOTE: Carry out all procedures in a 4°C cold room or on ice. Lyse cells and preclear the lysate 1. Incubate cells in Triton X-100 lysis buffer (5 × 107 cells/ml) for 1 hr at 4°C. 2. Centrifuge the lysate 10 min at 3000 × g to remove nuclei and save the supernatant. 3. Centrifuge the supernatant 1 hr at 100,000 × g and save the supernatant. Supernatants may also be prepared by microcentrifugation (10,000 × g) for 30 min. The supernatant must be used within several days or stored at −70°C. The length of storage is limited by autoradiolysis and the half-life of the isotope. 3H- and 14C-labeled samples can often be stored frozen for years. Storage of 125I-labeled samples is usually limited to 1 to 2 months because of autoradiolysis, while the usefulness of 35S-labeled samples is usually limited to 6 months because of half-life. Repeated freezing and thawing may disrupt antigenic determinants and dissociate some protein complexes, especially those that are noncovalently associated.

4. Preclear supernatant to be used in one batch by adding 10 µl activated, quenched (control) Sepharose per 200 µl supernatant. Shake on an orbital shaker 2 hr at room temperature or overnight at 4°C. Centrifuge 1 min at 200 × g and save supernatant. Preclearing removes nonspecifically absorbing material. Control Sepharose can be prepared without antibody or coupled with irrelevant (nonspecific) antibody. Irrelevant antibody is an antibody directed against an unrelated protein, and could also be whole IgG; it must not cross-react with the protein being immunoprecipitated. Affinity Purification


Supplement 18

Immunoprecipitate the antigen 5. Precoat 1.5-ml microcentrifuge tubes by filling with Triton X-100 lysis buffer 10 min at room temperature. Remove the solution by aspiration. Precoating minimizes antigen absorption to the tube.

6. Add 105 to 106 cpm of radiolabeled (125I or 35S) supernatant containing antigen (from step 4) to a precoated microcentrifuge tube and bring the volume to 200 µl with dilution buffer. The recommended amount of radioactivity is appropriate for eukaryotic cells with >1000 molecules of antigen/cell. It is assumed that detection on slab gels of 125I-labeled proteins will be carried out with enhancing screens and 35S-labeled proteins with fluorography. For nonradiolabeled samples, use 0.2 to 1 ml of precleared lysate.

cell containing unlabeled or radiolabeled protein antigens

1

1

2

2

1 lyse (detergent) 2 immunoprecipitate 3 wash antigen-specific 4 dissociate monoclonal (or polyclonal) antibody-Sepharose

antigen-specific monoclonal antibody

anti-lg antibody

3

3

4

4

(unlabeled) (radiolabeled) analyze by electrophoresis and silver stain

Immunoprecipitation

analyze by electrophoresis and autoradiography or colorimetric detection

Figure 9.8.2 Schematic representation of the stages of the immunoprecipitation protocols using either antibody-Sepharose (left, see Alternate Protocol 5) or anti-Ig serum (right, see Alternate Protocol 6). (1) Cell lysis. (2) Immunoprecipitation using specific antibodies coupled covalently to Sepharose beads (left) or specific antibodies combined with anti-Ig serum (right). (3) Washing. (4) Dissociation of the antigen-antibody complex in sample buffer for electrophoresis.



7. Add ∼10 µl of a 1:1 slurry of Ab-Sepharose/dilution buffer and shake 1.5 hr at 4°C on an orbital shaker. The antibody coupled to Sepharose is antigen specific. As described in the following support protocol, 5 mg/ml antibody per milliliter Sepharose is coupled, and the amount actually coupled can be estimated as described in step 10 of the support protocol. Shaking must be vigorous enough to suspend the Sepharose. Shaking may be extended to 3 hr; longer periods may increase background.

Wash, dissociate, and analyze the immunoprecipitate 8. Wash the Ab-Sepharose with 1 ml of the buffers listed below. After each wash, centrifuge 1 min at 200 × g or microcentrifuge 5 sec. Then, carefully aspirate the supernatant with a fine-tipped Pasteur pipet and leave 10 µl of fluid above the pellet. After the fourth wash, centrifuge again to bring down any residual drops on the side of the tube, aspirate, and leave 10 µl over the pellet. First wash: dilution buffer Second wash: dilution buffer Third wash: TSA solution Fourth wash: 0.05 M Tris⋅Cl, pH 6.8. Prepare a fine-tipped Pasteur pipet by pulling the pipet in a flame, scoring with a diamond pen, and breaking at the score.

9. Add 20 to 50 µl of 2× SDS sample buffer. Because the sample buffer has a higher density than the wash solution, it will sink into the Sepharose; do not vortex, because Sepharose may stick to side of tube above buffer level. Cap the tube securely and incubate 5 min at 100°C. 10. Vortex and centrifuge 1 min at 200 × g or microcentrifuge 5 sec. Load the supernatant, carefully avoiding the Sepharose, into a gel lane and analyze by SDS-PAGE (UNIT 10.1). 11. Detect labeled proteins by autoradiography (UNIT 10.11) with an enhancing screen (125I), by fluorography (35S, 14C, or 3H), or by colorimetric or chemiluminescent detection (biotinylated proteins; UNIT 3.6). PREPARATION OF ANTIBODY-SEPHAROSE This protocol details the procedure for covalently linking an antibody to Sepharose (an insoluble, large-pore-size chromatographic matrix) using the cyanogen bromide activation method. It is necessary to first prepare the antibody and Sepharose separately. Next, the Sepharose is activated with cyanogen bromide (alternatively, CNBr-activated Sepharose can be purchased from Amersham Pharmacia Biotech and used according to the manufacturer’s instructions). Finally, the CNBr-activated Sepharose is coupled to the antibody. Materials 1 to 30 mg/ml antigen-specific monoclonal or polyclonal antibody 0.1 M NaHCO3/0.5 M NaCl Sepharose CL-4B (or Sepharose CL-2B for high-molecular-weight antigens; Amersham Pharmacia Biotech) 0.2 M Na2CO3 Cyanogen bromide (CNBr)/acetonitrile (see recipe) 1 mM and 0.1 mM HCl, ice-cold (APPENDIX 2E) 0.05 M glycine (or ethanolamine), pH 8.0 Tris/saline/azide (TSA) solution (see recipe)

SUPPORT PROTOCOL



Supplement 18

Dialysis tubing (molecular weight cutoff >10,000) Whatman no. 1 filter paper Buchner funnel Erlenmeyer filtration flask Water aspirator Prepare the antibody 1. Dialyze 1 to 30 mg/ml antibody against 0.1 M NaHCO3/0.5 M NaCl at 4°C with three buffer changes during 24 hr. Use a volume of dialysis solution that is 500 times the volume of antibody solution. Dialysis is performed to remove all small molecules containing free amino or sulfhydryl groups (see UNIT 4.4 and APPENDIX 3C).

2. Centrifuge 1 hr at 100,000 × g, 4°C, to remove aggregates. Save the supernatant. Removal of aggregates is important. Because only some of the antibody molecules in an aggregate will be directly coupled to the Sepharose, the noncoupled antibody molecules may leach out during elution.

3. Measure the A280 of an aliquot of the solution and determine the concentration of the antibody (mg/ml IgG = A280/1.44). Dilute with 0.1 M NaHCO3/0.5 M NaCl to 5 mg/ml (or to the same concentration as desired for Ab-Sepharose) and keep at 4°C. Measure the A280 of this solution for later use in step 11. Prepare the Sepharose 4. Allow the Sepharose slurry to settle in a beaker and decant and discard the supernatant. Weigh out the desired quantity of Sepharose (assume density = 1.0). 5. Set up a filter apparatus using Whatman no. 1 filter paper in a Buchner funnel and an Erlenmeyer filtration flask attached to a water aspirator. Wash the Sepharose on the filter apparatus with 10 vol water. Sintered-glass funnels are traditionally recommended but rapidly become clogged unless coarse-porosity funnels are used.

Activate Sepharose with cyanogen bromide 6. Transfer Sepharose to 50-ml beaker and add an equal volume of 0.2 M Na2CO3. 7. Activate Sepharose at room temperature using 3.2 ml CNBr/acetonitrile per 100 ml Sepharose. Add CNBr/acetonitrile dropwise with a Pasteur pipet over 1 min, while slowly stirring the slurry with a magnetic stirrer. Continue stirring slowly for 5 min. Excessive and vigorous stirring may fracture the Sepharose beads. The protocol uses 2 g CNBr/100 ml Sepharose. Two to four grams of CNBr/100 ml Sepharose can be used to couple 1 to 20 mg of antibody/ml Sepharose. CAUTION: Activation should be carried out in a fume hood.

8. Rapidly filter the CNBr-activated Sepharose as in step 5. Aspirate to semidryness (i.e., until the Sepharose cake cracks and loses its sheen). 9. Wash with 10 vol ice-cold 1 mM HCl, then with 2 vol of ice-cold 0.1 mM HCl. Hydrate the cake with enough ice-cold 0.1 mM HCl so the cake regains its sheen, but so there is no excess liquid above the cake. Washing is most efficient if the wash solution is added evenly over the surface of the cake at about the same rate as the solution is removed by filtration. CNBr-activated Sepharose is very unstable at the alkaline pH necessary for activation; it is much more stable in dilute HCl. CNBr-activated Sepharose can be purchased premade from Amersham Pharmacia Biotech, but the coupling capacity will be lower. Immunoprecipitation



Couple antibody to CNBr-activated Sepharose 10. Immediately transfer a weighed amount of Sepharose (assume density = 1.0) to a beaker. Add an equal volume of a solution of antibody dissolved in 0.1 M NaHCO3/ 0.5 M NaCl (from step 2). Stir gently with a magnetic stirrer or rotate end over end 2 hr at room temperature or overnight at 4°C. 11. Add 0.05 M glycine (or ethanolamine), pH 8.0, to saturate the remaining reactive groups on the Sepharose and allow the slurry to settle. Remove an aliquot of the supernatant, centrifuge to remove any residual Sepharose, and measure A280. Compare absorbance to that of the A280 of the antibody solution from step 2 to determine the percentage coupling. 12. Store the Ab-Sepharose in TSA solution. IMMUNOPRECIPITATION OF RADIOLABELED ANTIGEN WITH ANTI-Ig SERUM


This protocol relies on the formation of soluble immune complexes between a protein and an antigen-specific antibody, followed by immunoprecipitation of the immune complexes by antibodies contained in anti-immunoglobulin (Ig) serum. This procedure is usually only used with radiolabeled or biotinylated antigen, as the unlabeled antibody remains in the precipitate and greatly complicates the use of any other detection method. Additional Materials (also see Alternate Protocol 5) Normal serum Anti-Ig serum (Zymed Laboratories) Antigen-specific antiserum or antigen-specific purified monoclonal antibody or antigen-specific hybridoma culture supernatant Follow the procedures in Alternate Protocol 5, with the following modifications at the indicated steps: 4a. Preclear by adding normal serum at a concentration of 2 µl/ml radiolabeled antigen. Add the proper amount of anti-Ig serum and let stand 12 to 18 hr at 4°C. Centrifuge 10 min at 1000 × g and reserve supernatant. Normal serum is the source of carrier Ig. The proper amount of anti-Ig serum must be determined by titration with radiolabeled antigen or Ig. For high-titered anti-Ig serum, this amount would be 20× to 40× the volume of antigen-specific antiserum, 2 to 4 µl/µg purified MAb, or one-third the volume of hybridoma culture supernatant.

7a. Add 1 µl antigen-specific antiserum, 3 µg antigen-specific purified MAb, or antigenspecific hybridoma culture supernatant (30 µl cloned line or 100 µl uncloned line). Vortex and allow to stand 2 hr at 4°C. Then add the proper amount of anti-Ig serum, vortex, and allow to stand 12 to 18 hr at 4°C. 8a. Wash the immunoprecipitate (see Alternate Protocol 5, step 8), except centrifuge 7 min at 1000 × g. 9a. Add 20 to 50 µl of 2× SDS sample buffer. Do not vortex, as immunoprecipitates may stick to side of tube above buffer level. Cap the tube securely. For immunoprecipitates, incubate first 1 hr at 56°C and then 5 min at 100°C. The initial 56°C incubation enhances the dissolution of the immunoprecipitates by reducing irreversible aggregation which occurs when precipitated protein is rapidly heated to 100°C. Proteolytic degradation has never been noted, probably because of the high IgG protein concentration.



Supplement 18

BASIC PROTOCOL 2

IMMUNOPRECIPITATION-RECAPTURE Once an antigen has been isolated by immunoprecipitation, it can be dissociated from the beads and reimmunoprecipitated (“recaptured”) either with the same antibody used in the first immunoprecipitation or with a different antibody (Fig. 9.8.3). Immunoprecipitationrecapture with the same antibody allows identification of a specific antigen in cases where the first immunoprecipitation contains too many bands to allow unambiguous identification. By using a different antibody in the second immunoprecipitation, immunoprecipitation-recapture can be used to analyze the subunit composition of multi-protein complexes (Fig. 9.8.4). The feasibility of this approach depends on the ability of the second antibody to recognize denatured antigens. Dissociation of the antigen from the beads is achieved by denaturation of antigen-antibody-bead complexes at high temperature in the presence of SDS and DTT. Prior to recapture, the SDS is diluted in a solution containing Triton X-100, and the DTT is neutralized with excess iodoacetamide. Recapture is then performed as in the first immunoprecipitation (see Basic Protocol 1, steps 6 to 26). Materials Elution buffer (see recipe) Beads containing bound antigen (see Basic Protocol 1, step 26) 10% (w/v) BSA Nondenaturing lysis buffer (see recipe) Heating block set at 95°C (Eppendorf Thermomixer 5436 or equivalent) 1. Add 50 µl elution buffer to 15 µl beads containing bound antigen. Mix by vortexing. The DTT in the elution buffer reduces disulfide bonds in the antigen and the antibody, and the SDS contributes to the unfolding of polypeptide chains.

2. Incubate 5 min at room temperature and 5 min at 95°C in a heating block. Cool tubes to room temperature. 3. Add 10 µl of 10% BSA. Mix by gentle vortexing. BSA is added to prevent adsorption of antigen to the tube, and to quench nonspecific binding to antibody-conjugated beads.

4. Add 1 ml nondenaturing lysis buffer. The iodoacetamide in the nondenaturing lysis buffer reacts with the DTT and prevents it from reducing the antibody used in the recapture steps. The presence of PMSF and leupeptin in the buffer is not necessary at this step.

5. Incubate 10 min at room temperature. 6. Clear the lysate and perform second immunoprecipitation (see Basic Protocol 1, steps 6 to 26). REAGENTS AND SOLUTIONS Use deionized or distilled water in all recipes and protocol steps. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

CNBr/acetonitrile To 25 g of cyanogen bromide (CNBr should be white, not yellow, crystals), add 50 ml acetonitrile to make a 62.5% (w/v) solution. This may be stored indefinitely at −20°C in a desiccator over silica. Allow to warm before opening. Immunoprecipitation

CAUTION: CNBr is a highly toxic lachrymator; handle in fume hood.



antigen 2 antigen 1 antibody 1 antibody 2

protein A– agarose bead

1

protein A–agarose bead

denaturation (see Basic Protocol 2)

2

antibody 2 binding to protein A–agarose bead (see Basic Protocol 1, steps 8 to 14)

3 recapture

wash and analysis

Figure 9.8.3 Scheme showing the stages of immunoprecipitation-recapture. (1) Dissociation and denaturation of the antigen: an antigen immunoprecipitated with antibody 1 bound to protein A–agarose beads is dissociated and denatured by heating in the presence of SDS and DTT. (2) Immobilization of the second antibody: antibody 2 is bound to protein A–agarose beads. (3) Recapture: the denatured antigen 2 (striped oval) is recaptured on antibody 2 bound to protein A–agarose beads. Alternatively, antibody 1 can be used again for further purification of the original antigen (square).



Supplement 18

Ab to:

1st

IP

2nd

IP

BSA

σ3

σ3

µ3

BSA

1

2

3

4

5

200

98 66 46

30 22

15

Figure 9.8.4 Example of an immunoprecipitation-recapture experiment. Human M1 fibroblasts were labeled overnight with [35S]methionine (UNIT 3.7) and extracted with nondenaturing lysis buffer (see Basic Protocol 1). The cell extract was then subjected to immunoprecipitation with antibodies to BSA (irrelevant antibody control; lane 1) and to the AP-3 adaptor (σ3; lane 2), a protein complex involved in protein sorting. Notice the presence of several specific bands in lane 2. The AP-3 immunoprecipitate was denatured as described in Basic Protocol 2 and individual components of the AP-3 complex were recaptured with antibodies to two of its subunits: σ3 (Mr ∼22,000; lane 3) and µ3 (Mr ∼47,000; lane 4). An immunoprecipitation with an antibody to BSA was also performed as a nonspecific control (lane 5). The amount of immunoprecipitate loaded on lanes 1 and 2 is ∼1⁄10 the amount loaded on lanes 3 to 5. Notice the presence of single bands in lanes 3 and 4. The positions of Mr standards (expressed as 10−3 × Mr) are shown at left. IP, immunoprecipitation.

Denaturing lysis buffer 1% (w/v) SDS 50 mM Tris⋅Cl, pH 7.4 (APPENDIX 2E) 5 mM EDTA (APPENDIX 2E) Store up to 1 week at room temperature (SDS precipitates at 4°C) Add the following fresh before use: 10 mM dithiothreitol (DTT; from powder) 1 mM phenylmethylsulfonyl fluoride (PMSF; store 100 mM stock in 100% ethanol up to 6 months at −20°C) 2 µg/ml leupeptin (store 10 mg/ml stock in H2O up to 6 months at −20°C) 15 U/ml DNase I (store 15,000 U/ml stock solution up to 2 years at −20°C)

Immunoprecipitation

1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF), added fresh from a 0.1 M stock solution in H2O, can be used in place of PMSF. AEBSF stock can be stored up to 1 year at −20°C.



Detergent-free lysis buffer PBS (APPENDIX 2E) containing: 5 mM EDTA (APPENDIX 2E) 0.02% (w/v) sodium azide Store up to 6 months at 4°C Immediately before use add: 10 mM iodoacetamide (from powder) 1 mM PMSF (store 100 mM stock in 100% ethanol up to 6 months at −20°C) 2 µg/ml leupeptin (store 10 mg/ml stock in H2O up to 6 months at −20°C) 1 mM AEBSF, added fresh from a 0.1 M stock solution in H2O, can be used in place of PMSF. AEBSF stock can be stored up to 1 year at −20°C.

Dilution buffer TSA solution (see recipe) containing: 0.1% Triton X-100 (store at room temperature in dark) 0.1% bovine hemoglobin (store frozen) Elution buffer 1% (w/v) SDS 100 mM Tris⋅Cl, pH 7.4 (APPENDIX 2E) Store up to 1 week at room temperature 10 mM DTT (add fresh from powder before use) Nondenaturing lysis buffer 1% (w/v) Triton X-100 (store at room temperature in dark) 50 mM Tris⋅Cl, pH 7.4 (APPENDIX 2E) 300 mM NaCl 5 mM EDTA (APPENDIX 2E) 0.02% (w/v) sodium azide Store up to 6 months at 4°C Immediately before use add: 10 mM iodoacetamide (from powder) 1 mM PMSF (store 100 mM stock in 100% ethanol up to 6 months at −20°C) 2 µg/ml leupeptin (store 10 mg/ml stock in H2O up to 6 months at −20°C) 1 mM AEBSF, added fresh from a 0.1 M stock solution in H2O, can be used in place of PMSF. AEBSF stock can be stored up to 1 year at −20°C.

Tris/saline/azide (TSA) solution 10 mM Tris⋅Cl, pH 8.0 (APPENDIX 2E) 140 mM NaCl 0.025% NaN3 CAUTION: Sodium azide (NaN3) is poisonous; wear gloves.

Triton X-100 lysis buffer TSA solution (see recipe) containing: 1% Triton X-100 (store at room temperature in dark) 1% bovine hemoglobin (store frozen) 1 mM iodoacetamide (from powder) Aprotinin (0.2 trypsin inhibitor U/ml) 1 mM PMSF (store 100 mM stock in 100% ethanol up to 6 months at −20°C) Prepare fresh 1 mM AEBSF, added fresh from a 0.1 M stock solution in H2O, can be used in place of PMSF. AEBSF stock can be stored up to 1 year at −20°C. Affinity Purification


Supplement 18

Wash buffer 0.1% (w/v) Triton X-100 (store at room temperature in dark) 50 mM Tris⋅Cl, pH 7.4 (APPENDIX 2E) 300 mM NaCl 5 mM EDTA (APPENDIX 2E) 0.02% (w/v) sodium azide Store up to 6 months at 4°C COMMENTARY Background Information

Immunoprecipitation

The use of antibodies for immunoprecipitation has its origin in the precipitin reaction (Nisonoff, 1984, and references therein). The term precipitin refers to the spontaneous precipitation of antigen-antibody complexes formed by interaction of certain polyclonal antibodies with their antigens. The precipitation arises from formation of large networks of antigen-antibody complexes, due to the bivalent or polyvalent nature of immunoglobulins and to the presence of two or more epitopes in some antigens. This phenomenon was quickly exploited to isolate antigens from protein mixtures; however, its use remained limited to antibodies and antigens that were capable of multivalent interaction. In addition, the efficiency of precipitate formation was highly dependent on the concentrations of antibody and antigen. Thus, the precipitin reaction was not generally applicable as a method for immunoprecipitation. A significant improvement was the use of secondary anti-immunoglobulin reagents (generally anti-immunoglobulin serum) to crosslink the primary antibodies, thus promoting the formation of a precipitating network. In the 1970s, immunoprecipitation became widely applicable to the study of cellular antigens as a result of several technological advances. A critical development was the introduction of methods for the production of monoclonal antibodies (Köhler and Milstein, 1975). The ability to produce unlimited amounts of antibodies with specificity against virtually any cellular antigen had a profound impact in many areas of biology and medicine. The fact that preparation of monoclonal antibodies did not require prior purification of the antigens accelerated the characterization of cellular proteins and organelles, a process in which immunoprecipitation protocols played a major role. To this day, monoclonal antibodies produced in mice or rats continue to be among the most useful tools in biology. Another important development was the discovery of bacterial Fc receptors, proteins found

on the surface of bacteria that have the property of binding a wide range of immunoglobulins. Two of the most widely used bacterial Fc receptors are protein A from Staphylococcus aureus (Kessler, 1975) and protein G from group G streptococci. Protein A and protein G bind both polyclonal and monoclonal antibodies belonging to different subclasses and deriving from different animal species (Table 9.8.1). Protein A was initially used to adsorb immunoglobulins as part of fixed, killed S. aureus particles. Both protein A and protein G are now produced in large quantities by recombinant DNA procedures and are available coupled to solid-phase matrices such as agarose. In most cases, the binding of polyclonal or monoclonal antibodies to immobilized protein A (or G) avoids the need to use a secondary antibody to precipitate antigen-antibody complexes. Because of their broad specificity and ease of use, protein A–agarose and protein G–agarose (and related products) are the state-of-the-art reagents for the isolation of soluble antigen-antibody complexes in immunoprecipitation protocols. Recent progress in the field of antibody engineering (reviewed by Rapley, 1995; Irving et al., 1996) promises to make antibody production a less time-consuming and haphazard process. Antibody fragments with high affinity for specific antigens can now be selected from phage display antibody libraries. Selected recombinant antibodies can then be produced in large quantities in Escherichia coli. Techniques have been developed for producing antibodies in soluble, secreted form. Affinity tags are added to the recombinant antibody molecules to facilitate purification, detection, and use in procedures such as immunoprecipitation. While attractive in principle, the production of recombinant antibodies has been plagued by technical difficulties that so far have limited their widespread use in biology. However, as technical problems are overcome, recombinant techniques will progressively replace immunization of animals as a way of producing anti-



bodies for immunoprecipitation and for other applications.

Critical Parameters Extraction of antigens Isolation of cellular antigens by immunoprecipitation requires extraction of the cells so that the antigens are available for binding to specific antibodies, and are in a physical form that allows separation from other cellular components. Extraction with nondenaturing detergents such as Triton X-100 (see Basic Protocol 1, Alternate Protocol 1, and Alternate Protocol 5) or in the absence of detergent (see Alternate Protocol 3) allows immunoprecipitation with antibodies to epitopes that are exposed on native proteins. Other nondenaturing detergents such as Nonidet P-40, CHAPS, digitonin, or octyl glucoside are also appropriate for extraction of native proteins (APPENDIX 1B). Some of these detergents (e.g., digitonin) preserve weak protein-protein interactions better than Triton X-100. If the antigen is part of a complex that is insoluble in nondenaturing detergents (e.g., cytoskeletal structures, chromatin, membrane “rafts”) or if the epitope is hidden within the folded structure of the protein, extraction under denaturing conditions is indicated (see Alternate Protocol 2). Alternate Protocol 2 may also be indicated for in vitro–translated products, which often tend to form aggregates (Anderson and Blobel, 1983). The number of cells necessary to detect an immunoprecipitated antigen depends on the cellular abundance of the antigen and on the efficiency of radiolabeling. The protocols for radiolabeling (UNIT 3.7) and immunoprecipitation described in this book are appropriate for detection of antigens that are present at low to moderate levels (10,000 to 100,000 copies per cell), as is the case for most endogenous integral membrane proteins, signal transduction proteins, and transcription factors. For more abundant antigens, such as cytoskeletal and secretory proteins or proteins that are expressed by viral infection or transfection, the quantity of radiolabeled cells used in the immunoprecipitation can be reduced accordingly. Production of antibodies Immunoprecipitation can be carried out using either polyclonal or monoclonal antibodies (see discussion of selection below). Polyclonal antibodies are most often prepared by immunizing rabbits, although polyclonal antibodies produced in mice, guinea pigs, goats, sheep,

and other animals are also suitable for immunoprecipitation. Antigens used for polyclonal antibody production can be whole proteins purified from cells or tissues, or whole or partial proteins produced in bacteria or insect cells by recombinant DNA procedures. Another useful procedure is to immunize animals with peptides conjugated to a carrier protein. Production of polyclonal antibodies to recombinant proteins and peptides has become the most commonly used approach to obtain specific probes for immunoprecipitation and other immunochemical techniques, because it does not require purification of protein antigens from their native sources. The only requirement for making these antibodies is knowledge of the sequence of a protein, which is now relatively easy to obtain as a result of cDNA library production and genomic DNA sequencing projects. Polyclonal antibodies can be used for immunoprecipitation in the form of whole serum, ammonium sulfate–precipitated immunoglobulin fractions, or affinity-purified immunoglobulins. Although all of these are suitable for immunoprecipitation, affinity-purified antibodies often give lower backgrounds and are more specific. Most monoclonal antibodies are produced in mice or rats. The sources of antigen for monoclonal antibody production are the same as those for production of polyclonal antibodies, namely proteins isolated from cells or tissues, recombinant proteins or protein fragments, and peptides. A significant advantage of using monoclonal antibodies is that antigens do not need to be purified to serve as immunogens, as long as the screening method is specific for the antigen. Another advantage is the unlimited supply of monoclonal antibodies afforded by the ability to grow hybridomas in culture or in ascitic fluid. Many monoclonal antibodies can now be produced from hybridomas deposited in cell banks or are directly available commercially. Ascitic fluid, cell culture supernatant, and purified antibodies are all suitable sources of monoclonal antibodies for immunoprecipitation. Ascitic fluid and purified antibodies should be used when a high antibody titer is important. Cell culture supernatants have lower antibody titers, but tend to give cleaner immunoprecipitations than ascitic fluids due to the lack of contaminating antibodies. Selection of antibodies: Polyclonal versus monoclonal What type of antibody is best for immunoprecipitation? There is no simple answer to this



Supplement 18

Immunoprecipitation

question, as the outcome of both polyclonal and monoclonal antibody production protocols is still difficult to predict. Polyclonal antibodies to whole proteins (native or recombinant) have the advantage that they frequently recognize multiple epitopes on the target antigen, enabling them to generate large, multivalent immune complexes. Formation of these antigenantibody networks enhances the avidity of the interactions and increases the efficiency of immunoprecipitation. Because these antibodies recognize several epitopes, there is a better chance that at least one epitope will be exposed on the surface of a solubilized protein and thus be available for interaction with antibodies. Thus, the likelihood of success is higher. These properties can be a disadvantage, though, as some polyvalent antibodies can cross-react with epitopes on other proteins, resulting in higher backgrounds and possible misidentification of antigens. Because they are directed to a short peptide sequence, anti-peptide polyclonal antibodies are less likely to cross-react with other proteins. However, their usefulness is dependent on whether the chosen sequence turns out to be a good immunogen in practice, as well as on whether this particular epitope is available for interaction with the antibody under the conditions used for immunoprecipitation. Unfractionated antisera are often suitable for immunoprecipitation. However, there is a risk that serum proteins other than the antibody will bind nonspecifically to the immunoadsorbent, and in turn bind proteins in the lysate that are unrelated to the antigen. For instance, transferrin can bind nonspecifically to immunoadsorbents, potentially leading to the isolation of the transferrin receptor as a contaminant (Harford, 1984). Polyclonal antisera can also contain antibodies to other antigens (e.g., viruses, bacteria) to which the animal may have been exposed, and these antibodies can also crossreact with cellular proteins during immunoprecipitation. Affinity-purified antibodies are a better alternative when antisera do not yield clean immunoprecipitations. Affinity-purification can lead to loss of high-affinity or low-affinity antibodies; however, the higher specificity of affinity-purified antibodies generally makes them “cleaner” reagents for immunoprecipitation. The specificity, high titer, and limitless supply of the best immunoprecipitating monoclonal antibodies are unmatched by those of polyclonal antibodies. However, not all monoclonal antibodies are useful for immunopre-

cipitation. Low-affinity monoclonal antibodies can perform acceptably in immunofluorescence microscopy protocols but may not be capable of holding on to the antigen during the repeated washes required in immunoprecipitation protocols. The use of ascitic fluid has the same potential pitfalls as the use of polyclonal antisera, as ascites may also contain endogenous antibodies to other antigens and proteins such as transferrin that can bind to other proteins in the lysate. In conclusion, an informed empirical approach is recommended in order to select the best antibody for immunoprecipitation. In general, it is advisable to generate and/or test several antibodies to a particular antigen in order to find at least one that will perform well in immunoprecipitation protocols. Antibody titer The importance of using the right amount of antibody for immunoprecipitation cannot be overemphasized. This is especially the case for quantitative immunoprecipitation studies, in which the antibody should be in excess of the specific antigen. For instance, in pulse-chase analyses of protein degradation or secretion (UNIT 3.7), it is critical to use sufficient antibody to deplete the antigen from the cell lysate. This is particularly important for antigens that are expressed at high levels, a common occurrence with the growing use of high-yield protein expression systems such as vaccinia virus (UNITS 5.11-5.15) or replicating plasmids in COS cells. Consider, for example, a protein that is expressed at high levels inside the cell, and of which only a small fraction is secreted into the medium. If limiting amounts of antibody are used in a pulse-chase analysis of this protein, the proportion of protein secreted into the medium will be grossly overestimated, because the limiting antibody will bind only a small proportion of the cell-associated protein and a much higher proportion of the secreted protein. The same considerations apply to degradation studies. Thus, it is extremely important in quantitative studies to ensure that the antibody is in excess of the antigen in the cell samples. This can be ascertained by performing sequential immunoprecipitations of the samples (see Basic Protocol 1, annotation to step 21). If the second immunoprecipitation yields only a small amount of the antigen relative to that isolated in the first immunoprecipitation (10%, either more antibody or less antigen should be used. Too much antibody can also be a problem, as nonspecific immunoprecipitation tends to increase with increasing amounts of immunoglobulins bound to the beads. Thus, titration of the antibody used for immunoprecipitation is strongly advised. Immunoadsorbent If cost is not an overriding issue, the use of protein A– or protein G–agarose is recommended for routine immunoprecipitation (see Basic Protocol 1). Protein A– or protein G– agarose beads (or equivalent products) have a very high capacity for antibody binding (up to 10 to 20 mg of antibody per milliliter of gel). Both protein A and protein G bind a wide range of immunoglobulins (Table 9.8.1). Backgrounds from nonspecifically bound proteins are generally low. Protein A– and protein G– agarose beads are also stable and easy to sediment by low-speed centrifugation. A potential disadvantage, in addition to their cost, is that some polyclonal or monoclonal antibodies bind weakly or not at all to protein A or protein G (Table 9.8.1). This problem can be solved by using an intermediate rabbit antibody to the immunoglobulin of interest. For example, a goat polyclonal antibody can be indirectly bound to protein A–agarose by first incubating the protein A–agarose beads with a rabbit antigoat immunoglobulin, and then incubating the beads with the goat polyclonal antibody. Antiimmunoglobulin antibodies (e.g., rabbit anti– goat immunoglobulins) coupled covalently to agarose can also be used for indirect immunoprecipitation in place of protein A– or protein G–agarose. A less expensive alternative to protein A– or protein G–agarose is the use of anti-Ig serum to crosslink the primary antibody (see Alternate Protocol 6). This procedure can result in very low backgrounds, although it requires proper titration of the anti-Ig serum. Protein A– agarose can also be substituted by fixed Staphylococcus aureus particles (Pansorbin). They have a lower capacity, can give higher backgrounds, and take longer to sediment. However, they work quite well in many cases. In order to establish if they are appropriate for a particular experimental setup, conduct a preliminary comparison of the efficiency of protein A– agarose with Staphylococcus aureus particles as immunoadsorbent. Specific antibodies coupled covalently to various affinity matrices can also be used for

direct immunoprecipitation of antigens (see Alternate Protocol 5). After binding to protein A–agarose, antibodies can be cross-linked with dimethylpimelimidate (Gersten and Marchalonis, 1978). Purified antibodies can also be coupled directly to derivatized matrices such as CNBr-activated Sepharose (see Support Protocol). This latter approach avoids having to bind the antibody to protein A–agarose. Covalently bound antibodies should be used when elution of immunoglobulins from the beads complicates further analyses of the complexes. This is the case when proteins in immunoprecipitates are analyzed by one- or two-dimensional gel electrophoresis (UNITS 10.1-10.4) followed by Coomassie blue or silver staining, or are used for microsequencing. Also, the released immunoglobulins could interfere with detection of some antigens by immunoblotting (UNIT 10.10) following immunoprecipitation. The support protocol for coupling protein antigens to CNBr-activated Sepharose is a modification of the methods of Cuatrecasas (1970) and March et al. (1974). As originally described, the washing was done at alkaline pH. Because activated Sepharose is very unstable at this pH, it was originally recommended that washing, adding the protein ligand, and mixing be done in 99%. Higher amounts of CNBr may result in multipoint attachment of IgG molecules to the matrix, thereby reducing accessibility to antigen. Most investigators purchase CNBr-activated Sepharose, while others, to achieve a higher coupling efficacy or to avoid the expense of the commercial product, prefer to prepare it themselves. Quantities and ratios recommended in these protocols have been found to work well with several hundred monoclonal antibodies and more than 40 different antigens. However, titration of Ab-Sepharose or sandwich reagents versus the protein antigen may further optimize a given immunoprecipitation. Nonspecific controls For correct interpretation of immunoprecipitation results, it is critical to include appropriate nonspecific controls along with the spe-



Supplement 18

additional wash with 0.1% SDS/0.1% DOC

PI I

PI I

200

46

30

22

15 1

2

3

4

Figure 9.8.5 Lowering background by washing with SDS and sodium deoxycholate (DOC). In this experiment, BW5147 cells (mouse thymoma) labeled with [35S]methionine for 1 hr were extracted with nondenaturing lysis buffer (see Basic Protocol 1). The extracts were subjected to immunoprecipitation with protein A–agarose beads incubated with either preimmune (PI) or immune (I) serum from a rabbit immunized with the ribosomal protein L17 (doublet at Mr ∼22,000). Lanes 1 and 2 correspond to immunoprecipitates obtained using the protocols described in this unit. Notice the presence of nonspecific bands and/or associated proteins in lane 2. Lanes 3 and 4 correspond to beads that were washed an additional time with 0.1% (w/v) SDS and 0.1% (w/v) DOC. Notice the disappearance of most of the nonspecific bands and/or associated proteins. The positions of Mr standards (expressed as 10−3 × Mr) are shown at left.

Immunoprecipitation

cific samples. One type of control consists of setting up an incubation with an irrelevant antibody in the same biochemical form as the experimental antibody (e.g., serum, ascites, affinity-purified immunoglobulin, antibody bound to protein A–agarose or directly conjugated to agarose), and belonging to the same species and immunoglobulin subclass as the experimental antibody (e.g., rabbit antiserum, mouse IgG2a). For an antiserum, the best control is preimmune serum (serum from the same animal obtained before immunization). Nonimmune serum from the same species is an accept-

able substitute for preimmune serum in some cases. “No-antibody” controls are not appropriate because they do not account for nonspecific binding of proteins to immunoglobulins. In immunoprecipitation-recapture experiments, control immunoprecipitations with irrelevant antibodies should be performed for both the first and second immunoprecipitation steps (Fig. 9.8.4). Another type of control is to perform an immunoprecipitation from cells that do not express a specific antigen in parallel with immunoprecipitation of the antigen-expressing cells. For instance, untransfected cells are a



perfect control for transfected cells. In yeast cells, null mutants that do not express a specific antigen are an ideal control for wild-type cells. Order of stages In the immunoprecipitation procedure described in Basic Protocol 1, the antibody is prebound to protein A–agarose before addition to the cell lysate containing the antigen. This differs from other methods in which the free antibody is first added to the lysate and the antigen-antibody complexes are then collected by addition of the immunoadsorbent. Although both procedures can give good results, the authors prefer the protocol described here because this method allows better control of the amount of antibody bound to the immunoadsorbent. Prebinding antibodies to the immunoadsorbent beads allows removal of unbound antibodies. The presence of unbound antibodies in the incubation mixture could otherwise result in decreased recovery of the antigen on the immunoadsorbent beads. Another advantage of the prebinding procedure is that most proteins other than the immunoglobulin in the antibody sample (e.g., serum proteins) are removed from the beads and do not come in contact with the cell lysate. This eliminates potential adverse effects of these proteins on isolation of the antigen. Washing The five washes described in Basic Protocol 1 (four with wash buffer and one with PBS) are sufficient for maximal removal of unbound proteins; additional washes are unlikely to decrease the background any further. The last wash with PBS removes the Triton X-100 that can lead to decreased resolution on SDS-PAGE. It also removes other components of the wash buffer that could interfere with enzymatic treatment of immunoprecipitates. It is not advisable to complete all the washes quickly (e.g., in 5 min), because this may not allow enough time for included proteins to diffuse out of the gel matrix. Instead, beads should be washed over ∼30 min, which may require keeping the samples on ice for periods of 3 to 5 min between washes. In order to reduce nonspecific bands, samples can be subjected to an additional wash with wash buffer containing 0.1% (w/v) SDS, or with a mixture of 0.1% (w/v) SDS and 0.1% (w/v) sodium deoxycholate (Fig. 9.8.5). This wash should be done before the last wash with PBS (in Basic Protocol 1) or before the wash with 0.05 M Tris⋅Cl, pH 6.8 (in Alternate Protocol 5).

Troubleshooting Two of the most common problems encountered in immunoprecipitation of metabolically labeled proteins are failure to detect specific antigens in the immunoprecipitates, and high background of nonspecifically bound proteins for antigens that were radiolabeled in vivo and analyzed by SDS-PAGE (UNIT 10.1) followed by autoradiography or fluorography (UNIT 10.11). When immunoprecipitates are analyzed by immunoblotting (UNIT 10.10), an additional problem may be the detection of immunoprecipitating antibody bands in the blots (Table 9.8.2).

Anticipated Results For antigens that are present at >10,000 copies per cell, the radiolabeling and immunoprecipitation protocols described in this book can be expected to result in the detection of one or more bands corresponding to the specific antigen and associated proteins in the electrophoretograms. Specific bands should not be present in control immunoprecipitations done with irrelevant antibodies. If antigens are labeled with [35S]methionine (UNIT 3.7), specific bands should be visible within 2 hr to 2 months of exposure. Due to the relatively low yield of the immunoprecipitation-recapture procedure (80%) pure protein in one chromatographic step, purification from insect and mammalian cells, which contain a higher percentage of His residues in their proteins than E. coli, can lead to significant background binding to immobilized metal ions. This may be circumvented by using stringent wash conditions (e.g., 5 to 10 mM imidazole), although a stringent wash may cause premature elution of the protein of interest. The location of the tag (N terminal, C terminal, or internal) can also have an effect on IMAC. If a change in tag location does not increase the effectiveness of IMAC, a denaturing purification can be attempted. Primary antibodies have also been developed for the detection of polyhistidine fusion proteins in vitro. Again, because of the predominance of histidine residues in mammalian and insect systems, anti-polyhistidine antibodies are notoriously promiscuous. Ni2+ resin can also be used to precipitate a polyhistidinetagged protein for the detection of protein-protein interactions.

Glutathione S-Transferase

Overview of Affinity Tags for Protein Purification

The pGEX E. coli expression vectors, which encode for N-terminal glutathione S-transferase (GST) molecules followed by protease cleavage sites, were first designed and used to express and purify antigens of the parasite Taenia ovis in 1988 (Smith and Johnson, 1988; Smith, 2000). Currently, pGEX vectors are available from Amersham Biosciences in all three reading frames and with three different protease cleavage sites (e.g., thrombin, factor Xa, and PreScission). GST fusion proteins can be purified by affinity chromatography (UNIT 6.6) on commercially available glutathione (γ-glutamylcysteinylglycine) Sepharose, which is affected by γ-glutamyl transpeptidase activity in crude cell lysates. Therefore, glutathione resin has a finite lifetime and can only be regenerated and reused between four and twenty times. Glutathione affinity chromatography is amenable to low concentrations of denaturing agents (2 to 3 M urea or guanidine hydrochloride), reducing agents (100 residues in length and readily dimerizes in solution. While the cost of this expression system (>$600.00) might discourage its use as a first-line affinity tag, the HisPatch ThioFusion tag might be useful when

other attempts at producing large amounts of soluble protein have failed.

NorpA The type II N-terminal PDZ domain (PDZ1) of InaD binds the C-terminus of no receptor potential A (NorpA), the relevant phospholipase C β isozyme in the Drosophila phototransduction pathway. The crystal structure of PDZ1 in complex with a peptide corresponding to the NorpA C-terminus shows that only the last five residues of NorpA contact PDZ1 and that a disulfide bond is the major intermolecular interaction (Kimple et al., 2001). The short PDZ1 binding motif of NorpA, coupled with its covalent yet dissociable interaction, led to the hypothesis that the NorpA C-terminal residues (Thr-Glu-Phe-Cys-Ala) could be used as an affinity tag for protein detection and purification by appropriately modified PDZ1



Supplement 36

medium quantity 0.1–1.0 mg purity tagged or untagged protein

tagged only

both tagged and untagged

untagged only

Is detection of tag important?

BCCP tag

intein-CBD tag

yes

no

HIS-tag

GST-tag

Figure 9.9.2 Flow chart describing general scheme for selecting an affinity tag for protein purification if a mid-range amount (100 µg to 1 mg) of highly pure tagged and/or untagged protein is needed. An example of an experimental method requiring this quantity of tagged protein is surface plasmon resonance (SPR), where a biotin moiety could be used to attach a protein ligand to a streptavidin-coated SPR chip. Other possible applications are in GST and Ni2+ precipitation assays, and mass spectrometry. Certain variants of the intein-CBD tag are useful for methods requiring a completely native protein, as it is possible to cleave off every intein residue from the fusion protein, leaving a native N or C terminus. Finally, the His- and GST-tags are useful because both tagged and untagged proteins can easily be, prepared during the same protein preparation, and there are many secondary reagents, such as antibodies, resins, SPR chips, and ELISA assays that are based on these fusion systems. Thick lines indicate the desired result has been reached.

small quantity (< 100 µg ) of moderate-tohigh purity tagged protein Will tag be used as a reporter?

Overview of Affinity Tags for Protein Purification

yes

no

AP or CAT

Epitope tag

Figure 9.9.3 Flow chart describing general scheme for selecting an affinity tag for purification of a small amount (70%); matrix compatible with nonionizing detergents and high salt, but not reducing agents; can increase expression of eukaryotic proteins in bacteria; anti-MBP antibodies specific; tag at N-term can decrease translation efficiency; very large size of tag may affect fusion protein properties

continued

Nilsson et al. (1997b), Terpe (2003)

Karp and Luminescent; can serve as a reporter immediately upon translation; useful for studies Oker-Blom (1999) involving in situ hybridization, RNA processing, RNA transfection or coupled in vitro transcription/translation, protein folding, and imaging; can be labeled with 35S; no more than five codons can be removed from the N- or C-term to maintain enzymatic activity; very large tag may affect properties of fusion

Tai et al. (1988)

Also known as β-galactosidase or β-Gal; enzymatic assay available for protein quantification; may increase proteolytic stability of fusion proteins to increase expression; fusion proteins may be insoluble; extremely large tag which forms tetramers in solution, potentially affecting properties of fusion protein

Detection, purification, and increased expression

Detection

Kwatra et al. (1995)

References

Short, linear recognition motif; antibody purification does not give high yields; low pH elution may irreversibly affect protein properties; matrix is of limited reusability

Comments

Detection and purification

Typical use(s)

9.9.15


Supplement 36

11 (CEQKLISE EDL)

5 (TEFCA)

495

5–6 (usually 5; RRRRR)

5–16 (DDDDD)

4 (CCCC)

NorpA

NusA

Polyarginine (Arg-tag)

Polyaspartate (Asp-tag)

Polycysteine (Cys-tag)

Lengthc (sequence)

N-term

C-term

C-term

N-term or C-term

C-term

N-term, C-term, or internal

Position pDual Expression System (Stratagene), PRO bacterial expression system (Clontech)

Commercial systems (supplier)

NA

NA

NA

Thiopropyl-Sepharo NA sethiol-containing reducing agent (e.g., DTT, β-ME)

Anion exchange resin/low-neutral pH salt gradient

Cation exchange resin/high pH salt gradient

NA

PDZ1/thiol-containi NA ng reducing agent (e.g., DTT, β-ME)

mAb/low pH

Matrix/elution

Characteristics of Protein Affinity Tagsa,b, continued

Myc

Tag

Table 9.9.1

Purification

Purification

Purification and immobilization

Increased expression and solubility

Detection, purification, and immobilization


Typical use(s)

Terpe (2003)

Terpe (2003)

Kimple and Sondek (2002)

Kolodziej and Young (1991)

References

Short, linear recognition motif; moderately pure protein in one step; purification must be performed in absence of thiol-containing reducing agents until elution step; reducing elution may disrupt properties of fusion protein

continued

Stevens (2000)

Short, linear recognition motif; polar tag may Stevens (2000) affect tertiary structure of protein and/or protein properties

Can immobilize targets on mica for microscopy studies; short, linear recognition motif; very pure protein in one step; charged tag may affect tertiary structure of protein and/or protein properties; limited success of tag cleavage by carboxypeptidase B

Anti-transcription termination factor; increases solubility and expression of fusion proteins; must be used in conjunction with another affinity tag for protein purification; large tag may affect properties of fusion protein

Short, linear recognition motif; alkaline phosphatase-coupled PDZ1 allows antibody-independent detection; PDZ1-NorpA interaction highly specific; PDZ1 can couple fusion proteins to SPR resonance chip; purification must be performed in absence of thiol-containing reducing agents until elution step; reducing elution may disrupt properties of fusion protein; tag must be at C-term to bind PDZ1

Short, linear recognition motif; anti-myc antibody somewhat promiscuous; antibody purification does not give high yields; low pH elution may irreversibly affect protein properties; matrix is of limited reusability

Comments

9.9.16

Supplement 36


9 (NANNPDWD F)

S1-tag

N-term or C-term

N-term or C-term

12

Protein C

N-term or C-term

Position

N-term

2–10 (usually 6; HHHHHH)

Lengthc (sequence)

NA

pXB, pBX, pXM, and pMX vectors (Roche)

mAb/Ca2+ buffer

mAb/low pH

NA

QIAexpress system (Qiagen), Selected pET directional TOPO, pBAD, and Gateway systems (Invitrogen)


Phenyl-Sepharose/ ethylene glycol

Divalent metal (i.e., Ni2+, Co2+, Cu2+, or Zn2+)/imidazole or low pH

Matrix/elution


Polyphenylalan 11 ine tag (FFFFFFFF (Phe-tag) FFF)

Polyhistidine (His-tag)

Tag

Table 9.9.1



Purification

Detection, purification and immobilization

Typical use(s)

Hepatitis B virus S1 region; short, linear recognition motif; AP1 antibody specific; has been tested in bacterial and mammalian expression systems; relatively pure protein in one step; antibody purification does not give high yields, low pH elution may irreversibly affect protein properties, and matrix is of limited reusability

Short, linear recognition motif; anti-PC antibody binds in Ca2+-dependent manner; elution by Ca2+ in physiological buffer conditions; antibody purification does not give high yields

Short, linear recognition motif; moderately pure protein in one step; nonpolar tag or ethylene glycol elution may disrupt properties of fusion protein

Most common purification tag; short, linear recognition motif; one-step purification of 20%–80% pure protein, depending on fusion protein expression levels; denaturing purification possible; matrix may be regenerated and reused indefinitely; can be used to immobilize fusion to Ni-NTA SPR chip, but significant dissociation complicates data analysis; Tag or elution may affect protein properties; detection antibodies highly promiscuous

Comments

continued

Berlot (1999)

Fritze and Anderson (2000)

Stevens (2000)

Bornhorst and Falke (2000)

References

9.9.17


Supplement 36

N-term or C-term

8–9 (WSHPQFEK or AWAHPQPG G)

159

Strep-tag

Streptavidin

N-term or C-term

N-term or C-term

Staphylococcal 280 protein G (Protein G)

N-term, C-term or internal

Position

N-term

15 (KETAAAKF ERQHMDS)

Lengthc (sequence) S-Tag system (Novagen)


Biotin/biotin or denaturation (e.g., heat, urea)

Strep-Tactin (modified streptavidin)/biotin or desthiobiotin

Amylose/low pH or amylose

NA

Strep-tag II system (Sigma-Genosys), pASK75 vector (Biometra)

NA

IgG/Low pH or IgG pEZZ 18 and pRIT2T vectors (Amersham Biosciences)

S-fragment of RNase A/low pH

Matrix/elution


Staphylococcal 280 protein A (Protein A)

S-tag

Tag

Table 9.9.1

Detection, purification, increased expression, and immobilization

Detection, purification, and immobilization

Purification and increased solubility

Purification and increased solubility


Typical use(s)

Skerra and Schmidt (2000)

Nilsson et al. (1997b)

Nilsson et al. (1997b)


References

continued

May increase proteolytic stability of fusion Sano et al. proteins to increase expression; extremely high (1998) affinity for biotin useful for immobilization of fusion on surfaces such as SPR chips; large size or tetramer formation may disrupt properties of fusion protein; fusion protein may not be released upon addition of free biotin, necessitating denaturing elution followed by refolding; newer streptavidin mutants that have lower affinities for biotin useful for purification

Short, linear recognition motif; matrix regenerable; useful for purification under anaerobic conditions, eukaryotic cell surface display, and immobilization to streptavidin-coated surfaces (e.g., SPR chips); specific binding conditions may be unsuitable for some fusions

Proteolytically stable; may increase solubility of fusion; fusion proteins secreted; purification does not give high yields; large tag size and/or low pH elution may irreversibly affect protein properties; matrix is of limited reusability

Proteolytically stable; may increase solubility of fusion; fusion proteins secreted; purification does not give high yields; large tag size and/or low pH elution may irreversibly affect protein properties; matrix is of limited reusability

Short, linear recognition motif; RNase S assay possible for quantitative assay of expression levels; colorimetric assays used for detection without antibody; tag or low pH elution may irreversibly affect protein properties; matrix is of limited reusability

Comments

9.9.18

Supplement 36


Lengthc (sequence)

38

260

109

25–336

76

6 (HTTPHH)

SBP

T7

Trx

TrpE

Ubiquitin

Universal

N-term or C-term or internal

N-term

N-term or C-term

N-term or C-term

N-term

C-term

Position

NA

NA


mAb/low pH

NA

mAb/low pH

NA

NA

NA

ThioFusion System Phenylarsinine oxide/thiol-containin (Invitrogen) g reducing agent (e.g. DTT, β-ME)

mAb/low pH

Streptavidin/biotin

Matrix/elution


Tag

Table 9.9.1


Increased solubility

Purification and increased expression

Increased solubility

Purification and increased expression

Purification and immobilization

Typical use(s)

References

Stevens (2000)

Terpe (2003)

Sequence HTTPHH is translated regardless of reading frame for ease in cloning; multiple tag copies increase antibody specificity; immobilized mAb can bind multiple tag copies in SPR studies; antibody purification does not give high yields; low pH elution may irreversibly affect protein properties; matrix is of limited reusability

continued

Nelson et al. (1999)

May increase solubility of proteins expressed in Stevens (2000) E. coli; not useful for expression in eukaryotic cells.

Larger constructs expressed may be targeted to inclusion bodies (allowing high-level expression of toxic genes; antibody purification does not give high yields; low pH elution may irreversibly affect protein properties; matrix is of limited reusability; large tag may affect properties of fusion protein

Heat stable; may increase solubility of fusion proteins; convenient purification of crude periplasmic extract from bacteria; purification must be done in absence of thiol-containing reducing agents until elution step; large tag or elution conditions may affect properties of fusion protein

May increase expression of fusion proteins; Stevens (2000) insoluble protein is targeted to inclusion bodies; denaturing purification of toxic proteins necessitates refolding

Relatively short recognition motif; Terpe (2003) immobilization of protein to various media (e.g., streptavidin-coated beads, SPR chips); tag at C-term only

Comments


11 (YTDIEMNR LGK)

Lengthc (sequence) C-term

Position mAb/low pH

Matrix/elution


pVB6, pBV, pVM6, and pMV vector set (Roche)

Commercial systems (supplier) Detection and purification

Typical use(s)

C-term residues of VSV-G; relatively pure protein in one step; antibody purification does not give high yields; low pH elution may irreversibly affect protein properties; matrix is of limited reusability

Comments


References

cSequence lengths are reported in amino acids.

bAdapted, with permission, from BioMedicalPDA (http://www.biomedicalpda.com), Professional PDA Publishing, LLC, Larchmont, N.Y.

aAbbreviations: ABP, albumin-binding protein; AP, alkaline phosphatase; APTG, p-amino-phenyl-beta-D-thiogalactosidase; β-ME, 2-mercaptoethanol; BCCP, biotin carboxyl carrier protein; CA, chloramphenicol acetyl transferase; C-term, C terminal; DHFR, dihydrofolate reductase; DTT, dithiothreitol; FRET, fluorescence resonance energy transfer GBP; galactose-binding protein; GFP, green fluorescent protein; GST, glutathione S-transferase; HA, hemagglutinin; HAT, histidine-affinity tag; HSV, Herpes simplex virus peptide; KSI, ketosteroid isomerase; mAb, monoclonal antibody; MBP, maltose-binding protein; NA, not applicable; N-term, N terminal; SBP, streptavidin-binding peptide; T7-tag, T7 gene 10 tag; Trx, thioredoxin; VSV-G, vesicular stomatitis virus glycoprotein peptide.

VSV-G

Tag

Table 9.9.1


9.9.19

Supplement 36

CHAPTER 10 Electrophoresis INTRODUCTION

E

lectrophoresis of protein samples in polyacrylamide gels is an indispensable analytical and, in some cases, preparative tool for the protein scientist. Electrophoresis can be used to separate and compare complex protein mixtures, evaluate purity of a protein during the course of its isolation, and provide estimates of physical characteristics such as subunit composition, isoelectric point, size, and charge. Each type of electrophoretic separation can be conducted in a variety of gel sizes ranging from microgels (e.g., Phast gels from Hoefer Pharmacia) only slightly larger than a postage stamp to giant gels much larger than this page (Garrels, 1979; Young et al., 1983). In general, the time required to electrophorese, stain, and destain small gels is very short, and a minimal amount of sample is consumed. In contrast, larger gels consume more reagents, sample, and time, but provide increased resolution. Small gels are therefore recommended for rapid screening, and larger gels are indicated when maximum resolution is required, as in analysis of complex mixtures or samples containing very similar components.

The most common one-dimensional gel methods utilize the detergent sodium dodecyl sulfate (SDS) to solubilize, denature, and impart a strong negative charge to proteins (UNIT 10.1). Although many one-dimensional SDS gel methods have been published over the past 30 years, the single most widely used method is that initially described by Laemmli (1970). SDS-based gel separations are very robust as most, but not all, proteins are readily solubilized in SDS solutions. Most proteins bind a uniform amount of SDS per microgram of protein, which imparts a uniform charge density per unit mass to provide a separation based on the mass of the polypeptide chain. An alternative one-dimensional separation methodology is based on the protein’s isoelectric point and is usually performed in the presence of denaturants such as urea (UNIT 10.2). Like one-dimensional SDS gel separations, this denaturing isoelectric focusing method is capable of separating a large number of components in a single dimension. A third one-dimensional method uses native conditions to separate proteins based on intrinsic charge (UNIT 10.3). Although this method has less resolving power than the previous two methods, it can be particularly useful for associating a specific electrophoretically separated component with a biological activity, since no denaturants are used. Two-dimensional electrophoresis (UNIT 10.4) involves the orthogonal combination of two different electrophoretic methods. Any two electrophoresis techniques can be combined in this manner to produce useful separations. The most common two-dimensional approach combines two individually high-resolution methods, consisting of isoelectric focusing (UNIT 10.2) followed by SDS gel electrophoresis (UNIT 10.1). Each of these methods is individually capable of resolving up to about 100 protein bands; when the two are combined, more than 1,000 proteins can be separated on a single two-dimensional gel (O’Farrell, 1975; Garrels, 1979). Hence this method can be used to analyze and compare complex protein mixtures, including whole-cell or tissue extracts. One important application of two-dimensional gels is the systematic analysis and quantitative comparison of whole-cell or tissue extracts for the study of the proteome, or complete protein profile, of Electrophoresis Contributed by David W. Speicher Current Protocols in Protein Science (1999) 10.0.1-10.0.3 Copyright © 1999 by John Wiley & Sons, Inc.


the cell or tissue. In recent years a growing emphasis has been focused on such studies, particularly for organisms whose complete genome sequence has been determined. Other applications of two-dimensional electrophoresis include assessment of the purity and homogeneity of a purified protein, including recombinant proteins (Chapter 7), and evaluation of some post-translational modifications (Chapters 12 and 13). After electrophoretic separation, proteins are usually detected using general protein staining methods, such as Coomassie blue staining or the more sensitive silver staining. These staining methods “fix” the proteins in the gel matrix by either chemical cross-linking or denaturation to prevent subsequent diffusion of the protein bands (UNIT 10.5). When proteins are to be eluted or extracted from the gels for subsequent analyses, however, fixation-based stains should usually be avoided. In these cases, methods for protein detection without fixation should be used (UNIT 10.6) when it is necessary to stain a gel to locate the protein of interest. A useful method of isolating proteins from gels for subsequent analysis involves electroblotting onto an inert support, such as polyvinylidene difluoride (PVDF) membranes (UNIT 10.7). As a result of the versatility of PVDF membranes, electroblotting has largely replaced electroelution as the method of choice for isolating proteins from most types of polyacrylamide gels (LeGendre and Matsudaira, 1988). PVDF membranes have very high protein binding capacities, good handling characteristics when either dry or wet, and are highly chemically inert. Hence they are compatible with many methods for subsequent analysis, including amino acid analysis, in situ chemical modification, in situ protease digestion, and N-terminal sequence analysis (Chapter 11). A wide range of general protein stains exist that are compatible with most blotting membranes (UNIT 10.8). Alternatively, radiolabeled proteins can be detected by autoradiography either in the gel or after transfer to a PVDF membrane (UNIT 10.11). The latter is normally preferable; as long as the proteins of interest are electrotransferred in high yield, autoradiography signals of weak β emitters such as 14C or 35S are usually higher on electroblotted membranes than in dried gels (even when fluorophores are impregnated in the gel), because the protein is primarily located on the surface of the membrane that was in contact with the gel during electroblotting. Historically, autoradiography was performed using medical X-ray films. More recently, a relatively wide range of commercially available autoradiography films and exposure methods have been developed that provide superior contrast and higher sensitivity for specific isotopes (UNIT 10.11). The dimensional stability and chemical resistance of PVDF membranes provides an optimal medium for combining multiple detection methods, either sequentially on the same membrane, or on parallel lanes that can be reassembled after different detection methods have been performed. For example, sets of lanes containing replicate samples can be run on a single gel separated by prestained standards, which are used as guides to cut the resulting membrane. Each lane set can be subjected to different detection methods such as protein staining and immunoblotting with different antibodies (UNIT 10.10). The analyzed membranes can then be precisely reassembled to determine the relationships between bands detected by different methods. As there is no shrinkage or swelling of the membrane, this approach is particularly valuable for detecting slight differences in migration between components detected by different methods.

Introduction

An approach that has recently emerged as a useful analytical method for separating peptides and, to a more limited extent, proteins is capillary electrophoresis (CE; UNIT 10.9). CE protocols have also been developed for preparative peptide isolation in adequate amounts for subsequent high sensitivity measurements such as micro-sequencing (Chapter 11) and mass spectrometry (Chapter 16). CE employs a fused-silica capillary column that may contain either free solution or a fluid matrix. One of CE’s limitations for protein



and peptide separations is the tendency of these molecules to interact with silanol groups on the capillary tubing surface. As described in UNIT 10.9, these interactions can be minimized through the use of coated capillaries, a separation pH several units above the protein pI, or buffer additives that minimize interactions with the silica walls. An interesting alternative approach is separation of proteins and peptides using capillary zone electrophoresis (CZE) in acidic amphoteric buffers (UNIT 10.13). The low conductivity of these buffers permits higher voltages to be used, resulting in faster separations, and the low pH minimizes ionization of silanol groups on underivatized capillaries. In this mode, nearly all proteins and peptides have strong positive charges and high mobilities. In contrast to the relatively simple equipment requirements of polyacrylamide-based electrophoretic methods, commercial CE instruments are similar in complexity and expense to high-performance liquid chromatography (HPLC) instruments, due to the necessity for a high-voltage supply, a high-performance cooling system, and a very sensitive detector. Because CE separates components by charge rather than by hydrophobicity as in reversedphase HPLC, it is a highly useful orthogonal separation method that is frequently used to complement HPLC peptide separation strategies. Compared with polyacrylamide methods, CE offers much higher resolution, flexibility, sensitivity, and speed for separating peptides (> Leu > Met >> His >>> Asn/Gly

Cleavage after Arg/Lys possible, but may be unspecific or the result of contaminating trypsin Inhibited with P1′ = Pro Enhanced cleavage seen with (1) P3, P1′, P2′, or P3′ = Arg (Lys) (2) P2 = Pro Reduced cleavage seen with (1) P2, P1′, or P2′ = Asp (Glu) (2) P3 or P2′ = Pro P1 = neutral or acidic amino Enhanced cleavage seen when acid (broad specificity) (1) P1′ = Gly (2) P2 = bulky/hydrophobic amino acid Inhibited with P2′ = Pro; not inhibited with P1 = Pro P1′ = Leu/Ile > Phe > Val >> Tyr > Ala Enhanced fragmentation with P1 = Phe/Tyr/Trp Reduced fragmentation with P1 = Glu/Asp — P1 or P1′ = Phe >> Leu >> Trp > Ala > other hydrophobic amino acids — P1 or P1′ = Ile > Val > Ala Gly/Ser (and other neutral, nonaromatic amino acids) P2 = hydrophobic amino Very broad specificity; extensive degradation acid Enhanced fragmentation with P1 = Lys/Arg

aCleavage site nomenclature: P - - - P - - - P - - - ↓ - - - P ′ - - - P ′ - - - P ′, where ↓ marks the site of cleavage. 3 2 1 1 2 3


tions (e.g., reduction and alkylation in the presence of guanidine⋅HCl, followed by dialysis or chromatography) are tedious and frequently result in heavy losses, particularly when working with protein quantities 1 µg/100 µl of reaction mixture under native conditions and >1 µg/50 µl of reaction mixture in the presence of guanidine⋅HCl, urea, or SDS. This restriction essentially determines the enzyme/substrate ratios, with the clear rule that those ratios should be kept as low as possible to avoid excessive autolytic digestion of the protease. Finally, it is imperative that the substrate be in solution before the protease is added. Never try to improve solubility (e.g., by heating, sonicating, or vortexing) in the presence of the enzyme, as this will result in loss of activity.

Protease-resistant substrates Proteases active in 2 M guanidine⋅HCl and 8 M urea provide efficient tools to digest substrates whose physical properties prevent normal enzymatic degradation. A brief survey of the literature along with additional tests indicate that ribonuclease, ADP-ribosyl cyclase, lysozyme, amylase, superoxide dismutase, triosephosphate isomerase, xylose isomerase, pancreatic trypsin inhibitor, and many other proteins are quite resistant to protease digestion (see Table 11.1.2). After heating in 6 M guanidine⋅HCl and subsequent dilution to 2 M urea, they apparently do not refold to a compact structure, rendering them amenable to digestion. Alternatively, 8 M urea can be used to denature the protein. All the protease-resistant substrates listed above have been digested successfully using one or the other technique, but not always by both (Vangrysperre et al., 1989; Fig. 11.1.2). When the guanidine⋅HCl or urea concentrations are lowered to 1 M or 4 M, respectively, the digests no longer proceed. This excludes the use of endoproteinase Glu-C and trypsin for these purposes. In general, guanidine⋅HCl in combination with endoproteinase Lys-C is the method of choice. Endoproteinase Lys-C and subtilisin undergo a substantial amount of autolysis in 2 M guanidine⋅HCl, 8 M urea, or 1% SDS. Care must be taken to not confuse these peaks with the real peptide map. Autolysis profiles are fairly reproducible. Thus, with enzyme/substrate ratios of ≤1:10 and appropriate enzyme blank experiments, mistakes can usually be avoided. Should an autolytic fragment be accidentally analyzed by sequencing or mass spectrometry, the error can be quickly traced by comparing the sequence with the known sequence of the protease (Table 11.1.4). Limited and partial digestion Although the usual goal of a proteolytic digest is to fully cleave all susceptible bonds and generate a complete peptide map, sometimes restricted digestion is advantageous. Cleavage of an artificially low number of bonds, each one to completion, will yield fewer (and bigger) fragments and result in less complicated chromatograms. Addition of chaotropes will sometimes lead to exactly such an effect. Mild digestion of a protein in its native state may provide useful information on the domain structure and surface topography (Marks et al., 1990). The protein is thereby kept soluble in a nondenaturing detergent (e.g., CHAPS) solu-


Table 11.1.4

Protease Sequences

Enzyme

Sequence

α-Trypsin (bovine) Chain 1

IVGGYTCGAN

TVPYQVSLNS

GYHFCGGSLI

NSQWVVSAAH

CYKSGIQVRL

GEDNINVVEG

NEQFISASKS

IVHPSYNSNT

LNNDIMLIKL

KSAASLNSRV

ASISLPTSCA

SAGTQCLISG

WGNTK (125)

SSGTSYPDVL

KCLKAPILSD

SSCKSAYPGQ

ITSNMFCAGY

LEGGKDSCQG

DSGGPVVCSG

KLQGIVSWGS

GCAQKNKPGV

YTKVCNYVSW

IKQTIASN (98)

IVGGYTCAAN

SIPYQVSLNS

GSHFCGGSLI

NSQWVVSAAH

CYKSRIQVRL

GEHNIDVLEG

NEQFINAAKI

ITHPNFNGNT

LDNDIMLIKL

SSPATLNSRV

ATVSLPRSCA

AAGTECLISG

WGNTK (125)

SSGSSYPSLL

QCLKAPVLSD

SSCKSSYPGQ

ITGNMICVGF

LEGGKDSCQG

DSGGPVVCNG

QLQGIVSWGY

GCAQKNKPGV

YTKVCNYVNW

IQQTIAAN (98)

GVSGSCNIDV

VCPEGDGRRD

IIRAVGAYSK

SGTLACTGSL

VNNTANDRKM

YFLTAHHCGM

GTASTAASIV

VYWNYQNSTC

RAPNTPASGA

NGDGSMSQTQ

SGSTVKATYA

TSDFTLLELN

NAANPAFNLF

WAGWDRRDQN

YPGAIAIHHP

NVAEKRISNS

TSPTSFVAWG

GGAGTTHLNV

QWQPSGGVTE

PGSSGSPIYS

PEKRVLGQLH

GGPSSCSATG

TNRSDQYGRV

FTSWTGGGAA

ASRLSDWLDP

ASTGAQFIDG

LDSGGGTP (268)

VILPNNDRHQ

ITDTTNGHYA

PVTYIQVEAP

TGTFIASGVV

VGKDTLLTNK

HVVDATHGDP

HALKAFPSAI

NQDNYPNGGF

TAEQITKYSG

EGDLAIVKFS

PNEQNKHIGE

VVKPATMSNN

AETQVNQNIT

VTGYPGDKPV

ATMWESKGKI

TYLKGEAMQY

DLSTTGGNSG

SPVFNEKNEV

IGIHWGGVPN

EFNGAVFINE

NVRNFLKQNI

EDIHFANDDQ

PNNPDNPDNP

NNPDNPNNPD

EPNNPDNPNN

PDNPDNGDNN

NSDNPDAA (268)

Chain 2 α-Trypsin (pig) Chain 1

Chain 2

Endoproteinase Lys-C (Achromobacter lyticus strain M497-1)

Endoproteinase Glu-C (Staphylococcus aureus strain V8)

Chymotrypsin A (bovine) CGVPAIQPVL A chain B chain

C chain

SGL (13)

IVNGEEAVPG

SWPWQVSLQD

KTGFHFCGGS

LINENWVVTA

AHCGVTTSDV

VVAGEFDQGS

SSEKIQKLKI

AKVFKNSKYN

SLTINNDITL

LKLSTAASFS

QTVSAVCLPS

ASDDFAAGTT

CVTTGWGLTR (130)

YTNANTPDRL

QQASLPLLSN

TNCKKYWGTK

IKDAMICAGA

SGVSSCMGDS

GGPLVCKKNG

AWTLVGIVSW

GSSTCSTSTP

GVYARVTALV

NWVQQTLAAN (100)

tion and the enzyme activity attenuated through reduced enzyme/substrate ratios (e.g., 90% dye content will increase detection of peptide fragments during reversed-phase HPLC. The key to the success of the procedure and quantitative recovery of peptides from both PVDF and nitrocellulose membranes is the use of RTX-100 in the buffer. This is desirable because non-hydrogenated Triton X-100 has several strong UV-absorbing contaminants (Fig. 11.2.1; Tiller et al., 1984). In addition, RTX-100 does not inhibit enzyme activity or interfere with peak resolution during HPLC as do ionic detergents such as SDS (Fernandez et al., 1992). Finally, the concentration of RTX100 can be decreased to 0.1% with no loss in peptide yield (Fernandez et al., 1994b). The addition of a second aliquot of enzyme after 4 to 6 hr initial digestion can improve peptide recovery (Best et al., 1994). A membrane should be cut into 1 × 1–mm pieces while keeping it wet to avoid buildup of static charge. These small pieces allow using the minimum amount of buffer to cover the membrane. The volume of digestion buffer used should be enough to cover the membrane (∼50 µl) but can be increased or decreased depending on the amount of membrane present. The enzyme solution should be selected based on any additional knowledge of the protein available, such as its amino acid composition and whether it is basic or acidic. If the protein is a complete unknown, endoproteinase Lys-C or Glu-C would be a good choice. The enzyme-to-substrate ratio should be ∼1:10; however, if the exact amount of protein is unknown, ratios of

1:2 through 1:50 are suitable for digestion and will not affect the quantitative recovery of peptides. After digestion, most of the peptides (∼80%) are recovered in the original buffer and the additional washes are performed to ensure maximum recovery. Microbore reversed-phase HPLC is the best isolation procedure for peptides.

Troubleshooting The greatest source of failure in obtaining internal sequence data is insufficient transfer of protein to the PVDF membrane, which leads to an inability to detect peptides during HPLC analysis. After staining the PVDF-bound protein, if the protein band cannot be detected by amido black staining but is observable with India ink (which is ∼10-fold more sensitive), the protein quantity may be insufficient for this procedure. Similarly, if the protein band is detectable by radioactivity or immunostaining but not by protein stain, the quantity may be insufficient for subsequent HPLC analysis. Amino acid analysis (UNIT 3.2), amino-terminal sequence analysis, or at the very least, comparison with stained standard proteins on the blot, should be performed to help determine if enough material is present. When a sufficient but small (less than 10 µg) amount of protein is available, problems may arise from misidentification of peptides on reversed-phase HPLC due to artifact peaks and contaminants. Although elimination of every contaminant is usually impossible, there are several strategic points and steps that can be taken to help reduce contamination. Simultaneous processing of a negative control (a protein-free segment excised from the PVDF membrane) will help to identify contaminants associated with the membrane and digestion buffer. The negative control must be processed through the same purification steps as the sample, including electroblotting and staining, and should be analyzed by HPLC immediately before or after the sample. A positive control (membrane-bound standard protein) is generally unnecessary but should be performed if the activity of the enzyme is in question or if a new lot number of enzyme is to be used. Major sources of contaminants include the stains used to visualize the protein on the PVDF membrane, the microcentrifuge tubes used for digestion, reagents used during digestion and extraction of peptides, and the HPLC itself. Stains are the greatest source of contaminants, and Coomassie brilliant blue in particular frequently gives problems. Amido black and Pon-


A 120

Absorbance at 220 nm (mAU)

80

40

B 120

80

40

20

40

60

80

Time (min)

Figure 11.2.1 HPLC profiles of digestion buffer blanks. (A) Blank for 50 µl of 1% hydrogenated Triton X-100 (RTX-100)/10% acetonitrile/100 mM Tris⋅Cl, pH 8.0. (B) Blank for 1% Triton X-100/10% acetonitrile/100 mM Tris⋅Cl, pH 8.0. Both samples were incubated 20 hr at 37°C. Sample volumes were brought to 200 µl with 150 µl of 0.1% TFA and samples were analyzed on a Vydac C18 column (2.1 × 250–mm) using chromatographic conditions previously described (Fernandez et al., 1992). Peaks eluting at 50 to 100 min in panel B are UV-absorbing contaminants present only in Triton X-100.

ceau S are generally the cleanest, and Coomassie brilliant blue-G which has been chromatographically purified with a dye content >90% (e.g., Aldrich) appears to generate fewer contaminants than other less pure Coomassie brilliant blue stains. Surprisingly, microcentrifuge tubes can produce significant artifact peaks, which seem to vary with supplier and lot number. A blank containing only digestion buffer from a microcentrifuge tube should be included

because some contaminants only appear after incubation in the RTX-100 buffer. The major concern with the digestion buffer is the hydrogenated Triton X-100 (see Figure 11.2.1), which is purchased as a 10% stock solution. Additional late-eluting peaks may be observed with certain lots of RTX-100, whereas other lots are completely free of UV-absorbing contaminants. Milli-Q water or water prepared as described by Atherton (1989) should be used

Chemical Analysis


A

Immobilon P

B

Immobilon Psq

C

ProBlott

20

Absorbance at 220 nm (mAU)

10

20

10

20

10

Time

Figure 11.2.2 Peptide maps of trypsin digestion of human transferrin bound to different types of membrane. (A) Immobilon P; (B) Immobilon Psq; and (C) ProBlott. Samples were prepared as described in the basic protocol. Four micrograms (∼53 pmol) of transferrin was analyzed by SDS-PAGE, electroblotted to PVDF, and stained with Ponceau S prior to digestion. Chromatographic conditions were as previously described (Fernandez et al., 1992).

Enzymatic Digestion of Proteins on PVDF Membranes

for all solution preparation. An HPLC blank (i.e., a gradient run with no injection) should always be performed to determine which peaks are related to the HPLC. As discussed in Critical Parameters, the concentration of RTX can be decreased without loss of peptide recovery. However, with a large amount of membrane this may not be the case. Previous procedures (Aebersold et al, 1987; Tempst et al., 1990; Bauw et al., 1989; Fernandez et al., 1992) required pretreatment of the membrane with PVP-40 to prevent any proteinase adsorption to the membrane. RTX-100 is essential for quantitative recovery of peptides from the membrane; however, RTX-100 also strips PVP-40 from the membrane, resulting in

a broad, large, UV-absorbing contaminant that can interfere with peptide identification. The PVP-40 contaminant does not depend on the age or lot number of PVP-40; making fresh solutions does not prevent the problem (Aebersold, 1993). This appears to be more prevalent with nitrocellulose and higher-binding PVDF (ProBlott and Immobilon Psq) than with lowerbinding PVDF (Immobilon P), and depends on the amount of membrane used. The PVP-40 contaminant also appears to elute earlier in the chromatogram as the HPLC column ages, becoming more of a nuisance in visualizing peptides. Therefore, using PVP-40 to prevent enzyme adsorption to the membrane should be avoided.


Peptide mapping by reversed-phase HPLC is described in detail in UNIT 11.6; however, there are a few considerations that are worth discussing here. A precolumn filter (Upchurch Scientific) must be used to prevent small membrane particles from reaching the HPLC column. Inspection of the pooled supernatants for visible pieces of PVDF can prevent clogs in the microbore tubing. Membrane fragments can be removed either with a clean probe (e.g., pointed tweezers, wire or thin pipet tip) or by spinning in a centrifuge and transferring the sample to clean vial.

Anticipated Results Peptide mapping by reversed-phase HPLC after digestion of the membrane-bound protein should result in several peaks on the HPLC. Representative peptide maps from trypsin digestion of human transferrin bound to different PVDF membrane types—Immobilon P, Immobilon Psq, and ProBlott—are shown in Figure 11.2.2. Peptide maps should be reproducible when performed under the same digestion and HPLC conditions as described in this unit, as demonstrated by Figure 11.2.2. In addition, the peptide maps from proteins digested on PVDF membranes are comparable if not identical to maps derived from proteins digested in solution, indicating that the same number of peptides are recovered from the membrane as from solution. The average peptide recovery is generally 40% to 70% based on the amount of protein analyzed by SDS-PAGE, and 70% to 100% based on the amount of protein bound to PVDF (as determined by amino acid analysis). Recovery of peptides from the membrane tends to be quantitative, and the greatest loss of sample seems to occur during electroblotting.

Time Considerations

The entire procedure can be done in ∼24 hr plus the time required for peptide mapping by reversed-phase HPLC (see UNIT 11.6). Cutting the membrane takes ∼10 min, incubation after the digestion buffer is added takes 5 to 30 min, digestion at 37°C takes 22 to 24 hr, and extraction of the peptides requires ∼20 min.

Literature Cited Aebersold, R. 1993. Internal amino acid sequence analysis of proteins after in situ protease digestion on nitrocellulose. In A Practical Guide to Protein and Peptide Purification for Microsequencing, 2nd Ed. (P. Matsudaira, ed.) pp. 105154. Academic Press, New York.

Aebersold, R.H., Leavitt, J., Saavedra, R.A., Hood, L.E., and Kent, S.B. 1987. Internal amino acid sequence analysis of proteins separated by oneor two-dimensional gel electrophoresis after in situ protease digestion on nitrocellulose. Proc. Natl. Acad. Sci. U.S.A. 84:6970-6974. Atherton, D. 1989. Successful PTC amino acid analysis at the picomole level. In Techniques in Protein Chemistry (T. Hugli, ed.) pp. 273-283. Academic Press, New York. Bauw, G., Van Damme, J., Puype, M., Vandekerckhove, J., Gesser, B., Ratz, G.P., Lauridsen, J.B., and Celis, J.E. 1989. Protein-electroblotting and -microsequencing strategies in generating protein data bases from two-dimensional gels. Proc. Natl. Acad. Sci. U.S.A. 86:7701-7705. Best, S., Reim, D.F., Mozdzanowski, J., and Speicher, D.W. 1994. High sensitivity sequence analysis using in situ proteolysis on high retention PVDF membranes and a biphasic reaction column sequencer. In Techniques in Protein Chemistry V (J. Crabb, ed.) pp. 205-213. Academic Press, New York. Fernandez, J., DeMott, M., Atherton, D., and Mische, S.M. 1992. Internal protein sequence analysis: Enzymatic digestion for less than 10 µg of protein bound to polyvinylidene difluoride or nitrocellulose membranes. Anal. Biochem. 201:255-264. Fernandez, J., Andrews, L., and Mische, S.M. 1994a. An improved procedure for enzymatic digestion of polyvinylidene difluoride-bound proteins for internal sequence analysis. Anal. Biochem. 218:112-118. Fernandez, J., Andrews, L., and Mische, S.M. 1994b. A one-step enzymatic digestion procedure for PVDF-bound proteins that does not require PVP-40. In Techniques in Protein Chemistry V (J. Crabb, ed.) pp. 215-222. Academic Press, New York. Mozdzanowski, J. and Speicher, D.W. 1990. Quantitative electrotransfer of proteins from polyacrylamide gels onto PVDF membranes. In Current Research in Protein Chemistry: Techniques, Structure, and Function. (J. Villafranca, ed.) pp. 87-94. Academic Press, New York. Tempst, P., Link, A.J., Riviere, L.R., Fleming, M., and Elicone, C. 1990. Internal sequence analysis of proteins separated on polyacrylamide gels at the submicrogram level: Improved methods, applications and gene cloning strategies. Electrophoresis 11:537-553. Tiller, G.E., Mueller, T.J., Dockter, M.E., and Struve, W.G. 1984. Hydrogenation of Triton X100 eliminates its fluorescence and ultraviolet light absorbance while preserving its detergent properties. Anal. Biochem. 141:262-266.

Chemical Analysis


Key References Fernandez et al., 1994a. See above. Describes digestion with and without PVP-40 and applies it to unknown proteins.

Contributed by Joseph Fernandez and Sheenah M. Mische The Rockefeller University New York, New York

Fernandez et al., 1994b. See above. Describes digestion procedure and emphasizes applicability to different types of PVDF membranes and the concentration of RTX-100 buffer.

Enzymatic Digestion of Proteins on PVDF Membranes


Digestion of Proteins in Gels for Sequence Analysis

UNIT 11.3

A high percentage of eukaroytic proteins have blocked amino termini, so it is usually necessary to cleave an “unknown” protein chemically or enzymatically to obtain the partial sequences needed for cDNA cloning. Because SDS-polyacrylamide gel electrophoresis (SDS-PAGE; UNIT 10.1) is the current method of choice for the final purification of the >25 pmol amounts of protein that are usually required for internal sequencing, procedures that can be used to digest proteins in situ in SDS-polyacrylamide gels are often the most useful and have the added benefit of eliminating losses that may occur during blotting. Two alternative strategies have been developed to respond to this need and to deal with the unique problems posed by SDS, which are that SDS inhibits trypsin, one of the enzymes that is most commonly used for internal sequencing studies, and also interferes with reversed-phase HPLC. In the Basic and Alternate Protocol 2, SDS is removed from the gel prior to enzymatic cleavage by the staining and subsequent washing steps. The Basic Protocol calls for an acetonitrile wash to remove residual SDS from the protein sample. A different detergent, Tween 20, is then added back to the sample to maintain the solubility of the denatured protein. In Alternate Protocol 2, the gel slices are washed with ammonium bicarbonate and the protein samples digested in the absence of detergent. Alternate Protocol 1 utilizes lysyl endopeptidase, an enzyme resistant to SDS; the digestion can therefore be carried out without prior removal of the SDS. Ultimately, SDS is removed via an anion-exchange precolumn that immediately precedes the reversedphase HPLC column. In all cases, the peptides resulting from in-situ digestion are extracted from the gel matrix, then separated via reversed-phase HPLC prior to amino acid sequencing. Before beginning the enzymatic digests described in these protocols, it is helpful to determine the amount of protein present in the sample via amino acid analysis, as described in Support Protocol 1. Reducing and alkylating proteins separated by SDSPAGE, as described in Support Protocol 2, facilitates the identification of cysteine residues during subsequent peptide sequencing reactions. DIGESTION OF PROTEINS IN GELS IN THE PRESENCE OF TWEEN 20 In this protocol, which is a slight modification of the Rosenfeld et al. (1992) procedure, an acetonitrile wash is used to remove residual SDS and Coomassie brilliant blue from the excised gel slice containing the protein of interest. The washed gel is then partially dried prior to rehydrating in the presence of the enzyme of choice—usually trypsin or lysyl endopeptidase—in a buffer containing Tween 20. The Tween 20 presumably helps both to remove residual SDS from the protein and to maintain the solubility of the denatured protein and its resulting cleavage fragments. After digestion, the peptides are alkylated with iodoacetic acid to facilitate identification of cysteine residues during amino acid sequencing (see Support Protocol 1).

BASIC PROTOCOL

After digestion, the peptides are extracted from the gel and separated on a C18 reversedphase HPLC column. Refer to UNIT 11.6 and to Stone et al. (1990, 1991, 1993) for further discussion of HPLC mapping and peptide isolation.

Chemical Analysis Contributed by Kathryn L. Stone and Kenneth R. Williams Current Protocols in Protein Science (1995) 11.3.1-11.3.13 Copyright © 2000 by John Wiley & Sons, Inc.

11.3.1 CPPS

Materials Protein sample separated on SDS-polyacrylamide gel (UNIT 10.1; include appropriate standard protein on gel) and stained with Coomassie brilliant blue (UNIT 10.5) 50% (v/v) acetonitrile in 0.2 M ammonium carbonate, pH 8.9 0.02% (v/v) Tween 20 (Sigma) in 0.2 M ammonium carbonate, pH 8.9 0.1 mg/ml modified trypsin in manufacturer’s dilution buffer (Promega; stable at least 2 years when stored at −20°C) 0.1 mg/ml lysyl endopeptidase (Achromobacter Protease I, Wako Chemicals USA) in 2 mM Tris⋅Cl, pH 8.0 (store >5 Å apart in space. Spin diffusion reduces cross-peak intensity and precludes quantitative geometric interpretation. Cross-peak intensity (or volume) increases roughly linearly with mixing time up to a point, and then starts to decay. The initial nOe build up is proportional to r−6, where r is the internuclear distance. In practice, a series of NOESY spectra are acquired with increasing mixing times, and the cross-peak volumes are plotted against the mixing time. Cross-peak volume ratios can be used to give quantitative relative distances if they are obtained from an experiment with a τm in the linear regime. Especially useful in this respect are the δ and ε aromatic protons of Tyr residues, whose mutual nOe cross-peak volume yields a reference distance for the calculation of other interproton distances. Figure 17.5.10B shows a 400-msec NOESY experiment recorded on the NPY[18-36] peptide. Intraresidue through-space interactions can be identified by comparison with the TOCSY data (Fig. 17.5.10A). Additional correlations yield the most useful information about the relative distances of different parts of the peptide. Sequential nOes are observed, for instance, from Ile-28 NH to the amide resonances of Tyr-27 and Asn-29, as marked on the spectrum. The two-dimensional ROESY experiment (Bothner-By et al., 1984; Bax and Davis, 1985) is analogous to NOESY in that correlations occur between protons that are close in space. However, in the ROESY method, all molecules behave as if they were in the extreme narrowing limit, despite molecular size or correlation time, and thus there is no zero crossing point. True ROESY correlations (i.e., through-space interactions) are always positive (opposite in phase to the diagonal), but of lower absolute intensity than the analogous NOESY correlations. How-

ever, negative correlations may be observed in a ROESY spectrum, and arise from TOCSYtype transfer or from chemical exchange. The latter process is a consequence of the exchange of (chemical) environment, for instance, during the mixing time of the experiment. The phenomenon of chemical exchange can be of great utility in the determination of dynamic processes of proteins in solution. For examples, see Fejzo et al. (1991) and Lian and Roberts (1993).

NMR Characteristics of Secondary Structural Elements The identification of secondary structure in a protein can be achieved by several complementary methods. The chemical shift perturbations of the alpha proton (and of Cα and Cβ, if available) of an amino acid can be characteristic, as seen above (see Chemical Shift). Amide NH protons involved in hydrogen bonding (the interaction that stabilizes α helices and β sheets) usually have much slower proton exchange rates than those in regions of little structure; these rates of exchange are usually measured by dissolving the protein in 2H2O and rapidly recording spectra as a function of time. Hydrogenbonded amide protons show a lower temperature dependence of chemical shift than those that are not hydrogen bonded. The NH-Hα coupling constant is a function of the dihedral angle between them (φ); this coupling is small (3 to 5 Hz) in α-helical structures, whereas it is usually much larger (8 to 12 Hz) in β sheets. As seen in the above section and in Figure 17.5.12, certain short-range nOes are characteristic of α helices and β sheets, and it is the observation of these correlations that gives the most definitive method of characterizing the location and boundaries of secondary structure. It is common practice to summarize the NMR data (as it defines secondary structure) as shown in Figure 17.5.13.

Toward the Structure: Putting It All Together Each resonance assignment experiment can also yield information about the three-dimensional structure of the protein itself. For instance, coupling constants revealed by one-dimensional experiments or two-dimensional methods based on COSY (e.g., DQF-COSY) constrain the dihedral angle between the coupled protons. The main structural constraints revealed by NMR are nOes, J coupling, and chemical shift. nOes reveal much about the overall protein folding (as well as secondary structure; see above discussion of NOESY and

Structural Biology


Supplement 19

sequence A D F Q R RS T A R E K R E A I S V Q R A QD D S E E N S F Q Y R H E D A N 3J u 69445u 4544799 7 u 989uu99 8 7 8u9 445 444 489 8 (Hz) NH,α NH exchange

dαN(i,i+3)

dαβ(i,i+3)

dαN(i,i +1) dNN(i,i +1) dβN(i,i+ 1) chemical shift index secondary structure

1 –1 α helix

β sheet

u

α helix

slow exchange fast exchange strong medium weak unobservable

Figure 17.5.13 Usual presentation of NMR information regarding secondary structure, illustrated using a fictitious peptide and data. Coupling constants between amide NH and Hα (intraresidue) are shown in Hz. Fast or slow amide exchange rates are shown with open or closed circles, respectively. Long-range (i,i+3) nOes between two residues are shown using horizontal bars connecting those residues, with line thickness used to indicate the relative size of these correlations. Shorter-range (i,i+1) nOes are shown as blocks representing large, medium, and small magnitudes. The chemical shift index represents the deviation of Hα shift from that found in random coil structures; a sequence of +1 values (representing an actual deviation of +0.3 and above) is representative of β-sheet structure, whereas the converse is representative of α helices.

Introduction to NMR of Proteins

ROESY), and are generally not used to determine accurate interatomic distances, but rather to provide upper distance bounds. Powerful software programs are available that calculate three-dimensional structures based on the satisfaction of NMR constraints. Usually, after all available constraints have been employed in the calculation, a “family” of structures is obtained. Each member of this family will have a slightly different structure, but will satisfy the NMR constraints within a certain tolerance. The quality of structures generated by these methods is generally quoted in terms of the maximum deviation within the family from the “average” structure (actually given as the root mean square deviation, RMSD). “Good” quality structures will have backbone RMSDs on the order of 0.5 Å. The interested reader is referred to the following articles for a more detailed

discussion of this topic: Sutcliffe (1993), Weber (1996), and Güntert (1997).

SPECTRAL ASSIGNMENT USING DOUBLE-RESONANCE HETERONUCLEAR METHODS Assignment of larger (>10 kDa) proteins is not well suited to homonuclear methods. The number of resonances increases monotonically with the number of residues, and even correlations observed in two-dimensional NMR spectra frequently overlap, precluding assignment. The overlap problem is exacerbated for proteins with a high degree of α-helical content, in which the dispersion of the Hα proton resonances is generally low. The widths of the proton signals increase with the size of the protein, due to the shortening of T2 relaxation times, as the rotational correlation time (τc) increases, leading to



reduced sensitivity for homonuclear correlation experiments, particularly COSY. Incorporation of NMR-active stable isotopes allows the use of large heteronuclear couplings to transfer magnetization (see Fig. 17.5.3), leading to experiments with high sensitivity. Generally, NMR correlation experiments that rely on a coupling of J Hz require a delay of (2J)−1 sec to fully transfer magnetization. Small couplings (i.e., nJHH) result in large (2J)−1 times, during which magnetization (i.e., signal) will be attenuated by T2 relaxation. Two-dimensional NMR techniques can be extended into three or more dimensions, any of which could be a heteronuclear frequency dimension. Spreading correlations into the heteronuclear dimension partially circumvents the resonance overlap problem, as the chemical shift dispersion of heteronuclei is generally much greater than that of protons. NMR of proteins uniformly and completely labeled with the nitrogen-15 isotope (U-15N) can exploit the large one-bond 15N-1H coupling of ∼90 Hz to establish correlations. Applications include isotope-edited correlation experiments, such as three-dimensional 15N/1H-TOCSYHMQC and NOESY-HMQC, which will be discussed below (see Isotope Editing: Two-Dimensional Heteronuclear Chemical Shift Correlations). These three-dimensional experiments have been successfully used in favorable cases—i.e., for smaller proteins (∼100 residues) or for larger proteins rich in β sheets (Marion et al., 1989)—to obtain nearly complete assignments of the protein backbone 15N and 1H resonances (Gronenborn et al., 1989). NMR of proteins labeled with both 13C and 15N (U-13C/15N protein) allows many additional heteronuclear couplings to potentially contribute to magnetization transfer (Fig. 17.5.3). The one-bond couplings are generally much larger than multiple bond 1H-1H couplings and greater than or comparable to 13C and 15N linewidths for most medium-sized proteins (i.e., 900 nonexchangeable protons, significant spectral changes can be observed. However, at this stage interpretation on the atomic or structural level is not possible; this would require characterization of both Ca2+-free and Ca2+-ligated forms of the protein. Nevertheless, several aspects of the Ca2+binding process can be understood in a qualitative way. Ca2+ binding occurs in two distinct phases. The first phase, characterized by the addition of up to two equivalents of Ca2+, causes resonances characteristic of apocalmodulin (Ca2+-free) to disappear, while a distinct set of “new” resonances appear. This additional set of resonances belongs to (Ca2+)2-calmodulin, as there is no distinct (Ca2+)1 species. This is well illustrated by the His-107 Hε1 resonance and by the Tyr-138 Hδ protons marked on Figure 17.5.25. This well-characterized behavior is indicative of slow exchange kinetics—that is, strong binding between protein and ligand. Changes accompanying the second phase of Ca2+ binding, that is between two and four equivalents bound to calmodulin, are more subtle. Resonances change their position gradually from being characteristic of the (Ca2+)2- to the (Ca2+)4-calmodulin complex, as again there is no distinct (Ca2+)3 form. This can be clearly seen for the Phe-65 Hδ protons, marked on Figure 17.5.25. This behavior, in contrast to that observed in the first phase of Ca2+ binding, is called



approximate equivalents of calcium

F-65δ

H-107ε1

ε

Y-138 δ 4.00

D-64,T-26α

3.50 3.00

2.50 2.00 1.75 1.50

1.25 1.00 0.75 0.50 0.25 0.00

8.0

7.5

7.0

6.5

6.0

5.5

ppm Figure 17.5.25 Low-field region of proton spectra obtained during the course of a titration of Trypanosome calmodulin with calcium at 360.13 MHz and 310K. Each spectrum was obtained after the addition of an aliquot of a 200 mM CaCl2 solution in 2H2O, corresponding to 0.25 calcium equivalents. The bottom-most spectrum represents calcium-free calmodulin and the top-most spectrum the fully calcium-ligated protein. The calculated molar ratio of calcium to calmodulin is shown on the figure. Notice the two resonances of His-107 Hε1 in the spectra obtained with 0.75 to 1 equivalent of calcium. This is due to slow exchange between calcium-ligated and calcium-free calmodulin, and means that the binding of calcium to calmodulin is very strong. The His-107 Hε1 resonance characteristic of calcium-free calmodulin (8.25 ppm) has completely disappeared after the addition of 2 equivalents of calcium. No further changes to the spectra were observed following addition of >4 equivalents of calcium. The appearance and disappearance of resonances is represented by up and down arrows, respectively; the dotted line indicates a resonance that changes shift.

Structural Biology


Supplement 19

approximate equivalents Y-138 calmidazolium ε1,2 δ1,2 D-64α 2.0

1.7 1.5 1.3 1.0 0.6 0.3 ε1 8.4

H-107

δ2

0.0

8.2 8.0 7.8 7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 6.0 5.8 5.6 5.4 5.2 ppm

Figure 17.5.26 Low-field region of proton spectra obtained during the course of a titration of calciumsaturated Phe-2H5 Trypanosome calmodulin with R-(−)-calmidazolium at 500.13 MHz and 310K. Each spectrum was obtained after the addition of an aliquot of 25 mM R-(−)-calmidazolium in methanol-2H4. It can easily be seen that the His-107 Hε1 proton acts as a marker for the titration, due to the very tight binding of calmidazolium to calmodulin, showing distinct chemical shifts for the apo- and calmidazoliumsaturated forms of the protein. After the addition of one molar equivalent of calmidazolium, the His-107 Hε1 resonance is split into two signals of equal area. This represents an equimolar mixture of calmodulin and the calmodulin-[calmidazolium]2 complex. The Tyr-138 ε1,2 protons exhibit similar behavior. All resonances (apart from the single His-107 δ2 proton) between 6.65 and 7.80 ppm are due to the aromatic protons of calmidazolium.

fast exchange kinetics. Although not shown in Figure 17.5.25, no further changes in chemical shift are observed following the addition of more than four Ca2+ equivalents. Calmodulin consists of two domains, each containing two Ca2+ binding sites. The C-terminal domain contains the higher-affinity Ca2+ binding sites, which are occupied first. This accounts for the observed slow exchange kinetics involving primarily C-terminal residues. The binding is cooperative (the binding of the first equivalent increases the affinity of the unoccupied site for Ca2+), explaining the absence of a (Ca2+)1 intermediate species. The lower-affinity Ca2+ binding sites that are located in the N-terminal domain are filled only after the C-terminal sites are occupied. The lower affinity of these sites for Ca2+ is reflected in the fast exchange kinetics observed.

Small Molecule Interactions

Introduction to NMR of Proteins

Using the techniques of isotope editing and filtering, it may be possible to observe, at will, the NMR resonances of both the protein and the ligand if one is isotopically normal but the other

is, for example, labeled with 13C. As part of a study to investigate the interaction between calmodulin from Trypanosoma brucei rhodiesiense and the very potent antagonist calmidazolium (Fig 17.5.21), a different, but nevertheless illustrative, labeling strategy was adopted. The only aromatic amino acids in Trypanosome calmodulin are nine Phe residues and single Tyr and His residues. Specific labeling with Phe-2H5 calmodulin effectively removes 90% of the protein aromatic resonances from the 1H NMR spectrum. The antagonist itself is mostly aromatic, so that the majority of its protons will resonate in the 7 to 8 ppm (aromatic) region of the 1H spectrum. Thus, the spectrum of a complex of calmidazolium with Phe-2H5 calmodulin will contain mostly antagonist signals in the aromatic region. Additionally, by using 2H2O as solvent, the NH signals, which would also resonate in the aromatic region, are absent. Figure 17.5.26 shows the 500 MHz one-dimensional proton spectrum of Phe-2H5 calmodulin in 2H2O during the course of titration with R-(−)-calmidazolium. The bottom spectrum in Figure 17.5.26 shows the protein in the



absence of ligand, clearly demonstrating the spectral simplification obtained using the Phe2H label (compare with the top-most spectrum 5 of Fig. 17.5.25 at 360 MHz or with Fig. 17.5.16C at 500 MHz) with only His-107 ε1, δ2 and Tyr-138 ε1,2, δ1,2 protons visible in the aromatic region. The number of calmidazolium equivalents is shown in the figure. Calmidazolium shows slow exchange binding kinetics, exemplified by the behavior of the His-107 ε1 proton and, to a more limited extent, by the Tyr-138 ε1,2 protons. Distinct changes in protein chemical shifts over the entire spectrum are observed throughout the titration. It can be deduced from Figure 17.5.26 that calmodulin binds two equivalents of calmidazolium, and that the spectrum obtained after the addition of one calmidazolium equivalent is indicative of a 1:1 mixture of antagonist-free calmodulin and the calmodulin-[calmidazolium]2 complex. Two-dimensional homonuclear methods (e.g., TOCSY and NOESY) can be used to assign the resonances of the bound calmidazolium molecule. The NOESY spectrum will also show intermolecular nOes between the bound calmidazolium molecule and the protein. These nOes will be observed from calmidazolium protons to those protons of calmodulin that are involved in or around the drug binding site. As an example of the use of these methods, Figure 17.5.27 shows three regions taken from a 200-msec NOESY experiment on the calmodulin-calmidazolium complex. Figure 17.5.27A shows intermolecular nOes between calmidazolium and calmodulin, the majority of which occur to the methyl groups of hydrophobic residues. Figure 17.5.27B shows nOes between calmidazolium aromatic and aliphatic protons. Figure 17.5.27C shows intramolecular nOes between calmidazolium aromatic protons. The nOe data not only aid in the assignment of calmidazolium resonances, but also yield information about the conformation of the molecule when bound to the protein. However, it is not adequate to completely assign those protein residues involved in complex formation. To accomplish this, a drug-protein complex obtained from doubly labeled calmodulin is required, which can be subjected to the very powerful isotope-editing NMR techniques introduced above (see Isotope Editing: Two-Dimensional Heteronuclear Chemical Shift Correlations). For a more detailed discussion of the use of NMR in the analysis of protein-ligand interactions, the reader is referred to the following articles and references therein: Feeney and Bird-

sall (1993), Cooke (1996), and Craik and Wilce (1997).

OTHER PROTEIN NMR TECHNIQUES AND FUTURE DIRECTIONS As the molecular weight of a protein increases, it becomes more and more challenging to generate data amenable to interpretation and assignment. The most obvious effect of increasing molecular weight is the increased number of resonances in discrete spectral regions, yielding chemical shift degeneracy. In addition, the correlation time (τc) increases with molecular weight, resulting in more efficient T2 relaxation and hence broader lines. As a result, correlation experiments that rely on small J couplings become extremely inefficient. The dipolar interaction between protons also contributes significantly to increased rates of relaxation and to spin diffusion in nOe experiments. The dipolar interaction between protons and 13C and 15N also becomes more efficient, resulting in lower sensitivity for double- and triple-resonance experiments. The dipole-dipole interaction between protons can be reduced by decreasing the “concentration” of protons in a protein. This may be achieved by substituting a certain proportion of protons with deuterons. Fractional deuteration is an isotope labeling technique producing a heterogeneous mixture of isotopomers in which each position within the protein has the same probability of being deuterated (Sattler and Fesik, 1996). Although the number of protons giving a signal is reduced, the concomitant decrease in linewidth more than compensates, resulting in an overall increased signal-to-noise ratio. This labeling technique is particularly powerful when combined with U-13C and/or 15N labeling. In this case, triple-resonance methods become far more efficient as the T2 values of the heteronuclei, which are dominated by one-bond dipolar interaction with 1H, become much longer. The mutual interaction between two dipoles is proportional to the distance and angle between them. The angular dependence normally averages to zero in isotropic liquids. If molecules have preferred orientations in space relative to B0, this produces new signal splittings in the case of uncoupled nuclei, or changes in splitting of coupled nuclei, because the dipolar interaction no longer averages to zero. Adding the protein to a low concentration of self-orienting particles such as phospholipids or even viruses, which can form oriented bicelles in solution, a degree

Structural Biology


Supplement 19

A

1.0

2.0

B

ppm

4.5

5.0

5.5

C 6.8

7.0

7.2

7.4

7.6

7.6

7.4

7.2

7.0

6.8

ppm

Figure 17.5.27 Expansions of a 200-msec NOESY spectrum obtained on the Phe-2H5 Trypanosome calmodulin-[calmidazolium]2 complex recorded at 500.13 MHz at a temperature of 310K. Panels A through C are each drawn with the same horizontal scale. (A) Intermolecular nOes between the aromatic protons of calmidazolium and calmodulin. The majority of such nOes are observed to hydrophobic residues of calmodulin, e.g., terminal methyl groups of Val, Ala, Ile, and Leu, and the ε methyl groups of Met residues. The assignment of these intermolecular nOes can potentially reveal the drug-binding sites on the protein. (B) Intramolecular nOes between the aromatic and aliphatic protons (i.e., H6, 7, and 8) of calmidazolium, again yielding both assignment and conformational information about the bound drug. (C) Mutual intramolecular nOes are observed between the aromatic protons of calmidazolium. These correlations contribute to the resonance assignment of these signals, and also give an indication of the conformation of the bound drug.



of alignment is imposed on the protein (Bax and Tjandra, 1997). A comparison of J couplings observed in the absence and presence of such additives yields the magnitude and sign of the dipolar coupling (Jobs = J + D, where D is the dipolar coupling). The contribution of the dipolar coupling to the observed splitting is a function of the relative orientation of the bond with respect to the B0 field. This information yields constraints that can be used in addition to nOes and torsion angles to further refine protein structure (Tjandra et al., 1997). A further source of nuclear relaxation is chemical shift anisotropy (CSA), an effect that is proportional to the correlation time of the protein and to the square of the applied magnetic field. This relaxation mechanism becomes particularly prominent for the 15N-1H amide group in high magnetic fields. It is possible to record the 15N-1H HSQC experiment such that the heteronuclear spin coupling is retained in f1 and f2. This results in four correlations for each 15N-1H pair, each of which relax at a different rate. For one of these multiplet components, the dipole-dipole (DD) and the CSA mechanisms contribute to relaxation in an opposing sense, and balance exactly at high enough magnetic field (the DD relaxation is field independent), so that one of the four multiplet lines becomes sharp. It is calculated for 15N-1H that the reduction in linewidth will be a maximum at a magnetic field of ∼1 GHz, 23.5 T. An experiment called transverse relaxation optimized spectroscopy (TROSY; Pervushin et al., 1997) selects only the sharp multiplet component. Using this as a building block for triple-resonance experiments (Salzmann et al., 1999), it may be possible to determine the structure of very high-molecular-weight proteins. Thus, the quest for magnets with increasingly high field strength is not only desirable for increased resolution and sensitivity, but also for its potential to produce highquality narrow-line spectra for very high-molecular-weight proteins. Further general discussion of the topics covered here can be found in Dötsch and Wagner (1998) and Wüthrich (1998).

LITERATURE CITED Bagby, S., Harvey, T.S., Kay, L.E., Eagle, S.G., Inouye, S., and Ikura, M. 1994. Unusual helixcontaining Greek keys in development-specific Ca2+-binding protein S. 1H, 15N, and 13C assignments and secondary structure determined with the use of multidimensional double and triple resonance heteronuclear NMR spectroscopy. Biochemistry 33:2409-2421.

Basus, V.L. 1989. Proton nuclear magnetic resonance assignment. In Nuclear Magnetic Resonance, Part B (T.L. James and N.J. Oppenheimer, eds.) pp. 132-149. Academic Press, New York. Bax, A. and Davis, D.G. 1985. Practical aspects of two-dimensional transverse NOE spectroscopy. J. Magn. Reson. 63:207-213. Bax, A. and Tjandra, N. 1997. High-resolution heteronuclear NMR of human ubiquitin in an aqueous liquid crystalline medium. J. Biomol. NMR 10:289-292. Bothner-By, A.A., Stephens, R.L., Lee, J., Warren, C.D., and Jeanloz, R.W. 1984. Structure determination of a tetrasaccharide: Transient nuclear Overhauser effects in the rotating frame. J. Am. Chem. Soc. 106:811-813. Braunschweiler, L. and Ernst, R.R. 1983. Coherence transfer by isotropic mixing: Application to proton correlation spectroscopy. J. Magn. Reson. 53:521-528. Cooke, R.M. 1996. Protein-ligand interactions: Examples in drug design. In NMR in Drug Design (D.J. Craik, ed.) pp. 245-274. CRC Press, Boca Raton, Fla. Craik, D.J. and Wilce, A. 1997. Studies of proteinligand interactions by NMR. In Protein NMR Techniques (D.G. Reid, ed.) pp. 195-232. Humana Press, Totowa, N.J. Croasmun, W.R. and Carlson, R.M.K. 1994. TwoDimensional NMR Spectroscopy. Applications for Chemists and Biochemists. VCH Publishers, New York. Derome, A.E. 1987. Modern NMR Techniques for Chemistry Research. Pergamon Press, Oxford. Dötsch, V. and Wagner, G. 1998. New approaches to structure determination by NMR spectroscopy. Curr. Opin. Struct. Biol. 8:619-623. Edwards, A.J. 1998. An NMR isotope labelling analysis of calmodulin interactions with high affinity chiral inhibitors. Ph.D. thesis, University of Hertfordshire, U.K. Evans, J.N.S. 1995. Biomolecular NMR Spectroscopy. Oxford University Press, Oxford. Feeney, J. and Birdsall, B. 1993. NMR studies of protein-ligand interactions. In NMR of Macromolecules: A Practical Approach (G.C.K. Roberts, ed.) pp. 183-215. Oxford University Press, Oxford. Fejzo, J., Westler, W.M., Macura, S., and Markley, J.L. 1991. Strategies for eliminating unwanted cross-relaxation and coherence-transfer effects from two-dimensional chemical-exchange spectra. J. Magn. Reson. 92:20-29. Friebolin, H. 1993. Basic One- and Two-Dimensional NMR Spectroscopy. VCH Publishers, New York. Gronenborn, A.M., Bax, A., Wingfield, P.T., and Clore, G.M. 1989. A powerful method of sequential proton resonance assignment in proteins using 15N-1H multiple quantum coherence spectroscopy. FEBS Lett. 243:93-98. Structural Biology


Supplement 19

Güntert, P. 1997. Calculating protein structures from NMR data. In Protein NMR Techniques (D.G. Reid, ed.) pp. 157-194. Humana Press, Totowa, N.J.

Neuhaus, D. and Williamson, M.P. 1989. The Nuclear Overhauser Effect in Structural and Conformational Analysis. VCH Publishers, New York.

Ikura, M., Kay, L.E., and Bax, A. 1990. A novel approach for sequential assignment of 1H, 13C, and 15N spectra of larger proteins: Heteronuclear triple-resonance three-dimensional NMR spectroscopy. Application to calmodulin. Biochemistry 29:4659-4667.

Neuhaus, D., Wagner, G., Vasák, M., Kägi, J.H.R., and Wüthrich, K. 1985. Systematic application of high-resolution, phase-sensitive two-dimensional 1H-NMR techniques for the identification of the amino-acid-proton spin systems in proteins. Eur. J. Biochem. 151:257-273.

Ikura, M., Clore, G.M., Gronenborn, A.M., Zhu, G., Klee, C.B., and Bax, A. 1992. Solution structure of a calmodulin–target peptide complex by multidimensional NMR. Science 256:632-638.

Pervushin, K., Riek, R., Wider, G., and Wüthrich, K. 1997. Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. U.S.A. 94:12366-12371.

IUPAC-IUB Commission on Biochemical Nomenclature. 1970. Abbreviations and symbols for the description of the conformation of polypeptide chains. J. Mol. Biol. 52:1-17. Jerala, R., Almeida, P.F.F., Ye, Q., Biltonen, R.L., and Rule, G.S. 1996. 1H,15N and 13C resonance assignments and secondary structure of group II phospholipase A2 from Agkistrodon piscivorus: Presence of an amino-terminal helix in solution. J. Biomol. NMR 7:107-120. Kumar, A., Ernst, R.R., and Wüthrich, K. 1980. A two-dimensional nuclear Overhauser enhancement (2D NOE) experiment for the elucidation of complete proton-proton cross-relaxation networks in biological macromolecules. Biochem. Biophys. Res. Commun. 95:1-6. Lian, L. and Roberts, G.C.K. 1993. Effects of chemical exchange on NMR spectra. In NMR of Macromolecules: A Practical Approach (G.C.K. Roberts, ed.) pp. 153-182. Oxford University Press, Oxford. Lukin, J.A., Gove, A.P., Talukdar, S.N., and Ho, C. 1997. Automated probabilistic method for assigning backbone resonances of (13C,15N)-labeled proteins. J. Biomol. NMR 9:151-166. Macomber, R.S. 1998. A Complete Introduction to Modern NMR Spectroscopy. John Wiley & Sons, New York. Marion, D., Driscoll, P.C., Kay, L.E., Wingfield, P.T., Gronenborn, A.M., and Clore, G.M. 1989. Overcoming the overlap problem in the assignment of 1H NMR spectra of larger proteins by use of three-dimensional heteronuclear 1H-15N Hartmann-Hahn-multiple quantum coherence and nuclear Overhauser-multiple quantum coherence spectroscopy: Application to interleukin 1b. Biochemistry 28:6150-6156. Markley, J.L. and Kainosho, M. 1993. Stable isotope labelling and resonance assignments in larger proteins. In NMR of Macromolecules: A Practical Approach (G.C.K. Roberts, ed.) pp. 101-152. Oxford University Press, Oxford. Mossakowska, D.E. and Smith, R.A.G. 1997. Production and characterisation of recombinant proteins for NMR structural studies. In Protein NMR Techniques (D.G. Reid, ed.) pp. 325-335. Humana Press, Totowa, N.J. Introduction to NMR of Proteins

Piotto, M., Saudek, V., and Sklenár, V. 1992. Gradient-tailored excitation for single-quantum NMR spectroscopy of aqueous solutions. J. Biomol. NMR 2:661-665. Redfield, C. 1993. Resonance assignment strategies for small proteins. In NMR of Macromolecules: A Practical Approach (G.C.K. Roberts, ed.) pp. 71-99. Oxford University Press, Oxford. Richarz, R. and Wüthrich, K. 1978. Carbon-13 NMR chemical shifts of the common amino acid residues measured in aqueous solutions of the linear tetrapeptides H-Gly-Gly-C-L-Ala-OH. Biopolymers 17:2133-2141. Roberts, G.C.K. 1993. NMR of Macromolecules: A Practical Approach. Oxford University Press, Oxford. Salzmann, M., Wider, G., Pervushin, K., Senn, H., and Wüthrich, K. 1999. TROSY-type triple resonance experiments for sequential NMR assignments of large proteins. J. Am. Chem. Soc. 121:844-848. Sattler, M. and Fesik, S.W. 1996. Use of deuterium labeling in NMR: Overcoming a sizeable problem. Structure 4:1245-1249. Stockman, B.J. 1996. Preparation of 2H, 13C and 15N isotopically enriched proteins for NMR spectroscopic investigations. In NMR Spectroscopy and its Application to Biomedical Research (S.K. Sarkar, ed.) pp. 159-185. Elsevier/NorthHolland, Amsterdam. Sutcliffe, M.J. 1993. Structure determination from NMR data II. Computational approaches. In NMR of Macromolecules: A Practical Approach (G.C.K. Roberts, ed.) pp. 359-390. Oxford University Press, Oxford. Sweeney, P.J. 1990. NMR analysis of drug interactions with isotopically labelled calmodulin. Ph.D. thesis, Hatfield Polytechnic, Hertfordshire, U.K. Tjandra, N., Omichinski, J.G., Gronenborn, A.M., Clore, G.M., and Bax, A. 1997. Use of dipolar 1H-15N and 1H-13C couplings in the structure determination of magnetically oriented macromolecules in solution. Nat. Struct. Biol. 4:732738.



Wang, Y., Frederick, A.F., Senior, M.M., Lyons, B.A., Black, S., Kirschmeier, P., Perkins, L.M., and Wilson, O. 1996. Chemical shift assignments and secondary structure of the Grb2 SH2 domain by heteronuclear NMR spectrsocoopy. J. Biomol. NMR 7:89-98. Weber, P.L. 1996. Protein structure determination from NMR data. In NMR Spectroscopy and Its Application to Biomedical Research (S.K. Sarkar, ed.) pp. 187-239. Elsevier/North-Holland, Amsterdam. Wishart, D.S., Sykes, B.D., and Richards, F.M. 1996. The chemical shift index. A fast and simple method for the assignment of protein secondary structure through NMR spectroscopy. Biochemistry 31:1647-1651.

Wüthrich, K. 1986. NMR of Proteins and Nucleic Acids. John Wiley & Sons, New York. Wüthrich, K. 1998. The second decade—into the third millenium. Nat. Struct. Biol. 5(NMR Suppl.):492-495.

Contributed by Andrew J. Edwards and David Reid SB Pharmaceuticals Welwyn, Herts, United Kingdom

The authors would like to thank their colleagues at SmithKline Beecham Pharmaceuticals for their assistance, especially Drs. Lesley K. MacLachlan and Julia Hubbard for providing the spectra of NPY. In addition, the authors would like to thank Prof. John Walker of the University of Hertfordshire for his collaboration on the Trypanosome calmodulin project.

Structural Biology


Supplement 19

Probing Protein Structure and Dynamics by Hydrogen Exchange–Mass Spectrometry

UNIT 17.6

Determining protein 3-D structure to a resolution of 2 to 3 Å provides a good starting point in understanding protein structure-function relationships. However, protein molecules are highly dynamic, which may be essential for many of their natural functions. For example, the structures of viral coat proteins may change extensively when viruses bind to cellular receptors and disassemble, releasing their genetic material into the cell. To understand structural biology at the molecular level requires a variety of analytical methods for detecting changes in the structures and dynamics of proteins. Low resolution experimental methods, such as circular dichroism (CD; UNIT 7.6) and fluorescence (UNIT 7.7), have been used to detect protein structural changes induced by changing temperature and pH or by adding denaturants. Although the exact locations of structural changes cannot be determined by these methods, they do provide convenient means for showing that changes have occurred. High resolution methods, such as NMR (UNIT 17.5) and X-ray crystallography (UNIT 17.3), are used to determine the location of every atom in a molecule. Hydrogen exchange rates at backbone peptide amide linkages of proteins have been used for many years as a sensitive probe for detecting changes in protein structure and dynamics (Hvidt and Nielsen, 1966; Woodward et al., 1982; Englander and Kallenbach, 1984). This approach is based on the fact that the rate constants at which amide hydrogens in protein exchange with deuterium or tritium in the bulk solvent are highly dependent on protein conformation. Hydrogen exchange rates are correlated with the solvent accessibility of amide hydrogens and their intramolecular hydrogen bonding. Even minor changes in protein conformation may cause rearrangement in the hydrogen-bonding network and alter the solvent accessibility of certain amide hydrogens. These changes often alter the hydrogen exchange rates at some amide linkages. Experimental procedures discussed in this unit are relevant only to hydrogens located at peptide amide linkages in polypeptides. Exchangeable hydrogen located in sidechains, as well as at the N- and C-termini, exchange too fast to be measured by these methods (Englander et al., 1985; Bai et al., 1993). Nuclear magnetic resonance (NMR; UNIT 17.5) has been particularly useful for hydrogen exchange measurements because exchange rates at individual peptide amide linkages can be determined. However, this approach is limited by requirements for large quantities of sample, high protein solubility, and relatively low molecular weight proteins. Numerous studies have demonstrated that mass spectrometry can also be used to detect deuterium levels at peptide amide linkages in proteins labeled under a variety of conditions (Katta and Chait, 1991; Thévenon-Emeric et al., 1992; Miranker et al., 1993; Zhang and Smith, 1993). When proteins are fragmented by acid proteases or collision-induced dissociation (CID MS/MS; UNTI 16.6), hydrogen exchange–mass spectrometry (HX-MS) can be used to determine deuterium levels in short segments of proteins. Particularly important advantages of mass spectrometry for detecting hydrogen exchange include the ability to study large proteins with high sensitivity and at very low concentrations; and the ability to determine the intermolecular distribution of deuterium. This unit provides a brief discussion of the interplay between protein dynamics and hydrogen exchange, a detailed prescription for designing and performing typical HX-MS experiments, and common procedures for processing and interpreting HX-MS data.

Structural Biology Contributed by Lintao Wang and David L. Smith Current Protocols in Protein Science (2002) 17.6.1-17.6.18 Copyright © 2002 by John Wiley & Sons, Inc.


SOME BACKGROUND ON HYDROGEN EXCHANGE The details of hydrogen exchange have been reviewed extensively elsewhere (Woodward et al., 1982; Englander and Kallenbach, 1984; Englander et al., 1985; Bai et al., 1993; Miller and Dill, 1995; Englander et al., 1997; Smith et al., 1997). The following discussion provides the basic hydrogen exchange background required to understand the general HX-MS protocols and to interpret HX-MS data. Hydrogen Exchange in an Unstructured Polypeptide Isotopic hydrogen exchange at each peptide amide linkage in an unstructured peptide is catalyzed by acid and base, as illustrated by the following equation: k int = k H [H + ] + kOH [OH - ] Equation 17.6.1

where kint is the intrinsic exchange rate constant, and kH and kOH are acid- and base-catalyzed rate constants, respectively. The linear dependence of kint on H+ and OH− shows that hydrogen exchange rates are highly dependent on pH. The pH dependence of amide hydrogen exchange rate constant calculated for polyalanine (see Fig. 17.6.1) is slowest at pH 2.4 to 3.0. Above this range, the exchange rate changes by one order of magnitude for each unit of pH. This high sensitivity to pH is an important consideration in all HX-MS experiments. For example, labeling may be performed under physiological conditions, then quenched by decreasing the pH to 2.5 prior to analysis. Temperature is another important factor that affects amide hydrogen exchange rates. The temperature dependence of exchange rates can be predicted using the effective activation energies for kH and kOH, which are 14 and 17 kcal/mol, respectively (Bai et al., 1993). Aside from the effects of pH and temperature, the exchange rate at each individual amide linkage is also dependent on the side chains of their neighboring amino acid residues. These effects, which are generally inductive and steric in nature, appear to be additive. Exchange results from model peptides can be used to predict hydrogen exchange rates in unstructured peptides at any pH and temperature (Molday et al., 1972; Bai et al., 1993). These calculated rate constants for exchange from an unstructured polypeptide are often

Probing Protein Structure and Dynamics by HX-MS


Figure 17.6.1 Dependence on pH of the rate constant for amide hydrogen exchange in polyalanine (Bai et al. 1993). Current Protocols in Protein Science

compared with measured rate constants for exchange from a folded polypeptide to determine the extent the folded structure decreases isotope exchange. Although side-chain effects in unstructured polypeptides alter the amide hydrogen exchange rates by as much as 30-fold, secondary, tertiary, and quaternary structural features of folded proteins may decrease amide exchange rates by as much as 108. It is this large reduction that makes hydrogen exchange a sensitive probe of the folded structures of proteins. Hydrogen Exchange in Folded Proteins Hydrogen exchange under conditions where a protein is generally folded may be described using a two-process model (Kim and Woodward, 1993; Bai et al., 1994; Li and Woodward, 1999), as illustrated in the following equations: kex f

, → f (D) f (H) 

Equation 17.6.2

k

k

k−1

k1

−1 kint 1   → u(D)    → f (D) → u(D) ← f (H) ←  

Equation 17.6.3

where f and u refer to folded and unfolded forms, respectively, and H and D refer to hydrogen and deuterium, respectively. Exchange of amide hydrogens directly from the folded form of a protein is described by Equation 17.6.2. The rate constant for this form of isotope exchange can be described by Equation 17.6.4: kex,f = βk int Equation 17.6.4

where β is the probability for exchange from folded forms and kint is the rate constant for isotope exchange from the totally unfolded polypeptide. It has been suggested that β represents highly localized, low amplitude structural changes (Elber and Karplus, 1987; Kim and Woodward, 1993). In the process illustrated by Equation 17.6.3, hydrogen exchange occurs only after a segment of protein unfolds, exposing several amide hydrogens to the bulk solvent. When exposed to solvent, the amide hydrogens exchange as if they were in an unfolded polypeptide. The unfolding and refolding rate constants are designated here by k1 and k-1, respectively. For most proteins at neutral pH and in the absence of denaturants, k-1 >> kint. In this case, the rate constant for exchange through momentary unfolding of a protein is expressed as kex,u =

k1 kint = K unf kint k-1

Equation 17.6.5

Structural Biology


Supplement 28

where Kunf is the equilibrium constant for momentary unfolding. When k-1 >> kint, a segment of any molecule must unfold and refold many times before exchange within it is complete. This unfolding may involve a short segment or the entire polypeptide backbone. Exchange at individual amide hydrogens may occur through a combination of these two models. Hence, the rate constant for hydrogen exchange, kex can be expressed as the sum of two rate constants, as illustrated in Equation 17.6.6. kex = kex,f + kex,u = (β + K unf )kint

Equation 17.6.6

This expression is particularly important because it links the experimentally measured deuterium exchange rate constant, kex, to important parameters describing protein dynamics, β and Kunf. The variation of kex with low levels of denaturants has been used to separate these two processes (Bai et al., 1993; Kim et al., 1993). Exchange by both processes leads to a random intermolecular distribution of isotope peaks and a single envelope of isotope peaks in mass spectra (Gross et al., 1996; Zhang et al., 1996). It may be noted that other protein dynamics, such as those found when k-1 7 (Yang and Smith, 1997). Although labeling may be continued for days or weeks, as required for detectable H/D exchange at the slowest exchanging amide linkages, protein aggregation and bacterial growth may present problems. If rate constants are to be derived from exchange results, samples should be analyzed at 4 to 5 time points for each decade of time. When performing continuous-labeling experiments, a “master” solution of protein is usually prepared. This solution contains enough protein for LCMS analysis of all samples. For example, if analysis required 10 time points with 300 pmol consumed for each LCMS analysis (typical for a microbore column, 1 × 50–mm), a total of 3 nmol of protein is required. To minimize sources of random error, it is prudent to analyze 2 to 3 samples for each time point. Quenching Exchange Reaction Once hydrogen exchange is initiated, a timer is started to indicate the exchange time. At various time points (10 sec, 30 sec, 1 min, and so on), an aliquot of sample containing enough protein for a single LCMS analysis is removed from the labeling solution. Isotope exchange in this aliquot is quenched by decreasing the pH to 2 to 3 and the temperature to 0°C. If the solution already contains phosphate buffer, the pH can be decreased by simply adding the appropriate quantity of acid. Hydrochloric, phosphoric, and trifluoroacetic (TFA) acids work well. If a buffer effective at pH 2.5 is not present, one must be added. The volume of quench solution to be added depends on several factors, which include volume, pH, and buffer concentration of the labeling solution, as well as the pH and buffer concentration of the quench solution. To minimize dilution of the sample, experiments should be designed so the volume of quench solution is 200 peptides. Because pepsin cleavage sites cannot be predicted with certainty, the peptides cannot be identified reliably by nominal molecular mass alone. However, these peptides may be readily identified by a combination of exact molecular mass, CID MS/MS or C-terminal sequencing using various carboxy-peptidases (McCloskey, 1990). HPLC/ESI-MS Analysis of Deuterated Peptides Although various strategies for measuring deuterium levels in polypeptides by mass spectrometry without the aid of HPLC have been reported, use of directly coupled HPLC offers many advantages. Because the response of ESI to specific peptides often depends on whether other peptides are present, fractionation of peptides prior to analysis by ESI-MS (UNITS 16.8 & 16.9) can substantially increase the number of peptides detected. In addition, separation of the peptides reduces the probability that peaks for peptides with similar molecular masses will overlap. Use of HPLC also allows considerable freedom in the choice of buffers, salts, and denaturants during sample preparation. Because gradient HPLC also serves as a solid phase extractor, it allows analysis of very dilute solutions (Wang et al., 2001). Finally, it is important to note that HPLC is performed with protic solvents that effectively remove deuterium from the side chains (N- and C-termini) via back exchange. As a result, the measured deuterium level reflects the number of deuteriums located at peptide amide linkages. In addition, removal of deuterium from the rapidly exchanging side chains may be important for minimizing deuterium losses during the electrospray ionization process. Although a wide variety of reversed-phase HPLC approaches has been used for HX-MS experiments, minimizing deuterium loss from the peptide-amide linkages is usually the principal concern. When choosing HPLC conditions, it is important to note that the half-life for deuterium loss from peptide amide linkages under typical quench conditions is 20 to 500 min (Bai et al., 1993). In addition to maintaining hydrogen exchange quench conditions (pH 2 to 3, 0°C), one normally uses short gradients and high flow rates to minimize the analysis time. The low temperature is maintained by submerging the injector and column in an ice bath, while the appropriate pH may be maintained using 0.05% TFA or 0.25% formic acid in both mobile phases. Small columns are generally preferred because they minimize sample dilution, thereby increasing sensitivity. However, the low flow rates used with small columns may create delays in formation of the gradient if the

Structural Biology


Supplement 28

HPLC system was not designed for such flow rates. In addition, low flow rates may substantially increase the residence time of peptides in the region following the column. To avoid excessive loss of deuterium in this region, which is not cooled, the residence should be less than ∼1 sec. A typical HPLC protocol suitable for most HX-MS experiments uses a 1 × 100–mm C18 column (e.g., Microtech/Vydac). Columns packed with materials that are not designed for large molecules should be avoided because they are easily plugged by undigested sample or pepsin. Mobile phases consisting of water and acetonitrile, both with 0.05% TFA, are used to elute the peptides within 10 min. A flow rate of 40 to 50 µl/min is a good compromise between separation and speed. Injection of 400 pmol is usually adequate for modern, high-sensitivity mass spectrometers. When the required sample volume is large (e.g., greater than ∼50 µl), the flow rate can be increased to shorten the sample loading time. The upper pressure limit of the column should not be exceeded. To decrease deuterium loss, the loading syringe should be cooled before sample injection. Depending on the salt content of the sample, 1 to 3 min of desalting on the column prior to elution may be advisable. Mass spectra of peptides eluting from the HPLC are recorded continuously. Several different types of ESI mass spectrometers have been used successfully for HX-MS measurements. The authors’ experience is based on four instruments, a Micromass Autospec magnetic sector, a Finnigan LCQ ion-trap, a Micromass VG platform singlequadruple, and a Micromass QTOF. To minimize deuterium loss in the electrospray source, the source tuning parameters, especially the source temperature, must be optimized. For example, significant deuterium loss in the source was detected in the Autospec when the source was operated at temperatures >60°C. Whether significant deuterium loss occurs in any part of the procedure, including inside the ESI source, the instrument should be monitored routinely using completely exchanged peptides. Processing of LCMS Data The next step following data acquisition is the identification of scans that were acquired during elution of each peptide of interest. Spectra from these scans are combined and analyzed to give the average molecular mass of the peptide, from which the number of deuteriums present can be determined. The ions representative of a peptide will have a range of molecular masses depending on the number of heavy atoms (e.g., 13C, 18O, 15N, and D) present. For most studies where the hydrogen exchange processes described by Equations 17.6.2 and 17.6.3 dominate, the intermolecular distribution of deuterium is random, which gives a single envelope of isotope peaks. Linkage between the width of this distribution and the structure/dynamics of the protein has been discussed (Gross et al., 1996; Zhang et al., 1996). Mass spectra of proteins with multiple, long-lived populations may have multiple envelopes of isotope peaks (Miranker et al., 1993; Zhang et al., 1996; Deng and Smith, 1999). The average mass of a peptide, which takes into account both the abundance (Ii) and the mass (mi) of each isotopic form of the peptide, is defined by Equation 17.6.7:

∑ I i mi average mass = i ∑ Ii i


Equation 17.6.7



The software provided with most mass spectrometers can be used to determine the average molecular mass of polypeptides whose isotope peaks are not resolved, as is usually the case for proteins. Additional software, such as MagTran (Zhang) or Excel (Microsoft) may be required to determine the average molecular masses of peptides whose isotope peaks are resolved. Typical HX-MS results obtained using a Micromass Autospec mass spectrometer equipped with an ESI source and array detector are given in Fig. 17.6.3. Approximately 400 pmol of brome mosaic virus (BMV) coat protein (mol. wt. 20 kD) peptic digest was loaded on a C18 reversed-phase column (1 × 50–mm, flow rate 40 µl/min, 2% to 60% acetonitrile in 6 min, 0.05% TFA). The total ion current (TIC), which is similar to a UV chromatogram, is shown in Fig. 17.6.3A. This chromatogram shows that most of the peptides eluted within 6 to 7 min. The selected ion plot for m/z 391-393, representing the peptic fragment 183-188 of BMV capsid protein, shows that this peptide eluted in a 30-sec time interval with an average retention time of 3.7 min (Fig. 17.6.3B). Although many

A

Relative intensity

100

50

0 2:24

B

total ion current (TIC)

3:36

4:48 min

6:00

7:12

100 Relative intensity

TIC of m/z 391-393

0 2:24


C

50

3:36

4:48 min

6:00

7:12

391.7

mass spectrum of scans from 3:30-3:58

50 639.5

430.4 0 320

782.8

720

520 m/z


D

50

391.7

mass spectrum of m/z 391.7

392.2

centroid = 391.85 charge state = 2

392.7 0 390

394

392

396

m/z region selected to determine the average mass of this peak

Figure 17.6.3 Representative HPLC/ESI-MS spectra in a typical HX-MS analysis showing data processing steps involved in determining the average mass of a particular peptide.

Structural Biology


Supplement 28

other peptides eluted in the same time interval (see Fig. 17.6.3C), their ions were well resolved in the mass spectra. The average molecular mass of this peptide was determined from the centroid of the entire envelope of isotope peaks multiplied by the charge-state (Fig. 17.6.3D). The deuterium level in this peptide was determined by subtracting the average mass of the non-deuterated form of the same peptide, which was determined in a separate experiment. Adjustment for Artifactual Exchange Although quench conditions may decrease isotope exchange by a factor of 105, a significant amount of exchange usually occurs during digestion and HPLC. To compare differences in the structure and dynamics of two closely related forms of a protein, one may simply compare the deuterium levels found for the two forms. That is, it is not necessary to adjust measured deuterium levels for exchange that occurred during digestion and analysis. However, more detailed analysis, which may include quantitative comparison of found and calculated deuterium levels, requires such adjustments. The amount of deuterium lost from a partially deuterated peptide may be estimated from the amount of deuterium lost from a totally deuterated peptide (i.e., totally exchanged in D2O), which can be determined by analyzing a sample of the totally deuterated protein (Zhang and Smith, 1993). This sample, the 100% reference, may be prepared most easily

A

120 100% control <m100%> = 393.46 D = 3.22

Relative intensity

100 80 60 40 20 0 390

B

391

392

393

394

labeled peptide <m> = 392.84 D = 1.98

Relative intensity

80 60 40 20 0 390

391

392

393

394

Relative intensity

395

396

120 0% control <m0% > = 391.98 D = 0.26

100 80 60 40 20 0 390


396

120 100

C

395

391

392

393

394

395

396

m/z

Figure 17.6.4 Mass spectra of a peptide fragment from BMV capsid protein that are (A) fully deuterated, (B) partially deuterated, and (C) nondeuterated.



by incubating the protein in D2O for 24 hr at pD 2 to 3 and elevated temperature. Denaturants may be required to achieve complete exchange. The mass spectrum of the segment including residues 183-188 of BMV capsid protein completely exchanged in D2O is given in Figure 17.6.4A. The centroid of this envelope of isotope peaks shows that this segment had 3.22 deuteriums when it arrived at the mass spectrometer. This segment has 4 exchangeable amide hydrogens (6 residues, one of which is Pro). Finding 3.22 deuteriums shows that 80.5% of the deuterium at peptide-amide linkages was retained during digestion and analysis. The deuterium recovery, 80.5%, may be used to adjust the deuterium level found in the same peptide when it is only partially deuterated at the time exchange was quenched. The mass spectrum of the 183-188 segment derived from partially deuterated BMV capsid protein (Fig. 17.6.4B) indicates that this segment had 1.98 deuteriums when it arrived at the mass spectrometer. Adjusting for the 80.5% recovery indicates that this segment actually had 2.46 deuteriums at the time exchange was quenched. Because deuterium loss is sequence specific (Bai et al., 1993), the deuterium recovery for each peptide must be determined. Furthermore, because the recovery is based on the sum of deuterium levels found at all positions, the adjustment is only an approximation of the recovery in partially deuterated forms of the same peptide. When the recoveries are relatively high, adjustments for artifactual deuterium losses are small. In such cases, errors in the adjustment may be insignificant. For optimized experimental conditions, one may expect deuterium recoveries of 65% to 95% for analysis of peptic fragments and 85% to 95% for intact proteins. Failure to consider deuterium exchange-in during digestion is another source of error in adjustments based only on deuterium recovery. A more accurate method, which adjusts for deuterium exchange-in during digestion, has been described (Zhang and Smith, 1993). This approach requires analysis of a 0% reference sample that was digested in a solution under the same conditions (including H/D ratio) as used for all other samples. The mass spectrum of the 183-188 segment from a 0% reference sample of BMV capsid protein (Fig. 17.6.4C) indicates that this segment had 0.26 deuteriums. These deuteriums exchanged-in during quench and digestion. The deuterium level in this segment at the time H/D exchange was quenched can be estimated from Equation 17.6.8: D=

< m > − < m0% > ×N < m100% > − < m0% > Equation 17.6.8

where D is the adjusted number of deuteriums in a peptide, N is the total number of amide linkages, and <m>, <m0%>, and <m100%> are the average masses of a peptide from a partially deuterated sample, the 0% control, and the 100% control, respectively. The application of Equation 17.6.8 to segment 183-188 indicates that this segment had 2.32 deuteriums at the time exchange was quenched. Errors in such adjustments due to different exchange rates at different linkages have been discussed (Zhang and Smith, 1993). Experimental Error The accuracy with which deuterium levels can be determined is of critical importance because it sets the threshold for detecting structural changes in a protein. For example, significant structural changes may alter deuterium levels by as little as 0.1 deuteriums. Experimental errors come from two major sources: sample preparation and mass measurement. The most important sources of error in sample preparation include time, pH, and temperature at all steps (e.g., exchange-in, digestion, and HPLC). It is helpful to note that the exchange rate changes ten fold for each pH unit and approximately three fold for

Structural Biology


Supplement 28

each 10°C change in temperature (Bai et al., 1993). The accuracy of the mass measurement depends on the signal-to-noise ratio and the particular instrument. However, one may expect the uncertainty in mass measurement to be 10 to 500 ppm. Samples to be compared with one another should be prepared in the identical buffer and analyzed under identical conditions. Deuterium in solvents and buffers may impose isotope effects on measured hydrogen exchange behavior (Bai et al., 1993; Connelly et al., 1993). The isotope effects on amide hydrogen exchange rates in peptides are generally small. However, the solvent isotope effect on the glass electrode used to measure pH (pD) may be important for some types of analyses. The authors’ laboratory, as well as many others, normally reports the value actually read from the pH meter. HX-MS DATA INTERPRETATION The deuterium levels in proteins or protein fragments may be used to detect changes or differences in protein structure and dynamics. For the simplest interpretation of hydrogen exchange results, finding different deuterium levels for the same labeling time is strong evidence that a change has occurred. For a more detailed interpretation, time-course exchange-in data can be fitted into a sum of first-order rate expressions to obtain the distribution of rate constants for exchange at amide linkage in the protein or peptic fragment. Changes in the exchange rate distribution are used to characterize changes in protein structure and dynamics. The models for H/D exchange (Equations 17.6.2 and 17.6.3) provide physical models for relating H/D exchange data and protein structure/dynamics. The discussion below illustrates both approaches. Detecting Structural Changes in Protein by Comparing Deuterium Levels Changes in the non-covalent structures of proteins often involve rearrangement of the intra-molecular hydrogen bonding network and alteration of the solvent accessibility of peptide amide hydrogens. Such structural changes are reflected by changes in the parameters β and Kunf defined in Equation 17.6.6. Furthermore, such structural changes may be reflected by changes in the deuterium levels found in peptides if the measurements are performed under conditions where kint is unchanged. Many applications of HX-MS have followed this approach (i.e., conditions where the pH and temperature are constant). Because the dependence of kint on pH and temperature is known, H/D exchange can also be used to detect changes in structure/dynamics that are induced by changes in pH and temperature. To cancel the effect of pH on the intrinsic rate of H/D exchange requires consideration of the labeling time. The deuterium level at any peptide-amide linkage (D) is given in Equation 17.6.9:

D = 1 − e− kex t Equation 17.6.9

where t is the time the protein was exposed to D2O. According to Equations 17.6.1 and 17.6.6, kex can be expressed as in Equation 17.6.10:


kex = (β + K unf ) kOH [OH − ] Equation 17.6.10



Combining Equations 17.6.9 and 17.6.10 results in Equation 17.6.11:

D =1− e

−(β+ Kunf ) kOH [OH − ]t

Equation 17.6.11

which indicates that deuterium levels may be used to detect pH-induced structural changes in folded proteins if different exposure times are used for different pH such that the product [OH−]t is constant. This general approach of using different exposure times to compensate for changes in intrinsic rates can also be used to compare hydrogen exchange results obtained at different temperatures (Zhang and Smith, 1993; Liu and Smith, 1994). This approach has been used to study the pH-induced swelling of the BMV capsid. Table 17.6.1 lists the exposure times used at different pD (D2O buffer), to determine structural changes in the virus capsid when the pH was increased from 5 to 7. These labeling times were chosen to keep the product [OH–]t constant. To satisfy this requirement, the labeling time for pD 5.43 was 74.1 times greater than the labeling time at pH 7.30. The general experimental procedure used for this study is illustrated in Figure 17.6.2. The differences in the deuterium levels found in the peptic fragments from BMV capsid protein–labeled at pD 5.43 and 7.30 are shown in Figure 17.6.5. A set of 16-peptide fragments were reported to cover nearly the entire backbone of the BMV capsid protein (20 kD, 187 residues). The change in deuterium level found for each fragment is expressed in both relative units (the percentage of the total number of amide linkages, given as the y-axis) and absolute units (the number of deuteriums, given above each bar). Both presentation modes have specific merits. Expressing the deuterium difference as the average difference per amide linkage facilitates comparing deuterium changes in fragments of different sizes. On the other hand, showing the difference in the absolute number of deuteriums facilitates comparing these changes with the error of the measurement. The uncertainty of measurement for each segment, expressed as percent, is indicated by the error bars in Figure 17.6.5. These results indicate that H/D exchange was greater in most segments of the capsid protein when the pD was 7.3, suggesting increased flexibility with increasing pH. The largest increases in deuterium level were found in segments including residues 70-90 and 118-151 (shaded dark gray), suggesting that these regions are most susceptible to destabilization with increasing pH. The X-ray crystal structure of cowpea chlorotic mottle virus (CCMV; Speir et al., 1995), a virus closely related to BMV, indicates that these structural changes likely occur at the center of the icosahedral asymmetric trimer.

Table 17.6.1 Exchange Times Used to Label Intact Brome Mosaic Virus in D2O at pD 5.43 and 7.30a,b (Wang and Smith 1999)

pD 7.30 (min) 0.33 1.00 5.50 30.00 90.00

pD 5.43 (min) 24.5 74.1 407.6 2223.0 7113.6

aThe labeling time at pD 5.43 was 74.1 times longer than the labeling time at pD 7.30. bpD values were taken directly from the pH meter without correction.

Structural Biology


Supplement 28

50 40 30

4.0

0.9

3.2 4.1

20

1.0

6.1

10

–1.1 –0.8

–0.5 166-170 156-165

–30

183-188

171-182

deuterium change at pD 7 _ + 0

152-155 149-151

136-148

123-135

109-118

118-121

97-108

91-96

70-90

48-61

62-69

–20

0.2

0.3

0 –10

1.1

1.0

0.5

1-47

Deuterium difference (%)

0.9

1.1

Protein fragment

Figure 17.6.5 Changes in deuterium levels found in segments of the BMV capsid protein following labeling of the intact virus particles at pD 5.43 or 7.30. The changes are plotted as rectangular boxes, where the height indicates the change expressed as a percentage of the total number of NHs in the peptide and the width indicates the length of the backbone segment. The change in the absolute number of deuteriums is given on the top of each bar. Shading indicates the relative degree of change found in each region. Adapted from Wang and Smith (2001).

Probing Protein Dynamics by Comparing Hydrogen Exchange Rate Distribution More detailed analysis of time-course data leads to the distribution of exchange rates within a particular segment and information on localized dynamics. The deuterium level measured in a peptic fragment is the sum of the deuterium levels at each peptide amide linkage. The levels at these linkages may span a range of 106 to 108. Although exchange rates at specific linkages cannot be determined from time-course data, the distribution of exchange rates within a peptic fragment can be estimated by fitting the data to a simple model. For example, the four-compartment model illustrated by Equation 17.6.12 has been used to determine the number of amide hydrogens with very fast, fast, slow, and very slow exchange rates (Engen et al., 1999). D = N − [( n1 (e − mi t ) + n2 (e − m2 t ) + n3 (e − m3t ) + n4 (e − m4 t ) Equation 17.6.12


Variable parameters n1, n2, n3, and n4 indicate the number of linkages where the H/D exchange rates are fast, intermediate, or slow (m1, m2, m3, and m4, respectively). Although the model used here is a major simplification of reality, it does allow one to distinguish between conformations where all linkages exchange at the same rate and conformations where there is a wide distribution of exchange rates. Other useful approaches for determining the distribution of exchange rates within peptic fragments of proteins have been described (Zhang and Smith, 1993; Zhang et al., 1997).



16.0

Deuterium level

12.0

8.0

4.0

0.0 0.01

0.1

1 10 Exchange time (min)

100

1000

Figure 17.6.6 Time-course exchange-in plot of segment 7-27 of free SH2 (solid triangles) or SH2 bound with a short peptide (open circles). Adapted from Engen et al. (1999).

Application of Equation 17.6.12 is illustrated using exchange-in data for a peptic fragment of the SH2 domain of hematopoietic cell kinase (Hck). The SH2 domain was labeled for 10 sec to 8 hr in D2O buffer (pH 6.9) with and without a 12-residue peptide known to bind to SH2. Deuterium levels found in the peptic fragment including residues 7-27 are presented in Figure 17.6.6 for both free and bound SH2. These results show that the bound form always has less deuterium than the free form. Application of Equation 17.6.12 to these data (see Table 17.6.2) shows that rate constants for H/D exchange within this short segment span a range >0.001 to 4 min−1. For the free form of SH2, this analysis shows that the hydrogens at 9 amide linkages exchange with rate constants >4 min−1 while three exchange with rate constants 4 min−1 has decreased from 9 to 8, and the number of linkages where exchange occurred with kex 4.0) 8.0 (>4.0)

2.5 (2.1) 3.0 (1.6)

4.5 (0.004) 1.5 (0.006)

3.0 ( Cys 458 > Cys 267 > Cys 287 > Cys 169 > Cys 635 > Cys 290 = Cys 496

Figure 17.8.12 Tailspike cysteines ranked according to the strengths of S–H⋅⋅⋅X hydrogen bonding. Structural Biology


Supplement 33

a highly versatile portal subunit structure capable of undergoing path-dependent switching between different conformational states (Bazinet and King, 1985; Valpuesta and Carrascosa, 1994). Crystallography and imaging methods reveal many architectural features of isolated portals (Valpuesta et al., 1999; Simpson et al., 2000; Orlova et al., 1999; Droge and Tavares, 2000), but provide little insight into molecular mechanisms of subunit folding, assembly, or stability. Recent biochemical (Moore and Prevelige, 2001) and Raman studies (Rodriguez-Casado et al., 2001; Rodriguez-Casado and Thomas, 2003) of the P22 portal protein provide a starting point for elucidating the complex structural and functional roles of the portal subunit. Raman spectra of unassembled (monomeric) and assembled (dodecameric) forms of the P22 portal protein are shown in Figure 17.8.13. The data indicate an α/β fold that is not appreciably altered by subunit assembly. Conversely, portal assembly generates major changes in the Raman markers that are diagnostic of side chain local environments. Most striking are changes observed in the 2520 to 2600 cm−1 region (Fig. 17.8.13, right panel), which indicate alterations in cysteine S–H hydrogen-bonding environments. In the case of the monomer, the Raman S–H profile is relatively broad with a central peak at 2562 cm−1 and a prominent shoulder of nearly equal intensity at 2557 cm−1. The positions, intensities, and widths of these overlapping bands suggest that although all four cysteines (Cys 153, Cys 173, Cys 283, Cys 516) of the monomer engage in relatively robust S–H⋅⋅⋅X hydrogen bonds, they partition into two sub-populations characterized by markers near 2557 and 2562 cm−1, respectively. For the dodecamer, the Raman S–H bands are collectively shifted to higher wavenumber values, indicative of general weakening of S–H⋅⋅⋅X hydrogen-bond strengths with assembly (Rodriguez-Casado et al., 2001). To further characterize the roles of portal cysteines in subunit folding and dodecamer assembly pathways, Raman spectroscopy was employed in combination with site-directed mutagenesis to measure Raman S–H signatures of the four single-site Cys → Ser mutant portals (Fig. 17.8.14; Rodriguez-Casado and Thomas, 2003). The data of Figure 17.8.14 resolve the residue-specific Raman sulfhydryl bands and provide an internally consistent S–H⋅⋅⋅X classification scheme for the portal monomers, as given in Table 17.8.3. On the basis of the previously established correlations (Li and Thomas, 1991; Li et al., 1992), the results show that the composite S–H Raman band envelope of the WT monomer comprises the following: (1) moderate S–H⋅⋅⋅X interactions from Cys 173 and Cys 283, and (2) strong S–H⋅⋅⋅X interactions from Cys 153 and Cys 516. Corresponding analysis of the data from mutant portal assemblies leads to the following additional conclusions: (3) Each portal variant exhibits a Raman S–H signature that is greatly perturbed by assembly; (4) all assemblies exhibit a Raman peak at 2570 cm−1 or higher, despite the absence of such a marker in the monomeric form; (5) with the exception of C516S, no mutant exhibits an S–H group that is promoted to a stronger hydrogen-bonding state with assembly.

Raman Spectroscopy of Proteins

The difference spectra computed between mutant and WT portal assemblies are complex. In previous work on Cys → Ser mutant tailspikes of phage P22, Raso and co-workers (2001) were able to deduce the unique S–H signatures of all eight cysteines of the WT tailspike. The tailspike Raman data indicated that Cys signatures are uncoupled to one another, i.e., mutation of one Cys site did not perturb the S–H⋅⋅⋅X hydrogen-bonding state of any other Cys site (Raso et al., 2001). While Figure 17.8.14A shows this to be also true for the portal monomers, Figure 17.8.14B indicates that cysteine signatures of the assemblies are not similarly conserved. A tabulation of plausible assignments and S–H⋅⋅⋅X hydrogen-bonding schemes for the mutant portal assemblies is given in the right column of Table 17.8.3.



A

800

1200 cm–1

1600

2587 2559

1648 1656 1678 1669

1548 1558

1469

1400

2587

2557 2562

1653 1672

2565

1000

1211 1232 1259 1278 1305 1354 1334

1080

1006

1064

852 892

759

difference

600

2566

1606 1617 1655 1672

1551

1554 1606 1617

1031

1080

827 852 901 934 957

757 756

828 852 901 935 959

monomer 620 642

Raman intensity

620 642

dodecamer

1002 1031 1079 1103 1102 1126 1126 1173 1174 1207 1207 1252 1244 1268 1269 1318 1320 1340 1340 1400 1400 1423 1423 1448 1448 1460 1459

B

1800 2520 2560 cm–1

2600

Figure 17.8.13 (A) Raman spectra (600-1800 cm−1, 532 nm excitation) of the P22 portal protein monomer (middle trace) and dodecamer (top trace) and the digital difference spectrum (second trace from bottom) obtained by subtracting the spectrum of the monomer from that of the ring. Also shown (bottom trace) is an approximate three-fold amplification of the difference spectrum. Proteins were dissolved to ∼40 mg/ml in 20 mM Tris⋅Cl (pH 7.5)/100 mM NaCl/2 mM EDTA. Spectra were corrected for contributions of the buffer solution and the gently sloping background typical of protein Raman spectra. Identical spectra were obtained for protein concentrations in the range 25 to 100 mg/ml. (B) Raman spectra (2520-2600 cm−1, 532 nm excitation) of the P22 portal protein monomer (middle trace) and dodecamer (top trace) and the digital difference spectrum (bottom trace) obtained by subtracting the spectrum of the monomer from that of the dodecamer. Data are adapted with permission from Rodriguez-Casado et al., 2001.

In addition to assembly-related changes to the hydrogen-bonding state of the portal cysteines, assembly-related perturbations are observed for portal tyrosines and tryptophans. The portal subunit contains 26 tyrosines (Eppler et al., 1991). Most occur within the central 50% of the sequence (residues 223 to 571) and none occurs in the C-terminal segment (572 to 725). The bottom trace of Figure 17.8.13 shows that the tyrosine marker at 852 cm−1 undergoes the largest relative intensity change in the Raman spectrum. This implicates the central region of the sequence in the intersubunit interface. A similar conclusion is reached from consideration of perturbed Trp markers (Rodriguez-Casado and Thomas, 2003). In contrast to the considerable perturbations to side chain environments accompanying portal assembly, changes in secondary structure are small. In the case of amide I (1600 to 1750 cm−1, Fig. 17.8.13), only small intensity changes (∼2%) in α-helix (1656 cm−1) and β-strand (1669 cm−1) markers are observed. The difference spectrum is consistent

Structural Biology


Supplement 33

Raman intensity

C516S C283S C173S

differences 2555 2563

monomer profiles 2558 2563

A

C516S-WT

C283S-WT

C173S-WT

C153S C153S-WT

WT

2555

assembly profiles 2563 2574 2586

B

2590

2555

2590

differences

C516S-WT

Raman intensity

C516S C283S C173S

C283S-WT

C173S-WT

C153S

2555

2590

2555 2590

C516S

2563

2586

assembly-monomer differences 2559 2563

C

2586

WT

2563

C153S-WT

2587

Raman intensity

C283S

C173S C153S

2563

WT

2555

2590

cm–1

Raman Spectroscopy of Proteins

2555

2590

cm–1

Figure 17.8.14 Raman profiles of WT and mutant portal proteins in the region of cysteine S–H stretching vibrations (2520-2620 cm−1). Conditions are as given in Figure 17.8.13. (A) The left panel compares S–H profiles of the monomeric proteins (WT and mutants, as labeled). The right panel shows the S–H signature specific to each cysteine of the WT monomer, obtained by subtracting the WT spectrum from the corresponding mutant spectrum. (B) The left panel compares S–H profiles of the assembled proteins (WT and mutants, as labeled). The right panel shows the S–H signature specific to each cysteine of the WT assembly, obtained by subtracting the WT spectrum from the corresponding mutant spectrum. (C) The left panel shows the computed assembly-minusmonomer difference spectrum for each variant, as labeled. The right panel demonstrates the non-additivity of cysteine signatures for the assembled proteins. Thus, the Raman signature of C516 that is obtained by subtracting the C516S mutant spectrum from the WT spectrum (B, top right) is not identical to the C516 signature (dashed-line trace) generated by subtracting the sum of signatures of C153, C173 and C283 (solid-line trace) from the WT spectrum (dotted-line trace). Data are adapted with permission from Rodriguez-Casado and Thomas, 2003.



Table 17.8.3 Cysteine Sulfhydryl Raman Markers and S–H⋅⋅⋅X Hydrogen-Bond Strengths of the P22 Portal Proteina

Cysteineb

Monomeric form

Assembled form

cm−1

H-bond

cm−1

H-bond

Cys 153 Cys 173

2558c 2563c

strong moderate

Cys 283

2563c

moderate

Cys 516

2558c

strong

2563c 2563d (50%) 2586d (50%) 2563d (27%) 2586d (73%) 2563 (50%) 2586 (50%)

moderate moderate very weak moderate very weak moderate very weak

aHydrogen-bond strengths are based upon model compound studies of Li and Thomas (1991). bAmino acid residue in wild-type portal protein. cPeak of Raman S-H stretching band. dComponent of closely spaced doublet. The percentage of the total Raman intensity contributed by each

component of the doublet is given in parentheses.

with curve fitting results (Berjot et al., 1987), which suggest only a marginal change in portal subunit secondary structure with assembly. It is interesting that this type of Raman difference fingerprint parallels that observed with P22 procapsid maturation (Prevelige et al., 1993). Such a different signature may be diagnostic of relatively strong intersubunit interactions. In summary, Raman studies of the P22 portal protein show that assembly leads to only a small perturbation of subunit secondary structure but large changes in the local environments of subunit side chains. In particular, portal assembly results in a major reorganization of cysteine S–H hydrogen-bonding states, substantial alteration of tyrosine O–H hydrogen-bonding states and large changes in tryptophan hydropathic environments. The Raman-monitored changes in side chains indicate that the N-terminal third and central third of the portal sequence are involved in intersubunit recognition. Other applications to icosahedral viruses In other recent Raman applications to icosahedral viruses, the method has been employed to probe the coat protein recognition domain of the P22 scaffolding protein (Tuma et al., 1998b), identify secondary and tertiary structures and enzymatic activity of the RNAtranslocating enzyme (P4) of the φ6 nucleocapsid (Juuti et al., 1998; Jenkins et al., 1999), investigate capsid assembly in PRD1 (Tuma et al., 1996a), identify subunit-specific interactions of the φ6 procapsid (Benevides et al., 2002), and characterize conformations and interactions of the packaged viral DNA genomes within capsids of bacteriophages T7, PRD1, and P22 (Aubrey et al., 1992; Tuma et al., 1996b; Overman et al., 1998). Proteins of Filamentous Viruses Bacteriophages fd, f1, and M13 are structurally identical members of the Ff group of class I filamentous viruses, which infect strains of Escherichia coli. The cylindrical Ff filament (∼880-nm length, ∼6-nm diameter) contains a covalently-closed ssDNA genome of 6410 nucleotides, sheathed by ∼2750 copies of a 50-residue α-helical capsid subunit (protein pVIII) plus a few copies of minor proteins at the filament ends. The capsid subunit sequence (1AEGDDPAKAA 11FDSLQASATE 21YIGYAWAMVV 31VIVGATIGIK 41 LFKKFTSKAS) is essentially identical among members of the Ff class (Day et al., 1988). The structure of the native Ff assembly has been studied by methods of solution

Structural Biology


Supplement 33

and fiber spectroscopy and fiber X-ray diffraction, which collectively provide many details of capsid structure and filament architecture (Day et al., 1988; Marvin, 1998). Overall, the results show that the ssDNA core of Ff is coated by a superhelical array of pVIII subunits arranged with five-fold rotational symmetry and an approximately twofold screw axis (C5S2 symmetry), and that the capsid subunit is a continuous α-helix tilted by a small angle from the virion axis. The class II filamentous viruses, represented below by Pseudomonas phages Pf1 and Pf3, differ in filament symmetry (C1S5.4) and other details from the class I particles (Welsh et al., 1998; Welsh et al., 2000). Although a complete three-dimensional structure is not known for any filamentous virus, assembly models have been proposed for Ff, Pf1, and Pf3 on the basis of available data. Because ssDNA constitutes only a small percentage (120 unique membrane protein structures solved to date (http://blanco.biomol.uci.edu/Membrane Proteins xtal.html). The major bottlenecks in this field are the expression of sufficient quantities of functional membrane proteins and the growth of well ordered, three-dimensional crystals for X-ray analysis. Although it is necessary to screen more parameters to obtain good membrane protein crystals than for soluble protein crystals, the approach is systematic and benefits from the availability of a wide variety of homogeneous detergents, the implementation of nanoliter pipetting robots, and the use of numerous 96-condition screens. This unit illustrates the steps required for crystallization of a bacterial outer membrane protein starting from a purified sample in detergent. For a description of expression and purification procedures used to isolate an E. coli outer membrane protein, see Yue et al. (2003). Crystallization procedures will vary depending on the membrane protein of interest, but the procedures described in this unit are generally applicable to inner membrane proteins as well. Support Protocol 1 describes the exchange of the purified membrane protein into a detergent solution suitable for crystallization, using size-exclusion chromatography. Support Protocol 2 describes preparation of the sample for crystallization, while Basic Protocols 1 to 3 describe crystallization procedures to identify and optimize conditions for the production of highly ordered membrane protein crystals.

ROBOTIC SCREENING OF CRYSTALLIZATION CONDITIONS For robotic screening, the hanging drop method is preferred over the sitting drop method because small (0.4 µl), detergent-containing sitting drops tend to adhere to the side of the sitting drop well, making visualization much more difficult.

BASIC PROTOCOL 1

Materials 96-condition screen (Table 17.9.1) 96-well U-bottom microplates (Greiner Bio-One) 8- or 12-channel multi-channel pipet Nanoliter pipetting robot (e.g., Mosquito, TTP LabTech) 96-well format crystallization sheets (e.g., CrystalClene Sheets HT, Molecular Dimensions) Constant temperature incubators (e.g., Rumed, Molecular Dimensions) 1. Prepare the screens by transferring 50 µl of each well solution from a 96-condition commercial screen to a 96-well plate using a multi-channel pipet. To prevent dust from entering the wells and to minimize evaporation, stack the finished plates, placing an empty plate on top. These plates should be used on the day they are prepared.

2. Initialize and calibrate the nanoliter pipetting robot according to the manufacturer’s instructions. The authors use a dispensing program that pipets 0.2 µl of protein onto all 96 positions of a CrystalClene sheet, and then pipets 0.2 µl of well solution onto the corresponding protein drops. Structural Biology Contributed by Travis J. Barnard, Jeremy L. Wally, and Susan K. Buchanan Current Protocols in Protein Science (2007) 17.9.1-17.9.15 C 2007 by John Wiley & Sons, Inc. Copyright


Table 17.9.1 Suggested 96-Condition Screens for Initial Crystallization Trials of Membrane Proteinsa

Hampton Research Crystal Screen HT Index HT Salt Rx HT Nextal Biotechnologies The Classics Lite The PEGs The AmSO4 The Anions The Cations The MPDs The pHClear The MbClass Suite The MbClass II Suite Molecular Dimensions Memstart/MemSys Emerald Biosystems Wizard Bloc (I&II) Cryo Bloc (I&II) OZMA 1-4 OZMA 8-10 a These screens are listed by manufacturer in random order.

Many robots do not have the capability to pipet hanging drops as described above, but instead are designed to pipet sitting drops. In this case, sitting drops can be used with alternative microplates and coversheets described in the Alternate Protocol, using a robot to pipet 0.2 µl protein plus 0.2 µl well solution.

3. Once the robot has finished the pipetting program, carefully apply the CrystalClene sheet to the 96-well microplate so the drops are centered over the wells. Firmly smooth the coversheet over the plate to seal the wells. Remove the protective backing from the cover sheet so the drops can be visualized. The Mosquito robot comes with a convenient adaptor that facilitates centering the coversheet over the plate. The plate is now complete.

4. Repeat the pipetting program and step 3 for the remaining plates. Incubate the plates at the desired temperature(s), usually 21◦ C and/or 4◦ C. If the membrane protein is known to be unstable at room temperature, it may be advisable to incubate the plates at 4◦ C; however, crystallization reagents sometimes stabilize membrane proteins, such that samples requiring purification at 4◦ C may still form crystals at 21◦ C. These two temperatures have yielded the most crystals of all types, but other temperatures may also be tried.

Crystallization of Integral Membrane Proteins

With either Basic Protocol 1 or the Alternate Protocol, the plates should be evaluated after 2 days, again after 1 week, and finally after 2 weeks. The drops usually reach equilibrium within 1 week, therefore if no crystals are observed after 2 weeks, it is advisable to discard the plates. If crystals are obtained, the conditions should then be scaled up and optimized in 24-well plates using larger volumes, as described in Basic Protocols 2 and 3. If no crystals are obtained, the following modifications can be made to the initial 96-condition screens. (1) Adjust the protein concentration. If the majority of



the drops are still clear after 2 weeks, try a higher protein concentration. If the majority have precipitated (characterized by diffuse, amorphous material in the drop), then try lowering the protein concentration. (2) Adjust the incubation temperature. If no crystals are obtained, try 15◦ C or 10◦ C. Also try higher temperatures, such as 25◦ C or 30◦ C, as it is unclear how temperature will affect the solubility of a protein and/or its propensity to aggregate. Note that the critical micelle concentration (cmc) of the detergent will also change with temperature, affecting crystallization results. (3) Change the detergent. The authors routinely screen new membrane proteins in every detergent in which they are stable (with reference to Table 17.9.4), and this large-scale screening is done early in a project. All 17 kits are screened for a single membrane protein in each detergent being tested to identify detergents that may yield useful membrane protein crystals. (4) If all else fails, try multiple homologs (Chang et al., 1998), a different construct (Locher et al., 2002), or complexing the membrane protein to a soluble protein (such as a natural ligand or protein partner, or an antibody fragment; Dutzler et al., 2003). The large detergent micelle surrounding a membrane protein may make crystallization difficult, so having a soluble protein bound to it may provide additional crystal contacts by expanding the hydrophilic surface.

MANUAL SCREENING OF CRYSTALLIZATION CONDITIONS If a nanoliter pipetting robot is not available, the same 96-condition kits can be screened manually, using 96-well sitting-drop plates. This method is very fast but requires significantly more protein because 1-µl drops of membrane protein are mixed with 1-µl drops of well solution. To screen all 17 kits as described in Basic Protocol 1, at least 1700 µl of concentrated (10 mg/ml) membrane protein sample is required.

ALTERNATE PROTOCOL

Materials 96-condition commercial screens (Table 17.9.1) 96-well conical sitting-drop microplates (Corning/Hampton Research) 8- or 12-channel multi-channel pipet (50-µl volume) 8- or 12-channel multi-channel pipet (1-µl volume) 1-µl repeater pipet ClearSeal Film (Hampton Research) Microplate centrifuge Constant temperature incubators (e.g., Rumed, Molecular Dimensions) 1. Prepare the screens by transferring 50 µl of each well solution from a 96-condition commercial screen to a 96-well sitting drop plate using a 50-µl volume 8- or 12-channel multi-channel pipet. 2. Using a 1-µl multichannel pipet, pipet 1 µl from each well into each sitting drop sub-well (small depression where the sample drop will reside). 3. Using the 1-µl repeater pipet, quickly pipet 1 µl of concentrated membrane protein sample into each of the 96 sitting-drop sub-wells. To avoid cross-contamination, do not mix the droplets at this point; centrifugation of the microplates will ensure mixing.

4. Cover the microplate with a ClearSeal film, using light pressure to seal it. 5. Centrifuge the microplate according to manufacturer’s recommendations for 2 min to merge precipitant and protein droplets. 6. Repeat steps 1 to 5 for all other 96-condition kits, as appropriate. 7. Incubate the plates at the desired temperature(s), usually 21◦ C and/or 4◦ C. See annotation of Basic Protocol 1, step 4, for guidance on optimization of conditions.

Structural Biology


Supplement 47

BASIC PROTOCOL 2

PREPARATION OF 24-WELL PLATES FOR OPTIMIZATION OF CRYSTALLIZATION CONDITIONS Once promising crystallization conditions have been identified using the screens described in Basic Protocol 1 or the Alternate Protocol, these conditions are repeated using larger drop volumes and expanded into finely sampled grid screens, usually varying two parameters (such as precipitant concentration and pH) on a given 24-well plate. This protocol describes how to set up 24-well plates in the context of working with membrane proteins in detergent.

Materials 10× salt stock solutions 10× buffer stock solution Precipitant stock solutions (concentrations will vary) Cryoprotectants (e.g., glycerol), optional 24-well plates (e.g., VDX plates with sealant, Hampton Research) Repeater pipet (optional) Orbital platform shaker (optional) 22-mm siliconized square cover slides (Hampton Research) Constant temperature incubators (e.g., Rumed, Molecular Dimensions) 1. Prepare the well solutions. Sequentially add the 10× salt solution, 10× buffer solution, and precipitant stock solutions to each well of a 24-well plate to attain a final volume of 1.0 ml in each well. If a cryoprotectant, such as glycerol, is to be added, include it in the precipitant solution. It is not necessary to add detergent to the well solutions, because the protein contains detergent at a concentration of at least twice the cmc so that dilution with an equal amount of precipitant will still result in sufficient detergent to maintain the protein in solution. All solutions or their components should be made with the highest quality water available and passed through 0.22-µm filters to remove particulates. Hampton Research (as well as many other companies) provides many salt, buffer, and precipitant stock solutions that are ready-to-use. Plates should be set up in a clean area as free of dust as possible. Normally, 0.1 ml of the salt and buffer stock solutions is added per well, which has a total volume of 1 ml. The precipitant stock solutions (a separate solution is made for each precipitant concentration) are made such that 0.8 ml can be added per well to yield the correct final concentration. If necessary, cryoprotectant is added to the precipitant stock solutions to help prevent damage to the crystals during flash-freezing in liquid nitrogen or liquid nitrogen–cooled propane. The authors normally use glycerol as a cryoprotectant at a final well concentration of 5% to 15% (v/v). Since these components are often very viscous, quantities are measured by weight according to density and not by volumetric pipetting. A repeater pipet will greatly reduce the time necessary to dispense the stock solutions into individual wells. In addition, these pipets operate using positive displacement, handling viscous solutions better than traditional pipets. It is very important that the preparation procedure for the well solutions be reproducible, because if crystals grow, more wells of the same condition will need to be prepared.

2. Mix the well solutions on an orbital platform shaker for at least 1 hr.


Shake the well solutions as vigorously as possible without spillage. It is important that the well solutions are thoroughly mixed for reproducible results. This can be difficult for highly concentrated PEG solutions. Check for proper mixing by stirring a well containing the highest precipitant concentration. If no Schlieren effects are observed, the plates are ready. The plates can be wrapped in plastic wrap and stored overnight at 4◦ C or used immediately. If a suitable orbital shaker is not available, each well can be mixed by stirring with a pipet tip.



3. Set up the 22-mm siliconized square cover slides on a clean, lint-free surface. Place 1.0 µl of protein on a siliconized cover slide, then add 1.0 µl of well solution to each protein drop, pipetting up and down four times for each drop to ensure proper mixing. Avoid introducing air bubbles and try to make the drops as hemispherical as possible. Change tips between each well. Place the cover slide over the appropriate well and press down to seal. If only one drop per slide is planned, six slides (one row of a VDX plate) can be prepared at a time. Care should be taken to work as quickly as possible to avoid evaporation. The detergent (some more than others) will cause the drop to spread on the slide. As a result, cover slides should always be siliconized to keep the surface area of the drop minimized. In addition, it is not uncommon to find batches of poorly siliconized cover slides as evidenced by drop spreading. These slides should not be used for membrane proteins. Place the slides over the appropriate wells and gently push them down to seal. Care should be taken not to get grease near the center of the slide as this will complicate visualization of the drop.

4. Once the plate is complete, incubate it at a constant temperature in an environment as free of vibrations as possible.

REFINEMENT OF INITIAL CRYSTALLIZATION CONDITIONS IN 24-WELL FORMAT

BASIC PROTOCOL 3

This protocol describes a 24-well format to refine initial crystallization conditions from commercial screens. In this protocol, precipitant concentration and pH are optimized because these parameters greatly affect crystal formation and growth and are good starting points for refinement. However, there are many other parameters that affect crystal growth (Table 17.9.2) and these often need to be optimized as well. Conditions should be refined for each crystal type observed because the type that diffracts best will not be known until diffraction data are collected. Thus, during initial optimization, crystals are judged by their appearance. Once conditions that produce crystals are identified, the Hampton Additive Screen (Table 17.9.3) can be used to identify small molecules that may improve crystal growth. Finally, for the best conditions and additives, multiple drops are prepared and the plates are left undisturbed until the crystals have reached their optimum size. Table 17.9.2 Variables Affecting Membrane Protein Crystal Growtha

Protein primary sequence and conformational flexibility Protein purity, stability, and concentration Temperature and pH Precipitant type and concentration Detergents, additives, and lipids Specific ions and ionic strength Presence or absence of ligand Presence or absence of affinity tags a All variables may substantially affect crystallization, so equal consideration should be

given to each item.

Structural Biology


Supplement 47

Table 17.9.3 Additive Screen Reagent Formulation (Hampton Research)


1

0.1 M barium chloride

42

0.1 M trimethylamine HCl

2

0.1 M cadmium chloride

43

1 M guanidine HCl

3

0.1 M calcium chloride

44

0.1 M urea

4

0.1 M cobaltous chloride

45

0.1 M nicotinamide adenine dinucleotide

5

0.1 M cupric chloride

46

0.1 M adenosine-5-triphosphate disodium salt

6

0.1 M magnesium chloride

47

0.1 M TCEP hydrochloride

7

0.1 M manganese (II) chloride

48

0.1 M L-cysteine

8

0.1 M strontium chloride

49

0.1 M EDTA sodium salt

9

0.1 M yttrium chloride

50

5% (w/v) polyvinylpyrrolidone K15

10

0.1 M zinc chloride

51

30% (w/v) dextran sulfate

11

0.1 M ferric (III) chloride

52

40% (v/v) pentaerythritol ethoxylate

12.

0.1 M nickel (II) chloride

53

10% (w/v) polyethylene glycol 3350

13

0.1 M chromium (III) chloride

54

30% (w/v) D(+)-glucose

14

0.1 M praseodymium (III) acetate

55

30% (w/v) D (+)-sucrose

15

1 M ammonium sulfate

56

30% (w/v) xylitol

16

1 M potassium chloride

57

30% (w/v) sorbitol

17

1 M lithium chloride

58

12% (w/v) inositol

18

2 M sodium chloride

59

30% (w/v) D-trehalose

19

0.5 M sodium fluoride

60.

30% (w/v) D-galactose

20

1 M sodium iodide

61

30% (v/v) ethylene glycol

21

2 M sodium thiocyanate

62

30% (v/v) glycerol

22

1 M potassium sodium tartrate

63

3 M NDSB-195

23

1 M tri-sodium citrate

64

2 M NDSB-201

24

1 M cesium chloride

65

2 M NDSB-211

25

1 M sodium malonate pH 7.0

66

2 M NDSB-221

26

0.1 M L-proline

67

2 M NDSB-256

27

0.1 M phenol

68

15% (w/v) 1,2,3-heptanetriol

28

30% (v/v) dimethyl sulfoxide

69

20% (w/v) benzamidine HCl

29

0.1 M sodium bromide

70

5% (w/v) LDAO

30

30% (w/v) 6-aminocaproic acid

71

5% (w/v) n-octyl-b-D-glucoside

31

30% (w/v) 1,5-diaminopentane di-HCl

72

5% (w/v) n-dodecyl-b-D-maltoside

32

30% (w/v) 1,6-diaminohexane

73

30% (w/v) trimethylamine N-oxide

33

30% (w/v) 1,8-diaminooctane

74

30% (w/v) 1,6 hexanediol

34

1 M glycine

75

30% (v/v) 2-methyl-2,4-pentanediol

35

0.3 M glycyl-glycyl-glycine

76

50% (w/v) polyethylene glycol 400

36.

0.1 M taurine

77

50% (v/v) jeffamine M-600 pH 7.0

37

0.1 M betaine hydrochloride

78

40% (v/v) 2,5 hexanediol

38

0.1 M spermidine

79

40% (v/v) (+)-1,3 butanediol

39

0.1 M spermine tetra-HCl

80

40% (v/v) polypropylene glycol P 400

40

0.1 M hexamine cobalt (III) chloride

81

30% (v/v) dioxane

41

0.1 M sarcosine

82

30% (v/v) ethanol

17.9.6

continued

Supplement 47


Table 17.9.3 Additive Screen Reagent Formulation (Hampton Research), continued

83

30% (v/v) isopropanol

90

40% (v/v) n-propanol

84

30% (v/v) methanol

91

5% (v/v) ethyl acetate

85

40% (v/v) 1,4 butanediol

92

40% (v/v) acetone

86

40% (v/v) tert-butanol

93

0.25% (v/v) dichloromethane

87

40% (v/v) 1,3 propanediol

94

7% (v/v) n-butanol

88

40% (v/v) acetonitrile

95

40% (v/v) 2,2,2-trifluoroethanol

89

40% (v/v) g-butyrolactone

96

40% (v/v) hexafluoro-2-propanol

Materials Commercial screens with crystal formation (see Basic Protocol 1 or Alternate Protocol) 10× salt stock solutions 10× buffer stock solution Precipitant stock solutions (concentrations will vary) 24-well plates (e.g., VDX plates with sealant, Hampton Research) Light microscope (4× to 20× magnification) Hampton Additive Screen (optional) 22-mm siliconized square cover slides (Hampton Research) Constant temperature incubators (e.g., Rumed, Molecular Dimensions) 1. From the results obtained using Basic Protocol 1 or Alternate Protocol, select the conditions that produced crystals or crystalline precipitate as a guide for setting up the 24-well plates. If no promising conditions were observed, it may be necessary to modify the protein through mutagenesis or to try homologs. If small needles or plates were observed, switching the detergent may produce high quality three-dimensional crystals. For one outer membrane protein, the authors observed needles in N,N-dimethyldodecylamine-N-oxide, thin plates in octyl tetraoxyethylene, and three-dimensional crystals in octyl-β-D-glucoside. Thus, detergent screening is a good approach to try early in the process, especially if initial optimization attempts only yield needles or thin plates.

2. From the chosen conditions, set up plates that screen a range of precipitant concentration and pH. Observe each plate directly after it is set up to ensure that all the drops are not precipitated. If all are precipitated, set the plate up again using a lower precipitant concentration range. The goal is to have clear drops at the lowest precipitant concentration and fully precipitated drops at the highest concentration, once the plate has reached equilibrium. For example, in a commercial screen, small crystals were observed in 30% PEG 1000, 200 mM NaCl, and 0.1 M Bis-Tris, pH 6.5. Using these conditions as a guide, the plates shown in Figure 17.9.1A were set up with the same protein concentration and incubated at the same temperature as the commercial screen. For the conditions at pH 6.5, two drops were set up per slide: control drops containing buffer (including detergent and any additives) without protein and drops containing buffer with protein. If no crystals appear in the control drops, this is a good indication, but not absolute proof, that the crystals obtained in sample drops are protein. In this example, no cryoprotectant is present (PEG 1000 is not a good cryoprotectant, whereas PEG 400, glycerol, and some other reagents offer significant cryoprotection; Garman and Doublie, 2003). Thus, if there is sufficient protein, the plates in Figure 17.9.1A

Structural Biology


Supplement 47

Figure 17.9.1 Optimization of pH and precipitant concentration. (A) Initial crystal optimization using three 24-well plates, with the top plate containing pH 3.5 to 5.0, the middle plate containing pH 5.5 to 7.0, and the bottom plate containing pH 7.5 to 9.0. Each plate uses the same six precipitant concentrations. The best condition in this example is marked ‘X’. (B) Two 24-well plates are designed using the best conditions found in A.

would be set up in duplicate, adding glycerol to all of the well solutions of one set to a final concentration of 10% (v/v). If the presence of glycerol did not adversely affect crystal formation, it would be included in all well solutions throughout the optimization process and, in subsequent purifications, the buffer used for size-exclusion chromatography in Support Protocol 1. Including glycerol in the size-exclusion buffer simplifies setting up plates that have equal concentrations of glycerol in the drop and the well. This can be important because glycerol gradients can significantly affect the rate of equilibration. Crystallization of Integral Membrane Proteins

Sometimes, crystals obtained from commercial screens are not reproducible in 24-well plates. This is likely due to different physical and chemical conditions between the 96-well and 24-well formats. Samples in 96-well plates equilibrate much more rapidly than in 24-well plates, due to the smaller volumes used.



3. Score plates from step 2 and refine the best conditions. Examine the plates under a microscope every 24 hr for the first few days after they are set up and then every other week for 2 to 3 months. Scoring plates is described on the Hampton Research Web site (http://www.hamptonresearch.com/) and by McPherson (1999). Briefly, indicate whether a drop is clear, contains some sort of precipitate, or contains crystals. If a precipitate is discrete and granular, it should be examined under high magnification for the presence of microcrystals. For drops containing crystals, indicate their general shapes and sizes. Often, different crystal morphologies are seen at different pH values. Optimize conditions for each crystal type because the best looking crystals do not always diffract strongly or to high resolution.

4. Further refine precipitant concentration and pH (or other variables) to optimize crystal growth. In the example illustrated in Figure 17.9.1, only one crystal type was found for the plates set up in Figure 17.9.1A and the best condition contained 28% PEG 1000 and 10% glycerol at pH 5.0. Figure 17.9.1B shows how these conditions would be further refined by more finely sampling precipitant and pH conditions.

5. (Optional) Select a single (reproducible) crystallization condition and multiply the precipitant and salt concentrations by 1.1 (to account for subsequent dilution with the additive). Set up four 24-well plates with this modified condition using the 96-condition Hampton Research Additive Screen according to manufacturer’s instructions. Pipet 0.9 µl protein onto a siliconized cover slide, add 0.2 µl additive, then add 0.9 µl well solution. Repeat the Additive Screen plates for other optimized crystallization conditions as needed. This screen contains a variety of small molecules that can affect membrane protein crystallization (Table 17.9.3). Additives are often used in combination with one or more detergents to optimize the size of the detergent micelle for crystallization (Table 17.9.4). If well-formed, single crystals are obtained using the additive screen, re-optimize, beginning at step 2 and include the additive(s) for all conditions.

6. Score the optimized plates from steps 4 and 5. Set up the best conditions multiple times to maximize the number of large, single crystals. For conditions that give large, well-formed crystals, set up at least four wells (one column of a VDX plate) with three drops per cover slide. Not all of the drops will behave identically due to random pipetting errors, but a large number of crystals suitable for X-ray diffraction should grow. In addition, leave these optimized plates undisturbed in the incubator for the amount of time it takes for crystals to grow to their optimum size. This time frame can be approximated from the behavior of previous plates. At this stage, excessive handling of the plates should be avoided because sudden temperature changes, shocks, or vibrations can induce nucleation, giving rise to showers of small crystals. The goal now is to limit nucleation and maximize crystal growth. Finally, in this example, the path to macroscopic membrane protein crystals required only optimizing the precipitant concentration and the pH. While these are important parameters, there are many more conditions that can and usually must be optimized (Table 17.9.2). See McPherson (1991), Michel (1991), McPherson (1999), and Iwata (2003) for a more comprehensive list of parameters that affect both soluble and membrane protein crystal growth, as well as a more complete discussion of crystal growth refinement.

DETERGENT EXCHANGE BY SIZE-EXCLUSION CHROMATOGRAPHY The starting membrane protein must be purified in a detergent in which it is stable and active, such as dodecyl maltoside (DDM). If possible, the protein should show no contaminants as judged by SDS-PAGE (UNIT 10.1). Because affinity tags used for purification could either help or hinder crystallization, they can be removed prior to detergent exchange if desired. The starting protein concentration is not important as long as the

SUPPORT PROTOCOL 1

Structural Biology


Supplement 47

total volume contains ≥1 mg of protein, but the sample must be concentrated to ≤5 ml for the chromatography column described in this protocol. The sample should be eluted from the size-exclusion column at a low flow rate for optimal detergent exchange to occur. An analysis of the peak shape gives an indication of the homogeneity of the sample, as well as separating different oligomeric states and removing large aggregates. A homogeneous sample will exhibit a symmetrical, Gaussian peak shape, whereas an unhomogeneous sample may display tailing on the leading or trailing side of the peak (making the peak asymmetric) or multiple peaks, reflecting the presence of higher molecular weight aggregates or multiple oligomeric states, such as monomers, dimers, and tetramers.

Materials Column buffer (see recipe) Purified membrane protein sample (≥1 mg) Sephacryl S300 HR 16/60 size exclusion column (GE Healthcare) ¨ Akta prime chromatography system (GE Healthcare) or equivalent 15-ml Centriprep concentrators (Millipore) 0.22-µm syringe filter 1. Attach a Sephacryl S300 HR 16/60 size exclusion column to the chromatography system according to the manufacturer’s instructions. Equilibrate the column at a flow rate of 0.5 ml/min with 240 ml (two column volumes) of column buffer. 2. Place up to 15 ml purified membrane protein in a 15-ml Centriprep concentrator, choosing the molecular weight cut-off of the ultrafiltration membrane according to the size of the protein. Centrifuge at 4◦ C according to manufacturer’s instructions. Refill the concentrator if necessary and repeat until the sample volume measures 3 to 5 ml. Filter the concentrated sample using a 0.22-µm syringe filter. It may be necessary to use a smaller molecular weight cut-off Centriprep membrane to avoid loss of sample. Additional information on choosing ultrafiltration devices of the appropriate molecular weight cut-off can be found in Critical Parameters and Troubleshooting.

3. Apply the membrane protein sample to the equilibrated Sephacryl column and elute the sample at a rate of 0.3 ml/min. If the sample elutes as a single, Gaussian peak, determine the protein concentration of the pooled peak fractions (UNITS 3.1 & 3.4). If the peak shape is not Gaussian, a decision must be made to either take fractions from the major species (avoiding leading or trailing peak shoulders) or to discard the sample entirely.

4. Repeat steps 1 to 3 for new protein samples in each detergent that may be useful for crystallization. Any of the detergents in Table 17.9.4 (and possibly other homogenous, monodisperse detergents) yielding a Gaussian peak shape for the protein of interest is a candidate for crystallization (Lemieux et al., 2003).


While dynamic light scattering (UNIT 7.8) is sometimes used to assess the monodispersity of soluble protein samples, this technique is difficult to use for membrane proteins because the detergent micelle (in which the membrane protein is incorporated) contributes to light scattering and may itself be polydisperse. For this reason, the authors assess approximate monodispersity by size exclusion peak shape. Analytical ultracentrifugation (UNIT 7.5) can also be used in some cases, if care is taken to match the buoyant density of the detergent (Lebowitz et al., 2002).



PREPARATION OF MEMBRANE PROTEIN SAMPLE FOR CRYSTALLIZATION

SUPPORT PROTOCOL 2

Once a purified protein solution containing the desired detergent has been obtained, further preparation of the sample is necessary for crystallization trials. The sample needs to be concentrated to a concentration suitable for crystal trials and then filtered to remove particulate matter and aggregates. After concentration and filtration, the sample can be stored prior to setting up crystallization trials (see Basic Protocols 1 through 3).

Materials Detergent-exchanged membrane protein sample (see Support Protocol 1) 15-ml Centriprep concentrators (Millipore) 500-µl Microcon concentrators (Millipore) 500-µl Ultrafree 0.22-µm filtration devices (Millipore) 1. Concentrate the membrane protein sample to 10 mg/ml using the largest molecular weight cut-off membrane possible without losing sample. First concentrate the sample in a 15-ml Centriprep concentrator according to manufacturer’s instructions, and then transfer the sample to 500-µl Microcon concentrator(s) for further concentration, if necessary. It is important to use the highest molecular weight cut-off membranes possible to avoid concentrating the detergent more than necessary. Detergents with low cmc values, such as DDM, have large micellar molecular weights, often exceeding the molecular weight of the protein of interest. Ideally, the final concentrated sample will have a detergent concentration close to that of the size-exclusion buffer.

2. If one or more additives are being used in combination with the detergent chosen for crystallization, these should be added to the concentrated sample, taking care to mix gently to prevent the formation of excess air bubbles. Incubate the sample for 1 hr on ice. A list of additives successfully used to crystallize membrane proteins is found in Table 17.9.4. Screening for useful additives is described in Basic Protocol 3.

3. Transfer the sample to one or more Ultrafree filtration devices and centrifuge according to the manufacturer’s instructions. This step removes aggregates that might inhibit crystallization. An alternative method for removing aggregated protein is to spin the sample in an ultracentrifuge for 1 hr at ≥100,000 × g to pellet aggregated protein.

4. Store the concentrated, filtered sample on ice or at 4◦ C. The amount of time a concentrated membrane protein sample can be stored depends on the identity of the membrane protein, and must be determined empirically. Some membrane proteins lose activity within hours, while others are stable for several weeks. Do not freeze the sample, because many membrane proteins may not crystallize if they have been frozen, even if they retain full activity. Once the protein has been exchanged into the detergent of choice and concentrated, a broad screen of crystallization conditions is recommended. This is most easily done using a nanoliter pipetting robot (see Basic Protocol 1). A survey of the crystallization conditions that have led to published structures of membrane proteins (see http://www.mpibpfrankfurt.mpg.de/michel/public/memprotstruct.html) indicates that most have included polyethylene glycol (PEG) as the precipitant. Table 17.9.4 lists precipitants that have successfully yielded membrane protein crystals. Some recommended crystallization screens for membrane proteins are listed in Table 17.9.1. These screens have been chosen because Structural Biology


Supplement 47

they sample a wide range of conditions while retaining enough overlap to indicate the consistency of the results. The choice of which of these screens to use depends simply upon the volume of concentrated protein available; if enough protein is available, all 17 screens can be set up for each membrane protein sample. For a robot screen, a suggested drop volume of 0.2 µl protein solution plus 0.2 µl well solution minimizes losses due to evaporation (seen in smaller drops) but also conserves the protein sample, allowing a large number of experiments to be carried out for a given amount of concentrated membrane protein. Using the method described in Basic Protocol 1 and allowing for sample losses, each 96-condition screen requires 30 µl of a 10 mg/ml sample. The procedure for screening using a crystallization robot is given in Basic Protocol 1. Table 17.9.4 Detergents, Additives, and Precipitants that Have Been Used to Solve Crystal Structures of Outer and Inner Membrane Proteinsa

Outer membrane proteins

Inner membrane proteins

Detergents Octyl tetraoxyethylene

Octyl-β-D-glucoside

Octyl-β-D-glucoside

Dodecyl-β-D-maltoside

N,N-Dimethyldodecylamine-N-oxide

N,N-Dimethyldodecylamine-N-oxide

N,N-Dimethyldecylamine-N-oxide

Nonyl-β-D-glucoside

Decyl pentaoxyethylene

Undecyl-β-D-maltoside

Octyl-2-hydroxy ethylsulfoxide

Decyl-β-D-maltoside

Octyl polyoxyethylene

Dodecyl nonaoxyethylene

Octyl pentaoxyethylene

Dodecyl octaoxyethylene

Decyl-β-D-maltoside

Octyl-β-D-maltoside

Cyclohexylpentyl-β-D-maltoside

Dodecyl-α-D-maltoside

Additives Heptane-1,2,3-triol

Heptane-1,2,3-triol

N,N-Dimethylhexylamine-N-oxide

Benzamidine

Heptyl-β-D-glucoside

CHAPS

2-Propanol

HECAMEG

Dioxane

Dodecyl octaoxyethylene

Inositol

Dioxane Ethanol 2-Propanol 1,6-Hexanediol Dioleylphosphatidylcholine

Precipitants


Polyethylene glycol 2000


2-Methyl-2,4-pentandiol


Polyethylene glycol 2000 MME



Various phosphates




Ammonium sulfate

Ammonium sulfate


a In each category, items are listed from most to least successful (adapted from http://www.mpibp-frankfurt.mpg.de/

michel/public/memprotstruct.html).



REAGENTS AND SOLUTIONS Use Milli-Q-purified water or equivalent for the preparation of all buffers. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Column buffer 20 mM Tris·Cl, pH 7.5 150 mM NaCl 0.5 mM EDTA 0.02% (w/v) NaN3 Detergent of choice at a concentration two times its cmc (le Maire et al., 2000) Prepare fresh A list of detergents successfully used to crystallize membrane proteins is found in Table 17.9.4. Properties of detergents, such as cmc, aggregation number, formula weight, and others, can be found in the Anatrace catalogue (http://www.anatrace.com). General properties of detergents can be found in Zulauf (1991).

COMMENTARY Background Information X-ray crystallography has been used to determine the structures of a large number of proteins. As of 2005, there were >32,000 structures deposited in the Protein Data Bank (http://www.rcsb.org/pdb/), yet only 0.5% of these represent integral membrane proteins. Membrane protein structures are underrepresented for several reasons. Many membrane proteins, both prokaryotic and eukaryotic, are difficult to express in functional form in sufficient quantities for structure determination (Grisshammer and Tate, 1995). Extraction of membrane proteins from their native lipid environment can lead to inactivation, as can subsequent purification using detergents. Crystallization is accomplished in the presence of detergents and/or lipids, adding more parameters that must be screened to obtain well-ordered 3-D crystals. Despite these difficulties, the number of membrane protein structures solved each year is increasing exponentially (White, 2004) due to the availability of new detergents, as well as advances in small-scale crystallization equipment, X-ray data collection at synchrotrons, and software used to more rapidly compute new structures.

Critical Parameters and Troubleshooting The most critical parameter for obtaining well-ordered membrane protein crystals is a natively folded, fully functional membrane protein sample of high purity. Care must be taken to analyze the activity of the sample in the desired detergent(s). Stability, defined as the length of time that a membrane protein retains full activity in a detergent environment, is also crucial to the crystallization process.

While determination of residual lipid and/or number of detergent molecules bound would be useful for establishing uniformity of protein preparations, these parameters are difficult to assay in a standard biochemistry laboratory. As with soluble proteins, compact, homogeneous molecules are more likely to crystallize than flexible molecules of limited purity. In addition, some ectodomains of membrane proteins remain folded and soluble without detergent when the transmembrane sequences are removed. These domains may be regarded as soluble proteins and crystallized as described in Chapter 17. Care should be taken when choosing molecular weight sizes for ultrafiltration devices. When a protein must be concentrated in earlier stages of the purification protocol, the filtration membrane pore size should be selected that minimizes protein losses. Membrane proteins are often observed to require smaller molecular weight cut-off membranes than their actual masses would predict, and we try to avoid sample losses during concentration wherever possible. The final detergent concentration is not critical at these stages. However, detergent micelles can have very large molecular weights, and many detergents are concentrated along with the protein sample when using molecular weight cut-off membranes of 10,000, 30,000, and sometimes even 50,000. This means that a ten-fold concentration of protein may also concentrate the detergent tenfold. Increased detergent concentrations can negatively influence crystallization, resulting in poorly ordered crystals, or no crystals at all. Therefore, once detergent exchange has been accomplished as described in Basic Protocol 1, the largest cut-off membranes possible are

Structural Biology


Supplement 47

used to maximize protein retention while minimizing concentration of detergent. Ideally, the final detergent concentration of the concentrated protein sample will be approximately twice the critical micelle concentration of the detergent. One parameter that has a greater influence on membrane protein crystals than soluble protein crystals is temperature. Since membrane protein crystals have large amounts of detergent in the crystal lattice, small fluctuations in effective detergent concentration can easily destroy growing crystals. These fluctuations occur when the temperature increases or decreases, due to the dependence of the cmc of the detergent on temperature. It, therefore, makes sense to buy the best quality crystallization incubators available, with the most constant temperature control. One final difference between crystallizing membrane and soluble proteins lies in the strategies for improvement of crystal quality. For soluble proteins, if no crystals are obtained from commercial screens, one usually modifies the construct or tries a homolog. Therefore, the screening process for a given protein or protein fragment is rather brief. For membrane proteins, it is often not possible (or desirable) to remove part of the protein, and much more optimization is done using different detergents and additives for a single membrane protein construct. It is common to obtain membrane protein crystals that diffract to medium ◦ resolution (∼10 A), but refinement of conditions to higher resolution takes more time, more protein, sampling of more conditions, and more patience.

Anticipated Results


Crystallization trials generally show one (or more) types of behavior: clear drops, amorphous precipitate, granular precipitate, phase separation, microscopic crystals, or macroscopic crystals. Due to the increasing detergent concentration over time in a vapor-diffusion experiment (as described in this unit), phase separation is very common, and occasionally leads to crystal growth. Such a system is unstable, however, because areas of detergentrich and detergent-poor solution coexist in the same drop, and growing crystals can be destabilized by these differences. The best crystals are obtained in drops where no phase separation occurs, with slow growth of a few macroscopic crystals. For an example of how various parameters affect the growth of well-ordered crystals for a bacterial inner membrane pro-

tein, see Lemieux et al. (2003). Several reviews on the crystallization of integral membrane proteins are also available (Garavito et al., 1996; Loll, 2003; Wiener, 2004).

Time Considerations Once a purified membrane protein is available, it can usually be concentrated and screened with a robot in just a few days. Refinement of crystallization conditions may take weeks to months or even years, depending on the number of parameters that must be optimized. It is therefore important to produce enough protein routinely to set up numerous crystallization experiments so that traditional parameters such as precipitant and pH, as well as membrane-protein specific parameters such as detergent type and concentration, additives, and possibly lipids, can be evaluated for their influence on crystalline order. In the authors’ experience, the best ordered membrane protein crystals grow to completion in a time frame of 1 week to 1 month, with more rapidly growing crystals displaying poorer order.

Acknowledgement The work of T.J.B., J.L.W., and S.K.B. is supported by the Intramural Research Program of the NIH, National Institute of Diabetes and Digestive and Kidney Diseases.

Literature Cited Chang, G., Spencer, R.H., Lee, A.T., Barclay, M.T., and Rees, D.C. 1998. Structure of the MscL homolog from Mycobacterium tuberculosis: A gated mechanosensitive ion channel. Science 282:2220-2226. Dutzler, R., Campbell, E.B., and MacKinnon, R. 2003. Gating the selectivity filter in ClC chloride channels. Science 300:108-112. Garavito, R.M., Picot, D., and Loll, P.J. 1996. Strategies for crystallizing membrane proteins. J. Bioenerg. Biomembr. 28:13-27. Garman, E.F. and Doublie, S. 2003. Cryocooling of macromolecular crystals: Optimization methods. Methods Enzymol. 368:188-216. Grisshammer, R. and Tate, C.G. 1995. Overexpression of integral membrane proteins for structural studies. Q. Rev. Biophys. 28:315-422. Iwata, S. ed. 2003. Methods and Results in Crystallization of Membrane Proteins. La Jolla, International University Line. le Maire, M., Champeil, P., and Møller, J.V. 2000. Interaction of membrane proteins and lipids with solubilizing detergents. Biochim. Biophys. Acta 1508:86-111. Lebowitz, J., Lewis, M.S., and Schuck, P. 2002. Modern analytical ultracentrifugation in protein science: A tutorial review. Protein Sci. 11:20672079.



Lemieux, M.J., Song, J., Kim, M.J., Huang, Y., Villa, A., Auer, M., Li, X.D., and Wang, D.N. 2003. Three-dimensional crystallization of the Escherichia coli glycerol-3-phosphate transporter: A member of the major facilitator superfamily. Protein Sci. 12:2748-2756.

Internet Resources

Locher, K.P., Lee, A.T., and Rees, D.C. 2002. The E. coli BtuCD structure: A framework for ABC transporter architecture and mechanism. Science 296:1091-1098.

http://www.mpibp-frankfurt.mpg.de/michel/ public/memprotstruct.html Hartmut Michel’s listing of crystallized membrane proteins, including a statistical analysis of crystallization conditions and structure references.

Loll, P.J. 2003. Membrane protein structural biology: The high throughput challenge. J. Struct. Biol. 142:144-153. McPherson, A. 1991. Useful principles for the crystallization of proteins. In Crystallization of Membrane Proteins. (H. Michel, ed.) pp. 1-51. CRC Press, Boca Raton, Fla. McPherson, A. 1999. Crystallization of Biological Macromolecules. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. Michel, H. 1991. General and practical aspects of membrane protein crystallization. In Crystallization of Membrane Proteins. (H. Michel, ed.) pp. 73-88. CRC Press, Boca Raton, Fla. White, S.H. 2004. The progress of membrane protein structure determination. Protein Sci. 13:1948-1949. Wiener, M.C. 2004. A pedestrian guide to membrane protein crystallization. Methods 34:364372. Yue, W.W., Grizot, S., and Buchanan, S.K. 2003. Structural evidence for iron-free citrate and ferric citrate binding to the TonB-dependent outer membrane transporter FecA. J. Mol. Biol. 332:353-368.

http://www.rcsb.org/pdb Protein data base—the site where all protein structures (determined by X-ray crystallography, electron microscopy, and NMR) are deposited and made publicly available.

http://blanco.biomol.uci.edu/ Membrane Proteins xtal.html Stephen White’s listing of all solved membrane protein structures. This list is continuously updated and contains references to the individual structures. http://www.emeraldbiosystems.com/OnlineStore/ home.php http://www.hamptonresearch.com/ http://www.moleculardimensions.com/us/ index.ihtml http://www.nextalbiotech.com/ Web sites for crystallization instrumentation and reagents. http://www.anatrace.com/default.htm The Anatrace Web site for the widest variety of high purity detergents for membrane protein purification and crystallization.

Contributed by Travis J. Barnard, Jeremy L. Wally, and Susan K. Buchanan National Institutes of Health Bethesda, Maryland

Zulauf, M. 1991. Detergent phenomena in membrane protein crystallization. In Crystallization of Membrane Proteins. (H. Michel, ed.) pp. 5372. CRC Press, Boca Raton, Fla.

Structural Biology


Supplement 47

CHAPTER 18 Preparation and Handling of Peptides INTRODUCTION

A

n important component of the armamentarium of modern protein scientists is the ability to design and construct peptides of almost any sequence. Peptides may be employed for a wide range of applications, including as substrates for kinases, proteases, or glycosidases, or as antigens to produce antisera that may recognize a protein containing that sequence. Peptides also have many biological functions on their own, interacting with receptors to stimulate changes in cellular function, or altering the interactions between cells, for example. Chemists developed the means to synthesize peptides early in this century (UNIT 18.1). The ability to create a molecule that possessed biological activity by organic synthesis is certainly a milestone in our science. For many years, however, peptide synthesis was practiced only by highly skilled and dedicated individuals, and the lengthy time required to obtain even simple peptides limited their use as reagents for biological experiments. Difficulties in purification effectively eliminated the study of complex peptides of more than 5 to 10 amino acids.

Several developments have revolutionized this field, fortunately, and they have brought peptide science into the realm of standard biological techniques available to many scientists. First was the development of solid-phase methods by Bruce Merrifield and subsequent improvements in the reagents available for synthesis that are detailed in UNIT 18.1. Second was the design and construction of automated instruments to perform the repetitive steps of the solid-phase methodology, culminating in the 1980s with instruments that provided nearly global accessibility to synthetic peptides of ever-increasing length. Another advance was the optimization of reversed-phase HPLC as a separation method for the rapid purification and analysis of synthetic peptides (UNIT 11.6). A future unit in this chapter will focus on medium-to-large scale peptide purification by HPLC methods. In addition, recent advances in mass spectroscopy (Chapter 16) have provided another critical analytical tool to verify the products of synthesis. A convenient, low-cost method for creating large numbers of related peptides or for assembly of a set of peptides that together span an entire protein to search for the epitope recognized by an antibody that binds to the full-length protein is provided by the synthesis of peptides on a multipin apparatus (UNIT 18.2). Frequently, genomic studies lead to the discovery of an open reading frame (ORF), predicting a protein sequence that might have an activity of interest. One step that could follow would be the generation of an antibody to use as a reagent to identify the protein product of the gene in a sample isolated from the organism under study. A small peptide of 10 to 15 amino acids can frequently be used as an antigen to stimulate antiserum production. This requires picking the best sequence out of an ORF of hundreds of amino acids and preparing the conjugate for injection into an animal for generation of the antiserum (UNIT 18.3). As the procedures employed in solid-phase peptide synthesis have improved, researchers’ desire to attempt the preparation of longer sequences has grown. Methods to ligate Contributed by Ben M. Dunn Current Protocols in Protein Science (2001) 18.0.1-18.0.2 Copyright © 2001 by John Wiley & Sons, Inc.

Preparation and Handling of Peptides


medium- to large-sized peptides together through defined chemistry are presented in UNIT 18.4. Success in this approach has opened the door to a variety of new synthetic objectives.

The chemistry and strategy utilized to prepare peptide dendrimers (UNIT 18.5) for use as immunogens, designed protein mimics, new reagents for drug discovery, or new biomaterials has been an outgrowth of solid-phase peptide synthesis. Methods for both Boc and Fmoc syntheses are described here. In addition, several procedures for the formation of cyclic peptides are presented, including derivatization of Lys and Cys residues. Preparing bioactive peptides or peptides with a fixed conformation often involves creating disulfide bridges (UNIT 18.6). The strategy and reactions necessary to achieve selective linkages in multi-Cys-containing peptides is described here; the analytical techniques required to characterize the peptides at each stage of the process are highlighted as well. Units planned for future expansion of this chapter will cover synthesis of derivatives such as phosphopeptides or other modification that can be done at the resin-bound stage, the creation of derivatives in which selective peptide bonds have been replaced to create inhibitors or to stabilize against degradation in biological systems, methods for storage and handling of peptides, specialized methods for different types of resins for solid-phase synthesis, the creation of peptide libraries by chemical or biological approaches, the chemistry necessary to achieve the coupling of difficult or hindered peptides, and methods for the assembly of peptides of 100 amino acid residues in a single, long synthesis. Ben M. Dunn

Introduction



Introduction to Peptide Synthesis DEVELOPMENT OF SOLIDPHASE PEPTIDE-SYNTHESIS METHODOLOGY A number of synthetic peptides are significant commercial or pharmaceutical products, ranging from the dipeptide sugar substitute aspartame to clinically used hormones such as oxytocin, adrenocorticotropic hormone, and calcitonin. Rapid, efficient, and reliable methodology for the chemical synthesis of these molecules is of utmost interest. The stepwise assembly of peptides from amino acid precursors has been described for nearly a century. The concept is a straightforward one, whereby peptide elongation proceeds via a coupling reaction between amino acids, followed by removal of a reversible protecting group. The first peptide synthesis, as well as the creation of the term “peptide,” was reported by Fischer and Fourneau (1901). Bergmann and Zervas (1932) created the first reversible Nα-protecting group for peptide synthesis, the carbobenzoxy (Cbz) group. DuVigneaud successfully applied early “classical” strategies to construct a peptide with oxytocin-like activity (duVigneaud et al., 1953). Classical, or solution-phase methods for peptide synthesis have an elegant history and have been well chronicled. Solution synthesis continues to be especially valuable for largescale manufacturing and for specialized laboratory applications. Peptide synthesis became a more practical part of present-day scientific research following the advent of solid-phase techniques. The concept of solid-phase peptide synthesis (SPPS) is to retain chemistry that has been proven in solution but to add a covalent attachment step that links the nascent peptide chain to an insoluble polymeric support (resin). Subsequently, the anchored peptide is extended by a series of addition cycles (Fig. 18.1.1). It is the essence of the solid-phase approach that reactions are driven to completion by the use of excess soluble reagents, which can be removed by simple filtration and washing without manipulative losses. Once chain elongation has been completed, the crude peptide is released from the support. In the early 1960s, Merrifield proposed the use of a polystyrene-based solid support for peptide synthesis. Peptides could be assembled stepwise from the C to N terminus using Nαprotected amino acids. SPPS of a tetrapeptide

Contributed by Gregg B. Fields Current Protocols in Protein Science (2001) 18.1.1-18.1.9 Copyright © 2001 by John Wiley & Sons, Inc.

was achieved by using Cbz as an α-amino-protecting group, coupling with N,N′-dicyclohexylcarbodiimide (DCC), and liberating the peptide from the support by saponification or by use of HBr (Merrifield, 1963). SPPS was later modified to use the t-butyloxycarbonyl (Boc) group for Nα protection (Merrifield, 1967) and hydrogen fluoride (HF) as the reagent for removal of the peptide from the resin (Sakakibara et al., 1967). SPPS was thus based on “relative acidolysis,” where the Nα-protecting group (Boc) was labile in the presence of moderate acid (trifluoroacetic acid; TFA), while side-chain-protecting benzyl (Bzl)– based groups and the peptide/resin linkage were stable in the presence of moderate acid and labile in the presence of strong acid (HF). The first instrument for automated synthesis of peptides, based on Boc SPPS, was built by Merrifield, Stewart, and Jernberg (Merrifield et al., 1966). From the 1960s through the 1980s, Boc-based SPPS was fine-tuned (Merrifield, 1986). This strategy has been utilized for synthesis of proteins such as interleukin-3 and active enzymes including ribonuclease A and all-L and all-D forms of HIV-1 aspartyl protease. In 1972, Carpino introduced the 9-fluorenylmethoxycarbonyl (Fmoc) group for Nα protection (Carpino and Han, 1972). The Fmoc group requires moderate base for removal, and thus offered a chemically mild alternative to the acid-labile Boc group. In the late 1970s, the Fmoc group was adopted for solid-phase applications. Fmoc-based strategies utilized t-butyl (tBu)–based side-chain protection and hydroxymethylphenoxy-based linkers for peptide attachment to the resin. This was thus an “orthogonal” scheme requiring base for removal of the Nα-protecting group and acid for removal of the side-chain protecting groups and liberation of the peptide from the resin. The milder conditions of Fmoc chemistry as compared to Boc chemistry—which include elimination of repetitive moderate acidolysis steps and the final strong acidolysis step—were envisioned as being more compatible with the synthesis of peptides that are susceptible to acid-catalyzed side reactions. In particular, the modification of the indole ring of Trp was viewed as a particular problem during Boc-based peptide synthesis (Barany and Merrifield, 1979), which could be alleviated using Fmoc chemistry. One example of the potential advantage of Fmoc chemistry

UNIT 18.1



(King et al., 1990). Thus, the mild conditions of Fmoc chemistry appeared to be advantageous for certain peptides, as compared with Boc chemistry. One of the subsequent challenges for practitioners of Fmoc chemistry was to refine the technique to allow for construction of proteins, in similar fashion to that which had been achieved with Boc chemistry. Fmoc chemistry had its own set of unique problems, including suboptimum solvation of the peptide/resin,

for the synthesis of multiple-Trp-containing peptides was in the synthesis of gramicidin A. Gramicidin A, a pentadecapeptide containing four Trp residues, had been synthesized previously in low yields (5% to 24%) using Boc chemistry. The mild conditions of Fmoc chemistry dramatically improved the yields of gramicidin A, in some cases up to 87% (Fields et al., 1989, 1990). A second multiple-Trp-containing peptide, indolicidin, was successfully assembled in high yield by Fmoc chemistry

X

A

NH

O

CH C

+

OH

linker

resin

anchoring X

repetitive cycle

A

NH

O

CH C

linker

resin N α -deprotection

Y

NH

A

O

CH C

X

+

OH

H2 N

O

CH C

linker

resin

coupling

Y NH

A

O

CH C

X

NH

O

CH C

linker

resin α

(1) N -deprotection (2) cleavage (3) side-chain deprotection

Z H2 N

Introduction to Peptide Synthesis

O

CH C

R NH

Y

O

CH C

NH n

O

CH C

X

NH

O

CH C

NH2 OH

Figure 18.1.1 Generalized approach to solid-phase peptide synthesis. Symbols: A, Nα-protecting group; circle, side-chain protecting groups; R, X, Y, and Z, side-chain functionalities.



slow coupling kinetics, and base-catalyzed side reactions. Improvements in these areas of Fmoc chemistry (Atherton and Sheppard, 1987; Fields and Noble, 1990; Fields et al., 2001) allowed for the synthesis of proteins such as bovine pancreatic trypsin inhibitor analogs, ubiquitin, yeast actin-binding protein 539-588, human β-chorionic gonadotropin 1-74, minicollagens, HIV-1 Tat protein, HIV-1 nucleocapsid protein Ncp7, and active HIV-1 protease. The milder conditions of Fmoc chemistry, along with improvements in the basic chemistry, have led to a shift in the chemistry employed by peptide laboratories. This trend is best exemplified by a series of studies (Angeletti et al., 1997) carried out by the Peptide Synthesis Research Committee (PSRC) of the Association of Biomolecular Resource Facilities (ABRF). The PSRC was formed to evaluate the quality of the synthetic methods utilized in its member laboratories for peptide synthesis. The PSRC designed a series of studies from 1991 to 1996 to examine synthetic methods and analytical techniques. A strong shift in the chemistry utilized in core facilities was observed during this time period—i.e., the more senior Boc methodology was replaced by Fmoc chemistry. For example, in 1991 50% of the participating laboratories used Fmoc chemistry, while 50% used Boc-based methods. By 1994, 98% of participating laboratories were using Fmoc chemistry. This percentage remained constant in 1995 and 1996. In addition, the overall quality of the peptides synthesized improved greatly from 1991 to 1994. Possible reasons for the improved results were any combination of the following (Angeletti et al., 1997): 1. The greater percentage of peptides synthesized by Fmoc chemistry, where cleavage conditions are less harsh; 2. The use of different side-chain protecting group strategies that help reduce side reactions during cleavage; 3. The use of cleavage protocols designed to minimize side reactions; 4. More rigor and care in laboratory techniques. The present level of refinement of solidphase methodology has led to numerous, commercially available instruments for peptide synthesis (Table 18.1.1). The next step in the development of solidphase techniques includes applications for peptides containing non-native amino acids, posttranslationally modified amino acids, and pseudoamino acids, as well as for organic molecules in general. Several areas of solid-phase

synthesis need to be refined to allow for the successful construction of this next generation of biomolecules. The solid support must be versatile so that a great variety of solvents can be used, particularly for organic-molecule applications. Coupling reagents must be sufficiently rapid so that sterically hindered amino acids can be incorporated. Construction of peptides that contain amino acids bearing posttranslational modifications should take advantage of the solid-phase approach. Finally, appropriate analytical techniques are needed to assure the proper composition of products.

THE SOLID SUPPORT Effective solvation of the peptide/resin is perhaps the most crucial condition for efficient chain assembly during solid-phase synthesis. Swollen resin beads may be reacted and washed batch-wise with agitation, then filtered either with suction or under positive nitrogen pressure. Alternatively, they may be packed in columns and utilized in a continuous-flow mode by pumping reagents and solvents through the resin. 1H, 2H, 13C, and 19F nuclear magnetic resonance (NMR) experiments have shown that, under proper solvation conditions, the linear polystyrene chains of copoly(styrene1%-divinylbenzene) resin (PS) are nearly as accessible to reagents as if free in solution. 13C and 19F NMR studies of Pepsyn (copolymerized dimethylacrylamide, N,N′-bisacryloylethylenediamine, and acryloylsarcosine methyl ester) have shown similar mobilities at resin-reactive sites as PS. Additional supports created by grafting polyethylene glycol (polyoxyethylene) onto PS—either by controlled anionic polymerization of ethylene oxide on tetraethylene glycol–PS (POE-PS) or by coupling Nω-Boc– or Fmoc–polyethylene glycol acid or –polyethylene glycol diacid to amino-functionalized PS (PEG-PS)—combine the advantages of liquid-phase synthesis (i.e., a homogeneous reaction environment) and solid-phase synthesis (an insoluble support). 13C NMR measurements of POE-PS showed the polyoxyethylene chains to be more mobile than the PS matrix, with the highest T1 spin-lattice relaxation times observed with POE of molecular weight 2000 to 3000. Other supports that have been developed that show improved solvation properties and/or are applicable to organic synthesis include polyethylene glycol polyacrylamide (PEGA), cross-linked acrylate ethoxylate resin (CLEAR), and augmented surface polyethylene prepared by chemical transformation (ASPECT). As the solid-phase



Supplement 26

Table 18.1.1

Instruments for Solid-Phase Synthesis

Suppliera

Instrument model

Peptide synthesis systems Advanced ChemTech 90 Apex 396 348 357 Bachem Bioscience SP4000-LAB SP4000-PRO Gilson/Abimed AMS422 Perkin-Elmer ABI433A Pioneer-MPS Rainin PS3 Sonata/Pilot Symphony/Multiplex CS Bio CS100 CS336 036 136 CS536 CS936S CS936 Intavis AG AutoSpot Multiple organic synthesis units Advanced ChemTech 384 Vantage aFor

Fmoc Boc

Batch Flow Monitoring

Scale (mmol)

No. of peptides

Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes Yes No No Yes No Yes Yes No Yes Yes Yes Yes Yes Yes Yes No

Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

No No No No No No No No Yes No No No No No No No No No No No

No No No No No No Yes Yes Yes No No No No No No No No No No No

0.1-12 0.005-1 0.005-0.15 0.005-0.25 0.25-5 5-50 0.005-1 0.05-1 0.005-0.1 0.1-0.25 0.1-50 0.005-0.35 0.05-1 0.05-0.25 0.1-2.5 0.1-2.5 0.2-25 ≤600 ≤12,500 3-4 nmol/mm2

1-2 96 48 42 1 1 1 1 16 3 1 12 1 3 1 1 1 1 1 384

Yes Yes

Yes Yes

Yes Yes

No No

No No

— —

4 × 96 96

contact information, see SUPPLIERS APPENDIX.

method has expanded to include organic-molecule and library syntheses, the diversity of supports will enhance the efficiency of these new applications. Successful syntheses of problematic sequences can be achieved by manipulation of the solid support. In general, the longer the synthesis, the more polar the peptide/resin will become (Sarin et al., 1980). One can alter the solvent environment and enhance coupling efficiencies by adding polar solvents and/or chaotropic agents (Fields and Fields, 1994). Also, using a lower substitution level of resin to avoid interchain crowding can improve the synthesis (Tam and Lu, 1995). During difficult syntheses, deprotection of the Fmoc group can proceed slowly. By spectrophotometrically monitoring deprotection as the synthesis proceeds, one can detect problems and extend base-deprotection times and/or alter solvation conditions as necessary. Introduction to Peptide Synthesis

COUPLING REAGENTS The classical examples of in situ coupling reagents are N,N′-dicyclohexylcarbodiimide (DCC) and the related N,N′-diisopropylcarbodiimide (Rich and Singh, 1979). The generality of carbodiimide-mediated couplings is extended significantly by the use of either 1hydroxybenzotriazole (HOBt) or 1-hydroxy-7azabenzotriazole (HOAt) as an additive, either of which accelerates carbodiimide-mediated couplings, suppresses racemization, and inhibits dehydration of the carboxamide side chains of Asn and Gln to the corresponding nitriles. Protocols involving benzotriazol-1-ylo xy -tr is(d imethy lam ino )p hosphonium hexafluorophosphate (BOP), benzotriazol-1y l - o x y - t r i s ( p y r r o li d in o ) p h o s p h o n iu m hexafluorophosphate (PyBOP), 7 -azabenzotriazol-1-yl-oxytris(pyrrolidino)phosphoni um hexafluorophosphate (PyAOP), O-benzotriazol-1-yl-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU), O-(7-azaben-



Table 18.1.2

Coupling Reagents and Additives Used in Solid-Phase Peptide Synthesis and Suppliers

Reagent

Abbreviation

Supplier(s)a

N,N′-dicyclohexylcarbodiimide N,N′-diisopropylcarbodiimide

DCC DIPCDI

A, ACT, AO, CI, CN, F, PI, PL, Q, S A, ACT, AO, CI, F, PE, Q, S

O-benzotriazol-1-yl-N,N,N′,N′-tetramethyluronium hexafluorophosphate O-benzotriazol-1-yl-N,N,N′,N′-tetramethyluronium tetrafluoroborate O-(7-azabenzotriazol-1-yl)-N,N,N′,N′tetramethyluronium hexafluorophosphate

HBTU

A, ACT, AS, CI, CN, F, NS, PI, PL, Q, S

TBTU

A, ACT, B, CI, CN, F, NS, PE, PI, PL, Q, S

HATU

PE

Benzotriazol-1-yl-oxy-tris(dimethylamino) phosphonium hexafluorophosphate Benzotriazol-1-yl-oxy-tris(pyrrolidino)phosphonium hexafluorophosphate 7-azabenzotriazole-1-yl-oxy-tris(pyrrolidino)phophonium hexafluorophosphate

BOP

A, ACT, AO, B, CI, CN, F, NS, PL, PI, Q, S

PyBOP

A, ACT, AO, CI, CN, F, S

PyAOP

PE

Tetramethylfluoroformamidinium hexafluorophosphate 1-hydroxybenzotriazole

TFFH

ACT, PE

HOBt

A, ACT, AO, AS, CI, CN, NS, PE, PI, Q, S

1-hydroxy-7-azabenzotriazole N,N-diisopropylethylamine

HOAt DIEA

PE A, ACT, AO, CI, F, PE, Q, S

N-methylmorpholine

NMM

A, AO, CI, F, S

aAbbreviations: A, Aldrich; ACT, Advanced ChemTech; AO, Acros Organics; AS, AnaSpec; B, Bachem; CI, Chem-Impex; CN, Calbiochem-Novabiochem; F, Fluka; NS, Neosystem/SNPE; PE, Perkin-Elmer; PI, Peptides International; PL, Peninsula Laboratories; Q, Quantum Biotechnologies; S, Sigma. For contact information, see SUPPLIERS APPENDIX.

zotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HATU), and O-benzotriazol-1-yl-N,N,N′,N′-tetramethyluronium tetrafluoroborate (TBTU) result in coupling kinetics even more rapid than that obtained with carbodiimides. Amino acid halides have also been applied to solid-phase peptide synthesis (SPPS). Nα-protected amino acid chlorides have a long history of use in solution synthesis. Fmoc–amino acid chlorides and fluorides react rapidly under SPPS conditions in the presence of HOBt/N,N-diisopropylethylamine (DIEA) and DIEA, respectively, with very low levels of racemization. For convenience, tetramethylfluoroformamidinium hexafluorophosphate (TFFH) can be used for automated preparation of Fmoc–amino acid fluorides. Amino acid fluorides have been found to be especially useful for the preparation of peptides containing sterically hindered amino acids, such as peptaibols. All of the coupling reagents and additives discussed here are commercially available (see Table 18.1.2).

SYNTHESIS OF MODIFIED RESIDUES AND STRUCTURES Peptides of biological interest often include structural elements beyond the 20 genetically encoded amino acids. Particular emphasis has been placed on peptides containing phosphorylated or glycosylated residues or disulfide bridges. Incorporation of side-chain-phosphorylated Ser and Thr by solid-phase peptide synthesis (SPPS) is especially challenging, as the phosphate group is decomposed by strong acid and lost with base in a β-elimination process. Boc-Ser(PO3phenyl2) and Boc-Thr(PO3phenyl2) have been found to be useful derivatives, where hydrogen fluoride (HF) or hydrogenolysis cleaves the peptide/resin and hydrogenolysis removes the phenyl groups. Fmoc-Ser(PO3Bzl,H) and Fmoc-Thr(PO3Bzl,H) can be used in conjunction with Fmoc chemistry with some care. Alternatively, peptide/resins that were built up by Fmoc chemistry to include unprotected Ser or Thr side chains may be subject to “global” or post-assembly phosphorylation. Side-chain-phosphorylated Tyr is less susceptible to strong-acid decomposition and is not at all base-labile. Thus, SPPS



Supplement 26

Introduction to Peptide Synthesis

has been used to incorporate directly FmocTyr(PO3methyl2), Fmoc-Tyr(PO3tBu2), FmocTyr(PO3H2), and Boc-Tyr(PO3H2). Phosphorylation may also be accomplished on-line, directly after incorporation of the Tyr, Ser, or Thr residue but prior to assembly of the whole peptide. Methodology for site-specific incorporation of carbohydrates during chemical synthesis of peptides has developed rapidly. The mild conditions of Fmoc chemistry are more suited for glycopeptide syntheses than Boc chemistry, as repetitive acid treatments can be detrimental to sugar linkages. Fmoc-Ser, -Thr, -5-hydroxylysine (-Hyl), -4-hydroxyproline (-Hyp), and -Asn have all been incorporated successfully with glycosylated side chains. The side-chain glycosyl is usually hydroxyl-protected by either benzoyl or acetyl groups, although some SPPSs have been successful with no protection of glycosyl hydroxyl groups. Deacetylation and debenzylation are performed with hydrazine/methanol prior to glycopeptide/resin cleavage or in solution with catalytic methoxide in methanol. Disulfide-bond formation has been achieved on the solid-phase by air, K3Fe(CN)6, dithiobis(2-nitrobenzoic acid), or diiodoethane oxidation of free sulfhydryls, by direct deprotection/oxidation of Cys(acetamidomethyl) residues using thallium trifluoroacetate or I2, by direct conversion of Cys(9-fluorenylmethyl) residues using piperidine, and by nucleophilic attack by a free sulfhydryl on either Cys(3-nitro-2-pyridinesulfenyl) or Cys(S-carboxymethylsulfenyl). The most generally applicable and efficient of these methods is direct conversion of Cys(acetamidomethyl) residues by thallium trifluoroacetate. Intra-chain lactams are formed between the side-chains of Lys or Orn and Asp or Glu to conformationally restrain synthetic peptides, with the goal of increasing biological potency and/or specificity. Lactams can also be formed via side-chain-to-head, side-chain-to-tail, or head-to-tail cyclization (Kates et al., 1994). The residues used to form intra-chain lactams must be selectively side-chain deprotected, while all side-chain protecting groups of other residues remain intact. Selective deprotection is best achieved by using orthogonal side-chain protection, such as allyloxycarbonyl or 1-(4,4-dimethyl-2,6-dioxocyclohex-1-ylidene)ethyl protection for Lys and allyl or N-[1-(4,4,-dimethyl-2,6-dioxocyclohexylidene)-3-methyl butyl]aminobenzyl protection for Asp/Glu in combination with an Fmoc/tBu strategy. Cyclization is carried out most efficiently with BOP

in the presence of DIEA while the peptide is still attached to the resin. The three-dimensional orthogonal protection scheme of Fmoc/tBu/allyl protecting groups is the strategy of choice for head-to-tail cyclizations. An amide linker is used for sidechain attachment of a C-terminal Asp/Glu (which are converted to Asn/Gln) and the αcarboxyl group is protected as an allyl ester. For side-chain-to-head cyclizations, the N-terminal amino acid (head) can simply be introduced as an Nα-Fmoc derivative while the peptide-resin linkage and the other side-chain protecting groups are stable to dilute acid or carry a third dimension of orthogonality.

PROTEIN SYNTHESIS There are three general chemical approaches for constructing proteins. First is stepwise synthesis, in which the entire protein is synthesized one amino acid at a time. Second is “fragment assembly,” in which individual peptide strands are initially constructed stepwise, purified, and finally covalently linked to create the desired protein. Fragment assembly can be divided into two distinct approaches: (1) convergent synthesis of fully protected fragments, and (2) chemoselective ligation of unprotected fragments. Third is “directed assembly,” in which individual peptide strands are constructed stepwise, purified, and then noncovalently driven to associate into protein-like structures. Combinations of the three general chemical approaches may also be employed for protein construction. Convergent synthesis utilizes protected peptide fragments for protein construction (Albericio et al., 1997). The advantage of convergent protein synthesis is that fragments of the desired protein are first synthesized, purified, and characterized, ensuring that each fragment is of high integrity; these fragments are then assembled into the complete protein. Thus, cumulative effects of stepwise synthetic errors are minimized. Convergent synthesis requires ready access to pure, partially protected peptide segments, which are needed as building blocks. The application of solid-phase synthesis to prepare the requisite intermediates depends on several levels of selectively cleavable protecting groups and linkers. Methods for subsequent solubilization and purification of the protected segments are nontrivial. Individual rates for coupling segments are substantially lower then for activated amino acid species by stepwise synthesis, and there is always a risk of racemization at the C-terminus of each segment. Care-



ful attention to synthetic design and execution may minimize these problems. As an alternative to the segment condensation approach, methods have been developed by which unprotected peptide fragments may be linked. “Native chemical ligation” results in an amide bond being generated between peptide fragments (Muir et al., 1997). A peptide bearing a C-terminal thioacid is converted to a 5-thio-2-nitrobenzoic acid ester and then reacted with a peptide bearing an N-terminal Cys residue (Dawson et al., 1994). The initial thioester ligation product undergoes spontaneous rearrangement, leading to an amide bond and regeneration of the free sulfhydryl on Cys. The method was later refined so that a relatively unreactive thioester can be used in the ligation reaction (Dawson et al., 1997; Ayers et al., 1999). Safety-catch linkers are used in conjunction with Fmoc chemistry to produce the necessary peptide thioester (Shin et al., 1999).

SIDE-REACTIONS

The free Nα-amino group of an anchored dipeptide is poised for a base-catalyzed intramolecular attack of the C-terminal carbonyl. Base deprotection of the Fmoc group can thus release a cyclic diketopiperazine while a hydroxymethyl-handle leaving group remains on the resin. With residues that can form cis peptide bonds, e.g., Gly, Pro, N-methylamino acids, or D-amino acids, in either the first or second position of the (C → N) synthesis, diketopiperazine formation can be substantial. The steric hindrance of the 2-chlorotrityl linker may minimize diketopiperazine formation of susceptible sequences during Fmoc chemistry. The conversion of side-chain protected Asp residues to aspartimide residues can occur by repetitive base treatments. The cyclic aspartimide can then react with piperidine to form the α- or β-piperidide or α- or β-peptide. Aspartimide formation can be rapid, and is dependent upon the Asp side-chain protecting group. Sequence dependence studies of Asp(OtBu)-X peptides revealed that piperidine could induce aspartimide formation when X = Arg(2,2,5,7,8pentamethylchroman-6-sulfonyl; Pmc), Asn(triphenylmethyl; Trt), Asp(OtBu), Cys(Acm), Gly, Ser, Thr, and Thr(tBu) (Lauer et al., 1995). Aspartimide formation can also be conformation-dependent. This side-reaction can be minimized by including 0.1 M HOBt in the piperidine solution (Lauer et al., 1995), or by using an amide backbone protecting group (i.e., 2-hydroxy-4-methoxybenzyl) for the resi-

due in the X position of an Asp-X sequence (Quibell et al., 1994). Cys residues are racemized by repeated piperidine deprotection treatments during Fmoc SPPS. Racemization of esterified (C-terminal) Cys can be reduced by using 1% 1,8diazabicyclo[5.4.0]undec-7-ene in N,N-dimethylformamide (DMF). Additionally, the steric hindrance of the 2-chlorotrityl linker minimizes racemization of C-terminal Cys residues. When applying protocols for Cys internal (not C-terminal) incorporation which include phosphonium and aminium salts as coupling agents, as well as preactivation in the presence of suitable additives and tertiary amine bases, significant racemization is observed. Racemization is generally reduced by avoiding preactivation, using a weaker base (such as collidine), and switching to the solvent mixture DMF-dichloromethane (DCM) (1:1). Alternatively, the pentafluorophenyl ester of a suitable Fmoc-Cys derivative can be used. The combination of side-chain protecting groups and anchoring linkages commonly used in Fmoc chemistry are simultaneously deprotected and cleaved by TFA. Cleavage of these groups and linkers results in liberation of reactive species that can modify susceptible residues, such as Trp, Tyr, and Met. Modifications can be minimized during TFA cleavage by utilizing effective scavengers. Three efficient cleavage “cocktails” quenching reactive species and preserving amino acid integrity, are TFA-phenol-thioanisole-1,2-ethanedithiol-H2O (82.5:5:5:2.5:5) (reagent K) (King et al., 1990), TFA-thioanisole-1,2-ethanedithiol-anisole (90:5:3:2) (reagent R) (Albericio et al., 1990), and TFA-phenol-H2O-triisopropylsilane (88:5:5:2) (reagent B) (Solé and Barany, 1992). The use of Boc side-chain protection of Trp also significantly reduces alkylation by Pmc or 2,2,4,6,7-pentamethyldihydro-benzofuran-5sulfonyl (Pbf) groups.

PURIFICATION AND ANALYSIS OF SYNTHETIC PEPTIDES Each synthetic procedure has limitations, and even in the hands of highly experienced workers, certain sequences defy facile preparation. The maturation of high-performance liquid chromatography (HPLC) has been a major boon to modern peptide synthesis, because the resolving power of this technique facilitates removal of many of the systematic low-level by-products that accrue during chain assembly and upon cleavage. Peptide purification is most commonly achieved by reversed-phase HPLC



Supplement 26

(RP-HPLC; UNIT 11.6). Either alternatively to or in tandem with RP-HPLC, ion-exchange HPLC (UNIT 8.2) and gel-filtration HPLC (UNIT 8.3) can be used for isolation of desired peptide products. The progress of peptide purification can be monitored rapidly by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS; UNITS 16.2 & 16.3) or ion-trap electrospray MS (UNIT 16.8). The homogeneity of synthetic materials should be checked by at least two chromatographic or electrophoretic techniques, e.g., RP-HPLC (UNIT 11.6), ion-exchange HPLC (UNIT 8.2), and capillary zone electrophoresis (UNIT 10.9). Also, determination of a molecular ion by MS (see Chapter 16) using a mild ionization method is important for proof of structure. Synthetic peptides must be checked routinely for the proper amino acid composition, and in some cases sequencing data are helpful. The PSRC studies (see discussion of Development of Solid-Phase Peptide Synthesis Methodology) have allowed for a side-by-side comparison of a variety of analytical techniques. Efficient characterization of synthetic peptides best been obtained by a combination of RP-HPLC and MS, with sequencing by either Edman degradation sequence analysis or tandem MS (UNIT 16.1) being used to identify the positions of modifications and deletions. Proper peptide characterization by multiple techniques is essential.

LITERATURE CITED Albericio, F., Kneib-Cordonier, N., Biancalana, S., Gera, L., Masada, R.I., Hudson, D., and Barany, G. 1990. Preparation and application of the 5-(4(9-fluorenylmethyloxycarbonyl)aminomethyl3,5-dimethoxyphenoxy)valeric acid (PAL) handle for the solid-phase synthesis of C-terminal peptide amides under mild conditions. J. Org. Chem. 55:3730-3743. Albericio, F., Lloyd-Williams, P., and Giralt, E. 1997. Convergent solid-phase peptide synthesis. Methods Enzymol. 289:313-336. Angeletti, R.H., Bonewald, L.F., and Fields, G.B. 1997. Six year study of peptide synthesis. Methods Enzymol. 289:697-717. Atherton, E. and Sheppard, R.C. 1987. The fluorenylmethoxycarbonyl amino protecting group. In The Peptides, Vol. 9 (S. Udenfriend and J. Meienhofer, eds.) pp. 1-38. Academic Press, New York. Ayers, B., Blaschke, U.K., Camarero, J.A., Cotton, G.J., Holford, M., and Muir, T.W. 1999. Introduction of unnatural amino acids into proteins using expressed protein ligation. Biopolymers (Peptide Sci.) 51:343-354. Introduction to Peptide Synthesis

Barany, G. and Merrifield, R.B. 1979. Solid-phase peptide synthesis. In The Peptides, Vol. 2 (E. Gross and J. Meienhofer, eds.) pp. 1-284. Academic Press, New York. Bergmann, M. and Zervas, L. 1932. Über ein allgemeines Verfahren der Peptidsynthese. Ber. Dtsch. Chem. Ges. 65:1192-1201. Carpino, L.A. and Han, G.Y. 1972. The 9-fluorenylmethoxycarbonyl amino-protecting group. J. Org. Chem. 37:3404-3409. Dawson, P.E., Muir, T.W., Clark-Lewis, I., and Kent, S.B.H. 1994. Synthesis of proteins by native chemical ligation. Science 266:776-779. Dawson, P.E., Churchill, M.J., Ghadiri, M.R., and Kent, S.B.H. 1997. Modulation of reactivity in native chemical ligation through the use of thiol additives. J. Am. Chem. Soc. 119:4325-4329. duVigneaud, V., Ressler, C., Swan, J.M., Roberts, C.W., Katsoyannis, P.G., and Gordon, S. 1953. The synthesis of an octapeptide amide with the hormonal activity of oxytocin. J. Am. Chem. Soc. 75:4879-4880. Fields, C.G. and Fields, G.B. 1994. Solvents for solid-phase peptide synthesis. In Methods in Molecular Biology, Vol. 35: Peptide Synthesis Protocols (M.W. Pennington and B.M. Dunn, eds.) pp. 29-40. Humana Press, Totowa, N.J. Fields, G.B. and Noble, R.L. 1990. Solid phase peptide synthesis utilizing 9-fluorenylmethoxycarbonyl amino acids. Int. J. Peptide Protein Res. 35:161-214. Fields, C.G., Fields, G.B., Noble, R.L., and Cross, T.A. 1989. Solid phase peptide synthesis of 15Ngramicidins A, B, and C and high performance liquid chromatographic purification. Int. J. Peptide Protein Res. 33:298-303. Fields, G.B., Otteson, K.M., Fields, C.G., and Noble, R.L. 1990. The versatility of solid phase peptide synthesis. In Innovation and Perspectives in Solid Phase Synthesis: Peptides, Polypeptides and Oligonucleotides, Macro-organic Reagents and Catalysts (R. Epton, ed.) pp. 241-260. Solid Phase Conference Coordination, Ltd., Birmingham, U.K. Fields, G.B., Lauer-Fields, J.L., Liu, R.-q., and Barany, G. 2001. Principles and practice of solidphase peptide synthesis. In Synthetic Peptides: A User’s Guide, 2nd ed. (G.A. Grant, ed.) in press. W.H. Freeman, New York. Fischer, E. and Fourneau, E. 1901. Über einige Derivate des Glykocoils. Ber. Dtsch. Chem. Ges. 34:2868-2877. Kates, S.A., Solé, N.A., Albericio, F., and Barany, G. 1994. Solid-phase synthesis of cyclic peptides. In Peptides: Design, Synthesis and Biological Activity (C. Basava and G.M. Anantharamaiah, eds.) pp. 39-57. Birkhaeuser, Boston. King, D.S., Fields, C.G., and Fields, G.B. 1990. A cleavage method which minimizes side reactions following Fmoc solid phase peptide synthesis. Int. J. Peptide Protein Res. 36:255-266.



Lauer, J.L., Fields, C.G., and Fields, G.B. 1995. Sequence dependence of aspartimide formation during 9-fluorenylmethoxycarbonyl solid-phase peptide synthesis. Lett. Peptide Sci. 1:197-205. Merrifield, R.B. 1963. Solid phase peptide synthesis I: The synthesis of a tetrapeptide. J. Am. Chem. Soc. 85:2149-2154. Merrifield, R.B. 1967. New approaches to the chemical synthesis of peptides. Recent Prog. Hormone Res. 23:451-482. Merrifield, R.B. 1986. Solid phase synthesis. Science 232:341-347. Merrifield, R.B., Stewart, J.M., and Jernberg, N. 1966. Instrument for automated synthesis of peptides. Anal. Chem. 38:1905-1914. Muir, T.W., Dawson, P.E., and Kent, S.B.H. 1997. Protein synthesis by chemical ligation of unprotected peptides in aqueous solution. Methods Enzymol. 289:266-298. Quibell, M., Owen, D., Packman, L.C., and Johnson, T. 1994. Suppression of piperidine-mediated side product formation for Asp(OBut)containing peptides by the use of N-(2-hydroxy4-methoxybenzyl) (Hmb) backbone amide protection. J. Chem. Soc. Chem. Commun. 2343-2344. Rich, D.H. and Singh, J. 1979. The carbodiimide method. In The Peptides, Vol. 1 (E. Gross and J. Meienhofer, eds.) pp. 241-314. Academic Press, New York. Sakakibara, S., Shimonishi, Y., Kishida, Y., Okada, M., and Sugihara, H. 1967. Use of anhydrous HF in peptide synthesis I: Behavior of various protective groups in anhydrous HF. Bull. Chem. Soc. Jpn. 40:2164-2167.

Sarin, V.K., Kent, S.B.H., and Merrifield, R.B. 1980. Properties of swollen polymer networks: Solvation and swelling of peptide-containing resins in solid-phase peptide synthesis. J. Am. Chem. Soc. 102:5463-5470. Shin, Y., Winans, K.A., Backes, B.J., Kent, S.B.H., Ellman, J.A., and Bertozzi, C.R. 1999. Fmocbased synthesis of peptide-αthioesters: Application to the total chemical synthesis of a glycoprotein by native chemical ligation. J. Am. Chem. Soc. 121:11684-11689. Solé, N.A. and Barany, G. 1992. Optimization of solid-phase synthesis of [Ala8]-dynorphin A. J. Org. Chem. 57:5399-5403. Tam, J.P. and Lu, Y.-A. 1995. Coupling difficulty associated with interchain clustering and phase transition in solid phase peptide synthesis. J. Am. Chem. Soc. 117:12058-12063.

KEY REFERENCES Atherton, E. and Sheppard, R.C. 1989. Solid Phase Peptide Synthesis: A Practical Approach. IRL Press, Oxford. An extensive collection of Fmoc-based synthetic methods and techniques. Barany and Merrifield, 1979. See above. The definitive, comprehensive overview of the solidphase method. Fields, G.B. 1997. Solid-phase peptide synthesis. Methods Enzymol. Vol. 289. A contemporary collection of SPPS techniques and applications.

Contributed by Gregg B. Fields Florida Atlantic University Boca Raton, Florida



Supplement 26

Synthesis of Multiple Peptides on Plastic Pins

UNIT 18.2

Scanning protein sequences by bioassay for smaller bioactive peptide sequences requires a source of many peptides homologous with the parent protein sequence. This unit deals with one of the synthetic methods for making such sets of peptides (see Fig. 18.2.1). The key to preparing large numbers (hundreds to thousands) of synthetic peptides in a short time and at minimal cost is to use a parallel synthesis technique which is efficient and can be done on a small scale. The multipin technology is suitable because it can be performed without expensive synthesizers and it uses equipment available to most laboratories. Prior experience with organic synthesis techniques or peptide chemistry is useful but not essential. The products of synthesis by multipin technology are unpurified peptides which are useful as screening reagents and may also be used to prepare purified peptide on a small scale. Most multipin techniques exploit the conventional 8 × 12 matrix layout of common microtiter equipment which simplifies handling of the synthesis, the products (peptides), and the test results. Computer assistance with synthesis and data analysis also speeds the cycle from designing the experiment through analyzing the results. With multipin technology, peptides are synthesized in parallel on plastic “pins” (Fig. 18.2.2) to give sets of peptides suitable not only for B and T cell epitope scanning but also for other bioassays. Peptides can be either permanently bound to the surface of the plastic for direct binding assays, or they can be released into solution. There is a choice of N- and C-terminal peptide endings. For solution-phase peptides, the synthesis scale can be 1 or 5 µmol (for a 10-mer, ∼1 mg or 5 mg, respectively). The preferred coupling/deprotection chemistry used is the milder 9-fluorenylmethyloxycarbonyl (Fmoc) protection scheme rather than the older t-butyloxycarbonyl (t-Boc) protection scheme (see UNIT 18.1), thus reducing the level of chemical safety risk arising from synthetic peptide chemistry. This unit covers the strategy of the multiple peptide approach to biological scanning, the synthetic protocols, and the handling of peptides after synthesis—cleavage, preliminary purification, storage, and analysis (see Basic Protocol). It is specific for the multipin technique using equipment obtained from Chiron Technologies, although some of the approaches are applicable to other multiple synthesis techniques. Procedures for multipin equipment obtained from other suppliers may differ from the procedures described here, and the manufacturer’s literature should be consulted. This unit also includes protocols for preparing Fmoc–amino acid solutions (see Support Protocol 1) and for acetylating (see Support Protocol 2) or biotinylating (see Support Protocol 3) synthesized peptides. STRATEGIC PLANNING For a protein whose primary structure is known, the conceptually simplest method of locating all the bioactive linear peptide sequences is to make all possible peptide subsets of the protein sequence and test them. If only selected parts of the sequence are synthesized, or only the predicted active parts, bioactive sequences could be missed. The use of a set of highly overlapping peptides likewise reduces the possibility that the most bioactive sequences might be missed because they are absent from the set. A set of all overlapping 20-mers offset along the sequence by one residue at a time should capture the entire set of, for example, helper T cell epitopes, and this is a much more reliable approach than trying to predict motifs. In reality, a synthetic peptide scan through a protein is a compromise between the cost and effort in making and screening all peptides and the Contributed by Stuart J. Rodda Current Protocols in Protein Science (1997) 18.2.1-18.2.19 Copyright © 1997 by John Wiley & Sons, Inc.


18.2.1 Supplement 9

need for completeness. Thus, one worker may choose to make all overlapping 8-mers to find the linear (continuous) B cell epitopes, and another may make 12-mers offset along the sequence by five residues for the same purpose. In each case, all sequences of eight residues from the protein are present in at least one peptide, but the latter approach requires only one-fifth the number of peptides.

decide on type of peptides to be made: e.g., noncleavable, cleaved, biotinylated, acetylated (see Table 18.2.1)

obtain kit with the required pin type and install software

Fmoc deprotect the pins

(n cycles) generate synthesis schedule using Pepmaker software based on protein sequence or individual peptide sequences

pipet activated amino acids (Support Protocol 1) into the wells of the reaction trays and place freshly deprotected blocks of pins into the appropriate trays; incubate

position the required number of pins in the pinholder, following the information on the synthesis schedule

deprotect the side chains of peptides; for the MPS kit, cleave the peptides from the pins

wash the peptides, either as a precipitate in a tube (MPS kit) or still on the pins (NCP, GAP, and DKP kits)

wash pins

Fmoc deprotect the pins

acetylate (Support Protocol 2 ) or biotinylate (Support Protocol 3) the N-terminus of the peptides before side chain deprotection or cleavage as required

test pin-bound peptides with conjugate alone, prior to testing with specific antiserum

carry out assays with pin-bound peptides Synthesis of Multiple Peptides on Plastic Pins

cleave the peptides (GAP or DKP kit)

carry out assays with cleaved peptides

Figure 18.2.1 Flow chart for multipin peptide synthesis.

18.2.2 Supplement 9


A

stem

gear

pin with gear

macrocrown

pin with macrocrown

leg

B

Figure 18.2.2 Apparatus for multipin peptide synthesis. (A) Assembled synthesis block with 96 gears (left) or 96 macrocrowns (right). (B) Components of the pin assembly. Components are either push-fitted together (e.g., legs or stems into the pin holder) or clipped on (gears or macrocrowns onto stems). All components are solvent-resistant plastic, either polyethylene, polypropylene, or copolymers of these two monomer types.

Planning the Synthesis Synthetic peptides are assembled by solid-phase synthesis one amino acid at a time, commencing with the C-terminal end of the peptide on the solid phase (see UNIT 18.1). The assembly process, or coupling, requires activation of the α-carboxyl group of each incoming amino acid to make it reactive with the α-amino group of the growing peptide chain. To prevent unwanted polymerization or side reaction, reactive groups in each amino acid must be temporarily protected, and the protecting group removed before further reaction can be carried out. The protecting group on the α-amino function of the most recently added amino acid must be removed before another amino acid can be coupled to it, so the α-amino protection must be labile under conditions that do not remove side-chain protection. Later, the side-chain-protecting groups must be removable under conditions that do not attack the peptide bonds. The two common protecting group “schemes” are known as t-butoxycarbonyl (t-Boc) or 9-fluorenylmethyloxycarbonyl (Fmoc). The protecting group scheme currently recommended for multipin peptide synthesis is the milder Fmoc scheme, which is the only scheme described in this chapter. Before beginning to plan the actual synthesis in detail, a choice needs to be made regarding how the peptides will eventually be presented in the bioassay. The options available to investigators are listed in Table 18.2.1. For noncleavable peptide (NCP) kits, peptides are permanently bound on the solid phase (pin surface) and can be used for direct binding assays but not for interaction with living cells or other complex (e.g., multicomponent) systems. In this case, the peptides must be



Supplement 9

Table 18.2.1

Types of Pins for Multipin Peptide Synthesisa

Name

Linkerb

Physical formatc Loading

Final form of peptide

NCP MPS MPS DKP GAP

Noncleavable AA ester Rink amide DKP-forming Glycine ester

Gear Macrocrown Macrocrown Gear Gear

(N-capping)-PEPTIDE-linker-pin (N-capping)-PEPTIDE-acid (N-capping)-PEPTIDE-amide (N-capping)-PEPTIDE-DKP (N-capping)-PEPTIDE-glycine-acid

50 nmol 5 µmol 5 µmol 1 µmol 1 µmol

aAbbreviations: DKP, diketopiperazine; GAP, glycine acid peptide; MPS, multiple peptide synthesis; NCP, noncleavable

peptide; (N-capping), a free amine, acetyl group, or biotin; PEPTIDE, the sequence of the peptide being made. bNature of linker between peptide and graft polymer on the pin: noncleavable linker, β-alanine-hexamethylenediamine;

DKP, diketopiperazine; AA ester, amino acid ester; Rink amide, Rink amide–forming linker. cSee Figure 18.2.2B.

“regenerated” between repeat assays by disrupting the peptide-ligand interaction without damaging the peptide. The quantity of peptide made is very small (50 nmol), but it is sufficient to provide a high surface density of peptide for direct binding assays. In the other options, peptides are synthesized on pins and then released into solution. The mechanism of peptide release into solution affects the postsynthesis handling and thus the suitability of peptides produced by each cleavage method for various assay systems. For multiple peptide synthesis (MPS) kits, the released peptides have a “native” free acid or an amide carboxy terminus. To make free acid C-termini, it is necessary to use macrocrowns that already have the first (C-terminal) amino acid on them because the chemistry of forming the first (ester) link is too difficult for the inexperienced user. In contrast, the Rink amide linker allows formation of a peptide with a C-terminal amide of any amino acid by adding the C-terminal amino acid to the Rink handle macrocrown using the standard amino acid coupling protocol. A Rink amide linker is a linker that can accept an amino acid but then can be cleaved in trifluoroacetic acid (TFA) to release the amide form of that amino acid (Rink, 1987). Although acid or amine endings are often the most desirable peptide format to have, they are also the most complex to produce because the cleavage of the peptides from the pin is into neat TFA plus scavengers which needs to be evaporated to recover the peptide. The scale of peptide synthesis for MPS kits is 5 µmol (∼5 mg of a decamer). For glycine acid peptide (GAP) kits, peptides with a glycine at the carboxy terminus are cleaved as the free acid, so that the C-terminal residue is a natural amino acid (glycine) and is not blocked. The peptides are also relatively simple to release from the pin and require little postsynthesis handling. However, the presence of glycine at the C-terminus may be undesirable where the C-terminus plays an important role in peptide bioactivity. The scale of synthesis for GAP kits is 1 µmol (∼1 mg of a decamer). In diketopiperazine (DKP) kits, peptides are synthesized with a DKP group at the C terminus. The DKP group is a cyclic dipeptide formed from C-terminal lysine and proline residues during the facile cleavage of the peptide under the mildest possible conditions: neutral aqueous buffer. In applications where the presence of the DKP group is acceptable, this type of peptide can make the downstream processing of synthetic peptides very simple and fast. The peptides can be placed into a bioassay system immediately after completing the cleavage. The scale of synthesis for DKP kits is 1 µmol (∼1 mg of a decamer). Synthesis of Multiple Peptides on Plastic Pins

For these five kit options, it is also possible to choose a variety of N-terminal endings on the peptides. For example, it may be desirable to acetylate pin-bound peptides (see Support Protocol 2) to eliminate the positive charge that would otherwise be present on

18.2.4 Supplement 9


the α-amino group of the N-terminal residue, or to enhance the activity of a peptide in a T helper assay (Mutch et al., 1991). A handy option for cleaved peptides is to place a biotin group on the N-terminus (see Support Protocol 3) so the peptide can be captured using avidin or streptavidin. These additions must be made prior to side-chain deprotection of the peptides. There are other configurations for multiple peptide synthesis—e.g., the SPOTS or “peptides on paper” system (Zeneca/CRB), the RaMPS system (DuPont), and multi-synthesizer machines (e.g., Advanced ChemTech). Assessing Peptide Sequences Peptides differ so much in properties that it is important to assess the likely properties of the peptides before attempting to synthesize them. Peptide length and hydrophobicity are the two main attributes affecting successful synthesis. The longer the peptide, the lower will be the purity of the product, as each amino acid coupling cycle is never 100% efficient. Synthesis of peptides longer than 20 residues should be avoided unless special attention can be given to each sequence. Hydrophobic peptides may be difficult to synthesize, but more significantly they may be poorly soluble in aqueous buffers, restricting their ultimate usefulness in bioassays. Prior to beginning synthesis of a set of peptides, it is sensible to assess them all for hydrophobicity (Fauchere and Pliska, 1983; UNIT 2.2) and decide if all should be attempted as they stand. In many cases, it is possible to choose slightly different peptides (longer, shorter, or using a different starting and finishing point in the homologous protein sequence) that will have more user-friendly properties. As well as these general factors affecting peptides, particular peptide sequences may have characteristics that make them difficult to synthesize, or they may be problematic after synthesis. It is not feasible to discuss all the common problems here. To help assessment of peptide sequences, a software application called Pinsoft is available free from Chiron Technologies. This allows any sequence to be typed in, and an assessment is automatically reported. Generating Peptide Sequences Computer software (Pepmaker) supplied with synthesis kits allows sets of overlapping peptide sequences to be generated from a protein sequence computer file using the single-letter amino acid code. Alternatively, sequences can be created using a word processor and the resulting computer text file can then be used by Pepmaker to guide synthesis. The use of this software simplifies the otherwise complex and tedious task of adding the right amino acids to each reaction plate on each synthesis cycle. MULTIPIN SYNTHESIS OF PEPTIDES Derivatized pins with macrocrowns or gears are attached to a pin holder. Each peptide is built up on the reactive surface of one pin by successive cycles of amino acid coupling, followed by washing and removal of the 9-fluorenylmethyloxycarbonyl (Fmoc) aminoprotecting group to prepare for the next amino acid coupling cycle. A critical step is properly dispensing activated amino acid solutions into the appropriate wells of each reaction tray. A list of the well locations for dispensing of each amino acid is generated by the Pepmaker software for this purpose. When the peptides are complete, trifluoroacetic acid (TFA) that contains scavengers is used to remove the side-chain-protecting groups, and for MPS kits, to cleave the peptides from the pins. The manual provided with each type of kit (see Table 18.2.1) includes instructions and hints for kit-specific procedures. NOTE: All reagents should be of the highest grade possible, preferably peptide synthesis or analytical reagent grade.

BASIC PROTOCOL



Supplement 9

Materials 20% (v/v) piperidine/dimethylformamide (DMF; see recipe) DMF, analytical reagent grade Methanol, analytical reagent grade 100 mM activated 9-fluorenylmethyloxycarbonyl (Fmoc)–protected amino acid solutions (see Support Protocol 1) Side chain deprotecting (SCD) solution (see recipe) Acidified methanol: 0.5% (v/v) glacial acetic acid/methanol 1:2:0.003 (v/v/v) ether/petroleum ether/2-mercaptoethanol (2-ME) 1:2 (v/v) ether/petroleum ether 0.1 M NaOH 0.1 M acetic acid 0.1 M sodium phosphate buffer, pH 8.0 (APPENDIX 2E) Sonication buffer (see recipe) Peptide Synthesis Starter Kit (e.g., Chiron Technologies) of the desired type, containing: Pepmaker computer program and ELISA reading and plotting programs Manual Pins with gears or macrocrowns Storage boxes or sealable bags, polyethylene or polypropylene (ICN Biomedicals) Pipettor tips, polyethylene or polypropylene (ICN Biomedicals) 0.3- or 1.5-ml reaction trays, polyethylene or polypropylene (Chiron Technologies, Nunc, or Beckman) Sonicator with power output of ∼500 W Dry nitrogen Rack containing 96 1-ml polypropylene tubes (Bio-Rad) 10-ml capped conical polypropylene centrifuge tubes Additional reagents and equipment for N-terminal acetylation (see Support Protocol 2; optional) or biotinylation (see Support Protocol 3; optional) CAUTION: Perform all chemistry steps in a well-functioning chemical fume hood. Wear solvent-resistant gloves, safety glasses, and protective clothing. The reagents can be flammable, toxic, and/or carcinogenic. Avoid sources of contamination which may affect the pins, including direct contact with the bench surface or exposure to vapors. The reagents for multipin synthesis can be handled in unsealed systems, but the amount of time that these reagents are left exposed to the open air should be minimized by using capped containers for liquids or polyethylene bags for pins wherever practical. Local regulations for safe disposal of solvents and used reagents must be followed. Prepare synthesis schedule and equipment 1. Use the Pepmaker computer program according to the instructions to generate the required set of peptide sequences (Fig. 18.2.3). Generate the printouts, which show for each coupling cycle how much of each amino acid solution, catalyst, and activating agent needs to be prepared (see Fig. 18.2.4) and where each amino acid solution is to be added to the reaction tray (see Fig. 18.2.5).

Synthesis of Multiple Peptides on Plastic Pins

The standard microtiter plate layout is an 8 × 12 matrix, in which the eight rows are identified as A through H and the twelve columns are identified as 1 through 12. However, the Pepmaker software uses a designation in which the column number is given first followed by a number designation for the row, beginning with row H, given in parentheses—i.e., 1(1) for well H1, 1(2) for well G1, 2(1) for well H2, and 12(8) for well A12 (see Fig. 18.2.6).

18.2.6 Supplement 9


2. Label each pin holder block indelibly on the top (e.g., Synthesis #1, Block A, Synthesis #1, Block B, and so forth), preferably by scratching into the plastic with a sharp tool. Place the label where it will help orient the block so that the pins are not accidentally placed into amino acid solutions in an inverted orientation. For example, keep pin H1 and well H1 at the lower left corner of the block (Fig. 18.2.6). Ink labels will run or disappear with exposure to solvents. The multipin system is based on the standard microtiter plate layout. The block is the complete unit and consists of the pin holder, which is the support that holds 96 pins (in an 8 × 12 array with standard ELISA microtiter plate spacing), and five legs to support the device and correctly position the active surfaces. A pin consists of an inert stem that supports either a gear or a macrocrown, both of which have an active surface on which the peptide is synthesized (see Figure 18.2.2). A gear is a detachable gear-shaped unit that fits on the thin end of a stem. A macrocrown is a detachable, vaned tip that fits on the thin end of a stem. It is made of high-density polyethylene and the surface is derivatized to give a solvent-compatible polymer matrix. Macrocrowns are provided in two forms: one has a linker that cleaves to give peptides with an amide at the carboxy terminus; the other has a linker that cleaves to give the free acid at the carboxy terminus and is supplied with an amino acid already attached to the linker. The reaction tray used for the synthesis is a polyethylene or polypropylene tray consisting of 96 wells in the standard microtiter plate 8 × 12 matrix. Shallow reaction trays (0.3-ml) are used with gears and deep trays (1.5-ml) are used with macrocrowns.

3. Remove any pins that are not required for the first cycle of amino acid coupling and store them dry in a plastic bag in the refrigerator until needed. Some pins need to be removed when the peptides in the synthesis are of various lengths because the software is designed to arrange all peptides to complete their synthesis on the same (final) coupling cycle. This approach eliminates unnecessary Fmoc–deprotection cycles for pins that are designated to carry the shorter peptides. The synthesis printout from the Pepmaker software shows which pins need to be added for each cycle of amino acid addition (Fig. 18.2.5). Pins (stems) can be pushed out from the top side of the pin holder. In the case of the MPS kit, where the first amino acid is already on the macrocrown as supplied, choose and mount the correct macrocrown for each position on the block.

Deprotect α-amino groups 4. Add 20% piperidine/DMF to a bath and place the pins in the bath so that the tips (macrocrowns or gears) are covered. Cover and let stand for 20 min at room temperature. CAUTION: Piperidine is flammable. The volume of reagent needed for all the bath steps depends on the shape of the bath, the critical factor being that all surfaces of the pins (i.e., the gears or macrocrowns) bearing the peptide need to be totally covered. A small bath suitable for gears is the upturned polypropylene lid of a pipettor tip box. For macrocrowns, deeper baths or deep-well polypropylene trays as supplied with the kit can be used.

5. Remove the block from the bath, shake off the excess liquid, and then wash the pins in a DMF bath for 2 min at room temperature. Again, the DMF must fully cover the tips.

6. Shake off the excess DMF and immerse the block completely in a deep bath of methanol for 2 min so that all surfaces of the block are washed. CAUTION: Methanol is flammable and toxic. In a shallower bath the block can be turned over so that the pin holder part is washed as well.



Supplement 9

A

GENERAL NET SYNTHESIS SCHEDULE NUMBER : 1

Page 1

Description: Example of a scan through Sperm Whale Myoglobin 8-mer peptides based on the sequence MBN-SW Peptide spacing increment is 1 Segment 1: 146 peptides starting at residue 1 First peptide: [VLSEGEWQ] Last peptide: [YKELGYQG] Protein sequence: MBN-SW (153 residues) 1: VLSEGEWQLV LHVWAKVEAD VAGHGQDILI RLFKSHPETL EKFDRFKHLK 51: TEAEMKASED LKKHGVTVLT ALGAILKKKG HHEAELKPLA QSHATKHKIP 101: IKYLEFISEA IIHVLHSRHP GNFGADAQGA MNKALELFRK DIAAKYKELG 151: YQG Amino Acid set to be used - AASET1 aaset 1:Free acid L-Fmoc amino acids - DIC/HOBt chemistry Number of copies of each peptide 1 Schedule based on a 250 microliter fill/well (Well concentration is 100 mM)

Figure 18.2.3 (above and at right) A portion of the synthesis schedule worksheets generated by Pepmaker software for Schedule no. 1 for synthesis of a set of all possible overlapping octamers of sperm whale myoglobin. (A) This page of the synthesis schedule is a summary of the features of the protein including its sequence. (B) This page of the synthesis schedule shows the sequences of the first 96 peptides that will be synthesized, the first two of which are the controls, PLAQGGGG and GLAQGGGG. Peptide sequences are shown in the conventional amino-to-carboxy-terminal direction (from left to right), with a “200 peptides), it is important that the dispensing be fast and accurate so that three couplings can be carried out per day.

Literature Cited Burrows, S.R., Gardner, J., Khanna, R., Steward, T., Moss, D.J., Rodda, S., and Suhrbier, A. 1994. Five new cytotoxic T cell epitopes identified within Epstein-Barr virus nuclear antigen 3. J. Gen. Virol. 75:2489-2493. Carter, J.M., VanAlbert, S., Lee, J., Lyons, J., and Deal, C. 1992. Shedding light on peptide synthesis. Bio/Technology 10:509-513. Fauchere, J.L. and Pliska, V. 1983. Hydrophobic parameters of amino acid side chains from the partitioning of N-acetyl-amino-acid amides. Eur. J. Med. Chem. 18:369-375. Geysen, H.M., Meloen, R.H., and Barteling, S.J. 1984. Use of peptide synthesis to probe viral antigens for epitopes to a resolution of a single amino acid. Proc. Natl. Acad. Sci. U.S.A. 81:3998-4002.

Geysen, H.M., Rodda, S.J., Mason, T.J., Tribbick, G., and Schoofs, P.G. 1987. Strategies for epitope analysis using peptide synthesis. J. Immunol. Methods 102:259-274. Maeji, N.J., Bray, A.M., and Geysen, H.M. 1990. Multi-pin peptide synthesis strategy for T cell determinant analysis. J. Immunol. Methods 134:23-33. Mutch, D.A., Rodda, S.J., Benstead, M., Valerio, R.M., and Geysen, H.M. 1991. Effects of end groups on the stimulatory capacity of minimal length T cell determinant peptides. Pept. Res. 4:132-137. Reece, J.C., Geysen, H.M., and Rodda, S.J. 1993. Mapping the major human T helper epitopes of tetanus toxin: The emerging picture. J. Immunol. 151:6175-6184. Reece, J.C., McGregor, D.L., Geysen, H.M., and Rodda, S.J. 1994. Scanning for T helper epitopes with human PBMC using pools of short synthetic peptides. J. Immunol. Methods 172:241254. Rink, H. 1987. Solid-phase synthesis of protected peptide fragments using a trialkoxydiphenylmethylester resin. Tetrahedron Lett. 28:37873790. Stewart, J.M. and Young, J.D. 1984. Solid Phase Peptide Synthesis, 2nd ed. Pierce Chemical Co., Rockford, Ill. Valerio, R.M., Bray, A.M., Campbell, R.A., Dipasquale, A., Margellis, C., Rodda, S.J., Geysen, H.M., and Maeji, N.J. 1993. Multipin peptide synthesis at the micromole scale using 2-hydroxyethyl methacrylate grafted polyethylene supports. Int. J. Pept. Protein Res. 42:1-9.

Contributed by Stuart J. Rodda Chiron Technologies Pty. Ltd. Victoria, Australia



Supplement 9

Synthetic Peptides for Production of Antibodies that Recognize Intact Proteins

UNIT 18.3

Antibodies that recognize intact proteins can be produced through the use of synthetic peptides based on short stretches of the protein sequence, without first having to isolate the protein. The procedure for selecting stretches of protein sequence likely to be antigenic is relatively straightforward. However, no procedure will identify a single sequence guaranteed to be effective, nor will it usually identify the best single sequence to use. Rather, several sequences will be identified that have a higher-than-average probability of producing an effective antigen. The steps to produce an effective antibody include: (1) designing the peptide sequence based on the sequence of the protein; (2) synthesizing the peptide; (3) preparing the immunogen either by coupling the synthetic peptide to a carrier protein or through the use of a multiple antigenic peptide (MAP); (4) immunizing the host animal; (5) assaying antibody titer in the host animal’s serum; and (6) obtaining the antiserum and/or isolating the antibody. This unit covers steps 1 and 3; step 2 requires a laboratory with expertise in peptide synthesis. Peptide synthesis services are widely available both academically and commercially. The best method to select potentially effective sequences is via a computer-assisted strategy (see Basic Protocol 1). An alternative manual method is also described (see Alternate Protocol 1) but is not recommended to replace the use of algorithms if there is a choice. A small synthetic peptide is usually insufficiently immunogenic on its own, and two methods have been developed to solve this problem. The first (see Basic Protocol 2) involves chemically coupling the synthetic peptide to a carrier protein to boost the immune response. The second method (see Alternate Protocol 2) entails direct synthesis of a MAP covalent multimer of the simple peptide sequence. Both methods have proven effective and it is a matter of personal preference which to use. Coupling to a carrier protein requires additional chemical manipulations after synthesis of the peptide, while the MAP is complete and ready for immunization at the conclusion of the synthetic protocol. Disadvantages of MAPs are that they are more difficult to produce homogeneously and to analyze postsynthetically. They also may be more prone to insolubility problems. A carrier protein is a relatively large molecule capable of stimulating an immune response independently. A synthetic peptide coupled to a carrier protein acts as a hapten and produces antibodies specific for the hapten (antibodies against the carrier protein are also produced). The most commonly used carrier proteins are keyhole limpet hemocyanin (KLH) and bovine or rabbit serum albumin (BSA or RSA). KLH is usually preferred, because it tends to elicit a stronger immune response and is evolutionarily more remote from mammalian proteins. A common problem with KLH, however, has been its solubility. Pierce Chemical Company sells a preparation of KLH purported to have better solubility properties (see below). Alternatively, peptides can be coupled to carrier proteins through either their amino (see Alternate Protocol 3) or carboxyl groups (see Alternate Protocol 4). These two alternate protocols are not recommended as a first choice for coupling, but are included because they have been used successfully and may be advantageous for certain special applications discussed in the Commentary. Also presented are methods for assaying free sulfhydryl content and for reducing disulfide bonds in synthetic peptides (see Support Protocols 1 and 2).

Contributed by Gregory A. Grant Current Protocols in Protein Science (2002) 18.3.1-18.3.19 Copyright © 2002 by John Wiley & Sons, Inc.



Once the coupling procedure has been performed, it is possible to determine the approximate degree of coupling by amino acid analysis (see Support Protocol 3). However, in most instances this is unnecessary and the product can be used directly. BASIC PROTOCOL 1

COMPUTER-ASSISTED SELECTION OF APPROPRIATE ANTIGENIC PEPTIDE SEQUENCES An antibody produced in response to a simple linear peptide will most likely recognize a linear epitope in a protein. Furthermore, that epitope must be solvent-exposed to be accessible to the antibody. The general features of protein structure that correspond to these criteria are turns or loop structures, which are generally found on the protein surface connecting other elements of secondary structure, and areas of high hydrophilicity, especially those containing charged residues. As a consequence, computer algorithms that predict protein hydrophilicity and tendency to form turns are very useful. Several analytic programs or algorithms that attempt to do this have been developed. Although the choice of method may rely on availability or personal preference, there tends to be a high level of agreement among them. As stated earlier, none of the methods will identify the one single sequence guaranteed to produce an effective antibody against any given protein. Rather, the methods will offer several good candidates, one or several of which can be used. Many of these algorithms may already be available on a local computer system. They are included in many commercial software packages such as GCG (Genetics Computer Group; see SUPPLIERS APPENDIX). The ExPASy Web site of the University of Geneva offers free access to a variety of different programs over the Internet at http://expasy.org/tools. The following protocol utilizes the hydropathy index developed by Kyte and Doolittle (1982) and the secondary structure prediction method for β turns developed by Chou and Fasman (1974) found in the tool “Protscale” at the ExPASy Internet address. 1. Using the selected algorithms, compute the hydropathy index and the tendency for β-turns of the protein sequence. Use a window size of 7 or 9 and give equal weight to each amino acid. Record the results in either graphical or numerical form, or both. As an example, the graphical representation of these results for the protein sequence shown in Figure 18.3.1 is presented in Figure 18.3.2. A window size determines the number of amino acids to be used in computing a value for the amino acid at the center of the window. For example, a window size of 9 includes 4

10

20

30

40

50

60

MAKVSLEKDK IKFLLVEGVH QKALESLRAA GYTNIEFHKG ALDDEQLKES IRDAHFIGLR SRTHLTEDVI NAAEKLVAIG CFCIGTNQVD LDAAAKRGIP VFNAPFSNTR SVAELVIGEL 120 LLLLRGVPEA NAKAHRGVWN KLAAGSFEAR GKKLGIIGYG HIGTQLGILA ESLGMYVYFY 180 DIENKLPLGN ATQVQHLSDL LNMSDVVSLH VPENPSTKNM MGAKEISLMK PGSLLINASR 240 GTVVDIPALC DALASKHLAG AAIDVFPTEP ATNSDPFTSP LCEFDNVLLT PHIGGSTQEA 300 QENIGLEVAG KLIKYSDNGS TLSAVNFPEV SLPLHGGRRL MHIHENRPGV LTALNKIFAE 360 QGVNIAAQYL QTSAQMGYVV IDIEADEDVA EKALQAMKAI PGTIRARLLY

Selecting Synthetic Peptides for Production of Antibodies

Figure 18.3.1 The amino acid sequence of a 410-residue protein analyzed by the method presented in Basic Protocol 1. The results are shown in Figure 18.3.2.



amino acids on each side of the central amino acid. The value computed for the central amino acid is the simple average of the values for each amino acid in the window.

2. Compare the results of the two analyses and look for areas of sequence that are high in turn tendency and high in hydrophilicity (low in hydrophobicity). In Figure 18.3.2, these areas correspond to positive peaks in the Chou-Fasman analysis and negative peaks in the Kyte-Doolittle analysis. The three best areas in terms of amplitude and correlation are shaded. These correspond to the sequences underlined in Figure 18.3.1. (Note the alignment of these peak optima as compared to the peaks around residue 300.)

A

1.3 1.2

Score

1.1 1.0 0.9 0.8 0.7 0.6

B

3

2

Score

1

0

–1

–2 50

100

150

200

250

300

350

400

Position

Figure 18.3.2 Graphical representation of the results generated by a computer algorithm for the sequence in Figure 18.3.1, analyzed by the method presented in Basic Protocol 1. The shaded areas represent three regions in the sequence meeting criteria for selection as potential immunogens. (A) Analysis for β turns (Chou and Fasman, 1974). (B) Analysis for hydrophobicity (Kyte and Doolittle, 1982).



Supplement 28

3. Examine the sequences for glycosylation site motifs and discard any sequences that contain them unless it is known that the protein is not glycosylated. Amino acids in glycosylated regions may be shielded from presentation to an antibody by masking carbohydrates. Amino-linked carbohydrate chains can occur at Asn-X-Ser or Asn-X-Thr sequences. Hydroxyl-linked carbohydrate chains do not appear to have a set motif. A program to assist in the prediction of mucin-type GalNAc O-glycosylation sites in mammalian lipoproteins is found in the tool “NetOGlc” at the Expasy site (http://expasy.org/tools). However, before using read the documentation carefully and keep in mind that such prediction methods cannot always be successful.

4. Select the best sequences resulting from this analysis to use as antigenic peptides. These are sequences where the largest positive values (peaks with positive deflection) for turn propensity correspond in position to the largest negative values (peaks with negative deflection) for hydrophobicity. The values obtained in these analyses are relative and dependent on the individual protein’s composition, so it is not possible to set an arbitrary minimum value as a cutoff for rejecting a particular peak. Rather, always select the peaks of greatest magnitude in any given sequence. In addition, the immediate amino-terminal and carboxyl-terminal regions of proteins are often exposed to solvent. If these areas appear to be hydrophilic in nature, they are also acceptable candidates. Thus each analysis may provide several potential sequences. How many peptides to make (see Anticipated Results) is a matter of individual choice. ALTERNATE PROTOCOL 1

MANUAL INSPECTION TO SELECT APPROPRIATE PEPTIDE SEQUENCES If computer algorithms are not available, it is possible to select potential sequences by manual inspection. Although there is no evidence that a manual method is any less effective than the use of computer algorithms, there is a greater probability of overlooking potentially important areas of sequence. It is therefore recommended that computer analysis be used whenever it is available. Although it can be done, it would be very time consuming and labor intensive to manually calculate values for every overlapping peptide offset by a single amino acid in the same way that the algorithms do. For this reason, areas rich in polar residues are selected for manual calculation of hydrophilicity and turn propensity. 1. Visually inspect the protein sequence and select areas that contain at least two to three charged residues (Lys, Arg, His, Asp, Glu) within a 10- to 15-residue span. If this criterion cannot be met, select sequences with the greatest number of charged residues.

2. From the sequences identified in step 1, select a subset of sequences that are the highest in Ser, Thr, Asn, Gln, Pro, and Tyr content. 3. Calculate average hydrophilicity and turn propensity for each amino acid in the selected sequences using the values given in Table 18.3.1 and a window of 9 residues (see Basic Protocol 1, step 1). Be sure to include the residues flanking the selected sequence for calculation of values for the residues at the ends of the selected sequence. In other words, do not use different size windows.

4. Plot the values for each amino acid of a chosen sequence. Selecting Synthetic Peptides for Production of Antibodies

Sequences whose optimal values for hydrophilicity and turn propensity correspond (as in Fig. 18.3.2) are considered good candidates.

5. Inspect sequences for glycosylation motifs and discard these candidates (see Basic Protocol 1, step 3).



Table 18.3.1

Hydrophobic and β-Turn Indices of Amino Acids

Amino acid

Symbols

Arginine Lysine Aspartic acid Glutamic acid Asparagine Glutamine Histidine Proline Tyrosine Tryptophan Serine Threonine Glycine Alanine Methionine Cysteine Phenylalanine Leucine Valine Isoleucine

Arg (R) Lys (K) Asp (D) Glu (E) Asn (N) Gln (Q) His (H) Pro (P) Tyr (Y) Trp (W) Ser (S) Thr (T) Gly (G) Ala (A) Met (M) Cys (C) Phe (F) Leu (L) Val (V) Ile (I)

aKyte

Hydrophobicity β-turn valuea propensityb −4.5 −3.9 −3.5 −3.5 −3.5 −3.5 −3.2 −1.6 −1.3 −0.9 −0.8 −0.7 −0.4 1.8 1.9 2.5 2.8 3.8 4.2 4.5

0.95 1.01 1.46 0.74 1.56 0.98 0.95 1.52 1.14 0.96 1.43 0.96 1.56 0.66 0.60 1.19 0.60 0.59 0.50 0.47

and Doolittle (1982). and Fasman (1974).

bChou

Amino acids of glycosylated regions may be masked in native proteins, so an antibody raised against them would be ineffective.

6. Select the best sequences (see Basic Protocol 1, step 4 for criteria), choosing a high turn-propensity-to-hydrophobicity ratio. DESIGNING A SYNTHETIC PEPTIDE FOR COUPLING TO A CARRIER PROTEIN

BASIC PROTOCOL 2

Although there is no direct evidence to show that the state of the termini of the peptide affects its ability to produce antibodies that will react with the protein, most procedures suggest that the termini of the peptide should mimic their native state. Thus, sequences whose terminal residues normally are in peptide linkage in the protein can have their amino-terminal and carboxyl-terminal groups modified by acetylation and amidation, respectively, during synthesis. Modification of the amino or carboxyl termini will decrease the polarity of the peptide in solution and could have a significant effect on the peptide’s solubility. If the peptide lacks sufficient protonatable side chains, modification of the termini can be omitted. A general rule to predict solubility is that the total number of charges at a given pH should be at least 20% of the number of residues in the peptide. 1. Choose a sequence of 10 to 15 amino acid residues for the synthetic peptide. Longer peptides are more difficult and expensive to make, and they are usually unnecessary.



Supplement 28

Try to choose a stretch of sequence that contains some charged residues such as Arg, Lys, His, Glu, and Asp. In addition to the high likelihood of these amino acids being located on the surface of the protein, they aid handling of the synthetic product by promoting solubility.

2a. If the selected sequence does not contain an internal cysteine: Place a cysteine on either the amino or carboxyl terminus for use in coupling to a carrier protein with a heterobifunctional cross-linking reagent such as MBS, m-maleimidobenzoyl-N-hydroxysuccinimide ester (see Basic Protocol 3). Cross-linking with heterobifunctional reagents is the recommended procedure for most peptides (see Basic Protocol 3). As an alternative to using a chemical cross-linking reagent, any peptide, regardless of amino acid content, can also be photochemically linked to a carrier protein if p-benzoyl benzoic acid is added to the peptide during synthesis (see Alternate Protocol 5). Although photochemical cross-linking is effective, it is not widely used. If the sequence includes the immediate amino or carboxyl terminal sequence of the protein, the cysteine should be placed on the end that would normally be engaged in the internal peptide bond. For sequences internal to the protein, the cysteine may be placed at either end according to the preference of the synthetic chemist. However, if amino-terminal capping (acetylation) is used after the coupling of each amino acid during synthesis, it is preferable to place the cysteine on the amino-terminal end of the peptide since then only the full-length peptide will contain the cysteine residue. In this way, if synthetic difficulties are encountered, only the full-length peptide will couple to the carrier. If placed on the carboxyl terminal end, the cysteine residue tends to racemize during synthesis unless a chlorotrityl resin is used. However, this should not have an effect on the rest of the peptide or the generation of antibodies.

2b. If the sequence contains an internal cysteine residue: Do not add a terminal cysteine for MBS cross-linking. Rather, use an alternative coupling procedure (see Alternate Protocols 3, 4, or 5) or synthesize a multiple antigenic peptide (MAP; see Alternate Protocol 2). Internal cysteine sulfhydryl groups will also cross-link to the carrier protein, and multiple cysteines will result in a peptide attached at multiple points. If the sulfhydryl will eventually be important for antibody recognition of the protein, the immunization may not produce effective antibodies. Furthermore, the additional constraint produced by the existence of multiple points of coupling may affect the ultimate ability of the antibody to recognize the protein.

3a. For sequences whose terminal residues are in peptide linkage within the protein: If the peptide is coupled using a heterobifunctional reagent such as MBS (see Basic Protocol 3), modify the amino and carboxyl ends by acetylation and amidation, respectively, during the synthetic procedure. If coupling is performed with a homobifunctional reagent that reacts with amino groups such as glutaraldehyde (see Alternate Protocol 3) or by the photochemical method (see Alternate Protocol 5), only amidate the carboxyl terminus. If coupling is performed with EDC (see Alternate Protocol 4), only acetylate the amino terminus. 3b. For sequences that are amino or carboxyl terminal to the protein:


i. If the sequence is the immediate amino or carboxyl terminal sequence of the protein and the peptide will be coupled with a heterobifunctional reagent such as MBS (see Basic Protocol 3), leave the end that is not in peptide linkage (and the end that does not contain the additional cysteine residue for MBS coupling) as the free amino or carboxyl group unless it is known that they are normally blocked. ii. If the sequence is the immediate amino terminal sequence of the protein and the peptide will be coupled with a homobifunctional reagent that reacts with amino



NH 2

aan C

O NH

H N

aa3 O

HO

O

C

O C

C

NH

NH β-Ala

aa2 O C N

O C

N

O

ε

Lys

H C N

H

C O H

O aa3

O

C

H N

C

Lys H Nε

aa2

O

NH

H

aa1

aan

NH 2

aa3

C

H2N

aan

O

H N α

aa1

O

O

H

C

Nα

C N

C

Lys

NαH

NεH

C O

C O

aa1

aa1

NH

NH

C O

C O

aa2

aa2

H N

O C

aa3

H N

C

aan

NH2

O

Figure 18.3.3 Representation of a four-branched multiple antigenic peptide (MAP). The MAP core is usually synthesized with a β-alanine residue attached to the solid phase support followed by a scaffold of lysine residues. The first lysine residue, attached through its α-carboxyl group to β-alanine, provides two primary amino groups (α and ε) for chain elongation. The next level of the lysine scaffold consists of two lysine residues which in turn provide four primary amino groups for chain elongation. MAPs with eight branches are formed by providing one additional layer of lysine to that depicted here.

groups such as glutaraldehyde (see Alternate Protocol 3) or photochemically (see Alternate Protocol 5), amidate the C-terminus. iii. If the sequence is the immediate carboxyl terminal sequence of the protein and the peptide will be coupled with glutaraldehyde (see Alternate Protocol 3) or photochemically (see Alternate Protocol 5), leave both termini free. 4. If the peptide will be coupled with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC; see Alternate Protocol 4), block the amino terminus by acetylation. This should be done even if the sequence is at the immediate amino terminus of the protein, since treatment with EDC can result in the production of covalent multimers of the peptide through reaction with the N-terminal amino group. Some procedures recommend using citraconylation to temporarily protect the amino group. However, this also adds a free carboxyl group that could result in multiple site attachment of the peptide to the carrier.



Supplement 28


DESIGNING A SYNTHETIC MULTIPLE ANTIGENIC PEPTIDE A multiple antigenic peptide (MAP; Posnett et al., 1988; Tam, 1988) is an effective alternative to coupling a simple linear peptide to a carrier protein. MAPs are covalent constructs consisting of a simple peptide sequence synthesized on a branched core (Fig. 18.3.3) with one copy of the peptide sequence on each of four or eight branches. One advantage of a MAP is that it is suitable for use as an immunogen at the conclusion of the synthetic process. On occasion, MAPs have produced effective protein antibodies when the conventional peptide coupled to carrier protein has not. Thus, MAPs represent an effective alternative approach for antiserum production. 1. Select a sequence between 10 and 15 residues in length. Longer sequences are unnecessary and increase the probability of synthetic problems. The presence of internal cysteine residues are not a concern with MAPs, but if present, take precautions to keep them reduced.

2. Synthesize the MAPs utilizing a four-branch core. Synthesize the MAP core de novo, or purchase resins for solid-phase peptide synthesis with four- or eight-branched cores (available commercially from Advanced ChemTech, Novabiochem, Applied Biosystems, AnaSpec, and Bachem Bioscience). Eight-branched cores are suitable if the peptide is no more than 12 to 14 residues and has a high degree of hydrophilicity. There are more synthetic problems with eight-branched MAPs, presumably due to the higher density of structure during synthesis: they are more difficult to characterize and probably raise a diverse antibody population against some synthetic artifacts (see Mints et al., 1997).

3. Optional. If the selected sequence was not the amino terminus of the protein, acetylate the new amino terminus. In the case of a MAP, the carboxyl terminus will remain in covalent linkage to the branched core.

4. Use the MAP directly as an immunogen. Coupling to a carrier protein as described in Basic Protocol 3 is usually not necessary. BASIC PROTOCOL 3

COUPLING SYNTHETIC PEPTIDES TO A CARRIER PROTEIN USING A HETEROBIFUNCTIONAL REAGENT If the synthetic peptide was designed with a cysteine residue at one terminus (see Basic Protocol 2, step 2a), the following procedure should be followed for coupling to keyhole limpet hemocyanin (KLH) or other carrier proteins. Care must be taken to assure that the cysteine sulfhydryl group has remained reduced. Under normal synthetic conditions, if the peptide was lyophilized and stored dry immediately after synthesis, the sulfhydryl usually remains in the reduced state. The presence of free sulfhydryl groups in the peptide can be determined with Ellman’s reagent (see Support Protocol 1) just prior to use; alternatively, high-resolution mass spectrometry can be used. If reduction is needed, follow the cysteine reduction procedure (see Support Protocol 2) before starting the coupling process.


The reagent most commonly used for this purpose is m-maleimidobenzoyl-N-hydroxysuccinimide ester (MBS). However, several related reagents (all available from Pierce) offer some additional features. Sulfo-MBS (Pierce) is a water-soluble alternative to MBS. Succinimidyl 4-(N-maleimidomethyl)-cyclohexane-1-carboxylate (SMCC) and its sulfonated analog sulfo-SMCC provide the same chemistry with a more pH-stable maleimide (see step 1). The MBS and SMCC reagents can be used interchangeably in this protocol (the sulfo reagents can be dissolved in aqueous solution, while the others must be dissolved in an organic solvent; the concentrations listed are appropriate for all four).



Materials Keyhole limpet hemocyanin (KLH; Pierce, Sigma, Calbiochem, or Boehringer Mannheim) 0.01 M sodium phosphate buffer, pH 7.5 (APPENDIX 2E) 10 mg/ml MBS in fresh N,N-dimethylformamide (DMF) 0.05 M and 0.1 M sodium phosphate buffer, pH 7.0 (APPENDIX 2E) Synthetic peptide with a reduced cysteine residue at either the N- or C-terminus 6 M guanidine⋅HCl (see recipe) Small glass vial with flat bottom ∼0.9 × 15–cm gel filtration column with Sephadex G-25 or G-50 (Pharmacia Biotech) or Bio-Gel P2 or P4 (Bio-Rad) resin; or prepacked PD-10 column (Pharmacia Biotech) Additional reagents and equipment for gel filtration chromatography (UNIT 8.3) NOTE: Do not use Tris or other buffers with primary amino groups in this procedure. CAUTION: MBS is a moisture-sensitive irritant. Read the Material Safety Data Sheet before use. 1. Dissolve 5 mg KLH in ∼0.5 ml of 0.01 M sodium phosphate buffer, pH 7.5, in a small, flat-bottomed vial. A pH range of 7.0 to 7.5 offsets competing reactions. Although the unprotonated form of the amine reacts with the N-hydroxysuccinimide ester and would be optimal at pH >8.0, hydrolysis of the ester bond and reaction of the maleimide group with amines is enhanced at higher pH.

2. Add 100 µl of 10 mg/ml MBS/DMF solution and stir gently with a micro stir-bar 30 min at room temperature. A small amount of precipitate may form during this procedure and is acceptable. However, if the precipitate is large, perform the procedure again with fresh components. As an alternative to performing this coupling procedure from scratch, it is possible to purchase MBS-activated KLH (Pierce; Boehringer Mannheim) or kits containing MBS-activated KLH and an alternate MBS-activated protein for use in ELISA assays (Pierce).

3. Separate MBS-activated KLH from free MBS on a ∼0.9 × 15–cm gel filtration column, equilibrating and eluting the column with 0.05 M sodium phosphate buffer, pH 7.0. Collect 0.5-ml fractions and read their absorbance at 280 nm (UNIT 8.3). The first peak to elute is the KLH-MBS conjugate. These fractions may appear cloudy. The second peak is uncoupled MBS.

4. Pool the KLH-MBS conjugate fractions in a separate tube. 5. Dissolve 5 mg of the synthetic peptide in 0.01 M sodium phosphate buffer, pH 7.0, immediately prior to use. If the peptide is poorly soluble, use 6 M guanidine⋅HCl. Maleimide groups react specifically with sulfhydryls at slightly acid to neutral pH.

6. Add the peptide solution to the KLH-MBS conjugate. Stir gently with a micro stir-bar 3 hr at room temperature. The coupling may be continued overnight.

7. Dialyze against 4 liters distilled water overnight at 4°C. Use for immunizations within 24 hr. Preparation and Handling of Peptides


Supplement 28


18.3.10

Supplement 28


O

O

N O C

O

NO2

OH (DTT)

OH

SH

COOH (Ellman's Reagent)

S S

O

HS CH2 CH CH CH2 SH

HOOC

O2N

C OH

O

N

O

R1 NH

OH

HOOC

O2N

R2

N

C O C

O

KLH

N OH

O

O

N

O

KLH

S S

NH2

OH

SH

S CH2 H C C H OH

S H2C OH

HOOC

O N C H

SH

KLH

KLH

(A = 412 nm)

C N H

O

O2N

SH

C R1 N N R2 H H

O

N C (CH2)3 C N H H

O

N C H

S SCH2 CH CH CH2 SH

O

O

KLH

O

N

O

O

S

Figure 18.3.4 Representations of the chemistries described in this unit. (A) Cross-linking with MBS; (B) cross-linking with glutaraldehyde; (C) cross-linking with EDC; (D) reaction of free sulfhydryls with Ellman’s reagent to produce a colorimetric product; (E) reduction of disulfides with DTT. Abbreviations: KLH, keyhole limpet hemocyanin; MBS, m-maleimidobenzoyl-N-hydroxysuccinimide ester; EDC, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide; DTT, N,N-dithiothreitol.

S

S

E

SH

D

O

O

NH2 HC (CH2)3 CH H2N

NH2

R1 N C N R2 (EDC)

C

KLH

B

KLH

A

ASSAY OF FREE SULFHYDRYLS WITH ELLMAN’S REAGENT Free sulfhydryl content of a peptide can be quantitatively determined with Ellman’s reagent, 5,5′-dithio-bis(2-nitrobenzoic acid). The molar extinction coefficient at 412 nm of thionitrobenzoate, the colored species generated when the reagent reacts with a free thiol, is 14,150 in 0.1 M sodium phosphate buffer (see Fig. 18.3.4). The sensitivity of the reaction is in the low nmol/ml range for sulfhydryl groups, making it well suited for synthetic peptides. By dry weight most synthetic peptides are only ∼60% to 75% peptide, the remainder consisting of counterions and water of hydration. Amino acid analysis (see UNIT 11.9) is needed to establish the actual peptide content unambiguously, but such precise measurement is not usually necessary for qualitative evaluation of the free sulfhydryl content of a peptide sample.

SUPPORT PROTOCOL 1

Materials Cysteine standard stock solution (see recipe) 0.1 M sodium phosphate, pH 8.0 (APPENDIX 2E) Peptide to be assayed Ellman’s reagent solution (see recipe) 13 × 100–mm glass test tubes 1. Prepare a cysteine standard curve by adding 25 µl, 50 µl, 100 µl, 150 µl, 200 µl, and 250 µl of cysteine standard stock solution to separate 13 × 100–mm tubes. Add ≤250 µl of each peptide to be tested to separate tubes. Bring the volume in each tube to 250 µl with 0.1 M sodium phosphate, pH 8.0. Add 250 µl of 0.1 M sodium phosphate, pH 8.0, to a blank tube. The cysteine content of the peptide to be assayed should fall within the range of the standard curve (37.5 to 375 nmol).

2. Add 50 µl Ellman’s reagent solution and 2.5 ml of 0.1 M sodium phosphate, pH 8.0, to each tube. Mix and incubate 15 min at room temperature. 3. Measure absorbance at 412 nm (A412). 4. Plot the A412 values of the standards after subtracting the value for the blank to produce a standard curve. Use this curve to determine the free sulfhydryl content of the peptides. REDUCING CYSTEINE GROUPS IN PEPTIDES When a peptide is synthesized with a terminal cysteine residue to be used for coupling with MBS (see Basic Protocol 3), the cysteine must be in the reduced state (present as free -SH rather than as a disulfide) in order to participate in the reaction with the coupling reagent. If peptides are lyophilized immediately after extraction from the resin cleavage cocktail or reversed-phase HPLC and used immediately after reconstitution, oxidation of cysteine side chains is usually not a problem. However, if oxidation to disulfides has occurred, the peptide can be reduced prior to use with the protocol presented here.

SUPPORT PROTOCOL 2

Dithiothreitol (DTT) is preferred to 2-mercaptoethanol (2-ME) as a reducing agent because its lower redox potential allows it to be effective at lower concentrations, and the reaction goes to completion because formation of the six-membered ring containing an internal disulfide is energetically favorable (see Fig. 18.3.4; also see APPENDIX 3A). To determine if reduction is necessary, quantitate the level of free sulfhydryl groups with Ellman’s reagent (see Support Protocol 1). Additional methods for reducing disulfides include using sodium borohydride (Gailit, 1993) and Tris(2-carboxyethyl)phosphine (TCEP; Getz et al., 1999).



Supplement 28

Materials Synthetic peptide 0.1 M sodium phosphate, pH 8.0 (APPENDIX 2E) 1 M aqueous dithiothreitol (DTT) 1 N HCl 100- or 250-µl polypropylene tubes Nitrogen gas source Additional reagents and equipment for reversed-phase HPLC of peptides (see UNIT 11.6) 1. Dissolve 5 to 10 mg of peptide in 0.1 M sodium phosphate, pH 8.0. 2. Add 100 µl of 1 M DTT. 3. Flush nitrogen over the surface of the liquid, seal the tube, and incubate 1 hr at 37°C. 4. Acidify with 1 N HCl and desalt by reversed-phase HPLC (UNIT 11.6). 5. Pool peptide fractions and lyophilize. Store lyophilized at 4°C until ready to use (up to several days). The oxidation state of the peptide can usually be followed by analytical monitoring of its elution position on reversed-phase HPLC. Disulfide-linked dimers of peptides generally elute later than the monomeric peptide. ALTERNATE PROTOCOL 3

COUPLING SYNTHETIC PEPTIDES TO A CARRIER PROTEIN USING A HOMOBIFUNCTIONAL REAGENT The available homobifunctional reagents couple compounds through primary amino groups. Therefore, peptides with internal lysine residues should not be used in this procedure. The reagent most commonly used for this procedure is glutaraldehyde, but it should not be used with peptides containing internal Cys, Tyr, or His residues. Other homobifunctional cross-linking reagents that can be used in the same way as glutaraldehyde, but do not cross react with Cys, Tyr, or His residues, are also available: disuccinimidyl suberate (DSS), disuccinimidyl glutarate (DSG), and bis(sulfosuccinimidyl) suberate (BS3; all available from Pierce). However, these reagents are not widely employed for coupling peptides to proteins and are not considered the method of choice, and methods for their use have not been formalized. This is probably because coupling of the synthetic peptide to itself and aggregation of the carrier protein can occur with homobifunctional reagents such as these. Glutaraldehyde, on the other hand, although also subject to this limitation, has generally been used successfully. Additional Materials (also see Basic Protocol 3) 50 mM sodium borate buffer, pH 8.0 (pH adjusted with HCl) Glutaraldehyde solution (see recipe) 1 M glycine in 50 mM sodium borate buffer, pH 8.0 NOTE: Do not use Tris or other buffers with primary amino groups in this procedure. 1. Dissolve 5 mg KLH in ∼1.0 ml of 50 mM sodium borate buffer, pH 8.0. 2. Add 5 mg synthetic peptide.


3. Slowly add 1 ml fresh glutaraldehyde solution with gentle mixing at room temperature. Allow to react for an additional 2 hr with gentle mixing. Formation of a yellowish color or milkiness is normal and does not affect the sample.



4. Add 0.25 ml of 1 M glycine to bind unreacted glutaraldehyde. A darker yellow to brown color may develop.

5. Dialyze the reaction mixture overnight at 4°C against 4 liters of 50 mM sodium borate buffer, pH 8.0, and then overnight against water. Use immediately. COUPLING SYNTHETIC PEPTIDES TO A CARRIER PROTEIN USING A CARBODIIMIDE


This procedure couples amino groups to carboxyl groups by way of activation of the carboxyl group with a water-soluble carbodiimide. Since the procedure is most easily performed in one step, peptides containing internal Asp, Glu, Lys, Tyr, or Cys residues should not be used. Also, in order to avoid making polymers of the peptide, the amino terminus should be blocked by acetylation during synthesis (see Basic Protocol 2). While this method has been used successfully, it is not considered to be the method of choice except in special situations. Additional Materials (also see Basic Protocol 3) 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC; Pierce), used fresh or stored desiccated and frozen 0.1 N HCl NOTE: Buffers containing amino or carboxyl groups should not be used in this procedure. According to some reports, buffers containing phosphate groups should also be avoided. Water is the safest choice as a solvent. 1. Dissolve 5 mg synthetic peptide in 1 ml water. 2. Add 25 mg EDC and carefully adjust pH to 4.0 to 5.0 by adding small amounts of 0.1 N HCl. Allow to react for 5 to 10 min at room temperature with gentle mixing. pH paper suffices to monitor this adjustment.

3. Dissolve 5 mg KLH in 0.5 ml water and add to solution from step 2. React 2 hr at room temperature with gentle mixing. 4. Dialyze against 4 liters of water overnight at 4°C. Use immediately. COUPLING SYNTHETIC PEPTIDES TO A CARRIER PROTEIN PHOTOCHEMICALLY


Most synthetic peptides today are made with 9-fluorenylmethyloxycarbonyl (Fmoc) chemistry (see Fig. 18.1.1), and therefore the following protocol links the photoactive group to the free amino terminus of the peptide. An alternative approach if t-butyloxycarbonyl (Boc) chemistry is used is to link it to an ε-amino group of a terminal lysine (Gorka et al., 1989). Additional Materials (see also Basic Protocol 3) 4-benzoyl benzoic acid (Sigma or Aldrich) Quartz spectrophotometry cuvettes 1. Have the synthetic chemist who is making the peptide attach a benzoyl benzoic acid group to the amino terminus simply by treating the reagent as the terminal residue during normal synthesis. This blocks the amino terminus.



Supplement 17

2. Dissolve 5 mg KLH in ∼0.5 ml of 0.01 M sodium phosphate buffer, pH 7.5. 3. Add 2 mg of synthetic peptide containing the benzoyl benzoate adduct. 4. Place in a 1-cm quartz cuvette and irradiate with 366-nm light for 3 hr at a distance of 0.5 cm. Use immediately. The peptide can be used directly for immunization. SUPPORT PROTOCOL 3

CALCULATION OF THE MOLAR RATIO OF PEPTIDE TO CARRIER PROTEIN The molar ratio of peptide to carrier protein coupling efficiency can be calculated to determine the level of substitution achieved by the coupling procedure. This information can be obtained using the results of amino acid compositional analysis. By performing the calculations presented in this protocol, the molecules of peptide in the conjugate per molecule of carrier protein in the conjugate can be determined. 1. Obtain the amino acid composition of the carrier protein, the peptide, and the peptide/carrier conjugate. Amino acid compositional analysis (of these hydrolysates) is usually available at sources that provide automated peptide synthesis (see UNITS 18.1 & 18.2). Be sure that the conjugate is free of unconjugated peptide (i.e., it should be well dialyzed). 2. Determine a scaling factor (SF) that relates the moles of protein in the unconjugated carrier protein to the moles of protein in the peptide/carrier conjugate. This is done by comparing the molar ratio of ≥3 amino acids present in the carrier protein and peptide/carrier conjugate but not present in the peptide. For example, if the peptide TGLRDSC (Table 18.3.2) is coupled to a carrier protein, choose A, K, and I. The calculation is done as: Table 18.3.2 Sample Calculation of the Extent of Coupling of the Peptide TGLRDSC to Carrier Proteina

Amino acid


D E G S T H P A M V F L I C Y K R Total pmol amino acid

Composition of carrier protein

Composition of peptide/carrier conjugate

Amount of carrier protein amino acids in conjugate

Amount of peptide amino acids in conjugate

80 110 95 65 70 10 25 103 5 60 22 55 65 7 13 65 75

185 222 215 150 163 19 51 206 11 118 45 133 135 22 25 125 177

160 220 190 130 140 20 50 206 10 120 44 110 130 14 26 130 150

25 — 25 20 23 — — — — — — 23 — 8 — — 27

925

2002

1850

151



SF = [(pmol A in conjugate/pmol A in carrier) + (pmol K in conjugate/pmol K in carrier) + (pmol I in conjugate/pmol I in carrier)]/3. For these amino acids, the carrier protein yields are as follows: A = 103 pmol, K = 65 pmol, and I = 65 pmol. For these same amino acids, the peptide/carrier conjugate yields: A = 206 pmol, K = 125 pmol, and I = 135 pmol. From these values, the relative amount of carrier protein in the conjugate versus the unconjugated carrier protein (SF) can be calculated as follows: (206/103 + 125/65 + 135/65)/3 = 2.0, indicating that there is twice as much carrier protein in the peptide/carrier conjugate hydrolysate as in the carrier-protein hydrolysate. 3. Calculate the moles of peptide present in the conjugate by subtracting the moles of amino acid present in the carrier from the moles of amino acid present in the conjugate. Choose ≥3 amino acids present in the peptide. The relative amount (SF) of protein present in the carrier protein versus the amount in the conjugate as calculated in step 2, must also be considered as follows: pmol peptide in conjugate = {[pmol G in conjugate − (SF × pmol G in carrier)]+ [pmol L in conjugate − (SF × pmol L in carrier)]+ [pmol R in conjugate − (SF × pmol R in carrier)]}/3

Therefore, the amount of peptide in the conjugate hydrolysate for the example shown in Table 9.4.1, calculated using the amino acids G, L, and R, is {[215 − (2 × 95)] + [133 − (2 × 55)] + [177 − (2 × 75)]}/3 = 25 pmol. 4. Calculate the number of moles of protein in the conjugate hydrolysate as follows:

pmol carrier protein in conjugate =

(total pmol carrier protein amino acids) × 110 molecular weight of carrier protein

where total pmol carrier protein amino acids = SF × (total amino acid composition of carrier in pmol) and 110 is the average molecular weight of an amino acid. In this example, there are 1850 pmol of carrier protein amino acids in the conjugate; therefore, 1850 pmol × (110/100,000) = 2.04 pmol carrier protein in conjugate. 5. Determine the ratio of peptide to carrier protein as follows: molecules peptide in conjugate/molecules carrier protein in conjugate = pmol peptide in conjugate/pmol carrier protein in conjugate. Using the values calculated in steps 3 and 4, the result is: 25 pmol peptide in conjugate/2.04 pmol carrier protein in conjugate = 12.2 molecules peptide in conjugate per molecule carrier protein in conjugate.



Supplement 28


Cysteine standard stock solution Dissolve 26.3 mg cysteine hydrochloride monohydrate in 100 ml of 0.1 M sodium phosphate, pH 8.0 (APPENDIX 2E). Prepare immediately before use. Ellman’s reagent solution Dissolve 4 mg Ellman’s reagent, 5,5′-dithio-bis-(2-nitrobenzoic acid) (Pierce), in 1 ml of 0.1 M sodium phosphate, pH 8.0 (APPENDIX 2E). Prepare immediately before use. Glutaraldehyde solution, 0.15% Add 30 µl of 25% aqueous glutaraldehyde solution to 5 ml of 50 mM sodium borate buffer, pH 8.0 (pH adjusted with HCl). Prepare fresh and use immediately. If the glutaraldehyde precipitates, check the pH. It should not be above 8.0; a slightly lower pH can be used (pH 7 to 8). CAUTION: Glutaraldehyde is a sensitizing agent that should be handled in a hood and only according to the recommendations in the Material Safety Data Sheet. When mixing solutions or performing reactions, keep the container covered to prevent vapors from escaping into the atmosphere.

Guanidine⋅HCl, 6 M Dissolve 1 g guanidine⋅HCl in 1 ml of 0.05 M sodium phosphate, pH 7.0 (APPENDIX 2E). Store up to several weeks at room temperature. The resulting 1.8-ml solution should be ∼0.025 M phosphate/6 M guanidine⋅HCl at pH 7.0.



Synthetic peptides are linear arrays of amino acids that in most instances possess a random structure in solution. While it is not difficult to produce antipeptide antibodies, it does not necessarily follow that the antibodies will recognize a protein containing the same stretch of sequence found in the peptide. In order for this to occur, the amino acids in the protein must be oriented to the antibody in a way similar to that of the synthetic peptide. This generally requires three basic features of the protein: (1) that the stretch of sequence be exposed to solvent; (2) that the sequence be a continuous stretch of amino acids; and (3) that it not possess a higherorder structure that renders it unrecognizable by the antibody population. The large number of model protein structures now available indicate that almost all of the ionized groups in water-soluble proteins are on the protein surface. Asp, Glu, Lys, and Arg residues, on the average, comprise 27% of the protein surface and only ∼4% of the protein interior. The fraction of residues that are at least 95% buried range from 0.36 to 0.60 for nonpolar residues and 0.01 to 0.23 for polar residues. Only 1% of Arg and 3% of Lys residues fall

into the 95% buried range (Creighton, 1993). Therefore, it is reasonable to expect solvent-exposed areas of proteins to display relatively high levels of polar and charged residues, particularly Arg and Lys. Proteins display three kinds of secondary structure: α-helices, β-sheets, and turns or loops (see UNIT 17.1). Turns or loops generally connect elements of α-helices and β-sheets, and can either fit one of several rather strict motifs with recognizable hydrogen bonding patterns or be of a more extended, random nature. These turn or loop structures appear to be most useful for antibody production because they tend to be found on the surface of proteins connecting larger arrays of helices and sheets, and they consist of continuous stretches of amino acids. Although many amino acid residues in helices and sheets are also exposed at the surface, the regular geometry of amino acids contained within them makes them less suitable for this purpose. For instance, in β-sheet structures the side chain of each successive amino acid in the β-sheet strand points in the opposite direction to the ones immediately preceding and following it. Thus, even if the amino acid side chains are not predominantly buried in the interior of



the protein, only every other side chain is exposed on the same surface of the sheet. This can hinder recognition by an antibody produced with a linear peptide capable of assuming a more random structure. A similar situation exists for α-helices. Although the change in direction of the side chains of successive amino acids is perhaps not as abrupt as in β-sheets, only approximately every third or fourth side chain is found on the same surface of the helix. Epitopes in proteins have been identified in amphipathic helices, but unless the synthetic peptide assumes a similar helical structure in solution, recognition by the antibody may be problematic. These considerations have led to more useful methods for predicting sequences that will produce antibodies recognizing intact proteins. A variety of different indices that predict hydrophilicity or hydrophobicity and secondary structure are available (see UNITS 2.2 & 2.3, respectively). In addition, predictive methods based on segmental mobility, side chain accessibility, and sequence variability (see Van Regenmortel et al., 1988) have also been proposed. All of these methods generally tend to yield similar results, but it must be noted that these procedures were developed for (and work best with) water-soluble proteins composed of a single globular structure. Additional complications can arise with multisubunit proteins, where normally exposed structures may be shielded by subunit interactions, or membrane proteins with large sections shielded from the solvent. The method presented in this unit utilizes the correlation between the hydrophilic character of a peptide sequence (Kyte and Doolittle, 1982) and its propensity to form β-turn structures (Chou and Fasman, 1974). Free access to these and many other algorithms is provided at the ExPASy Web site of the University of Geneva at http://expasy.org.tools. After selection of the peptide sequence, an effective immunogen is generally produced by coupling the peptide to a carrier protein or by synthesizing a multiple antigenic peptide (MAP), with four or eight identical peptides assembled simultaneously on the α and ε amines of the terminal lysines of a branched core (see Fig. 18.3.3).

Critical Parameters Analyzing protein sequences with algorithms or tables of assigned values for amino acids is a well-established procedure, but evaluating these results and selecting the candidate sequences requires some consideration. To take

full advantage of the results, choose areas of sequence that give the maximum values for the properties being evaluated and that also show the highest degree of residue-by-residue correlation. In other words, choose areas of maximum amplitude where the centers of the peaks correspond to the same sequence with a divergence of no more than two to three residues. Examples of this are given in Figure 18.3.2, which shows results from the method presented in Basic Protocol 1 for the sequence shown in Figure 18.3.1. The top panel in Figure 18.3.2 predicts β-turns as calculated by the method of Chou and Fasman (1974). The bottom panel is a prediction of hydrophobicity using the parameters of Kyte and Doolittle (1982). The data are analyzed by looking for areas of high turn propensity (maximum positive deflection in the top panel) and high hydrophilicity (maximum negative deflection in the bottom panel). The shaded areas in Figure 18.3.2 designate three segments that meet these criteria. Note that the maximum and minimum values of these three stretches of protein sequence correlate very well. Additional areas of high hydrophilicity (bottom panel) are found near residues 64, 132, 137, 149, and 345, although the β-turn values of these secondary candidates are not as high as those of the three shaded areas. Two equally hydrophilic areas at residues 49 and 299 correspond to downward deflections in the β-turn profile and are thus not good candidates based on this analysis. Many different chemistries are available for coupling synthetic peptides to carrier proteins to produce effective immunogens (Van Regenmortel et al., 1988). In many cases, however, side reactions or incompatibilities in chemistry between the coupling agent and the residues present in the peptide can be problematic. In order to simplify the process and present the greatest probability of success in most cases, only a few coupling methods are presented in this unit. In this regard, the recommended coupling procedure is cross-linking of the peptide via cysteine residues to keyhole limpet hemocyanin (KLH) with the heterobifunctional reagent m-maleimidobenzoyl-N-hydroxysuccinimide ester (MBS; see Basic Protocol 3). This effective method has enjoyed great success and can be used for virtually any peptide. The one caveat is that it is not recommended for peptides with internal cysteine residues, since they will also link to the carrier. It is also critical, when coupling with MBS through an added terminal cysteine residue, that the sulfhydryl group of



Supplement 28


the peptide be present in the free or reduced form (see Support Protocols 1 and 2). In addition to MBS coupling, other procedures commonly used (see Alternate Protocols 3 and 4) are included as alternatives for use in special situations, but these are not recommended as a general alternative to MBS because they are more restrictive and have the potential for undesirable side reactions. Glutaraldehyde coupling (see Alternate Protocol 3) should not be used with peptides containing internal Lys, Cys, Tyr, or His residues and, since it is a homobifunctional reagent, cross-linking of the peptide to itself and the carrier to itself can occur. The latter lowers antigenicity and can result in extensive aggregation and precipitation of the carrier. 1-ethyl-3(3-dimethylaminopropyl) carbodiimide (EDC; see Alternate Protocol 4) is a water-soluble carbodiimide and should not be used with peptides containing internal Lys, Glu, Asp, Tyr, or Cys residues. Alternate Protocol 5 describes a simple photochemical coupling strategy (Gorka et al., 1989). Another good alternative for most peptides is the production of a multiple antigenic peptide (MAP; see Alternate Protocol 2). With this method the composition of the peptide is not a concern beyond its potential solubility properties. In most cases, since hydrophilic sequences are selected, this also is not a major problem. Both four- and eight-branched MAPs have been found to be effective. However, four-branched MAPs are recommended because they are less prone to synthesis problems and are easier to characterize. As with any synthetic peptide, the product must be well characterized before use. If the peptide is not what it was intended to be, this decreases the probability of generating antibodies that will recognize the protein. At the very least, check synthetic peptides for homogeneity by analytical HPLC and correct mass by mass spectrometry (see Chapter 16). Characterization of MAP can be more problematic due to their multibranched nature (Mints et al., 1997): HPLC and mass spectrometric analysis can be compromised by the presence of four to eight peptide chains per molecule, each of which may have only a small percentage of modification at any particular residue but which in the aggregate contribute to broad spectra. However, this feature of MAPs usually does not tend to compromise their ability to form antigens of the proper peptide since the correct sequence is usually present in high enough concentration that a significant amount of specific antibody is produced among the polyclonal population. Amino acid analysis (UNIT 11.9), which is less sensitive to

multiple small differences, tends to give a reasonable assessment of the MAP integrity.

Anticipated Results The methods outlined in this unit produce an effective polyclonal antiserum against an intact protein from a single peptide sequence ∼50% to 70% of the time. Therefore, it is advisable to prepare two or three different peptides from a given protein to increase the probability of at least one of them being effective.

Time Considerations Computer-assisted analysis of a protein sequence and inspection of the data to select several candidate sequences takes from 5 to 30 min. Manual analysis of a protein sequence can take several hours but can certainly be accomplished in 50 residues from the N- or C-termini, then its chemical modification will be extremely difficult using a two-fragment ligation strategy, since this would require the chemical synthesis of a peptide >50 residues in length. This problem can be overcome by assembling the target protein from three polypeptide fragments using a sequential native ligation strategy. The basic method for sequential native ligation uses a N-terminal Cys polypeptide (C-terminal fragment), an α-thioester polypeptide (N-terminal fragment), and an Nα(methylsulfony)ethyloxycarbonyl-Cys, α-thioester polypeptide (central fragment). The procedure starts by ligating the middle fragment and the C-terminal fragment at pH 7.5 in the presence of thiol cofactors. The Nα(methylsulfony)ethyloxycarbonyl (Msc) group is then removed and the resulting polypeptide (central plus C-terminal fragment) is ligated with the N-terminal fragment to give the target protein, which is purified and refolded. The general approach is depicted in Figure 18.4.2. Materials Three polypeptide fragments to be ligated: Polypeptide with N-terminal Cys residue (see Support Protocols 1 and 2) Polypeptide with Nα(Msc)-Cys, α-thioester functionality (see Support Protocol 5) Polypeptide with α-thioester functionality (see Support Protocols 3 and 4) 1 M HCl Additional reagents and equipment for native chemical ligation of two polypeptides (see Basic Protocol 1)

Native Chemical Ligation of Polypeptides

1. Prepare the ligation buffer (see Basic Protocol 1, steps 1 and 2). Ligate the central [i.e., Nα(Msc)-Cys, α-thioester] and C-terminal (N-terminal Cys) polypeptide fragments [see Basic Protocol 1, steps 3 to 8, but use the Nα(Msc)-Cys, α-thioester polypeptide in the reaction mix in place of the α-thioester polypeptide].



2. When the ligation reaction is complete, remove the Nα-Msc protecting group by raising the pH of the crude ligation mixture to 13 with 1 M NaOH solution (see Basic Protocol 1, step 4). After 1 min, lower the pH to 5.0 to 7.0 with 1 M HCl (use ∼1.1 to 1.2 times the volume of 1 M NaOH required to raise the pH to 13). Usually, for a 1-ml ligation mixture, the pH can be raised by adding 75 µl of 1 M NaOH, then dropped by adding 95 µl of 1 M HCl. It is recommended that the pH be dropped by adding the acid all at once, instead of by titration, for reasons of speed. The presence of thiols improves the yield in the deprotection step, since they trap the still reactive methyl ethylenyl sulfone side product.

3. Analyze the crude reaction mixture (see Basic Protocol 1, steps 6 and 7) and examine to see if the Msc deprotection is complete. Purify the polypeptide fragment (see Basic Protocol 1, steps 9 to 11), which will consist of the central fragment ligated to the C-terminal fragment. 4. Starting with the lyophilized (central plus C-terminal) polypeptide fragment prepared in steps 1 to 3 above and the N-terminal polypeptide fragment (containing an α-thioester functionality), perform the ligation reaction to generate the target protein (see Basic Protocol 1, steps 3 to 10). 5. Purify and refold the target protein (see Basic Protocol 1, steps 11 to 15).

HS

HS

Msc-HN-Cys

peptide 2

CO-SR

+ H2N-Cys

– peptide 1 CO2

aqueous buffer, pH 7.5 thiophenol 0.5% benzyl mercaptan

ligation 1

HS

HS

Msc-HN-Cys peptide 2

CO-NH- Cys

removal of Msc group

HS

peptide 1

CO2–

pH 13 HS

H2N-HN-Cys peptide 2

CO-NH- Cys

peptide 1

CO2–

aqueous buffer, pH 7.5 thiophenol 0.5% benzyl mercaptan

ligation 2

+ H3N peptide peptide 33 HS + H3N-Cys

peptide 3

CO-NH-Cys peptide 22

CO-SR

HS CO-NH-Cys

peptide 1

CO2–

Figure 18.4.2 The principle of sequential native chemical ligation. The key to this approach is the reversible protection of the α-amino group of the central peptide fragment, thereby preventing self-reaction with the α-thioester moiety present in the same molecule. This can be accomplished with the base-labile Msc [Nα(methylsulfony)ethyloxycarbonyl] group, which can be easily removed by brief treatment with base after the first ligation step. The newly deprotected fragment is now ready for the next ligation step.



Supplement 15

SUPPORT PROTOCOL 1

CHEMICAL SYNTHESIS OF N-TERMINAL CYS-POLYPEPTIDES The chemical synthesis of N-terminal Cys-polypeptides can be easily carried out using standard SPPS with Boc or Fmoc Nα-protected amino acids and the appropriate resins (see Table 18.4.1 and UNIT 18.1). In both cases, the synthesis can be performed on automated solid-phase peptide synthesizers, which are available in the core facilities of many institutions. NOTE: Extreme care must be taken to avoid exposing N-terminal Cys-polypeptides to even trace amounts of carbonyl-containing compounds (e.g., acetone or formaldehyde). These chemicals react rapidly with the α-amino and thiol groups of the N-terminal Cys to give a very stable thiazolidine derivative which prevents the peptide from participating in subsequent native chemical ligation reactions. For the same reason, the groups benzyloxymethyl (Bom) and t-butyloxymethyl (Bum), used to protect the side-chain of His in Boc and Fmoc SPPS, respectively, should be avoided during the synthesis of these polypeptides since they release formaldehyde during the cleavage/deprotection step.

SUPPORT PROTOCOL 2

SUPPORT PROTOCOL 3


BIOSYNTHESIS OF N-TERMINAL CYS POLYPEPTIDES Biosynthesis represents a complementary approach to chemical synthesis (see Support Protocol 1) for obtaining N-terminal-Cys polypeptides, and is especially useful when the target polypeptide fragment is somewhat greater than 50 residues in length (i.e., too large to be chemically synthesized). Assuming that the cDNA for the gene is available, N-terminal Cys polypeptides can be obtained using a mutagenesis/proteolysis strategy (Erlandson et al., 1996). This involves constructing an expression vector in which a DNA sequence encoding the peptide Ile-Glu-Gly-Arg-Cys is inserted between a upstream affinity-purification handle (e.g., MBP, GST, or His tag) and the appropriate gene fragment of interest (for cloning and mutagenesis see UNIT 6.6). Following expression and affinity purification, the fusion protein is treated with Factor Xa to give the desired recombinant Cys-polypeptide, which can then be further purified if necessary. These recombinant Cys-polypeptides can be used in native chemical ligation reactions as per their synthetic counterparts. CHEMICAL SYNTHESIS OF α-THIOESTER POLYPEPTIDES This protocol describes the chemical synthesis of a 3-mercaptopropionamide α-thioester polypeptide on a 3-mercaptopropionamide-MBHA resin. This resin can be easily prepared from commercially available MBHA resin using a three-step solid-phase procedure (Fig. 18.4.3). The solid-phase synthesis of the polypeptide is achieved using Boc amino acid derivatives employing in situ neutralization/HBTU [2-(1H-benzotriazolyl)-1,1,3,3tetramethyluronium hexafluorophosphate] activation protocols for Boc-SPPS (Schnölzer et al., 1992). The corresponding α-thioester polypeptide, suitable for native chemical ligation, is obtained after cleavage with hydrogen fluoride and purification. Materials Methylbenzhydrylamine (MBHA) resin (Peptides International) Dimethylformamide (DMF, spectrophotometric grade; Fisher) 5% (v/v) diisopropylethylamine (DIEA, peptide synthesis grade; Perkin-Elmer Applied Biosystems) in DMF (store up to 1 month at room temperature) 97% 3-bromopropionic acid Dichloromethane (DCM, spectrophotometric grade; Fisher) 99% diisopropyl carbodiimide (DIPC, 99%; Aldrich) Diisopropylethylamine (DIEA, peptide synthesis grade; Perkin-Elmer Applied Biosystems)

continued



Br MBHA

NH2

CO2H, DIPC

O MBHA

NH

DCM

O

CH3-COSH, DIEA MBHA Br DMF

NH

NS

O MBHA

NH

S

CH3

OH, DIEA, DMF

O

Boc-AA-OSu MBHA

S-AA-Boc

O

NH

SH

DMF

Figure 18.4.3 Scheme showing the preparation of the Boc-aminoacyl-3-mercaptopropionamide-MBHA resin using solidphase transformations. Abbreviations: AA, aminoacyl; Boc, t-butyloxycarbonyl; CH3-COSH, thioacetic acid; DCM, dichloromethane; DIEA, diisopropylethylamine; DIPC, diisopropyl carbodiimide; DMF, dimethylformamide; MBHA, methylbenzhydrylamine; OSu, hydroxysuccinimide ester.

Ac2O/DIEA/DMF solution (see recipe) AcSH/DIEA/DMF solution (see recipe) BME/DIEA/DMF solution (see recipe) Boc–amino acyl–N-hydroxysuccinimide ester (Boc-AA-OSu; where AA is the first amino acid to be incorporated into the polypeptide; Bachem) Trifluoroacetic acid (TFA, BioGrade) Ninhydrin test reagents: monitor 1, monitor 2, and monitor 3 (Perkin-Elmer Applied Biosystems) HF/p-cresol solution (see recipe) Diethyl ether, cold 10% to 50% acetonitrile in H2O containing 0.1% TFA 15-ml manual peptide synthesis vessel (Peptides International) Black rubber tubing (1/4 in. i.d. × 5/8 in. o.d. × 3/16 in. wall thickness; Fisher), resistant to acids and organic solvents 2-liter side-arm flasks with rubber stoppers and glass tubing to fit Pasteur pipet containing glass wool for filtration 2-ml polypropylene column (Microcolumn X from Isolab) with Teflon stopcock 13 × 100–mm glass test tube 110°C heating block 60% ethanol HF cleavage apparatus (Peptides International) Additional material and equipment for solid-phase peptide synthesis (Schnölzer et al., 1992; UNIT 18.1), preparative C18 reversed-phase HPLC (see Basic Protocol 1, step 11), and ESI-MS (Chapter 16) Prepare the MBHA resin 1. Build a manual solid-phase peptide synthesis system (Fig. 18.4.4) consisting of a 15-ml manual peptide synthesis vessel attached by 12 in. of black rubber tubing through a rubber stopper to a 2-liter side-arm flask, which is in turn connected to a vacuum source.



Supplement 15

2. Place 1 g of MBHA resin (1.0 mmol/g) into the peptide synthesis vessel, add enough DMF to cover the dry resin (∼5 ml), and wait 30 min. This preswelling step is crucial for the success of any reaction carried out on a solid support, since it renders the functional groups on the polymer available for reaction.

3. Drain the resin. Cover the resin with 5% DIEA/DMF, leave for 1 min, then drain the resin. Repeat this process two more times, then perform three 20-sec flow washes with DMF—i.e., wash the resin with a continuous flow of DMF for 20 sec while keeping a constant volume of solvent above the resin bed (usually 3 to 4 mm), then drain the resin. A continuous flow wash is an extremely efficient way of exchanging solvent in a swollen polymer—much more so than the commonly used bulk wash (i.e., by adding solvent, mixing, and draining).

Prepare the symmetrical anhydride of 3-bromopropionic acid 4. Dissolve 1.222 g 3-bromopropionic acid (8 mmol) in a minimal volume of DCM (∼1 to 2 ml), then add 630 µl of 99% DIPC (4 mmol). Shake the solution vigorously and wait 10 min. A white precipitate corresponding to the diisopropyl urea will appear during the activation reaction. If no such precipitate forms, the reaction has not taken place and should be repeated. It is best to prepare the symmetrical anhydride fresh for each coupling reaction.

5. Filter the solution through a Pasteur pipet containing glass wool. It is very important to remove the diisopropyl urea by filtration before adding the symmetrical anhydride solution to the resin. The authors have found that if this step is not carried out the acylation reaction does not go to completion.

Couple the symmetrical anhydride of 3-bromopropionic acid to the MBHA resin 6. Add the filtered solution from step 5 to the MBHA resin in the peptide synthesis vessel. Without applying any suction, add 800 µl DIEA (4.5 mmol) and the minimal amount of DMF (1 to 2 ml) required to give a good slurry. 7. Leave the coupling reaction for 30 min with occasional stirring using a glass stirring rod.

solid-phase peptide reaction vessel vacuum resin stopcock

waste container


rubber tubing resistant to acids and organic solvents Figure 18.4.4 Suggested apparatus for the manual solid-phase synthesis of thioester peptides.



8. Drain the resin and wash thoroughly with three 20-sec DMF flow washes as in step 3. 9. Repeat the coupling process (steps 4 to 8) two more times. 10. Add 5 ml Ac2O/DIEA/DMF solution to the resin and wait 10 min. Drain the resin and wash three times with DMF as in step 3. This acetylation step ensures that the small amount of unreacted amine groups remaining on the MBHA resin cannot participate in any subsequent acylation steps.

Prepare the 3-mercaptopropionamide-MBHA resin 11. Add 5 ml AcSH/DIEA/DMF solution to the resin and wait 20 min. Drain the resin and wash three times with DMF as in step 3. Repeat this entire process (treatment with AcSH/DIEA/DMF solution, along with the washings) two more times. 12. Add 5 ml BME/DIEA/DMF solution to the resin and wait 20 min. Drain the resin and wash three times with DMF as in step 3. Repeat this entire process (treatment with BME/DIEA/DMF solution, along with the washings) two more times. The 3-mercaptopropionamide-MBHA resin which is the product of this step is susceptible to oxidation upon long-term storage. The authors therefore recommend that it be immediately acylated with the appropriate Boc amino acid derivative.

Couple the first Boc amino acid to the 3-mercaptopropionamide-MBHA resin 13. Dissolve 3 molar equivalents of the appropriate Boc-AA-OSu (where AA is the first amino acid to be incorporated in the synthesis) in ∼6 ml DMF, then add this solution to the 3-mercaptopropionamide-MBHA resin. Add 714 µl DIEA (4 molar equivalents, 4 mmol) and leave the coupling reaction for 3 to 4 hr with occasional stirring. Most of the Boc amino acid derivatives are commercially available as N-hydroxysuccinimide esters. If the required Boc amino acid derivative is not commercially available it can be readily prepared manually (Bodanszky and Bodanszky, 1994).

14. Drain the resin and wash three times with DMF as in step 3. Add 5 ml Ac2O/DIEA/DMF solution to the resin and wait 10 min, then wash the resin three times with dichloromethane (DCM) using 20-sec flow washes as described in step 3. Dry the resin under vacuum and store at −20°C. Determine the final substitution of the Boc-amino acyl-3-mercaptopropionamideMBHA resin 15. Place ∼5 mg of the dry Boc–amino acyl–3-mercaptopropionamide–MBHA resin from step 14 in a 2-ml polypropylene Microcolumn X column with a Teflon stopcock, connected to a vacuum source. 16. Add 1 ml TFA to the resin and wait 2 min. Drain the column, then flow wash first with DMF and then with DCM. Drain again and dry under vacuum. 17. Weigh an exact amount of the dried resin (∼3 to 5 mg) and place in a 13 × 100–mm glass test tube. 18. Add 2 drops of ninhydrin test monitor 1 reagent, 4 drops of ninhydrin test monitor 2 reagent, and 2 drops of ninhydrin test monitor 3 reagent, then incubate 5 min at 110°C in a heating block. 19. Dilute the blue solution with 60% ethanol to 25 ml and mix well. Measure the absorbance at 570 nm using 1-cm path-length cuvette. Calculate the substitution of the thioester resin (in mmol/g) as 1.67 × (A570/amount of resin in mg).



Supplement 15

For a resin with an initial loading of ∼1 mmol/g the final substitution for the corresponding acylated thioester resin is usually ∼0.2 mmol/g.

Perform solid-phase synthesis of the α-thioester polypeptide 20. Synthesize the polypeptide sequence using the in situ neutralization/HBTU activation protocols for Boc-SPPS (Schnölzer et al., 1992; UNIT 18.1). Solid-phase deprotection of the 2,6-dinitrophenyl (Dnp) and formyl (For) protecting groups (commonly used for the protection of the side chains of His and Trp residues, respectively) results in premature cleavage of α-thioester polypeptides from the resin. Therefore, these residues should either be deprotected after the ligation reaction is complete or incorporated as Boc-Trp-OH (no side-chain protection) or Boc-Trp(Hoc)-OH and Boc-His(Bom)-OH during SPPS .

21. Once the synthesis is complete, flow wash with DMF and DCM as in step 14. Dry the resin under vacuum and store at −20°C. Cleave and purify the α-thioester polypeptide 22. Using 200 mg of peptide α-thioester resin from step 21, cleave the peptide-resin by treating with 5 ml of HF:p-cresol solution for 1 hr at 4°C in an HF cleavage apparatus. CAUTION: Anhydrous HF is a highly toxic and corrosive gas and should only be manipulated in a fume hood using the commercially available specialized apparatus.

23. Remove the HF under vacuum and resuspend both peptide and resin in ∼40 ml cold diethyl ether with gentle stirring for 10 min. Filter the suspension on a glass Buchner funnel under vacuum, without letting air pass through the filter, as this could oxidize the cleaved peptide. 24. Wash the material in the filter (containing the cleaved peptide and the resin) three times, each time with 10 ml cold diethyl ether. Wash an additional three times, each time with 10 ml dichloromethane, then wash with 10 ml cold diethyl ether again. Discard the wash liquid. These washes will remove most of the scavenger byproducts.

25. Add 10 ml freshly degassed 50% acetonitrile in water containing 0.1% TFA to the filter, wait 10 min (to dissolve the cleaved peptide), then filter. Repeat this process three more times, then recover and lyophilize the filtrates. 26. Analyze a 50-µl aliquot by reversed-phase HPLC and ESI-MS (Chapter 16; also see Basic Protocol 1, step 7). 27. Purify the lyophilized propionamide α-thioester peptide by preparative C18 RPHPLC (see Basic Protocol 1, step 11). SUPPORT PROTOCOL 4


BACTERIAL EXPRESSION OF α-THIOESTER POLYPEPTIDES This protocol describes the preparation of α-thioester polypeptides using bacterial expression in E. coli. The desired polypeptide-encoding gene fragment is first cloned into a commercially available pTYB vector (Fig. 18.4.5). Following E. coli transformation, soluble expression of the polypeptide-intein-CBD (chitin binding domain) fusion protein is induced and the cells are harvested and lysed. The lysate is then loaded onto a chitin column and the fusion protein affinity purified. Finally, the target α-thioester polypeptide is cleaved from the column and eluted using a buffer containing ethanethiol. NOTE: Initial studies using an MBP-intein-CBD system indicate that the majority of amino acid residues, when located immediately before the intein N-terminal cysteine, allow both purification of fusion proteins and efficient cleavage with thiols. However, Pro,



Figure 18.4.5 pTYB1, one of the series of pTYB vectors useful for the cloning, expression and purification of recombinant α-thioesters polypeptides. (A) pTYB1 vector; (B) sequence of pTYB1 cloning/expression region. This vector uses a bacteriophage T7 promoter-driven system (see UNIT 5.1 and Ausubel et al., 1999). The target gene is cloned into the multiple cloning site (MCS) polylinker region to create an in-frame fusion between the C-terminus of the target gene and the N-terminus of the gene encoding a modified intein. The DNA encoding a small chitin binding domain (CBD) has been also added to the C-terminus of the intein gene to allow the resulting fusion protein to be purified by affinity chromatography. Following purification, the target α-thioester polypeptide is obtained through a transthioesterification reaction by treating the fusion protein with ethanethiol. The pTYB vectors are available from New England Biolabs. Preparation and Handling of Peptides


Supplement 15

Cys, and Asn residues were found to inhibit the in vitro cleavage with thiols. Note that partial cleavage during bacterial expression can sometimes be observed in certain systems, resulting in a decrease in the yield of fusion precursors. For example, in the MBP system, Asp, Arg, His, and Glu residues gave rise to in vivo cleavage (>50%). Materials Gene construct encoding target polypeptide pTYB1 vector (New England Biolabs) E. coli BL21 (or any other suitable E. coli strain) LB medium containing 100 µg/ml ampicillin (APPENDIX 4A) 1 M isopropyl-1-thio-β-D-galactopyranoside (IPTG), filter sterilized Chitin beads slurry: 50% (w/v) suspension of chitin beads (New England Biolabs) in 40% ethanol Column buffer (see recipe) Lysis buffer (see recipe) Cleavage buffer (see recipe) Refrigerated centrifuge with Beckman JA-10 rotor and centrifuge buckets to hold 1 liter of culture 1.5 × 10–cm glass or polypropylene column Refrigerated centrifuge with Beckman JA-17 rotor and appropriate centrifuge tubes Shaker Additional reagents and equipment for introducing plasmid vectors into bacterial cells (APPENDIX 4D), growth of bacteria in liquid medium (APPENDIX 4A), lysis of bacterial cells using a French press (UNIT 6.3), ESI-MS (UNITS 16.1 & 16.2) Express the fusion protein 1. Insert the gene construct encoding the target polypeptide into the pTYB1 vector (see manufacturer’s instruction for vector and Ausubel et al., 1999) so as to express the target protein as an intein-CBD fusion. Introduce the vector into the E. coli BL21 cells (see APPENDIX 4D and Ausubel et al., 1999). Inoculate 15 ml LB/ampicillin with E. coli BL21 containing a pTYB vector expressing the desired protein fusion. Grow overnight with shaking at 37°C (APPENDIX 4A). 2. Inoculate 1000 ml LB/ampicillin with 10 ml of the overnight culture and grow with shaking at 37°C to an OD600 of 0.5 to 0.6 (APPENDIX 4A). 3. Add 1 ml of 1 M IPTG (final concentration of 1 mM) to the culture and continue incubation for 1 to 6 hr at 37°C. The optimal induction conditions (i.e., incubation time, temperature, and final IPTG concentration) for soluble expression will depend on the in vivo properties of the overexpressed protein, and should be optimized for every system.

4. Centrifuge the cell culture 10 min at 8700 × g (7000 rpm in Beckman JA-10 rotor), 4°C. Discard supernatant. Proceed to step 5 immediately or freeze pellet at −70°C indefinitely. If extract preparation is to be be carried out immediately, the chitin column may be prepared during the centrifugation step.

Prepare the affinity column 5. Add ∼10 ml of chitin bead slurry to a 1.5 × 10–cm column and allow liquid to drain just to the top of the packed resin bed (∼5 ml). Native Chemical Ligation of Polypeptides

6. Wash the column with 100 ml column buffer at a flow rate of 1 ml/min.



Prepare the cell extract 7. If cell pellet was frozen, (step 4), thaw on ice. Resuspend pellet in 30 ml ice-cold lysis buffer. Lyse cells using a French press (UNIT 6.3). IMPORTANT NOTE: From this step on, all procedures should be carried out at 4°C.

8. Centrifuge lysate for 30 min (or until supernatant is clear) at 25,000 × g (14,000 rpm in Beckman JA-17 rotor), 4°C. Decant supernatant into a clean container on ice and discard pellet. The supernatant can be frozen and stored at −70°C indefinitely before continuing with the procedure. In order to monitor the expression, extraction, and purification steps, it is convenient to take small aliquots at every step and analyze them by SDS-PAGE (UNIT 10.1).

Purify the fusion protein 9. If extract is frozen, thaw on ice. Load onto the chitin column (at a rate no faster than 0.5 ml/min). Collect flowthrough and reapply to column, then repeat this process one more time. Chitin beads have a capacity of ∼2 mg of CBD-tagged protein per ml packed beads. Therefore, the amount of extract that can be loaded on the column will depend on the amount of soluble fusion protein in the extract.

10. Wash the column with 100 ml column buffer at a flow rate of 1 ml/min. Discard flowthrough. Be sure that all traces of crude extract have been washed off the sides of the column. Obtain the α-thioester polypeptide 11. Add 5 ml cleavage buffer to the column and gently shake the reaction slurry overnight at room temperature. 12. Drain the beads, then wash with 15 ml cleavage buffer (three times, each time with 5 ml). Pool all of the fractions. IMPORTANT NOTE: Thioesters are susceptible to hydrolysis under alkaline conditions; consequently purification and storage of α-thioester polypeptides should be performed at pH 6.0 or below.

Analyze and purify the α-thioester polypeptide 13. Analyze a 50-µl aliquot by reversed-phase analytical HPLC and ESI-MS (see UNITS 16.1 & 16.2; also see Basic Protocol 1, step 7). 14. Purify the target α-thioester polypeptide by reversed-phase preparative HPLC see Basic Protocol 1, step 11). CHEMICAL SYNTHESIS OF Nα(Msc)-CYS, α-THIOESTER POLYPEPTIDES This protocol describes how to introduce the Nα(methylsulfony)ethyloxycarbonyl (Msc) protecting group onto the α-amino group of an N-terminal Cys polypeptide synthesized on a 3-mercaptopropionamide-MBHA resin (see Support Protocol 3). The resulting Nα(Msc)-Cys, α-thioester polypeptides are used in sequential native chemical ligation reactions (see Basic Protocol 2). NOTE: Boc-SPPS of α-thioester polypeptides requires the use of the Bom side-chain protecting group for His. The Bom group releases fomaldehyde during its deprotection with HF; however the formation of the thiazolidine adduct with the N-terminal Cys cannot take place due to the presence of the NαMsc group.

SUPPORT PROTOCOL 5



Supplement 15

Additional Materials (also see Support Protocol 3) Fully protected Boc-polypeptide-3-mercaptopropionamide-MBHA resin (see Support Protocol 3, step 21), dried (Methylsulfonyl)-ethyl 4-nitrophenyl carbonate (Msc-ONp; Fluka) Prepare the Boc-polypeptide-3-mercaptopropionamide-MBHA resin 1. Place ∼0.5 g of fully protected Boc-polypeptide-3-mercaptopropionamide-MBHA resin in a peptide synthesis vessel attached to a vacuum source and swell with DMF (see Support Protocol 3, steps 1 and 2). 2. Deprotect the Boc group by adding ∼5 ml of TFA to the resin, waiting 1 min, then draining the resin. Repeat this process one additional time, then drain the resin and perform three 20-sec flow washes with DMF (see Support Protocol 3, step 3). 3. Neutralize the polypeptide-3-mercaptopropionamide-MBHA resin by treating with 5% DIAE/DMF and flow washing with DMF (see Support Protocol 3, step 3). Introduce the Msc group on the α-amino group of the polypeptide-3-mercaptopropionamide-MBHA resin 4. Dissolve 560 mg (2 mmol) of Msc-ONp in 4 to 5 ml DMF and add this mixture to the peptide thioester resin, then add 340 µl (2 mmol) of DIEA. Leave the coupling reaction for 3 hr, with occasional stirring using a glass stirring rod, then drain the resin and perform three 20-sec flow washes with DMF. 5. Place ∼1 mg of resin from step 4, in a column (see Support Protocol 3, step 15). Flow wash first with DMF and then with DCM, drain again, and dry under vacuum. Place the aliquot in a glass test tube and perform the ninhydrin test (see Support Protocol 3, step 18). If the color of the solution is strongly blue, repeat the coupling reaction (step 4) and test again; if not, continue with step 6, below. Sometimes the ninhydrin test can give false positives (i.e., strong blue coloration) even although the acylation reaction is complete. This is especially true when the α-amino group is protected with a base-labile group (i.e., Fmoc or Msc). Therefore, if the ninhydrin test is still positive after the third coupling, proceed to step 6.

6. Perform three 20-sec flow washes with DMF and then with DCM. Dry the resin under vacuum and store at −20°C until further use. 7. Cleave and purify 200 mg of peptide-thioester resin (see Support Protocol 3, steps 22 to 26). REAGENTS AND SOLUTIONS Use Milli-Q-purified water or equivalent for the preparation of all buffers. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Ac2O/DIEA/DMF solution 15 parts (v/v) 99% acetic anhydride (Ac2O) 15 parts (v/v) diisopropylethylamine (DIEA, peptide synthesis grade; Perkin-Elmer Applied Biosystems) 70 parts (v/v) dimethylformamide (DMF, spectrophotometric grade; Fisher) Prepare fresh


AcSH/DIEA/DMF solution 1 part (v/v) 96% thiolacetic acid (AcSH; Fluka) 1 part (v/v) diisopropylethylamine (DIEA, peptide synthesis grade; Perkin-Elmer Applied Biosystems) 8 parts (v/v) dimethylformamide (DMF, spectrophotometric grade; Fisher) Prepare fresh



BME/DIEA/DMF solution 1 part (v/v) 98% 2-mercaptoethanol (BME) 1 part (v/v) diisopropylethylamine (DIEA, peptide synthesis grade; Perkin-Elmer Applied Biosystems) 8 parts (v/v) dimethylformamide (DMF, spectrophotometric grade; Fisher) Prepare fresh Cleavage buffer 0.1 mM EDTA 200 mM sodium phosphate, pH 6.0 250 mM NaCl 3% (v/v) ethanethiol 0.1% (v/v) Triton X-100 Adjust pH to 6.0 with 1 M NaOH or HCl Store up to 6 months at 4°C Column buffer 0.1 mM EDTA 20 mM sodium phosphate, pH 7.2 250 mM NaCl 0.1% (v/v) Triton X-100 Adjust pH to 7.2 with 1 M NaOH or HCl Store up to 6 months at 4°C Guanidine⋅HCl buffer, 6 M 6 M guanidine⋅HCl (APPENDIX 3A) 0.1 M sodium phosphate 1 mM EDTA Adjust pH to 7.5 with 1 M NaOH Store up to 6 months at room temperature HF/p-cresol solution 96 parts (v/v) hydrogen fluoride (HF; anhydrous) 4 parts (v/v) p-cresol Prepare fresh CAUTION: Anhydrous HF is a highly toxic and corrosive gas and should be manipulated only in a fume hood, using the commercially available specialized apparatus (HF cleavage apparatus from Peptides International).

Lysis buffer 0.1 mM EDTA 1 mM PMSF 25 mM HEPES, pH 8.0 250 mM NaCl 5% (v/v) glycerol Adjust pH to 8.0 with 1 M NaOH Store up to 6 months at 4°C COMMENTARY Background Information The introduction of solid-phase peptidesynthesis (SPPS) by Bruce Merrifield revolutionized the chemical synthesis of peptides (Merrifield, 1963; also see UNIT 18.1). Despite the enormous impact of SPPS in the generation

and study of small bioactive peptides, it is now clear that the combination of incomplete acylation/deprotection reactions and other well documented side reactions places an intrinsic limit on the size of peptides accessible by efficient stepwise SPPS. Thus, polypeptides of up



Supplement 15


to ∼50 residues in length can be prepared by stepwise SPPS with reasonable confidence, but beyond this, the chances of success fall off precipitously. In order to overcome this size limitation, recent years have seen renewed interest in the use of convergent synthetic strategies—i.e., the synthesis of large polypeptides from smaller peptide building blocks which are themselves accessible via the SPPS approach. It should be noted that fragment condensation has a long and illustrious history in the peptide chemistry field; however, there have always been serious problems associated with the manipulation of the fully protected peptides in these classical convergent strategies. Although the use of minimal protection strategies represented a step in the right direction (see LloydWilliams et al., 1993 for an extensive review), it was not until the early 1990s that a truly practical way of performing fragment condensations was developed—i.e., chemical ligation (for reviews see Muir, 1995; Wallace, 1995). The original chemical ligation strategies were all based on the premise that an unnatural moiety could be used to covalently join two fully unprotected (and hence water-soluble) polypeptides, each bearing unique and mutually reactive groups. A number of different chemistries have been developed for this purpose, all of which give rise to an unnatural covalent structure at the ligation site (Muir, 1995; Wallace, 1995). In 1994, a second-generation ligation chemistry was introduced, known as “native chemical ligation,” which allows the preparation of proteins with native backbone structures from fully unprotected peptide building blocks (Dawson et al., 1994). This important extension of the chemical ligation concept makes use of the mild acylating power of the α-thioester functionality. The principle of native chemical ligation is depicted in Figure 18.4.1. The first step involves the chemoselective reaction which occurs between the free thiol group of an unprotected N-terminal Cys-polypeptide and a second, unprotected polypeptide containing an α-thioester group. This transthioesterification reaction gives rise to a thioester-linked intermediate which spontaneously rearranges to form a native peptide bond at the ligation site. The target full-length polypeptide is thus obtained without any further manipulation. Native chemical ligation reactions are performed in aqueous buffers at pH 7 to 7.5 in the presence of thiol cofactors. At this pH, the regioselectivity of the reaction is such that the reaction can to be performed in the presence of all the

functionalities commonly found in proteins. Even the presence of additional Cys residues in one or both fragments does not affect the regioselectivity of the ligation (Hackeng et al., 1997). Small proteins or protein domains ∼100 to 120 amino acids in length can be reliably generated from two peptide building blocks in a single chemical ligation step (for a few examples see Dawson et al., 1994, 1997; Lu et al., 1996; Hackeng et al., 1997; Camarero et al., 1998). The total chemical synthesis of larger protein targets (>100 residues) via the ligation of just two fragments becomes more and more problematic as the size increases. This is due to the difficulties associated with the direct stepwise SPPS of polypeptide segments bigger than 50 residues. This difficulty can be overcome by performing multiple ligation reactions using three or more synthetic peptides (e.g., Canne et al., 1995). Native chemical ligation has been extended to allow multiple ligation steps to be performed sequentially in a controlled and directed way (Muir et al., 1997; Camarero et al., 1998). The general approach is depicted in Figure 18.4.2. Key to this strategy is the temporary protection of the α-amino group of the central peptide with the base-labile 2-(methylsulfonyl)ethyloxycarbonyl (Msc) group (Tesser and Balvert-Geers, 1975). The presence of this protecting group prevents the N-terminal Cys from reacting in an intramolecular or intermolecular fashion with the α-thioester functionality present in the same polypeptide fragment. Once the first ligation reaction is finished, the Nα-Msc group can be efficiently removed, allowing the next ligation step to be performed (Fig. 18.4.2). The native chemical ligation approach can also be used in the semisynthesis of proteins (Muir et al., 1998; Severinov and Muir, 1998; Erlandson et al., 1996; Evans et al., 1998) from recombinant and synthetic polypeptide fragments. As described above, native chemical ligation of two polypeptides requires that one of the fragments possess an N-terminal Cys and that the other contain an α-thioester moiety. Polypeptides containing an N-terminal cysteine residue for use in litigation can be obtained using standard recombinant DNA expression methods (see Chapter 5). Importantly, biosynthetic methods are also now available for the generation of α-thioester polypeptides (Muir et al., 1998; Severinov and Muir, 1998; Evans et al., 1998). This is made possible using the IMPACT expression system, commercially available from New England Biolabs (Chong



A

HS intein

NH

N-extein

B

HS

O

NH NH2

O

clone protein gene in IMPACT system

C-extein express in E. coli

O

N→S acyl transfer

HS recombinant protein

N-extein

S O H2N

O intein

CBD

HS NH NH2

C-extein

Affinity purification on chitin beads

HS

O

transthioesterification

CO-NH-Cys intein*

recombinant protein

O

CO-NH-Cys intein*

CBD

N-extein HS H2N

S

O intein

C-extein

NH NH2

N→S Acyl transfer recombinant protein

O

CO-S

H2N-Cys intein*

S→N acyl transfer and succinimide formation transthioesterification H N

N-extein O HS H2N

+

CBD

i) 3% ethanethiol at pH 6 overnight ii) filtration

C-extein SH

recombinant protein

CO-SCH2 CH3

O intein NH O

Figure 18.4.6 Principles of the biosynthetic preparation of α-thioester polypeptides by recombinant techniques. (A) Scheme representing the proposed mechanism of protein splicing involving the intein from Saccharomyces cerevisiae VMA1 gene (Xu and Perler, 1996). (B) Expression, purification, and cleavage of polypeptide-intein*-CBD fusion protein (where the asterisk refers to the mutation Asn454→Ala in the intein element and CBD refers to the chitin binding domain) with an appropriate thiol to give the α-thioester polypeptide.

et al., 1997). This system utilizes a protein splicing element, an intein from the Saccharomyces cerevisiae VMA1 gene, in conjunction with a chitin binding domain (CBD) which allows purification by affinity chromatography (Fig. 18.4.6). The natural intein has been modified in the expression system (Asn454→Ala) in order to block the normal protein splicing reaction in midstream. This results in the formation of a thioester linkage between the polypeptide of interest and the intein (Fig. 18.4.6). Cleavage of the thioester can thus be induced by treatment of the expressed fusion protein with the appropriate

thiol to give the target α-thioester polypeptide (Muir et al., 1998; Severinov and Muir, 1998; Evans et al., 1998).

Critical Parameters and Troubleshooting As mentioned in Basic Protocol 1, pH is the most critical parameter in the native chemical ligation of two fully unprotected peptides in aqueous buffers. The efficiency of the ligation reaction is strongly dependent on the pH; above pH 8.0 the reaction loses its regioselectivity and below pH 6.0 the reaction is usually very slow.



Supplement 15

Another crucial factor is the concentration of the two reactants. The bimolecular nature of native chemical ligation (and ligations in general) means that the concentration of both reactants should be as high as possible for efficient reaction. Generally the use of chemical denaturants (GdmCl or urea) will allow high concentrations of both reactants to be achieved. Furthermore, the use of denaturing conditions helps to alleviate potential steric problems that may be associated with the use of folded polypeptides. Another important parameter for the success of the ligation reaction is the availability of the thiol and/or α-amino groups of the N-terminal Cys polypeptide during the ligation reaction. If one or both groups are chemically blocked, the ligation will not take place. It is well known that N-terminal Cys peptides can react rapidly with carbonyl-containing compounds (e.g., acetone and formaldehyde) to give the corresponding N-terminal thiazolidine adducts, which are totally unreactive in the native ligation process. It is thus crucial to avoid the use of these substances while handling all peptides and buffers. Note that the authors have found acetone, a commonly used solvent for washing glassware in many laboratories, to be particularly problematic in this regard.

Anticipated Results Native chemical ligation has been applied to the synthesis and semisynthesis of a large number of proteins and protein domains. These studies indicate that very high yields (80% or better) are typically obtained for the ligation step. Following purification, the total yield usually drops to 50% to 60%. In sequential ligations where the Nα-Msc group has to be deprotected in situ after the first ligation step, the typical total yield after purification is ∼30% to 40%. It is also important to note that the protein-folding step may further decrease the final yield of product.

Time Considerations


In Basic Protocols 1 and 2, the chemical ligation step typically requires 2 days, although in some cases the ligation reaction can proceed very rapidly (in a few hours) or somewhat more slowly (in 4 days). The purification step (RPHPLC or other liquid chromatographies) can be performed in half a day. Processing of the purified samples (lyophilization or concentration) can take 1 to 2 days (depending on the volume). Finally, folding of the protein (if necessary) can be achieved in 3 days.

The thioester resin can be prepared in 1 day. The time required for the chemical synthesis of an α-thioester polypeptide (see Support Protocol 3) will depend on its size. Typically, for peptides ∼50 residues in length, the solid-phase chain assembly can be carried out in 3 days manually or in 1 day using an automated synthesizer. The deprotection-cleavage and purification can be performed in 1 day, and lyophilization of peptide fractions can take 1 to 2 days (depending on the volume). Bacterial expression of a peptide-inteinCBD fusion protein and preparation of the crude cell extract requires 2 days for the system described here (see Support Protocol 4). Column preparation requires ∼2 to 3 hr. Loading and washing the column can be done in 3 hr. The cleavage of the α-thioester polypeptide from the affinity column requires 10 to 15 hr (overnight). Purification and processing of the polypeptide fractions requires 1 to 2 days.

Literature Cited Ausubel, F.A., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, K. (eds.). 1999. Current Protocols in Molecular Biology. John Wiley & Sons, New York. Bodanszky, M. and Bodanszky, A. 1994. The Practice of Peptide Synthesis, 2nd ed. Springer-Verlag, Berlin. Camarero, J.A., Cotton, G.J., Adeva, A., and Muir, T.W. 1998. Chemical ligation of unprotected peptides directly from a solid support. J. Pept. Res. 51:303-316. Canne, L.E., Ferre-D’Amare, A.R., Burley, S.K., and Kent, S.B.H. 1995. Total chemical synthesis of a unique transcription factor-related protein: cMyc-Max. J. Am. Chem. Soc. 117:2998-3007. Chong, S., Mersha, F.B., Comb, D.G., Scott, M.E., Landry, D., Vence, L.M., Perler, F.B., Benner, J., Kucera, R.B., Hirvonen, C.A., Pelletier, J.J., Paulus, H., and Xu, M.Q. 1997. Single-column purification of free recombinant proteins using a self-cleavable affinity tag derived from a protein splicing element. Gene 192:271-281. Dawson, P.E., Muir, T.W., Clark-Lewis, I., and Kent, S.B.H. 1994. Synthesis of proteins by native chemical ligation. Science 266:776-779. Dawson, P.E., Churchill, M.J., Ghadiri, M.R., and Kent, S.B.H. 1997. Modulation in native chemical ligation through the use of thiol additives. J. Am. Chem. Soc. 119:4325-4329. Erlandson, D.A., Chytill, M., and Verdine, G.L. 1996. The leucine zipper domain controls the orientation of AP-1 in the NFAT AP-1 DNA complex. Chem. Biol. 3(12):981-991. Evans, T.C., Benner, J.J., and Xu, M.-Q. 1998. Semisynthesis of cytotoxic proteins using a modified protein splicing element. Protein Sci. 7:22562264.



Hackeng, T.M., Mounier, C.M., Bon, C., Dawson, P.E., Griffin, J.H., and Kent, S.B. 1997. Total chemical synthesis of enzymatically active human type II secretory phospholipase A2. Proc. Natl. Acad. Sci. U.S.A. 94:7845-7850.

Schnölzer, M., Alewood, P., Jones, A., Alewood, D., and Kent, S.B.H. 1992. In situ neutralization in Boc-chemistry solid phase peptide synthesis: Rapid, high yield assembly of difficult sequences. Int. J. Pept. Protein Res. 40:180-193.

LLoyd-Williams, P., Albericio, F., and Giralt, E. 1993. Convergent solid-phase peptide synthesis. Tetrahedron 49:11065-11133.

Severinov, K. and Muir, T.W. 1998. Expressed protein ligation: A novel method for studying protein-protein interactions in transcription. J. Biol. Chem. 273:16205-16209.

Lu, W., Qasim, M.A., and Kent, S.B.H. 1996. Comparative total synthesis of turkey ovomucoid third domain by both stepwise solid phase synthesis and native chemical ligation. J. Am. Chem. Soc. 118:8518-8523. Merrifield, R.B. 1963. Solid phase peptide synthesis. I. The synthesis of a tetrapeptide. J. Am. Chem. Soc. 85:2149-2154. Muir, T.W. 1995. A chemical approach to the construction of multimeric protein assemblies. Structure 3:649-652. Muir, T.W., Dawson, P.E., and Kent, S.B.H. 1997. Protein synthesis by chemical ligation of unprotected peptides in aqueous solution. Methods Enzymol. 289:266-298. Muir, T.W., Sondhi, D., and Cole, P.A. 1998. Expressed protein ligation: A general method for protein engineering. Proc. Natl. Acad. Sci. U.S.A. 95:6705-6710.

Tesser, G.I. and Balvert-Geers, I.C. 1975. The methylsulfonylethyloxycarbonyl group: A new and versatile amino protective function. Int. J. Pept. Prot. Res. 7:295-305. Wallace, C.J.A. 1995. Peptide ligation and semisynthesis. Curr. Opin. Biotechnol. 6:403-410. Xu, M.-Q. and Perler, F.B. 1996. The mechanism of protein splicing and its modulation by mutation. EMBO J. 15:5146-5153.

Contributed by Julio A. Camarero and Tom W. Muir The Rockefeller University New York, New York



Supplement 15

Synthesis and Application of Peptide Dendrimers As Protein Mimetics

UNIT 18.5

Peptide dendrimers are branched polymers with peptides attached centrally to a template or core matrix. They are synthesized as defined dendritic structures. Their molecular weights increase geometrically as a function of generation branching of monomers. Usually, 2 to 16 peptidyl branches of the same or different sequences are used to form a peptide dendrimer. Architecturally, peptide dendrimers belong to the multichained, cascade-shaped polymer family. Thus, they contain multiple N- or C-terminal ends that can be exploited as a platform for multitasks. Such a molecular architecture is different from conventional polypeptides or proteins that contain a single unbranched contiguous structure with two terminal ends. Peptide dendrimers are also different from organic star-burst dendrimers, which are spherically shaped and often consist of >16 hyperbranched small organic monomers. However, the fundamental dendrimeric design consisting of monomers radiated from a center core is similar, conceptually, to the organic dendrimers. Peptide dendrimers come in different forms and sizes. Their forms are determined by the templates and peptide structures while their sizes are determined by the numbers and lengths of peptide monomers. Although their sizes vary over a wide range from 3 to >100 kDa, most peptide dendrimers reported to date are in the range of 6 to 20 kDa. A driving force in developing peptide dendrimers is the effort to mimic forms and functions of proteins. Thus far, peptide dendrimers have found applications in four active areas: (1) immunological applications as immunogens and antigens; (2) de novo design of artificial proteins; (3) protein mimetics as agonists and antagonists in drug discovery; and (4) new biopolymers and biomaterials. However, this unit will focus on the use of peptide dendrimers as protein mimetics. The use of peptides to mimic a portion of a protein structure is a challenging and powerful tool in the discovery of new drugs. In native proteins, discontinuous bioactive peptide surfaces are held together in a particular conformation by the structural rigidity of the protein. One approach to mimicking the structural surface on the protein is to bring the potential peptide sequences together by assembling the peptide chains on a template as a peptide dendrimer. Recently, small peptides have been demonstrated to adopt well defined conformations when they are attached on a template (Schneider and Kelly, 1995). In the past 15 years, templates suitable for protein engineering have been developed. These templates are flexible dendrimeric peptides (Tam, 1988; Wrighton et al., 1997) and cyclic peptides (Mutter and Vuilleumier, 1989), as well as more rigid organic molecules (Unson et al., 1984; Sasaki and Kaiser, 1989). In 1988, the multiple antigenic peptide (MAP) system was introduced in the authors’ laboratory (Tam, 1988; Tam and Lu, 1989) as a novel approach to preparing peptide immunogens. The MAP consists of an inner core matrix built up of a large layer of Lys residues and a surface of peptide chains attached to the core matrix (Fig. 18.5.1). Because of its dendrimeric structure, MAP can be very useful as a template for assembling potential peptide surfaces. Furthermore, the effect of MAP on secondary structures has also been studied (Shao and Tam, 1995). These study results showed that the peptide dendrimer has an increased ordered helical structure, while the peptide monomer and the tetravalent lysinyl core matrix do not have any ordered structure. The synthetic schemes for preparing MAPs suitable for protein mimetics are the same as those for generating MAPs as peptide immunogens. In addition to the applications in immunology, examples in the literature Contributed by James P. Tam and Jane C. Spetzler Current Protocols in Protein Science (1999) 18.5.1-18.5.35 Copyright © 1999 by John Wiley & Sons, Inc.



B

A

NH NH NH

NH NH

NH Ala-OH

Ala-OH

NH NH

NH NH

NH

NH

Figure 18.5.1 Schematic drawing of the MAP system. (A) Tetravalent MAP core matrix. (B) Octavalent MAP core matrix. The solid dots represent Lys and the ribbons with arrows represent the peptide.

have applied MAPs in areas such as inhibitors, artificial proteins, affinity purifications, and intracellular transport. The MAP, which is chemically defined and homogeneous, was first intended as an approach to overcoming the limitations of the conventional method for producing antipeptide antibodies. In the conventional approach, the peptide antigen is conjugated to a known large protein, or synthetic polymer carrier, to form a peptide-carrier conjugate. Although this approach has been used successfully in eliciting animal antibodies, it has several inherent limitations. First, only a small portion of peptide antigen is represented in the whole conjugate. Second, there is chemical ambiguity in the antigen composition and structure. Third, irrelevant epitopes and antibodies may be produced, and finally, carrier toxicity and carrier-induced epitope suppression may occur. The MAP systems have been used successfully to produce both polyclonal and monoclonal antibodies that specifically recognize native proteins. They have also been used to produce sera that have a significantly higher titer of antibodies than sera with antibodies against the same peptides conjugated to keyhole limpet hemocyanin (KLH) as a carrier protein (Tam, 1988).


Cyclic peptides are a powerful approach in the design of hormones and bioactive peptides, since favorable conformational constraints associated with biological activity can be fixed and conformational mobility diminished. In addition, the stability against biodegradation is increased. Cyclic peptides are also beneficial as peptide antigens, since they can elicit higher-quality antibodies than their linear counterparts. Based on these demands, the cyclic multiple antigen peptide (cMAP) approach (Spetzler and Tam, 1996; Zhang and Tam, 1997), which has branched multiple closed-chain architectures, was developed. The cMAP system is a superior approach for protein mimetics because the multiple constrained peptides can mimic bioactive conformations. Whether to select this approach over MAP depends on the properties of the peptides, but usually if the peptides are too small to adopt a stable conformation on their own, incorporation of a cyclic structure may be necessary. Two methods for preparing MAP systems are described in this unit. The direct approach for preparing MAP systems is presented in the first two protocols, including the procedure



A

B

stepwise peptide synthesis

indirect

direct functionalized peptide

peptide MAP core matrix

functionalized MAP core

cleavage and deprotection

MAPs

functionalized functionalized X Y free MAP core free peptide

conjugation MAPs

Figure 18.5.2 The direct and indirect approaches for preparing MAPs. Spheres represent the solid support; the MAP core matrix is Lys2-Lys-Ala-OH or Lys4-Lys2-Ala-OH; X represents nucleophile; Y represents electrophile.

for t-butyloxycarbonyl (Boc) chemistry (see Basic Protocol 1) and the procedure for 9-fluorenylmethyloxycarbonyl (Fmoc) chemistry (see Alternate Protocol). An indirect approach for preparing MAP systems, in which peptide and core matrix are synthesized separately and conjugated by several ligation methods, is then described (see Basic Protocols 2 and 3). The cMAP approach is also executed using either the direct or indirect approach, but requires an additional cyclization step to constrain the peptides after synthesis. The synthesis of cMAP is described (see Basic Protocols 4, 5, and 6), and the preparation of cyclic peptides is illustrated (see Support Protocols 4, 5, and 6). Support Protocol 1 describes the ninhydrin assay to assess the completeness of the coupling reaction. In most cases, MAP systems can be used directly after simple dialysis or desalting. Some immunological studies, however, require purified MAPs. Support Protocols 2 and 3 describe MAP system purification by dialysis and high-performance gel-filtration chromatography. STRATEGIC PLANNING Realizing the distinction between the direct and indirect approach is essential (Fig. 18.5.2) when synthesizing MAP systems. In the direct approach, the MAP core is formed by attaching several layers of lysine to the resin, which contains either Gly or Ala, using solid-phase peptide synthesis (Merrifield, 1963; UNIT 18.1). The peptide is then assembled onto this MAP core matrix (Basic Protocol 1 and Alternate Protocol). After the synthesis, the entire MAP is deprotected and cleaved from the resin. In the indirect approach, the MAP core and peptide are prepared and purified separately (Basic Protocols 2 to 3). Then, they are linked together in the absence of any side-chain protecting groups, after purification, to form a completed MAP. For preparing MAPs by either the direct or indirect approach, both the peptide synthesis and the ligation chemistry are easy to perform. Other features of the direct and indirect approach are discussed in the Commentary (see Background Information). However, the considerations for choosing a direct or indirect approach involve requirements for purity, properties of the peptide antigens, and time factors.



Supplement 17

Cyclic

peptidea

Typeb

Reactive endsc Electrophile Nucleophile

Ligation site (X)

O - CSH

Cys

Liu and Tam, 1997

O - CSHd

Cys

Zhang and Tam, 1997

CO

- CO2CH2CHOe

Cys

Botti et al., 1996

CO

- COCHOf

Cys

Botti et al., 1996

x e-e

Cys

e-e

Cys S HO

x x

x x

e-e

N CO

CO2H e-s

References

S HN

CO2H e-s - CO- CH=N - OCH2 CO-

-COCHOf NH2-OCH2CO-

Pallin and Tam, 1995

x CO2H s-s -CO- CH=N- OCH2 CO-

H2N x H2N

CO2H s-s

s-s

-COCHOf NH2-OCH2CO- Pallin and Tam, 1996 Cys

Cys

Tam et al., 1991

Figure 18.5.3 Available methods for preparing cyclic peptides from unprotected linear peptides. Notes: (a) arrow indicates the direction of the amide bond; (b) types of cyclic peptides are designated as e-e, end-to-end; e-s, end-to-side chain; and s-s, side-chain-to-side-chain; (c) an electrophile/nucleophile pair is placed at the ends or side chain of the linear peptide (e.g., for formation of e-e cyclic peptide, the amino nucleophile is Cys and the carbonyl electrophile can be a thioester); (d) R = (CH2)2-CONH2; (e) from periodate oxidation of a C-terminal glyceric ester; (f) from periodate oxidation of Ser and Thr attached to side chain of Lys.

For generating cMAPS, peptides on the core matrix need to be constrained as an additional step. A key tactic in the authors’ strategy for preparing cMAP is the use of unprotected peptides as precursors. In the past few years, the authors’ laboratory has developed a plethora of methods to form cyclic peptides from unprotected linear peptides in aqueous solution. Figure 18.5.3 lists those methods that are useful in preparing cMAPs. In planning of the cMAP synthesis, two questions need to be addressed: (1) how should the peptide antigen be constrained and (2) what are the suitable methods? The following are simple guidelines in the strategic planning for preparing cMAPs. There are three general methods for constraining peptides in the cMAP system: end-toend, end-to-side chain, and side chain-to-side chain (Figure 18.5.3). In one example (Basic Protocol 4) using the direct approach, cMAP is self-assembled via thiazolidine linkages after its unprotected precursor is synthesized by solid-phase synthesis. This procedure produces cyclic peptides that are linked in an end-to-side chain manner. A similar strategy can be applied for end-to-end or side chain-to-side chain methods by modifying the synthetic strategy accordingly. Peptides with a motif that favors a cyclic structure are highly recommended in this approach. Two other procedures (Basic Protocols 5 and 6) describe the use of an indirect approach for the synthesis of purified cMAP in two distinct steps. These protocols involve the synthesis of side chain-to-side chain and end-to-end cyclic peptides containing a free thiol, which are then linked to a MAP core matrix Synthesis and Application of Peptide Dendrimers As Protein Mimetics

A critical point in the preparation of both MAP and cMAP is the selection of the linear peptide sequences. Some of the available methods for identifying potential bioactive sequences include peptide library approaches (Tam et al., 1991; Houghten et al., 1991),



simultaneous synthesis of a high number of peptides (UNIT 18.2), and computer-assisted modeling (UNIT 18.3). The procedures described in this unit involve use of an automated peptide synthesizer (e.g., Perkin-Elmer Applied Biosystems). For Boc syntheses, a special hydrogen fluoride cleavage apparatus (Peptide Institute or Peninsula Laboratories) and hood are required. Manual procedures can also be used (Stewart and Young, 1984). A detailed discussion of peptide synthesizers and reagents is presented in UNIT 18.1. CAUTION: Some chemicals used in these procedures are harmful. N-N′-dicyclohexylcarbodiimide (DCC) can cause severe allergy in sensitive persons. Avoid any skin contact with DCC! Benzotriazolyl N-oxytrisdimethylaminophosphonium hexafluorophosphate (BOP) and 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyl uronium hexafluorophosphate (BBTU) can be used as alternatives to DCC. Trifluoroacetic acid (TFA) is a toxic, volatile organic acid and should only be used under a hood. Besides being toxic, hydrogen fluoride (HF) is also highly corrosive to glass instruments. Never use glass instruments or containers in HF cleavage. Thick polyvinyl gloves and a protective face mask should be worn. More details about these chemicals can be found in the literature (Stewart and Young, 1984). DIRECT Boc SYNTHESIS OF MAP SYSTEMS In the direct approach for preparing MAP systems, the MAP core is first formed by attaching several levels of lysine onto an amino acid coupled to a resin matrix. The peptide sequence is then assembled stepwise on this MAP core resin. After peptide assembly is complete, the MAP system is cleaved from the resin support (Tam, 1988). The procedure for preparing MAP systems by the direct approach using Boc chemistry is presented in Figure 18.5.4 and Table 18.5.1. Solid-phase peptide synthesis (Merrifield, 1963) using

BASIC PROTOCOL 1

X-Ala-O

synthesis of MAP core matrix

X-HN X-HN-Lys

Lys2 -Lys-Ala-O 4

peptide synthesis Peptide Peptide

Lys

Lys2 -Lys-Ala-O 4

cleavage

Peptide Peptide

Lys

Lys2 -Lys-Ala-OH 4

MAPs

Figure 18.5.4 Direct approach of preparing MAPs with an octabranched core matrix. Sphere represents solid support; X represents Boc or Fmoc.



Supplement 17

Table 18.5.1

Schedule for Boc Solid-Phase Peptide Synthesis

Reaction Description 1 2 3 4 5 6 7 8a 9 10 11 12 13c

DCM wash 50% (v/v) TFA/DCM prewash 50% (v/v) TFA/DCM deprotection DCM wash (5×) 5% (v/v) DIEA/DCM prewash 5% (v/v) DIEA/DCM neutralization DCM wash (5×) Boc-AA (3 eq) in DCM DCC (3 eq) in DCM Addition of DMF DMF wash (2 times) DCM wash (2 times) Ninhydrin test

Volume Repetitions (ml/g resin) and time 15 15 15 15 15 15 15 5 1 15 15 15 —

3 × 1 min 1 × 1 min 1 × 20 min 5 × 1 min 1 × 1 min 1 × 5 min 5×1 1 × 1 minb 1 × 30 minb 1 × 30 min 2 × 1 min 2 × 1 min 5 min

aIn case of coupling of Boc-Asn-OH, and Boc-Gln-OH, the DCC/HOBt coupling method should be

used to avoid dehydration of unprotected side chain amide groups. Alternatively use Boc-Asn(Trt)OH and Boc-Gln(Trt)-OH to minimize degradation side reactions. Boc-Arg(Tos)-OH should also be coupled with the DCC/HOBt method. bDo not drain until the completion of reaction 10. cIf ninhydrin test does not show completion of coupling reaction (>99.5%), repeat reactions 7 to 13.

Boc chemistry involves prewashing the resin, removing the Boc protecting group, neutralizing the resin, coupling the next Boc–amino acid residue, and then repeating this cycle until the complete peptide sequence is assembled. Finally, the peptide is cleaved from the resin support. Materials Synthesis reagents (Table 18.5.1; prepared as described in Stewart and Young, 1984): Dichloromethane (methylene chloride, DCM) 50% (v/v) trifluoroacetic acid (TFA) in DCM (TFA/DCM) 5% (v/v) diisopropylethylamine (DIEA) in DCM (DIEA/DCM) 1.0 M N,N′-dicyclohexylcarbodiimide (DCC) in DCM (DCC/DCM) N,N′-dimethylformamide (DMF) 0.1 mmol/g Boc-β-Ala-OCH2-PAM-resin (Bachem California) Boc-Lys(Boc) (Bachem California) Boc-amino acids (Boc-AA), with protected side-chain groups (Bachem California, Peninsula Laboratories) appropriate for synthesis of the desired peptide antigen, dissolved in dichloromethane (DCM). 10% (v/v) thiophenol in DMF or 10% (v/v) mercaptoethanol (2-ME)/5% (v/v) DIEA in DMF p-cresol p-thiocresol Dimethyl sulfide (DMS) Liquid hydrogen fluoride (HF), –78°C 99:1 (v/v) cold diethyl ether/2-mercaptoethanol (2-ME) 10% (v/v) acetic acid in water Synthesis and Application of Peptide Dendrimers As Protein Mimetics

Automated peptide synthesizer (e.g., Perkin-Elmer Applied Biosystems #ABI430A) Manual reaction vessel (Pierce) Hydrogen fluoride cleavage apparatus (Peptide Institute or Peninsula Laboratories)



50-ml round-bottom flask Coarse- and fine-porosity fritted glass funnels 50°C water bath Prepare the core matrix 1. Place 0.5 g of 0.1 mmol/g Boc-β-Ala-OCH2-PAM-resin in a reaction vessel in an automated peptide synthesizer. Begin the synthesis, using the synthesis reagents and directions outlined in Table 18.5.1, making the substitutions in reaction 8 of the table as described in step 2a, b, or c. Reactions 1 through 13 (Table 18.5.1) constitute one complete cycle; one cycle is required for each amino acid in the sequence. Some automated synthesizers may include the test to evaluate the coupling reaction in their cycle; if not, the ninhydrin test (reaction 13) will have to be performed manually (Support Protocol 1, Sarin et al., 1981). β-alanine is used as an internal standard for amino acid analysis which provides the molar ratios of other amino acids against β-alanine. Instead of β-alanine, other simple amino acids such as alanine can also be used.

2a. To prepare a MAP core matrix with two lysine branches: In reaction 8 of the first cycle, couple 0.15 mmol (52 mg) Boc-Lys(Boc) to the resin added in step 1. Proceed to step 3. 2b. To prepare a MAP core matrix with four lysine branches: Complete step 2a, then, in reaction 8 of the second cycle, couple 0.3 mmol (104 mg) Boc-Lys(Boc) to the resin from step 2a. Proceed to step 3. The substitution level doubles with each coupling of Boc-Lys(Boc).

2c. To prepare a MAP core matrix with eight lysine branches: Complete steps 2a and 2b, then, in reaction 8 of the third cycle, couple 0.6 mmol (208 mg) Boc-Lys(Boc) to the resin from step 2b. Proceed to step 3. Preformed MAP core matrix is also commercially available from Applied Biosystems, Bachem California, and Calbiochem-Novabiochem.

Synthesize the peptide on the MAP core matrix 3. Perform synthesis of the peptide on the MAP core matrix prepared in step 2a, b, or c by carrying out the reactions in Table 18.5.1 (one cycle of 13 reactions per amino acid), using, in reaction 8 of the table, 0.15 mmol of Boc-amino acids for a MAP core matrix with two branches, 0.3 mmol of Boc-amino acids for a MAP core matrix with four branches, or 0.6 mmol of Boc-amino acids for a MAP core matrix with eight branches, based on 0.05 mmol of starting Boc-β-Ala-OCH2-PAM resin. Select the appropriate Boc-amino acids for the synthesis of the peptide antigen. For descriptive purposes, the peptide antigen IEDNEYTARQG (IG-11) is used in this protocol. For this peptide, the Boc-amino acids would be Boc-Gly, Boc-Gln, Boc-Arg (Tos), Boc-Ala, Boc-Thr(Bz1), Boc-Tyr(BrZ), Boc-G1u(OBzl), Boc-Asn, Boc-Asp (OBzl), and Boc-Ile for the synthesis of IG-11-MAP. The peptide chain would be assembled in sequence G, Q, R, A, T, Y, E, N, D, E, I. The most commonly used side-chain-protecting groups for Boc amino acids are: Ser and Thr, benzyl ether (Bzl); Asp and Glu, benzyl ester (OBzl); Asn, and Gln, unprotected or trityl (Trt) amide; Met, unprotected; Tyr, 2-bromobenzyloxycarbonyl (BrZ); Arg, toluenesulfonyl (Tos); Lys, 2-chlorobenzyloxycarbonyl (CIZ); His, Nim-2,4-dinitrophenyl (Dnp); Trp, Nin-formyl, Cys, p-methylbenzyl (MeBzl), or acetamidomethyl (Acm). Some special side-chain-protecting groups can also be found in the literature listed at the end of this unit (see Literature Cited).

4. After synthesis of the peptide is complete, transfer the resin to a manual reaction vessel and remove the Dnp from peptide residues with Nim-Dnp protecting groups using 10% thiophenol in DMF (three 8-hr treatments, 5 ml each), or 10% 2-ME/5% DIEA in DMF (two 30-min treatments, 5 ml each).



Supplement 17

5. Remove N-terminal Boc groups by treating for 20 min with 10 ml of 50% TFA/DCM, neutralize the resin with 10 ml of 5% DIEA/DCM, and then wash the resin five times, each time with 10 ml DCM (procedure similar to “reactions 1 to 4” in Table 18.5.1). Place the reaction vessel in a desiccator, and dry the resin under a vacuum. Boc groups would be removed in HF cleavage. It is advisable, however, to remove these t-butyl protecting groups prior to HF cleavage to avoid generating long-lived, reactive t-butyl cations during HF cleavage.

Cleave the peptides 6. Place 300 to 500 mg of MAP resin in the reaction vessel of the hydrogen fluoride cleavage apparatus. Melt p-cresol and p-thiocresol by placing bottles in 50°C water bath. Add 0.75 ml of the p-cresol and 0.25 ml of the p-thiocresol to the reaction vessel. Once this mixture has cooled to room temperature and solidified, add a magnetic stir-bar and 6.5 ml DMS to the reaction vessel and mix by stirring for 2 min at low speed without splattering resin on the reaction vessel. 7. For low HF cleavage: Add liquid HF at −78°C to give a final volume of 10 ml 7.5/2.5/65/25 (v/v/v/v) p-cresol/p-thiocresol/DMS/HF. Equilibrate the mixture to 0°C by immersing the reaction vessel in an ice bath and stirring. Stir the mixture vigorously for 2 hr at 0°C, and remove HF and DMS under a vacuum. Proceed to step 8 for high HF cleavage, or work up reaction as in step 9. CAUTION: HF cleavage must be performed in a well ventilated chemical fume hood. Thick polyvinyl chloride gloves and a protective face mask must be used. The outside wall of the vessel should have a predetermined mark for 10 ml.

8. For high HF cleavage: Recharge the mixture with 14 ml liquid HF at −78°C to give a final total volume of 15 ml HF/p-cresol/p-thiocresol. Stir the reaction mixture 1 hr at 0°C. Remove HF under a vacuum using a water aspirator at 0°C. See precautions for working with HF (step 7 annotation)

9. Work up reaction: Wash the residue by adding 30 ml of cold 99:1 diethyl ether/2-ME to the reaction vessel to remove p-thiocresol and p-cresol. Filter the resin and the precipitated peptide with a fritted-glass funnel in a 50-ml round-bottom flask. Discard the ether wash (the peptide will be distributed partially in the reaction vessel and partially in the fritted-glass funnel). Dissolve the crude peptide on both the reaction vessel and the fritted-glass funnel with 100 ml of 10% acetic acid, collect the filtrates, and lyophilize the solution to dryness. Store at −70°C. ALTERNATE PROTOCOL


DIRECT Fmoc SOLID-PHASE SYNTHESIS OF MAP SYSTEMS Fmoc synthetic chemistry is another widely used method for solid-phase peptide synthesis. The procedure for preparing MAP systems (Defoort et al., 1992) by the direct approach using Fmoc chemistry is presented in Figure 18.5.4 and Table 18.5.2. The basic procedure in Fmoc solid-phase peptide synthesis involves prewashing the resin, removing Fmoc protecting groups, and coupling the next Fmoc-amino acid residue. This cycle is repeated until peptide elongation is completed, and the peptide has been cleaved from the resin support. Materials Fmoc-Ala-Wang-resin (p-alkoxybenzyl alcohol resin preloaded with Fmoc-alanine at 0.3 to 0.5 mmol alanine/g resin; Bachem California) Fmoc synthesis reagents N,N′-dimethylformamide (DMF) 20% (v/v) piperidine in DMF (Pip/DMF) 0.5% (w/v) 1-hydroxybenzotriazole (HOBt) in DMF (HOBt/DMF)



Table 18.5.2

Schedule for Fmoc Solid-Phase Peptide Synthesis

Reaction Description 1 2 3 4 5 6a 7 8 9 10c

DMF wash 20% (v/v) Pip/DMF prewash 20% (v/v) Pip/DMF deprotection DCM wash 0.5% HOBt/DMF wash Fmoc-AA (3 eq) in DMF HOBt (3 eq) in DMF DCC (3 eq) in DCM DMF wash Ninhydrin test

Volume Repetitions and (ml/g resin) time 15 15 15

2 × 1 min 1 × 3 min 1 × 20 min

15 12 12 1 1 15 —

6 × 1 min 2 × 1 min 1 × 1 minb 1 × 1 minb 1 × 60 min 5 × 1 min 5 min

aCoupling procedure for all amino acids. bDo not drain until the completion of reaction 7. cIf ninhydrin test does not show completion of coupling reaction (>99.5%), repeat reactions 5 to 7.

Fmoc-Lys(Fmoc) (Bachem California; Peninsula Laboratories) Acetic anhydride Fmoc-amino acids (Fmoc-AA) with protected side-chain groups (Bachem California; Peninsula Laboratories) appropriate for synthesis of the desired peptide antigen, dissolved in DMT Cleavage solution (see recipe), prepare fresh and prechill to 0°C Anhydrous methyl t-butyl ether, cold Anhydrous diethyl ether 10% (v/v) acetic acid in water Automated peptide synthesizer (e.g., Perkin-Elmer Applied Biosystems #ABI430A) Manual reaction vessel (Pierce) Coarse- and fine-porosity fritted glass funnels Prepare the resin 1. Place 0.5 g of 0.3 to 0.5 mmol/g Fmoc-Ala-Wang-resin in a reaction vessel in an automated peptide synthesizer. Couple 0.055 mmol (0.32 mg) Fmoc-Lys(Fmoc) onto the resin by carrying out reactions 1 to 9 in Table 18.5.2, adding the Fmoc-Lys(Fmoc) in reaction 6 of the table. Omit the ninhydrin test step. Reactions 1 through 10 (Table 18.5.2) constitute one complete cycle; one cycle is required for each amino acid in the sequence. Some automated synthesizers may include the test to evaluate the coupling reaction in their cycle; if not, the ninhydrin test will have to be performed manually (Support Protocol 1, Sarin et al., 1981). The commercially available Fmoc-Ala-Wang-resin usually has a loading of 0.3 to 0.5 mmol Fmoc-Ala/g resin. This content, however, is too high for synthesis of AMP (see Background Information). To prepare a resin with 0.1 mmol lysine/g resin at the first level in MAP core, only 0.055 mmol of Fmoc-Lys (Fmoc) should be used. This incomplete coupling will result in the loading of ∼0.1 mmol lysine/g resin.

2. Add 15 ml DCM and 0.75 ml acetic anhydride to the reaction vessel and incubate 1 hr to terminate unreacted amines. Wash the resin five times with 10 ml DMF (1 min each) as in Table 18.5.2, reaction 9. Preparation and Handling of Peptides


Supplement 17

As a result of the previous incomplete coupling step, an excess of free amino groups will be on the resin. This step will acylate (cap) free amino groups prior to coupling the next level of lysine.

Prepare the core matrix 3a. To prepare a MAP core matrix with four or eight branches: In reaction 6, couple 0.3 mmol (177 mg) Fmoc-Lys(Fmoc) to the resin from step 2 by following the reaction steps in Table 18.5.2. Proceed to step 3b (for core matrix with eight branches) or step 4 (for core matrix with four branches). 3b. To prepare a MAP core matrix with eight branches: In reaction 6, couple 0.6 mmol (354 mg) Fmoc-Lys(Fmoc) to the resin from step 3a by following the reaction steps in Table 18.5.2, adding the Fmoc-Lys(Fmoc) at reaction 6 in the table. Proceed to step 4. Preformed MAP core matrix can be purchased from Applied Biosystems or CalbiochemNovabiochem.

Synthesize the peptide on the MAP core matrix 4. Perform synthesis of the peptide on the MAP core matrix prepared in step 3a or b by carrying out the reactions in Table 18.5.2 (one cycle of 10 reactions per amino acid to be added) using, in reaction 6 of the table, 0.6 mmol of Fmoc-amino acids for a MAP core matrix with four branches or 1.2 mmol of Fmoc amino acids for a MAP core with eight branches. The most commonly recommended side-chain-protecting groups for Fmoc-amino acids are: Ser, Thr, and Tyr, t-butyl ether (tBu); Asp and Glu, t-butyl ester (OtBu); Asn, Gln, trityl (Trt) amide; Arg, 2,2,5,7,8-pentamethylchroman-6-sulfonyl (PMC) or 4-methoxy-2,3,6trimethylhenzenesulfonyl (Mtr); Lys, tert-butyloxycarbonyl (Boc); His, Nim-trityl (TrT) Trp, and Met, unprotected; Cys, trityl (Trt) thioether or acetamidomethyl (Acm). All are commercially available.

5. After completing the synthesis, perform reactions 1 to 4 in Table 18.5.2, to remove the Fmoc group on the amino group of N-termini (last coupling) with 20% pip/DMF. Wash the resin with 10 ml DMF (three times, 1 min each) and 10 ml DCM (three times, 1 min each). Cleave the peptides 6. Transfer peptide/resin mixture to a manual reaction vessel. Add prechilled cleavage solution to the peptide/resin at a ratio of 10 ml cleavage solution per 0.2 to 0.4 g peptide/resin. Stir mixture 1.5 to 2 hr at room temperature. Other cleavage solutions can be used. Reagent K [82.5:5:5:2.5:5 (v/v/v/v/v) TFA/thioanisole/phenol/1,2-ethanedithiol (EDT)/H2O] can be used for peptides that contain multiple Arg and Trp residues. For simple peptides, a cleavage solution of 5% H2O/95% (v/v) TFA can be used.

7. Filter the peptide/resin mixture through a coarse-porosity fritted glass funnel, and wash the resin three times, each time with 5 ml TFA. Concentrate the filtrate to a small volume (∼2 ml) under a vacuum. 8. Precipitate peptide with 30 to 50 ml cold anhydrous methyl t-butyl ether.


9. Filter the peptide precipitate through a fine-porosity fritted glass funnel, and wash the peptide three times, each time with 10 ml anhydrous diethyl ether. Dry the crude peptide (in the glass funnel) under a vacuum. 10. Dissolve the crude peptide in 10% acetic acid and lyophilize to dryness. Store at −70°C.



INDIRECT Boc AND Fmoc SYNTHESIS OF MAP SYSTEMS USING THIOL CHEMISTRY

BASIC PROTOCOL 2

MAP systems have been successfully prepared by the direct approach, but they can also be synthesized by an indirect approach. In this method, the peptide antigen and MAP core matrix are prepared and purified separately and then linked together through functional groups on both elements to form a MAP system (Fig. 18.5.2 and Fig. 18.5.5; Lu et al., 1991). The indirect approach combines the advantages of solid-phase peptide synthesis with those of conjugation in a solution phase. An unprotected, purified peptide can be used in ligation. The orientation of peptides, particularly those derived from sequences at or near carboxyl termini, can be arranged as in native protein by linkage through the N-terminus to the core matrix. A variety of functional groups have been used for ligation to guarantee site-specific, chemically unambiguous MAP systems (also see Background Information). This protocol describes ligation through a sulfhydryl group of Cys in the peptide and a chloroacetyl group on the MAP core matrix. The same ligation chemistry is used to attach cyclic peptides to the MAP core matrix (see Basic Protocol 6). Materials Chloroacetic acid (Aldrich) 1.0 M N,N′-dicyclohexylcarbodiimide (DCC) in DCM (DCC/DCM) Sulfhydryl reducing solution (see recipe) Nitrogen source 1 M Tris⋅Cl 1 M NaOH 0.1 M NH4HCO3 or 10% acetic acid Additional reagents and equipment for solid-phase peptide synthesis using Boc (see Basic Protocol 1) or Fmoc chemistry (see Alternate Protocol), purification

Reactive groups Peptide

directiona

C N

N C

N

C

N

C

N

N

Ligation site

Electrophileb

O S-CH2 C-

S

Nucleophilec References O X-CH2 CX=Cl or Br

Lu et al., 1991

OO HC-C-d

Shao and Tam, 1995

O O S C C N H O O -C-CH2 ON=CH-C-

O -C

O -C-CH2 ONH2

OO HC-C-d

Shao and Tam, 1995

C

O O -C-NHN=CH-C-

O -C-NHNH2

OO HC-C-d

Shao and Tam, 1995

C

O -C

OO HC-C-d

Spetzler and Tam, 1995

O O N=N-CH2 -C- -C

SH NH2

NHNH2

Figure 18.5.5 Ligation methods for preparing MAP or cMAP by the indirect approach. Notes: (a) The direction in which the peptide is attached to the core matrix; (b) the functional group is placed at the end of the peptide; (c) the functional group is placed at the end of the branched Lys residues; (d) Generated for NalO4 oxidation of Ser of Thr.



Supplement 17

of peptides via RP-HPLC (UNIT 11.6), dialysis (see Support Protocol 2), and high performance gel-filtration chromatography (see Support Protocol 3) Prepare peptide antigen by solid-phase peptide synthesis 1. Synthesize the peptide using either Boc or Fmoc chemistry as in Basic Protocol 1 or the Alternate Protocol. For the ligation of peptide to a chloroacetylated MAP core matrix, an additional Cys should be attached to either C-termini (for peptide antigens derived from internal or N-terminal regions) or N-termini (for peptide antigens derived from the C-terminal region) of the peptide.

2. Purify the peptide using the methods in UNIT 11.6 or in Support Protocols 2 and 3 of this unit. Prepare chloroacetylated MAP core matrix 3. Prepare a MAP core matrix with four or eight lysine branches (see Basic Protocol 1, step 2), or Alternate Protocol, step 3). 4. Remove the Nα-terminal protecting group (see Basic Protocol 1, step 5, or Alternate Protocol, step 5). 5. Couple chloroacetic acid (4 molar equivalents) to the free amino groups of core matrix using 4 molar equivalents of DCC/DCM as a common amino acid coupling. This is done as in reactions 4 to 12 of Table 18.5.1 for Boc chemistry or reactions 4 to 9 of Table 18.5.2 for Fmoc chemistry, except that chloroacetic acid is being used instead of a protected amino acid.

6. Cleave the MAP core matrix from resin (see Basic Protocol 1, steps 6 to 9, for Boc-amino acids—using high HF cleavage—or Alternate Protocol, steps 6 to 8, for Fmoc-amino acids—using 95% (v/v) TFA in water as the cleavage solution). 7. Purify MAP core matrix using the methods in UNIT 11.6 or in Support Protocols 2 and 3 in this unit. Conjugate peptide antigen with chloroacetylated MAP core matrix 8. Dissolve 28 µmol peptide (from step 2) in 5 ml sulfhydryl reducing solution. Keep this solution under nitrogen and adjust pH to 6.0 to 6.5 with 1 M Tris⋅Cl. Flush the solution with nitrogen for 30 min, then seal the solution (no air contact) for 2 hr. Before the peptide is conjugated to the chloroacetylated MAP core matrix, the disulfide bond of the terminal Cys should be reduced to a sulfhydryl group by treatment with tributylphosphine (Bu3P). Dithiothreitol (DTT) cannot be used for this purpose because DTT may react with the chloroacetylated core matrix. Note that this reaction is performed off-resin and in solution phase. On-resin conjugation can be performed by modifying the sequence of these reactions and the use of an aqueous-compatible resin support, e.g., PEG-modified resin, which is available commercially.

9. Add 4 µmol of the tetrabranched chloroacetylated MAP core matrix (from step 7) to the peptide solution (from step 8). Adjust the pH to 8.5 to 9.0 with 1 M NaOH. Stir the mixture 45 min at room temperature. Synthesis and Application of Peptide Dendrimers As Protein Mimetics

10. Dialyze the MAP system conjugate sequentially against 2 liters of 0.1 M NH4HCO3 or 10% acetic acid and then deionized water at room temperature, 8 hr each change. Lyophilize to dryness and store at −70°C.



This dialysis procedure is different from that described in Support Protocol 2 because the peptides have already been purified.

11. Remove any byproducts by purifying the MAP conjugate by high-performance gel-filtration chromatography (Support Protocol 3). INDIRECT Boc AND Fmoc SYNTHESIS OF MAP SYSTEMS USING CARBONYL CHEMISTRY

BASIC PROTOCOL 3

Another method for preparing the MAP system by the indirect approach is to exploit the selectivity of other weak bases, particularly in the condensation reaction with aldehydes (Fig. 18.5.5; Shao and Tam, 1995). Two types of weak bases can be used. The first type consists of conjugated amines whose basicities are lowered by neighboring electron-withdrawing groups such as hydroxylamine and substituted hydrazines. These react with aldehyde to form oxime from hydroxylamine and hydrazones from hydrazines. The second type contains the 1,2-disubstituted patterns such as 1,2-aminoethanethiol derivatives of N-terminal cysteine and 1,2-aminoethanol derivatives of N-terminal threonine. Proline-like rings such as thiazolidine or oxazolidine are formed from the reaction between an aldehyde and cysteine or threonine, respectively. An aldehyde moiety on a peptide or MAP core matrix can be obtained by NaIO4 oxidation of N-terminal Ser, Thr, or Cys under neutral conditions to give an α-oxoacyl group. The additional weak base available for conjugation is attached to the N-termini of the peptide so the orientation of the peptide is C to N. However, the weak base-aldehyde chemistry has also been used to ligate constrained peptides to a MAP core matrix in the preparation of cMAP (see Basic Protocol 5). Materials Boc-aminooxyacetic acid (see recipe) or Boc-succinic acid hydrazide (see recipe) Benzotriazolyl N-oxytrisdimethylaminophosphonium hexafluorophosphate (BOP) (Richelieu Biotechnologies) Diisopropylethylamine (DIEA; Aldrich) 0.01 M sodium phosphate buffer, pH 7 (APPENDIX 2E) Sodium periodate (Aldrich) Ethylene glycol (Aldrich) 0.4 M sodium acetate buffer; mix equal volumes 0.4 M sodium acetate and 0.4 M acetic acid Dimethyl sulfoxide (DMSO; Aldrich) Dimethyl formamide (DMF) 0.1 M NaOH or glacial acetic acid Argon source C8 RP-HPLC column (see UNIT 11.6) Additional reagents and equipment for solid-phase peptide synthesis using Boc chemistry (Basic Protocol 1) or Fmoc chemistry (Alternate Protocol), and purification of peptides by semipreparative RP-HPLC (UNIT 11.6) Prepare the peptide by solid-phase peptide synthesis 1. Synthesize the peptide using Boc or Fmoc chemistry as in Basic Protocol 1 or the Alternate Protocol. 2. Remove the N-terminal protecting group (see Basic Protocol 1, step 5, or Alternate Protocol, step 5). 3. Couple (on-resin) the Cys residue (for thiazolidine ligation; 4 molar equivalents) to the free amino group of the peptide using 4 molar equivalents of DCC/DCM as a



Supplement 17

common amino acid coupling (see Table 18.5.1, steps 4 to 12, for Boc chemistry, or Table 18.5.2, steps 4 to 9, for Fmoc chemistry). Couple Boc-aminooxyacetic acid (for oxime ligation) or Boc-succinic acid hydrazide (for hydrazone ligation; 4 molar equivalents) to the amino terminus of the peptide using 4 molar equivalents of 1:1 (molar:molar) BOP/DIEA, dissolved in DMF, as coupling reagent in reaction 9 of Table 18.5.1 (in place of the DCC). For thiazolidine ring formation, an additional Cys residue is attached to the N-terminus of the peptide. For coupling through an oxime or hydrazone bond, the weak base is also placed at the N-terminus of the peptide.

Prepare the glyoxyl-MAP core matrix 4. Prepare the tetravalent MAP core matrix (Lys2-Lys-Ala) by the stepwise solid-phase synthesis method using either Boc chemistry (see Basic Protocol 1 and Table 18.5.1) or Fmoc chemistry (see Alternate Protocol and Table 18.5.2). Normal resin loading, ranging from 0.3 to 0.9 mmol/g, is recommended since only four coupling steps are required.

5. Remove the Boc group with 50% TFA/DCM (see Basic Protocol 1) or the Fmoc group with 20% pip/DMF (see Alternate Protocol). 6. Couple the fully protected Ser residue to the MAP core matrix using DCC and HOBt as described in step 3. 7. Cleave the Ser4-MAP core matrix from the resin by the high HF procedure, HF/pcresol (9:1, v/v), as described for Boc chemistry (see Basic Protocol 1, steps 6 to 9) or 95% (v/v) TFA in water as described for Fmoc chemistry (see Alternate Protocol, steps 6 to 8), and lyophilize the intermediate without purification. 8. Dissolve 5.86 µmol (5 mg) Ser4-MAP core matrix (from step 7) in 2 ml 0.01 M sodium phosphate buffer, pH 7, and add 46.9 µmol (10 mg) sodium periodate in a 2-ml microcentrifuge tube. 9. Stir the mixture for 5 min at room temperature, then quench the reaction by adding 93.8 µmol ethylene glycol (5.3 µl). 10. Isolate the product by semipreparative RP-HPLC as described in UNIT 11.6. Conjugate peptide with glyoxyl-MAP core matrix 11. Prepare a 5 mM peptide stock solution in water using the peptide from step 3 and a 5 mM glyoxy-MAP stock solution in water using the purified MAP from step 10. 12. Add 400 µl of 0.4 M sodium acetate buffer and 40 µl of the 5 mM glyoxy-MAP stock solution (0.2 µmol glyoxy-MAP) to 400 µl of the 5 mM peptide stock solution (2 µmol peptide). Then, add 800 µl DMSO in the case of oxime and hydrazone ligation and 800 µl DMF in the case of thiazolidine ligation. 13. Adjust pH of the solution to 5.7 (for oxime and hydrazone ligation) or 4.5 (for thiazolidine ligation) using 0.1 M NaOH or glacial acetic acid. 14. Stir the mixture under argon for 8 hr at room temperature, and monitor the completion of the ligation on a C8 RP-HPLC column (UNIT 11.6). Synthesis and Application of Peptide Dendrimers As Protein Mimetics

15. Purify the MAP conjugate by semipreparative HPLC (UNIT 11.6). Lyophilize to dryness and store at −70°C.



NINHYDRIN TEST The ninhydrin assay is used to check for completeness of the coupling reaction (Sarin et al., 1981). It is performed at the end of each synthesis cycle, usually as described here.

SUPPORT PROTOCOL 1

Materials Ninhydrin test reagents: Solution A/B mixture and Solution C (see recipe) 60% (v/v) ethanol in H2O 0.5 M tetraethylammonium chloride in dichloromethane (DCM) 10 × 75–mm test tubes 100°C heating block Pasteur pipet containing glass wool plug 1. At the end of each synthesis cycle (see Table 18.5.1 and Table 18.5.2), remove ∼10 mg wet resin from the reaction vessel. Dry the resin 1 hr under vacuum in a desiccator. 2. Weigh a 3- to 5-mg sample of dry resin into a 10 × 75–mm test tube. Add 100 µl of the solution A/B mixture and 25 µl solution C. To another tube add only the ninhydrin test reagents without resin. Mix. 3. Heat both tubes 10 min in a heating block at 100°C. Cool tubes in cold water. 4. Add 1 ml of 60% ethanol to the tubes. Mix thoroughly and filter through a Pasteur pipet containing a glass wool plug. 5. Rinse the resin twice with 0.20 ml of 0.5 M tetraethylammonium chloride in DCM. Combine the filtrates. 6. Dilute the combined filtrate to 2 ml with 60% ethanol. 7. Measure the A570 of the sample filtrate against the reagent blank. For routine monitoring, an effective extinction coefficient of 1.5 × 104 may be used.

PURIFICATION OF MAP SYSTEM BY DIALYSIS The chemistry of solid-phase peptide synthesis has been significantly improved since it was first introduced in 1963. Now synthesis of peptides of 100 or more residues with satisfactory yield and purity is possible. In general, crude synthetic MAP systems (after the cleavage step) are pure enough for many purposes (e.g., for production of antisera). When highly purified MAP systems are required (e.g., for vaccines for humans), the crude synthetic product must be purified to near homogeneity. MAP systems can be purified by dialysis, gel-filtration chromatography, RP-HPLC (UNIT 11.6), and high-performance gel-filtration chromatography (Support Protocol 3). This protocol details the special procedure required for the sequential dialysis of MAP systems.

SUPPORT PROTOCOL 2

Materials Crude MAP system (see Basic Protocol 1, 2, or 3, or Alternate Protocol) 0.1 M NH4HCO3/(NH4)2CO3 (pH 8.0) in 8 M and 2 M urea 1 M acetic acid Dialysis tubing (e.g., Spectra/Por 6, MWCO 1000, Spectrum) 1. Dissolve crude MAP system in 100 ml of 0.1 M NH4HCO3/(NH4)2CO3 (pH 8.0) in 8 M urea. 2. Load the peptide solution into dialysis tubing.



Supplement 17

3. Dialyze the peptide solution sequentially against 2 liters of each of the following solutions, for 8 hr each, at room temperature: 0.1 M NH4HCO3/(NH4)2CO3 (pH 8.0) in 8 M urea 0.1 M NH4HCO3/(NH4)2CO3 (pH 8.0) in 2 M urea H2O 1 M acetic acid. 4. Remove peptide solution from dialysis tubing. Lyophilize to dryness. Store at −70°C. SUPPORT PROTOCOL 3

PURIFICATION OF MAP USING HIGH-PERFORMANCE GEL-FILTRATION CHROMATOGRAPHY In this protocol, a crude MAP system is dissolved in potassium phosphate and purified over a specialized gel-filtration column by HPLC. Materials 10 mg crude MAP system (Basic Protocol 1, 2, or 3, or Alternate Protocol) 0.1 M potassium phosphate, pH 7.0 (APPENDIX 2E) Column for high-performance gel-filtration chromatography (prepacked Bio-Sil TSK 250, 300-mm length × 7.5-mm i.d.; mol. wt. range for separation: 10 to 300 kDa; Bio-Rad #125-0062) Guard column (75-mm length × 7.5-mm i.d.; Bio-Rad #125-0061) Additional reagents and equipment for HPLC (UNIT 11.6) 1. Dissolve 10 mg crude MAP system in 1 ml of 0.1 M potassium phosphate, pH 7.0. 2. Set up HPLC system with the Bio-Sil TSK column linked to the guard column. Set UV detector at 225 nm. Equilibrate the gel-filtration column with 0.1 M potassium phosphate, pH 7.0, at a flow rate of 0.5 ml/min. 3. Inject 200 µl of the crude MAP solution (from step 1). Elute peptide at a flow rate of 0.5 ml/min. Collect the purified MAP product. 4. Repeat injection (step 3) four times to purify all of the material. Pool the purified MAP product. 5. Lyophilize MAP fraction to dryness. Store MAP at −70°C.

BASIC PROTOCOL 4


DIRECT SYNTHESIS OF END-TO-SIDE CHAIN cMAP This approach is suitable for preparing cMAP with multiple end-to-side chain cyclic peptides. It combines solid-phase (on-resin synthesis) with solution-phase synthesis (off-resin cyclization; Spetzler and Tam, 1996). First, stepwise solid phase synthesis is used to assemble the peptides linked to the MAP core matrix on resin. The cMAP precursor is then cleaved from the resin, and its protecting groups removed. The peptide antigens are cyclized in cMAP in aqueous solution. The chosen sequences were peptide antigens derived from the V3 loop of the surface protein gp120 of human immunodeficiency virus-1 (HIV-1), MN strain. Two peptides with the sequence GPGRAFYTTKNIIG and KRKRIHIGPGRAFYTTKNIIG were used. The essential motif GPGR was placed in the middle of the antigen sequence. The cMAP precursor contains a tripartite design (Fig. 18.5.6). The first two parts are constant regions. Part 1 at the C-terminus contains a short peptide which can be used for identification, delivery to give an MHC class I molecule or a T-helper epitope, followed by a tetralysyl core matrix. Part 2 contains a specific chemical cleavage site AspPro that gives a monomeric cyclic peptide to verify the yield of the intrachain cyclization. This part may not be necessary if verification is



Self-assembly of multiple cyclic antigen peptide

-Lys -Lys

Lys-Ala-

stepwise synthesis

Ser(tBu) Cys(StBu)

Lys-(Asp-Pro)-MPC-

MPC TFA

Ser Lys-(Asp-Pro)-MPC

Cys(StBu) H

O

O NaIO4

Lys-(Asp-Pro)-MPC

Cys(StBu) H

R3P

O

SH NH2-Cys S N H

pH 5.6

O

Lys-(Asp-Pro)-MPC

O Lys-(Asp-Pro)-MPC

= HOCH2

OCH2

= peptide

R = CH2 CH2CO2H

Figure 18.5.6 The scheme for preparing cMAP using the direct method.

not needed. The third part is the variable region and contains a peptide framed by a lysine at the COOH-terminus (ε-Lys) and a Cys at the amino terminus to give the end-to-side chain cyclic structure. The two mutually reactive functions are the α-Cys and a masked aldehyde attached to the side chain of the C-terminal lysine. Conversion to the aldehyde under mild aqueous conditions will permit an end-to-side chain cyclization via thiazolidine (Fig. 18.5.6). The protecting strategy for the two reactive moieties, Cys and aldehyde, is designed to be compatible with Fmoc chemistry (Alternate Protocol and Table 18.5.2) and the removal of these moieties under aqueous conditions. Sulfenyl tertbutyl is chosen for Cys as Cys(StBu) because it can be removed under aqueous conditions by trialkylphosphine and is stable to oxidation by periodate. The masked form of the aldehyde is the 1,2-aminoalcohol of a Ser residue anchored on the lysyl side chain. It can be converted to an aldehyde by periodate oxidation. The use of trialkylphosphine in the scheme has two advantages. First, it inhibits intermolecular disulfide formation as a side reaction, and second, it allows a one-pot reaction for unmasking the thiol protecting group and cyclization. Materials Fmoc-Lys[methyltrityl(Mtt)] (Calbiochem-Novabiochem) Fmoc-Lys(Fmoc) (Bachem California) Fmoc-Cys(StBu) (Calbiochem-Novabiochem) 1% (v/v) TFA/5% (v/v) triisopropylsilane (TIS; Aldrich) in DCM Cleavage solution: 92.5/2.5/2.5/2.5 (v/v) trifluoroacetic acid (TFA)/triisopropyl silane (TIS) (Aldrich)/thioanisole (Aldrich)/H2O, prepared fresh 0.01 M sodium phosphate buffer, pH 6.8 (APPENDIX 2E)



Supplement 17

Sodium periodate (Aldrich) Tris(2-carboxyethyl)phosphine hydrochloride (TCEP; Calbiochem-Novabiochem) 10 mM sodium phosphate, pH 6.8 (APPENDIX 2E) 10 mM sodium acetate buffer, pH 4.2 (APPENDIX 2E) 70% (v/v) formic acid in H2O Additional reagents and equipment for peptide synthesis (Alternate Protocol and Table 18.5.2), semipreparative HPLC (UNIT 11.6), and assay of free sulfhydryls with Ellman’s reagent (UNIT 18.3) Perform on-resin synthesis of the cMAP precursor 1. Synthesize the cMAP precursor starting from Fmoc-Ala-Wang resin (0.5 g, 0.1 mmol/g) manually or via an automated peptide synthesizer using the synthesis reagents and directions outlined in Table 18.5.2 and described in Alternate Protocol. The order of assembly of the precursor is: the MAP core matrix, the cleavage site Asp-Pro, Fmoc-Lys(Mtt), peptide sequence, and Fmoc-Cys(S-tBu). See Figure 18.5.6.

2. Remove (on-resin) the side-chain-protecting 4-methyltrityl (Mtt) group of Lys using 1% TFA/5% TIS in DCM using the procedure described in step 6 of the Alternate Protocol. Discard the TFA/TIS/DCM solution and neutralize the TFA-amine salt on-resin by performing reactions 4 to 7 of Table 18.5.1. 3. Couple Fmoc-Ser(tBu) to the lysyl side chain by DCC and HOBt procedure as a common amino acid coupling as in reactions 6 to 9 of Table 18.5.2. 4. Deprotect and cleave the precursor of cMAP from the resin by 92.5/2.5/2.5/2.5 (v/v) TFA/TIS/thioanisole/H2O as outlined in the Alternate Protocol, steps 5 to 10. Lyophilize the peptide. Avoid thiol scavengers due to the presence of Cys(S-tBu).

Perform off-resin cyclization to form cMAP 5. Dissolve 0.44 µmol cMAP precursor (from step 4) in 2 ml 0.01 M sodium phosphate buffer, pH 6.8, and add 3.52 µmol (0.75 mg) sodium periodate in 27 µl water. 6. Stir the solution for 2 min, then purify the peptide by semipreparative HPLC as described in UNIT 11.6. Lyophilize the oxidized cMAP precursor. 7. Add 17.6 µmol (1.5 mg) TCEP in 47 µl of 10 mM sodium acetate buffer, pH 4.2, to the solution of oxidized cMAP in 1.5 ml of 10 mM sodium acetate buffer, pH 4.2. 8. Stir the mixture at 22°C for 48 hr, and monitor the progress of the cyclization by assay of free sulfhydryls with Ellman’s reagent (UNIT 18.3) until all the thiols are consumed. 9. Purify cMAP by semipreparative HPLC as described in UNIT 11.6. 10. For determining the intrachain yield, add 0.5 ml 70% formic acid to 0.05 µmol of the lyophilized cMAP in a glass test tube. Incubate the solution either at 37°C for 48 hr or 60°C for 24 hr. Add 0.5 ml water and lyophilize the sample. This procedure will cleave the cyclic peptide at the AspPro site as a monomer.

11. Separate the monomeric sample by RP-HPLC (UNIT 11.6).


The intrachain yield is calculated from the amino acid hydrolysis of the cMAP and the cyclic monomer.



INDIRECT Fmoc SYNTHESIS OF SIDE CHAIN-TO-SIDE CHAIN cMAP USING CARBONYL CHEMISTRY

BASIC PROTOCOL 5

In this approach, a side chain-to-side chain cyclized peptide is attached to the core matrix through a lysyl side chain (Pallin and Tam, 1996). The sequence used is a peptide antigen chosen from the neutralizing determinant of the V3 loop of gpl20. The target amino acid sequence is IGPGP. The precursors, which contain three Lys residues, are introduced as Fmoc-Lys(Mtt) and Boc-Lys(Fmoc), respectively, for attaching three functional groups (hydroxylamine, Ser, and Cys(StBu) through the side chains (Fig. 18.5.7). An O-alkylhydroxylamine group is attached to the side chain of the first Lys before the peptide sequence. The second Lys is introduced after the peptide sequence, and a Ser moiety is coupled to the side chain as a masked aldehyde, which is oxidized to an aldehyde by NaIO4. A cyclic peptide is formed by using the intramolecular oxime formation from the reaction between O-alkylhydroxylamine and the aldehyde. A Cys(StBu) residue on the

A

ONHBoc stepwise synthesis

Ser(tBu)

O=C

Cys(StBu) ONH2 O=C TFA

-Lys-Ala-

Boc-Lys-Ala-Lys

1 Ser

Lys-Ala-Lys

Lys-Ala-OH

Cys(StBu) 2 ONH2 O=C NaIO4

Lys-Ala-Lys

H

O

O Lys-Ala-OH

Cys(StBu)

3 Lys-Lla-Lys O=C Cys ON=CHC=O

pH 5.5 TCEP

Lys-Ala-OH

4

B SH NH2

HCOCO HCOCO-Lys HCOCO-Lys

Lys-Ala-OH

4

Y Y-Lys Y-Lys

HCOCO

= HOCH2

Lys-Ala-OH

Y

OCH2

Y=

S HN O

SH NH2 4

Lys-Ala-Lys

=

O=C Cys ON=CHC=O Lys-Ala-OH

Figure 18.5.7 The scheme for preparing cMAP utilizing carbonyl chemistry and using the indirect method. (A) Oxime approach for preparing unprotected cyclic peptide antigen. (B) Attachment to MAP via thiazolidine linkage.



Supplement 17

side chain of the third Lys moiety is used to ligate the cyclic peptides to the aldehyde moieties of the MAP core. Materials Fmoc-Lys[methyltrityl(Mtt)] (Calbiochem-Novabiochem) 1% (v/v) TFA/5% (v/v) triisopropylsilane (TIS; Aldrich) in DCM 20% piperidine in DMF (pip/DMF) Boc-Ser(tBu; Bachem California) Boc-(aminooxy)acetic acid (Bachem California) Boc-Lys(Fmoc; Bachem California) 0.01 M sodium acetate buffer, pH 7 (APPENDIX 2E) Sodium periodate (NaIO4;Aldrich) Ethylene glycol 0.1 M NaOH Tris(2-carboxyethyl)phosphine hydrochloride (TCEP; Calbiochem-Novabiochem) 0.05 M sodium acetate buffer, pH 6 (APPENDIX 2E) Tetravalent MAP core containing aldehyde groups (see Basic Protocol 3) Acetic acid Additional reagents and equipment for peptide synthesis (see Alternate Protocol and Table 18.5.2), semipreparative and preparative HPLC (UNIT 11.6), and mass spectrometry (Chapter 16), Perform on-resin synthesis of the cyclic precursor 1. Synthesize the linear unprotected peptide by the stepwise solid phase strategy using the Fmoc chemistry protocol (see Alternate Protocol and Table 18.5.2) and couple the first Lys residue as Fmoc-Lys(Mtt). 2. Remove the Mtt protecting group using 1% TFA/5% TIS in DCM and couple Boc-Ser(tBu) to the side chain of Lys as described in step 2 of Basic Protocol 4. 3. Synthesize the peptide sequence using the Fmoc strategy as described in the Alternate Protocol. 4. Incorporate the second Lys residue as the Fmoc-Lys(Mtt) derivative. 5. Couple Boc-(aminooxy)acetic acid to its side chain. 6. Couple the third Lys as a Boc-Lys(Fmoc) derivative. 7. Remove the Fmoc group by 20% pip/DMF (see Alternate Protocol). 8. Couple Fmoc-Cys(StBu) to the side chain of lysine. 9. Cleave and purify the peptide precursor as described for Fmoc chemistry (Alternate Protocol). Perform off-resin cyclization to form side chain-to-side chain cyclic peptides via an oxime linkage 10. Dissolve the peptide in 0.01 M sodium acetate buffer, pH 7, to give a final concentration of 0.8 M.


11. Oxidize the 1,2-amino alcohol moiety to the glyoxyl derivative by addition of a 2 M excess of NaIO4 for 2 min at room temperature. Quench reaction with a >20-fold excess of ethylene glycol. 12. Purify the peptide by semipreparative HPLC (UNIT 11.6) to remove the formaldehyde byproduct.



13. Collect the product from semipreparative HPLC and adjust pH to 5.5 with 0.1 M NaOH, to effect the oxime cyclization. 14. Monitor the progress of the intramolecular cyclization by RP-HPLC (UNIT 11.6). The reaction is usually completed within 12 hr.

15. Characterize the cyclic peptide by mass spectrometry (Chapter 16). Attach cyclic peptides to the MAP core matrix via a thiazoline linkage 16. Add 1.5 µmol (0.13 mg) TCEP to a solution of 0.3 µmol cyclic peptide in 0.5 ml of 0.05 M sodium acetate buffer, pH 6. Stir the mixture for 1 hr at room temperature. 17. Add 0.036 µmol tetravalent MAP core containing aldehyde groups (Basic Protocol 3) to the above mixture and adjust pH to 5 with acetic acid. 18. Stir the solution for 36 hr at 18°C, and purify the product by preparative HPLC as described in UNIT 11.6. INDIRECT PREPARATION OF END-TO-END cMAP USING THIOL CHEMISTRY

BASIC PROTOCOL 6

In this protocol, the cyclic peptide is attached through its side chain to the core matrix (Zhang and Tam, 1997). The selected sequences were YGGFL and AVEIQFMHNLGK. First, the precursor of the unprotected cyclic peptide and the functionalized core matrix are synthesized by SPPS (Fig. 18.5.8). The cyclic peptide is generated by an intramolecular transthioesterification reaction in which the cyclization occurs under equilibration of ring chain tautomerization in aqueous solution. Sequentially, the end-to-end cyclic peptide is formed via an S- to N-acyl transfer reaction, and a Cys residue is generated. In this method, racemization is minimized, and the cyclization can occur in high concentration. The generated free thiol group of the Cys residue is then available for ligation. Preparation of the (chloroacetyl)lysyl MAP core matrix is described in Basic Protocol 2. In the second step, the thioether bond is formed between the cyclic peptide and the core matrix at slightly basic pH. Materials 3-thiopropionic acid (Aldrich) MBHA (p-methyl benzhydrylamine) or Gly-PAM resin (Calbiochem-Novabiochem) 1 mM dithiothreitol (DTT; Aldrich) in distilled water 20 mM Na2HPO4/10mM citric acid buffer, pH 7.5 Tris(2-carboxyethyl)phosphine, hydrochloride (TCEP; Calbiochem-Novabiochem) 0.2 M sodium phosphate buffer, pH 7.4 (APPENDIX 2E)/10 M ethylenediaminetetraacetic acid (EDTA; Aldrich) Argon source Tetravalent chloroacetyl(lysinyl) core matrix (see Basic Protocol 2) Dimethylformamide (DMF) Additional reagents and equipment for peptide synthesis (see Basic Protocol 1), preparative RP-HPLC (UNIT 11.6), and mass spectrometry (Chapter 16) Perform on-resin synthesis of the cyclic precursor 1. Couple 3-thiopropionic acid to 0.5 g MBHA or Gly-PAM resin at 1.1 mmol/g by DCC/DCM in a reaction vessel (see Basic Protocol 1). Boc-Gly-PAM resin must be deprotected to give a free α-amine. The idea of this step is to form a reversible linker containing a thioester at the C-terminus after HF cleavage. For



Supplement 17

A stepwise synthesis 4-MeBzl-S Boc-NH

O HF

HS H2N

1

O NH2

SR pH 7.2 O

S

O 3

2

O

O

SH NH O 4

B SH

ClCH2 CO ClCH2 CO-Lys Lys-Ala-OH ClCH2 CO-Lys ClCH2 CO

SCH2 CO SCH2 CO-Lys

4 pH 8.5

SCH2 CO-Lys SCH2 CO

= HS-(CH2)2 CO-MBHA resin

Lys-Ala-OH

5

O

SH

R = (CH2)2 CONH2

SH

= 4

NH O

Figure 18.5.8 The scheme for preparing cMAP utilizing thiol chemistry and using the indirect method. (A) Thioester approach for preparing unprotected cyclic peptide antigen. (B) Attachment to MAP via thioalkylation.

an alternative method to prepare this resin, see recipe for thioester resin (HS-CH2CH2COMBHA-resin 1) for C-terminal thioester.

2. Add 1 mM DTT to the resin for 1 hr. Disulfide bonds may have been formed during the acylation reaction.

3. Synthesize the peptide sequence by stepwise solid phase synthesis (also see Fig. 18.5.7) either with an automated peptide synthesizer or manually using the synthesis reagents and directions outlined in Table 18.5.1 and Basic Protocol 1. Avoid the use of base-mediated coupling reagents such as benzotriazol-1-yl-oxy-tris(dimethylamino)phosphonium hexafluorophosphate (BOB), 2-(IH-benzotriazol-1-yl)-1, 1, 3,3-tetramethyluronium tetrafluorohorate (TBTU), and 2-(IH-benzotriazol-1-yl)1,1, 3,3-tetramethyluronium hexafluorophosphate (HBTU), since the thioester is not stable to base. Synthesis and Application of Peptide Dendrimers As Protein Mimetics

4. Cleave the peptide thioester by the HF protocol as described in Basic Protocol 1. Purify the product by preparative RP-HPLC (UNIT 11.6). Lyophilize and characterize the product by mass spectrometry (Chapter 16).



If the His residue is present in the sequence, use Boc-His(Tos) instead of Boc-His(DNP) which requires thiolysis to release the DNP protecting group.

Perform off-resin cyclization to form an end-to-end cyclic peptide containing an XCys ligation site 5. Dissolve 3.5 µmol linear peptide thioester (from step 4) in a solution of 2 ml of 20 mM Na2HPO4/10mM citric acid buffer, pH 7.5, containing 3.5 µmol (1.04 mg) TCEP. 6. Stir the solution for 6 hr, and monitor the progress of the intramolecular cyclization by RP-HPLC (UNIT 11.6). 7. Purify the cyclic peptide by preparative RP-HPLC as described in UNIT 11.6. Characterize and lyophilize the product. Attach cyclic peptides to the MAP core via a thioether linkage 8. Dissolve 28 µmol cyclic peptide (from step 7) in 1.25 ml of 0.2 M sodium phosphate buffer containing 10 µM EDTA, pH 7.4. 9. Purge the solution with argon for 10 min. 10. Add 1.4 µmol (1.1 mg) tetravalent (chloroacetyl)lysinyl core matrix (prepared as described in Basic Protocol 2) predissolved in 0.75 ml DMF. 11. Stir the mixture for 24 hr at room temperature, then purify the cMAP by preparative RP-HPLC as described in UNIT 11.6. 12. Characterize cMAP by mass spectrometry (Chapter 16). PREPARING END-TO-END CYCLIC PEPTIDES VIA AN X-CYS LIGATION SITE

SUPPORT PROTOCOL 4

In this protocol, an unprotected cyclic peptide containing an X-Cys ligation site is prepared (Liu and Tam, 1997). The sequence is a peptide antigen derived from the V3 loop of the gp120 of IIIB. The peptide precursor is synthesized on a thioester benzhydryl resin which, after HF cleavage, generates Cα-thiocarboxylic acid. A Cys residue activated as Cys(Npys) is incorporated at the amino-terminus. First, the end-to-end peptide is cyclized from formation of a mixed acyl disulfide which undergoes an intramolecular S,N-acyl transfer reaction. The generated hydrodisulfide (S-SH) can be reduced by thiolysis to give the native Cys-residue. Materials Thiobenzhydryl resin for C-terminal thioacid (see recipe) Boc-Cys(Npys) (Bachem California) 25% acetonitrile/0.05% TFA in H2O, pH 2 Sodium acetate, solid Dithiothreitol (DTT; Aldrich) Additional reagents and equipment for peptide synthesis (see Basic Protocol 1 and Table 18.5.1) and RP-HPLC (UNIT 11.6) 1. Perform on-resin synthesis of the cyclic precursor containing an X-Cys ligation site. Synthesize the precursor of the cyclic peptide on a thiobenzhydryl resin using the Boc-chemistry protocol as described in Basic Protocol 1 and outlined in Table 18.5.1. Cap the amino-terminal with Boc-Cys(Npys), and cleave and deprotect the peptide by HF. The resulting precursor is a peptide with a Cys(Npys) residue at the N-terminal and a thiocarboxylic acid at the C-terminal. The other side chains are unprotected.



Supplement 17

2. Initiate the cyclization after HF cleavage (see Basic Protocol 1) in 25% acetonitrile in H2O containing 0.05% TFA, pH 2, giving a final concentration of 1.75 µM with respect to the peptide solution. Monitor the progress of the cyclization by RP-HPLC (UNIT 11.6). During the cyclization, a sulfur-sulfur exchange occurs first and a yellow color appears due to the release of Npys-H.

3. Adjust the solution to pH 6 by using solid sodium acetate to complete the cyclization reaction. 4. After 10 min, add 1 eq DTT to the solution to reduce the cysteinyl hydrodisulfide to cysteine, and monitor the reaction by RP-HPLC (UNIT 11.6). 5. Purify the cyclic peptide by preparative RP-HPLC as described in UNIT 11.6. SUPPORT PROTOCOL 5

PREPARING END-TO-END CYCLIC PEPTIDES VIA AN X-THIAPROLINE LIGATION SITE In this example, the unprotected peptide precursor is synthesized on an Fmoc-amino acid-O-CH2-cyclic acetal resin (Botti et al., 1996). The peptide sequence is as follows: GRAFVTIGKIG derived from the V3 loop of gpl20, HIV-1. During the TFA cleavage, a glyceric ester is generated at the C-terminal. The diol of the glyceric ester is converted to glycoaldehyde by oxidation with NaIO4. The precursor contains a Cys(StBu) residue at the N-terminus while all the other side chains are unprotected. Two steps are involved to form thiaproline. First, the thiazolidine ring is generated from the reaction between an aldehyde moiety and the 1,2-aminothiol of the N-terminal Cys. Then, an O- to N-acyl transfer reaction occurs through a tricyclic ring contraction to give a peptide bond. The thiazolidine ring formation is mediated after NaIO4 oxidation and removal of the StBu protecting group, and the intramolecular O- to N-acyl transfer reaction occurs upon readjusting the pH to 5.9. Materials Fmoc-Gly-O-CH2-cyclic-acetal resin (see recipe for acetal resin) Cleavage solution: 91:3:3:3 (v/v) TFA/H2O/thioanisole/anisole NaIO4 0.01 M sodium phosphate buffer, pH 5.5, 5.7 and 6 (APPENDIX 2E) Tributyl phosphine (Bu3P) Isopropyl alcohol Additional reagents and equipment for peptide synthesis (see Alternate Protocol 2 and Table 18.5.2) and RP-HPLC (UNIT 11.6) 1. Synthesize end-to-end cyclic peptides via a thiaproline ligation site. Assemble the peptide sequence on Fmoc-Gly-O-CH2-cyclic-acetal resin by Fmoc chemistry using stepwise phase synthesis as described in the Alternate Protocol and Table 18.5.2. Cap the amino-terminal with Boc-Cys(StBu). 2. Cleave and deprotect the precursor as in the Alternate Protocol, steps 6 to 10, using a cleavage solution containing 91:3:3:3 TFA/H2O/thioanisole/anisole and allowing the reaction procedure to proceed for 2.5 hr. Lyophilize the product (Alternate Protocol, step 10).


3. Add 30 µmol NaIO4 in 0.3 ml 0.01 M sodium phosphate buffer, pH 5.7, to 7 µmol peptide glyceric ester also dissolved in 1 ml 0.01 M sodium phosphate buffer, pH 5.7. Stir the solution for 30 min, and purify the product by semipreparative RP-HPLC (UNIT 11.6).



4. Dissolve 1.18 µmol peptide-glycoaldehyde (from step 3) in 0.2 ml 0.01 M sodium phosphate buffer, pH 5.5, and add 10.9 µmol tributyl phosphine (Bu3P) in 30 µl isopropyl alcohol (1:10 v/v dilution of Bu3P in isopropyl alcohol). Stir and monitor the progress of the cyclization by analytical RP-HPLC (UNIT 11.6). Purify and lyophilize the product. The starting product forms two products corresponding to the cis and trans isomers of the cyclic peptide.

5. Redissolve the purified peptide in 0.2 ml 0.01 M sodium phosphate buffer, pH 6, and monitor the amide formation by analytical RP-HPLC (the O- to N-acyl transfer can take up to several days). If the reaction is slow, warm up the solution to 40°C. Purify the product and analyze by HPLC. PREPARING END-TO-SIDE CHAIN CYCLIC PEPTIDES VIA A THIAZOLIDINE OR OXIME LINKAGE

SUPPORT PROTOCOL 6

In these two examples (Pallin and Tam, 1995; Botti et al., 1996), the precursor contains a Lys residue which is introduced as an Fmoc-Lys(Mtt) for attaching Ser through its side chain. The sequences are peptide antigens ranging from 5 to 26 amino acids in length and are derived from the V3 loop of gp120, HIV-1. The 1,2-amino alcohol function of Ser acts as a masked aldehyde and can be converted to glyoxyaldehyde after treatment with NaIO4. A Cys(StBu) moiety is incorporated at the N-terminus to generate a thiazolidine ring while Boc-(aminooxy)acetic acid is incorporated to form an oxime bond. All the other side chains are unprotected. The thiazolidine ring is formed by first oxidizing the amino-terminal Ser to an aldehyde moiety and then removing the StBu protection group by TCEP. In generating cyclic oxime peptides, cyclization occurs after oxidation to the aldehyde. Materials Fmoc-Lys(Mtt) 1% (v/v) TFA/5% (v/v) triisopropylsilane (TIS; Aldrich) in DCM Boc-Ser(tBu) Cleavage solution: 92.5/2.5/2.5/2.5 (v/v) TFA/TIS/thioanisol/H2O Sodium periodate 0.01 M sodium phosphate buffer, pH 6.8 (APPENDIX 2E) 0.01 M sodium acetate buffer, pH 4.2 (APPENDIX 2E) Tris(2-carboxyethyl)phosphine, hydrochloride (TCEP; Calbiochem-Novabiochem) Additional reagents and equipment for peptide synthesis (see Alternate Protocol and Table 18.5.2), RP-HPLC (UNIT 11.6), and mass spectrometry (Chapter 16) 1. Synthesize end-to-side chain cyclic peptides via a thiazolidine or oxime linkage. Assemble the peptide segment on a Wang resin by the Fmoc-chemistry protocol using stepwise peptide synthesis as described in the Alternate Protocol and outlined in Table 18.5.2. The order is as follows: X-protected peptide sequence-Lys[Boc-Ser(tBu)]βAla, where X = Boc-Cys(StBu) or Boc-NH-O-CH2-CO-.

2. Couple the Lys residue as Fmoc-Lys(Mtt) and remove the Mtt protection group with 1% TFA,/5% TIS in DCM. Couple Boc-Ser(tBu) to the side chain of Lys (Alternate Protocol). 3. Cleave and deprotect the peptide (see Alternate Protocol, steps 6 to 10) using 92.5/2.5/2.5/2.5 TFA/TIS/thioanisol/H2O and allowing the reaction to proceed for 1 hr. Lyophilize the crude product (Alternate Protocol, step 10).



Supplement 17

4. Add 1.55 µmol sodium periodate (0.33 mg) in 3.5 µl H2O to 0.76 µmol peptide precursor dissolved in 1 ml 0.01 M sodium phosphate buffer, pH 6.8. Shake the solution for 2 min, and purify the product by semipreparative RP-HPLC (UNIT 11.6). For oxime formation, the intramolecular cyclization occurs after oxidation to the aldehyde. Purify and characterize the product.

5. Dissolve 0.94 µmol peptide in 1 ml 0.01 M sodium acetate buffer, pH 4.2, and add 9.3 µmol TCEP (2.3 mg) dissolved in 23 µl 0.01 M sodium acetate buffer, pH 4.2. Stir the solution, and monitor the cyclization by analytical RP-HPLC (UNIT 11.6). After purification by semipreparative RP-HPLC (UNIT 11.6), characterize the product by mass spectrometry (Chapter 16). REAGENTS AND SOLUTIONS Use Milli-Q-purified water or equivalent for the preparation of all buffers. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Acetal resin Add 0.2 g sodium bicarbonate to a suspension of chloromethylated polystyrene (1% divinylbenzene) copolymer (Merrifield resin, purchased, e.g., from Bachem California; 0.76 mmol/g, 0.5 g) in 3 ml DMSO. After stirring for 7 hr at 155°C, wash the resin with water, methanol, DMF, DCM, and diethyl ether and then dry under vacuum. To confirm the conversion of the chloromethyl resin to the aldehyde resin, use FT-IR analysis, which should show the typical carbonyl absorption at 1700.5 cm–1, and microanalysis, which should show that the resin does not contain any chlorine. The benzaldehyde resin (0.59 mmol/g, 1.0 g) is suspended in 10 ml anhydrous dimethoxyethane with 1 g glycerol and 10 mg p-toluenesulfonic acid. After 24 hr at 90°C, wash the resin with a 1:1 (v/v) mixture of 3% sodium bicarbonate and 1,4-dioxane, followed by water, dioxane, DMF, methanol, DCM, and diethyl ether. Dry under vacuum. Use FT-IR analysis to monitor the disappearance of the carbonyl absorbance at 1700.5 cm–1. To functionalize the acetal resin with a Fmoc-amino acid, activate the Fmoc-amino acid, e.g., Fmoc-Gly-OH (5.58 mmol, 1.66 g) using DCC (3 mmol, 0.62 g), dimethylaminopyridine (0.28 mmol, 35 mg), and HOBt (0.28 mmol, 44 mg) for 20 min in 10 ml DMF. Insoluble dicyclohexylurea is then removed, and the solution is added to a suspension of the glycerol-acetal resin (1.0 g) in 5 ml DMF. After 18 hr, wash the resin three times each with dioxane, DMF, methanol, DCM, diethyl ether and then dry under vacuum. Microanalysis will indicate 0.71% of nitrogen, corresponding to a substitution of 0.5 mmol/g. UV absorbance of the cleaved Fmoc (wavelength 301 nm, I = 7840) will indicate a substitution of 0.5 mmol/g.

Boc-aminooxyacetic acid Add 10 mmol carboxymethylamine hemihydrochloride (Aldrich) to a mixture of 20 ml dioxane, 10 ml water, and 10 ml 1 N NaOH on ice. Add 11 mmol di-t-butyl pyrocarbonate (Aldrich) to the above solution at 0°C and stir the solution overnight at room temperature. Adjust pH to 4 using diluted KHSO4 and extract the product twice, each time for 30 min, with ethyl acetate. Dry the organic phase over MgSO4. Evaporate the organic phase under vacuum, and recrystallize Boc-(aminooxy) acetic acid (Boc-NHNH2) from ethyl acetate–hexane. Synthesis and Application of Peptide Dendrimers As Protein Mimetics

Boc-succinic acid hydrazide On ice, dissolve 42 mmol Boc-NHNH2 (Aldrich) in 20 ml dichloromethane (DCM) and add 40 mmol succinic anhydride. Then add 10 mmol diisopropylethylamine (DIEA) to the above solution at 0°C, and stir the solution for 6 hr at room



temperature. Remove DCM under vacuum, and dissolve the residue in 60 ml ethyl acetate. Wash the solution twice with 15 ml 5% KHSO4, twice with 10 ml saturated NaCl solution, and finally with 10 ml water. Dry the organic phase over MgSO4, and evaporate the organic phase under vacuum. Wash the oily residue three times, each time with 20 ml ether-hexane and dry under vacuum. Cleavage solution for Fmoc protecting group 90 parts trifluoroacetic acid (TFA) 6 parts thioanisole 3 parts 1,2-ethanedithiol (EDT) 1 part anisole Prepare fresh and chill to 0°C The above reagents are available from Aldrich

Ninhydrin test reagents Solution A: Dissolve 40 g reagent grade phenol in 10 ml absolute ethanol. Solution B: Dissolve 65 mg potassium cyanide in 100 ml water. Dilute 2 ml of this solution to 100 ml with freshly distilled pyridine. Mix solution A and solution B, and store indefinitely at 4°C. Solution C: Dissolve 2.5 g ninhydrin in 50 ml absolute ethanol. Store solution C indefinitely, in the dark. Pyridine from any commercial source may be used to prepare solution B and must be distilled before use. Reflux 500 ml pyridine with 2 g ninhydrin for 2 hr, then distill the pyridine and store several months at 4°C, in a brown bottle (away from light).

Sulfhydryl reducing solution 4.25 ml 6 M guanidine⋅HCl 0. 25 ml 5% (w/v) EDTA (50 mg EDTA/ml) 0.5 ml 2% (w/v) tributylphosphine (Bu3P) in propanol Prepare just before use CAUTION: Tributylphosphine (Bu3P) is a highly toxic, volatile reducing agent and should only be handled in a chemical hood.

Thiobenzhydryl resin 2 for C-terminal thioacid Add thiolacetic acid (CH3COSH, 30 mmol; Aldrich Chemical) and triethylamine (30 mmol; Aldrich Chemical) to a suspension of 4-methylbenzhydryl chloride resin (6 g, 3 mmol) in 60 ml DMF. React the mixture at 60°C for 12 to 18 hr until a negative silver chloride test (see below) of the resin indicates that no free chloride groups are present. Wash the resin three times each with DMF, DCM, CH3OH, DCM, and DMF. Treat resulting resin with a mixture of cysteine methyl ester hydrochloride (9 mmol), triphenylphosphine (9 mmol), and DIEA (9 mmol) in DMF/DCM (3:1, v/v, 60 ml). After 2 hr, wash the resin thoroughly three times each with DMF, DCM, and CH3OH and dry under vacuum. The yield from this procedure should be 6.3 g. For the silver chloride test, remove ∼20 mg of resin to a small glass test tube, add 0.5 ml of pyridine, and heat at 100°C in a heating block for 15 min in a well-ventilated hood. Cool and dilute with an equal volume of 0.1 N nitric acid. Add a few drops of 0.1 N silver nitrate solution. The presence of silver chloride precipitate indicates an incomplete reaction.

Thioester resin (HS-CH2CH2CO-MBHA-resin 1) for C-terminal thioester Add 3-thiopropionic acid (1.2 ml, 13 mmol), HOBt (1.8 g, 13 mmol), and DIC (2 ml, 13 mmol) sequentially to a suspension of 4-methylbenzhydrylamine (MBHA) resin (6 g, 3.2 mmol) in 60 ml DMF. Shake the mixture at room temperature for 30 min until a ninhydrin test (see Support Protocol 1) of the resin indicates that no free



Supplement 17

amino groups are present. Wash the resin three times each with DMF, DCM, CH3OH, DCM, and DMF. Treat resulting resin with a mixture of cysteine methyl ester hydrochloride (0.55 g, 3.2 mmol), triphenylphosphine (0.85 g, 3.2 mmol), and DIEA (0.68 ml, 3.9 mmol) in DMF/DCM (3:1, V/V, 60 ml). After 2 hr, wash the resin thoroughly three times each with DMF, DCM, and CH3OH and dry under vacuum. The yield from this procedure should be 6.3 g.



The multiple antigen peptide (MAP) system was originally designed to replace the protein carrier and to overcome the limitations of conventional conjugation approaches (Tam, 1988, 1996; Posnett et al., 1988). The MAP system consists of three structural elements. A simple amino acid (e.g., alanine) is bound through a proper linkage to a solid-phase polymer to initiate the synthesis and serve as an internal standard. An inner core matrix is constructed from several levels of lysine to generate multiple reactive ends (two to eight ends) to which peptides can be linked. The surface layer is composed of either multiple copies of linear or cyclic synthetic peptides attached to the lysine inner-core matrix (Fig. 18.5.1). The MAP makes use of a limited sequential propagation of a trifunctional amino acid to form a low-molecular-weight core matrix. Lysine is suitable for this purpose because both Nα- and Nε-amino groups are available as reactive ends. The sequential propagation of lysine generates 2n reactive ends, each of which can serve as an attachment site for peptides. The complete MAP has been shown to be large enough to stimulate immune responses. The lysine core matrix is small; the bulk of the complex is formed by up to 16 peptides layered around the inner core matrix. As an example, a peptide dendrimer with an octabranched core matrix and eight copies of the peptide, each containing 14 amino acid residues, has a total molecular weight of 13 to 14 kDa. Only 7% of the mass is due to lysine residues of the core matrix; 93% of the mass is due to peptide sequences. The core matrix is oligomeric and contains noncationic peptidyl and isopeptidyl lysine amide on both Nα- and Nε-termini of lysine. As a result, the MAP core matrix is totally bioinactive and functions only as a template to keep the multiple peptides together. It is the dendritic peptide chains on the core matrix that are likely to be mobile and contribute to the biological activity of the MAP system. Recently, Wrighton and co-workers (Wrighton et al., 1997) have used β-Ala-Lys as

a pseudosymmetrical dendrimeric template and have attached two copies of the erythropoietin mimetic peptide (EMP) to the ε-amino group and the α-amino moiety, respectively. The two intramolecular disulfide bridges in EMP were formed by using an orthogonal protecting group scheme. As a result, the dimer has 100-times higher affinity for the erythropoietin receptor (EPOR) and 10-fold better antagonistic activity in vivo than the monomer. The MAP system is chemically unambiguous, and the method of synthesis provides accurate knowledge of conformation and quantity of the peptide. The MAP system is also versatile. If half of the reactive lysine amino chains (i.e., ε-amino groups) are chemically blocked during synthesis, a peptide dendrimer containing two different peptide sequences can be constructed. For example, Nα- and Nε-amino groups of lysine at the last level are orthogonally protected with t-butyloxycarbonyl (Boc) and 9-fluorenylmethyloxycarbonyl (Fmoc), respectively. The first peptide chain can be assembled on the Nα-termini of this special lysine core matrix using Boc chemistry. The Fmoc groups of Nε-termini are not affected during the Boc synthesis. After synthesis of the first peptide is complete, the second peptide chain can be synthesized on the Nε-termini using Fmoc chemistry. The entire MAP system with two different peptide sequences is then cleaved from resin by hydrogen fluoride (HF) cleavage. One such MAP system was used as the model for synthetic hepatitis B vaccines (Tam and Lu, 1989). The simplicity and flexibility of the design and synthesis of MAP also permit incorporation of a hydrophobic built-in adjuvant such as tripalmitoyl-S-glyceryl cysteine (P3C; Defoort et al., 1992). Constrained peptides that mimic native protein conformations are highly desirable because of the advantages in improving their biological activity. To achieve such constrained peptides, linear peptide sequences need to be circularized and then attached to a lysyl MAP core matrix. This approach requires two types of synthetic manipulations: a cyclization strat-



egy of a linear precursor and a chemoselective method to attach to the template. The conventional approach to achieving two synthetic components is usually cumbersome and requires a scheme of multi-tiered protecting groups. For example, conventional cyclization to form cyclic peptides requires a protection scheme for selected functional side chains. The reactions are carried out in organic solvent (off-resin) or solid phase (on-resin). After cyclization, the side-chain protecting groups are removed. This unit describes two simplified approaches (the direct approach and the indirect approach; also see Fig. 18.5.2) to form cMAPs through unprotected constrained peptides generated from their linear unprotected peptides under aqueous conditions. The authors’ two approaches do not use protecting groups and produce cMAPs directly for a variety of purposes, including antigen scanning and generation of constrained peptide libraries. The direct and indirect methods for preparing cMAP are the same as for MAP, and include an extra cyclization step after the linear unprotected precursors are synthesized. Synthetic peptides have been used extensively for production of site-specific antibodies for confirming the identity of proteins derived from recombinant DNA, for exploring biosynthetic pathways, and for defining precursorproduct relationships (e.g., proenzyme and preproenzyme) and structural domains of proteins (Lerner, 1982). So far, the MAP approach has been used in the preparation of experimental vaccines (Amon and Horwitz, 1992) against hepatitis (Tam and Lu, 1989), malaria (Tam et al., 1990; Pessi et al., 1991), foot-and-mouth disease (Francis et al., 1991), and HIV (Defoort et al., 1992; Nardelli et al., 1992). Various templates derived from peptide and organic compounds containing several peptide chains have been able to induce protein structural motifs like α-helix, β-sheet, β-turn, and loop conformations (Schneider and Kelly, 1995). Since the MAP system also has an increased ordered helical structure (Shao and Tam, 1995) and its synthesis is well established, MAP is suitable as a template to mimic proteins. Many important questions, such as how a certain structural motif can be induced and how the secondary structures interact with each other, can be addressed by using the authors’ MAP system. Artificial enzymes can also be designed and generated because certain tertiary structures associated with a specific enzymatic activity can be mimicked on the surface of the MAP core matrix.

Critical Parameters Chemistry of synthesis The MAP system can be effectively synthesized by both Boc and Fmoc chemistry (Tam and Spetzler, 1997). The selection of the synthetic method depends on several factors, including peptide sequence, instrumentation requirements, cost, and chemical wastes. The most important factor is probably peptide sequence. For sequences very rich in amino acids that are susceptible to strong acid modifications [e.g., alkylation (Cys, Met, Trp, and Tyr), acylation (Glu), and dehydration (Asp-Gly, AspSer)], Fmoc chemistry may provide better results than Boc chemistry. However, use of the low-high hydrogen fluoride cleavage technique has greatly reduced these types of side reactions. Sequences rich in Arg, Trp, Tyr, and Met, as well as peptides with proline as one of the first two residues from the C-terminus, may do poorly in Fmoc chemistry. Occasionally, a peptide cannot be successfully prepared by either of these two methods. A failed synthesis may be caused by sequence-dependent side reactions. The second factor that affects the choice of synthetic method is the instrumentation requirement. In general, Fmoc chemistry is easier in operation and requires less equipment. Boc chemistry normally needs special equipment for the final hydrogen fluoride cleavage of peptide from resin. If cost is a consideration, Boc chemistry is preferable because Bocamino acid derivatives are less expensive than Fmoc-amino acid derivatives. The nature and quantity of chemical wastes produced during synthesis should also be considered. The solvent, coupling reagent, and cleavage reagent wastes generated in Boc chemistry are acidic (due to TFA) and contain halogens (TFA, DCM). The waste generated by Fmoc chemistry is basic (due to piperidine) and is not halogen-containing; the DCM used in reaction 8 of the procedure (Table 18.5.2) is negligible in this regard. For the synthesis of cMAPs, the preparation of their linear precursors are often limited to either Boc or Fmoc chemistry depending on the cyclization method used (see discussion of Method of Synthesis, below). However, the same precautions have to be taken as for the synthesis of MAP. Method of synthesis Both direct and indirect approaches for preparing MAPs are presented in this unit. In the direct approach, the linear peptide sequence



Supplement 17


and MAP core matrix are synthesized in a single operation, directly on the resin matrix, by stepwise solid-phase peptide synthesis (Basic Protocol 1 and the Alternate Protocol). The core matrix (two or three levels of lysine) is formed first by coupling lysine to a small amino acid attached to a resin support. The peptide sequence is then synthesized stepwise on this branched-lysine scaffolding. After peptide synthesis, the MAP is cleaved from the resin support. In the indirect approach, the MAP core matrix and peptide are synthesized separately, purified, and then combined to form the complete MAP by one of a number of different ligation methods illustrated in Figure 18.5.5 (Basic Protocols 2 to 3). One advantage of the indirect approach is that the components— MAP core matrix and peptide—are purified before conjugation. Any impurity generated during solid-phase peptide synthesis can be removed; this may be difficult in the direct approach because of the unique structural characteristic of MAP. Preparation of cMAP using the direct approach can be complicated by competing intramolecular versus intermolecular cyclization. The competing reactions can be minimized for single-chain peptides under dilute conditions, but this is not possible for cMAPs because of simultaneous multiple peptide copies on the branched structure. Furthermore, extensive intermolecular cyclization among the precursors can also give polymers. The authors’ solution for avoiding side products and for simultaneously obtaining a self-assembly process to give multiple cyclic peptide is based on ring-chain tautomerization. Ring-chain tautomerization is an entropy process and is conformation-driven as well as concentration-independent, to favor intra- over intermolecular reactions. The initial bond formation is reversible in ring-chain tautomerism. Thermodynamically stable products are formed under equilibrating conditions suitable for self-assembly of the multiple peptide chains on the dendrimer to form end-toside chain cyclic peptides. In the synthetic scheme of cMAP, the ring-chain tautomerization for intramolecular cyclization is a thiazolidine ring generated from an N-terminal Cys and an aldehyde attached to the side chain of Lys (Fig. 18.5.6). Peptides with a motif that favor a cyclic structure are highly recommended in this approach. The authors’ schemes for preparing cyclic peptides by the indirect approach result in ligation sites of peptide bonds like X-Cys and X-SPro and surrogate peptide bonds such as

thiazolidine and oxime that both are metabolically stable (Fig. 18.5.3). These schemes also produce peptides which are cyclized through end-to-end, end-to-side chain, or side chain to side chain (Pallin and Tam, 1995, 1996; Botti et al., 1996; Zhang and Tam, 1997), depending on the strategic placement of the two intended sites that must react with each other. Although peptide and nonpeptide bonds are described, the chemistries governing their syntheses are rather similar. To generate unprotected cyclic peptides with peptide bonds such as X-Cys and X-SPro from linear unprotected precursors, the reaction must be regioselective (Liu and Tam, 1994; Tam, 1995). A common strategy for obtaining regioselectivity is entropic activation achieved by capturing the Nα- and Cα-termini at close proximity as a covalent bond allowing an intramolecular acyl shift and forming a peptide bond. In cyclization reactions, the capturing reaction between an Nα-amine and a weakly activated Cα-acyl moiety has to be specific in the presence of many other side-chain functional groups, including Nε-amine. Two methods are available to form a Cys bond in the ligation site. In one method, cyclization is achieved via an intramolecular transthioesterification of the linear precursor containing an N-terminal Cys and a C-terminal thioester. The thioester intermediate subsequently undergoes a proximity-driven S- to N-acyl transfer to give the desired amide bond, and introduces a thiol moiety (Basic Protocol 5). In the other method, the cyclization occurs via an acyl disulfide–mediated intramolecular acylation. The linear precursor consists of an N-terminal Cys protected and activated as Cys(Npys) and a C-terminal thiocarboxylic acid. The mixed disulfide undergoes an S to N-acyl transfer through a six-member ring and a native Cys residue is obtained after thiolytic reduction (Support Protocol 4). Cyclic peptides with a thiol moiety are useful as building blocks in the indirect approach for preparing cMAPs (see discussion of Method of Ligation, below). Also, cyclic peptides containing an SPro residue to mimic Pro can be generated (Support Protocol 2). The cyclization is mediated by an amine-aldehyde condensation. The linear precursor contains an N-terminal Cys and a C-terminal glycoaldehyde generated from NaIO4 oxidation of the corresponding glyceric ester. The Cys residue is protected as Cys(StBu) because of its sensitivity to NaIO4 oxidation. First, a thiazolidine ring is formed when the StBu protecting group is removed by phosphine. Then a proximity-driven O to Nacyl transfer reaction occurs through a tricyclic



ring contraction to afford the peptide bond. Cyclic peptides through side-chain-to-end and end-to-end can be achieved. In the preparation of cyclic peptides with peptide bond surrogates from totally unprotected linear precursors, chemoselective ligation based on carbonyl chemistry is used. Carbonyl chemistry exploits the selectivity of weak bases in the condensation with an aldehyde under acidic conditions where the basic sidechain nucleophiles are protected by protonation. For the purposes here, weak bases such as thiols, hydrazides, and amino-oxy groups are available, since they are easy to incorporate into the peptide during solid-phase synthesis. An aldehyde is obtained by NaIO4 oxidation of N-terminal Ser, Thr, or Cys under neutral conditions to give an α-oxoacyl group. A thiazolidine ring is formed from the reaction between the aldehyde and the 1,2-aminoethanethiol of N-terminal Cys. Under the same conditions, an amino group will form a reversible Schiff base with the aldehyde. The Cys residue, which is incorporated as Cys(StBu), can play a dual function in the synthetic scheme. First, it can be used for cyclization where the cyclization is mediated after deprotection of the thiol of Cys, as already mentioned above. Second, it can be applied for ligating unprotected cyclic peptides to a core matrix when the Cys residue is deprotected after the cyclic peptide is formed. Another weak base is aminooxy, which affords oxime at low pH. In the authors’ scheme, the aldehyde group is attached to the side chain of Lys while the nucleophilic aminooxyl or cysteinyl moiety is incorporated to either the amino terminal of the peptide or through the side chain of Lys to afford end-to-side chain or side chain-to-side chain cyclic peptides. Method of ligation The advantage of the indirect approach is that conjugation between the unprotected peptide and MAP core is chemically unambiguous. The functional groups on the peptide chain and core matrix can be designed to avoid reacting with other functional groups on the peptide. A variety of conjugation methods and cross-linking reagents may be used for conjugation to avoid modification of the peptide. The methods described in this unit are some of many variations (Fig. 18.5.5). A pair of mutually reactive functional groups such as a nucleophile and an electrophile are introduced on the monomeric peptide and MAP core matrix, respectively. The two functional groups only react specifically with each other. The methods that use thiol and

carbonyl chemistries are suitable for the preparation of both MAP and cMAP with nonpeptide bonds. A thiol group can be selectively reacted in the presence of amine and other side-chain groups because it is a stronger nucleophile, especially in neutral or acidic conditions where the amine group is protonated. Haloacetyl (Fig. 18.5.5) is a functional group used for selectively reacting with a thiol group under neutral to slightly basic pH conditions. However, under these conditions, the thiol tends to be oxidized to disulfide. Both unprotected linear (Basic Protocol 2) and cyclic (Basic Protocol 6) peptides contain a free thiol that has been ligated to chloroacetyl groups incorporated on the lysine core matrix. The indirect approach using thiol chemistry also allows the peptides to be attached to the MAP core matrix through either the amino or the carboxyl terminus, increasing the flexibility of the design. However, other functional groups such as maleimide or aromatic thiols like 2-thiopyridyl, nitropyridyl sulfenyl, or 4-dithiopyridyl, which also react specifically with a thiol, can be applied to ligate unprotected peptides to the MAP core matrix. Carbonyl chemistry that exploits the selectivity of weak bases in condensation with an aldehyde has already been mentioned (see discussion of Method of Synthesis, above) and is shown in Figure 18.5.5. Thiazolidine ring formation has been used to ligate both linear (Basic Protocol 3) and constrained peptides (Basic Protocol 5) to a MAP core matrix containing α-oxoacyl groups. Other types of weak bases are hydroxylamine and hydrazine, which afford oxime and hydrazone bonds with an aldehyde moiety. Oxime and hydrazone bonds in the ligation site have been suitable for linear unprotected peptides (Basic Protocol 3). Both types of weak bases are added to the N-terminus, so the peptide sequence will be oriented with the amino terminal residues proximal to the core matrix and the carboxyl terminal residues distal to core matrix. The conjugation reactions listed in Figure 18.5.5 have some limitations. The haloacetyl derivative may react with amine or other nucleophilic groups in peptides in basic medium. This type of reaction will lead to chemical ambiguity of the conjugate. To prevent this reaction, the pH of the conjugating solution should not be above 9.0. Aldehyde groups reacting with N-terminal cysteine rapidly form a stable thiazolidine derivative at pH 4 to 6. The resulting thiazolidine is quite stable under normal manipulating conditions, but the aldehyde



Supplement 17

itself is unstable. However, this can be circumvented by inert gas protection. Hydroxylamine and hydrazine react with aldehydes, forming an oxime and a hydrazone linkage, respectively, at pH ∼5. Both linkages are stable at physiological pH (6 to 8), but the hydrazone derivative will be hydrolyzed at a lower or higher pH. The rate of hydrolysis is pH dependent. Stability of the linkage can be increased by reducing the hydrazone bond to hydrazine with sodium cyanoborohydride (NaBH3CN).

Troubleshooting


After the peptide sequence is selected, MAP or cMAP with this special sequence can be synthesized by solid-phase synthesis. As with solid-phase peptide synthesis, the reactions of each cycle (including deblocking and coupling reactions) must be very efficient and should be driven to completion in a short time. The following calculations demonstrate the importance of high yield from each cycle. If the mean yield of each cycle of synthesis is 95%, the final yield for a peptide with 20 amino acid residues would be only 35.8%. However, if the yield of each cycle is 98%, 99%, 99.5%, or 99.9%, the total yields for the synthesis of a peptide with 20 residues would be 66.8%, 81.8%, 90.5%, or 98.0%, respectively. Furthermore, the byproducts produced in peptide synthesis usually closely resemble the desired peptide product in sequence and structure and are very difficult to remove by normal separation methods. Because our MAP system is a dendrimer, the degree of heterogeneity of peptide products will be much higher than that of solid-phase peptide synthesis. For example, in the synthesis of a MAP system with four branches, a 99% yield from each cycle results in 96% of MAP with the correct structure. If the mean yield of each cycle of synthesis is 99%, the synthesis of a four-branched MAP system with a peptide antigen of 20 amino acid residues will result in final yield of 44.2% with the correct sequence, instead of 81.8% obtained in usual solid-phase peptide synthesis. The major side reactions during solid-phase peptide synthesis are: deletion, where one or more amino acid residues are skipped during synthesis; termination, where chain elongation stops at a point because of acylation or other side reaction; and racemization, where the amino acid is racemized during peptide elongation. All these side reactions result in heterogeneity of peptide product, and decrease the total yield of synthesis. To overcome these side reactions, excess amounts of reagents should

be used in the synthesis to force reactions to completion and to shorten the reaction time. Double coupling or even coupling at elevated temperatures (50° to 60°C) should be used if a single coupling fails to complete the reaction. Efficiency of synthesis should be closely monitored by the ninhydrin test (Support Protocol 1) or quantitative ninhydrin test (Sarin et al., 1981). The unusual properties of MAP systems (e.g., their multibranched structure and tendency to aggregate) should be considered to achieve a high-quality product. The synthesis should be started with a lower level of amino acid loading on the resin than is customary with usual solid-phase peptide synthesis, because peptide content will increase geometrically with each additional layer of lysine. Peptide synthesis will be difficult if the conventional loading of 0.3 to 0.8 mmol amino acid/g resin is used. In the Alternate Protocol for Fmoc chemistry, because resin with 0.1 mmol/g loading is not commercially available, synthesis starts with a resin of 0.3 to 0.5 mmol alanine/g resin, but only some of the amino groups are used for preparation of the dendrimer. Dimethyl formamide (DMF) is a more suitable solvent than dichloromethane (DCM) for avoiding potential aggregation of peptides. The peptide resin should not be dried at any stage of synthesis because resolvation of dried resin that contains MAP scaffolding is difficult. During HF cleavage in Boc synthesis, several precautions should be taken to obtain satisfactory results. The HF line must be kept clean. The line should be checked before each cleavage. Very often, heterogeneous and complex peptide products are caused by a dirty, contaminated HF line. The HF line should also be checked for leaks before each cleavage. A leaky HF line can result in the loss of HF and dimethyl sulfide (DMS) during the cleavage reaction; this can lead to unsatisfactory results. A leaky HF line may also allow water and carbon dioxide into the reaction vessel, diluting the HF cleavage mixture. The cleavage mixture should be mixed completely during the reaction. Sometimes the magnetic stir-bar is frozen by the p-cresol or p-thiocresol mixture during the introduction of the reagent. This can be avoided by adding the reagents in the following order: (1) peptide resin, (2) p-cresol and pthiocresol in melted form, then, after cooling and when the mixture has solidified, (3) the magnetic stirbar, and (4) DMS. In Fmoc synthesis, trifluoroacetic acid (TFA) cleavage should be prolonged if the peptide contains Arg(Pmc). Thioanisole can



accelerate the cleavage of the Pmc group. The deprotection of trityl (Trt) from cysteine residues is reversible, and scavengers such as thiophenol and 1,2-ethanedithiol (EDT) should be added to the cleavage mixture to prevent Trt reattachment to Cys. EDT also minimizes the alkylation reaction of Trp. The optimal volume of cleavage solution should be determined experimentally. In general, larger volumes of cleavage solution will, to some extent, suppress side reactions because most of the side reactions in the cleavage stage are bimolecular reactions. However, the more cleavage solution used, the larger the volume of TFA that must be removed, and the larger the amount of scavenger that will remain, making purification and analysis of the peptide more difficult. TFA can be reduced to a smaller volume under a vacuum by means of rotary evaporation, but the mixture should never be dried. A small-scale cleavage, using 20 to 30 mg of peptide resin, should be performed to determine optimal cleavage reaction conditions such as the combination and the ratio of scavengers, reaction time, and the volume of cleavage mixture. HPLC can be used to evaluate the results. Reaction conditions should be modified if the crude cleavage product is not as homogeneous as expected. The desired degree of purity should be ∼60% or a major peak in the HPLC. MAP systems prepared by the direct approach have been found to have an unusual tendency to aggregate after cleavage from the resin. The entire MAP should be dialyzed under basic, denaturing conditions using 8 M urea to remove unwanted additives from the cleavage reaction. Treatment with base under these conditions will also convert the possible strong acid–catalyzed O-acyl rearrangement product of serinyl and threonyl peptides back to N-acyl peptides. After dialysis, MAP systems can be further purified by either RP-HPLC or highperformance gel-filtration chromatography. In the additional cyclization step for preparing cyclic unprotected peptide building blocks and the entire cMAP by the direct approach, some of the protocols involve oxidation with NaIO4. The order of the synthetic scheme to generate aldehydes from the oxidation has to be followed carefully, since not all the functional groups in their unprotected form, suitable for cyclization, are stable under the condition of NaIO4. Perform the NaIO4 oxidation at neutral pH in the presence of a large excess of free methionine as a scavenger, to prevent oxidation of Met to Met(O). The oxidation rate increases

as pH decreases, but side reactions occur at lower pH. The aldehyde may decompose, and the absence of air is necessary. Flushing with nitrogen or argon for 30 min results in an oxygen-free solution. For cyclizations that include the presence of free sulfhydryl groups, the reactions must also be carried out in a solution devoid of oxygen to avoid dimerization. Also, addition of thiols or trialkylphosphine to the solution prevents the formation of dimers. The cyclization reactions can be carried out in a final concentration of up to 20 mM without affecting the yield of the desired product, since the reactions usually are concentration-independent. In the indirect approach where either linear or cyclic unprotected peptides are ligated to a MAP core matrix using carbonyl or thiol chemistry, the ligation reactions are also performed in an oxygen-free solution. Usually, the ligation is performed with a peptide concentration at 1.2- to 2-fold excess relative to core matrix. For the thiol alkylation reaction in thiol chemistry, iodoacetylated and bromoacetylated derivatives are known to be more reactive than the chloroacetylated derivative. However, the chloroacetylated derivative is more stable during the preparation of core matrix, especially during HF cleavage. The peptide and nonpeptide bonds formed in MAP and cMAP are stable at physiological pH (pH 6 to 8) as well as at low pH during analytical and preparative HPLC. However, for long-term storage, the peptides must be stored as a dry powder at a maximum temperature of −20°C.

Anticipated Results If synthesis of an eight-branched MAP with ∼15 amino acid residues starts from 0.5 g resin with a resin loading of 0.1 mmol/g, then >0.5 g of crude MAP can be obtained. Its purity depends on both the peptide sequence and the method used for synthesis. Usually the purity of crude products will be >60%, and after purification ∼40% to 50% can be recovered as purified MAP. This amount is usually more than sufficient for biological purposes. In general, the reaction should start with 0.3 to 0.5 g resin. For preparing cMAP using the direct approach, the overall yield can be up to 60% for cyclic peptides containing 17 and 24 amino acid residues, also starting with 0.5 g resin (loading is 0.1 mmol/g). In the direct approach, the average yield for MAP is between 50% and 60% using the ligation methods in Figure



Supplement 17

18.5.5. The yield of the pure cMAP system is ∼30% when cyclic peptides ranging from 6 to 14 amino acids are used. However, the site and structure, as well as the excess and concentration of the peptides influence the obtained yield. Both methods for generating MAP and cMAP systems are applicable for mimetic proteins. If HPLC of a crude MAP prepared by the direct approach shows multiple peaks without a major product, the complex may be difficult to purify, and the yield of MAP may not be satisfactory. In this case, the synthetic procedure (synthetic chemistry, protecting combination, coupling procedure, and/or cleavage conditions) should be modified. Also, the progress of the ligation reactions in the indirect approach is conveniently monitored by RP-HPLC (UNIT 11.6). If no product is formed, the peptides should be reanalyzed by mass spectrometry (Chapter 16) for the presence of nucleophilic and electrophilic moieties that have to react with each other. The peptide may even have to be resynthesized. Also, be aware of the formation of byproducts that can be generated from hydrolysis of the starting material at a given pH. Therefore, recheck the pH during the cyclization and assembling reactions.

Time Considerations


An automatic peptide synthesizer can assemble ∼15 residues in one day by a rapid procedure using a 15-min cycle or in 2 days with the the normal cycle of ∼1 hr. The synthesis of a MAP system with its peptide chain of 10 to 20 residues will usually take 1 to 2 days using the direct approach. Using a manual shaker for solid-phase peptide synthesis, assembly of as many as 6 to 12 amino acid residues onto the resin in one day is possible. The same MAP system can be prepared within 4 days by manual synthesis. Cleavage and subsequent purification can take from 2 days to 1 week, depending on the purity of the crude product as well as the required purity of the isolated MAP system. In preparing cMAP using the direct approach, 5 days should be added for the extra steps of cyclization, purification and characterization. In the indirect approach for preparing MAP systems, both the peptide and core matrix are prepared and purified before ligation. Ligation of the peptide with the core matrix requires an additional reaction step and adds 2 to 3 days to the process. For cyclization of the linear unprotected precursors, 3 extra days are needed.

Literature Cited Amon, R. and Horwitz, R.J. 1992. Synthetic peptides as vaccines. Curr. Opin. Immunol. 4:449453. Botti, P., Pallin, T.D., and Tam, J.P. 1996. Cyclic peptides from linear unprotected peptide precursors through thiazolidine formation. J. Am. Chem. Soc. 118:10018-10024. Defoort, J.-P., Nardelli, B., Huang, W., Ho, D.D., and Tam, J.P. 1992. Macromolecular assemblage in the design of a synthetic AIDS vaccine. Proc. Natl. Acad. Sci. U.S.A. 89:3879-3883. Francis, M.J., Hastings, G.Z., Brown, F., McDermed, J., Lu, Y.A., and Tam, J.P. 1991. Immunological evaluation of the multiple antigen peptide (MAP) system using the major immunogenic site of foot-and-mouth disease virus. Immunology 73:249-254. Houghten, R.A., Pinilla, C., Blondelle, S.E., Appel, J.R., Dooley, C.T., and Cuervo, J.H. 1991. Generation and use of synthetic peptide combinatorial libraries for basic research and drug discovery. Nature 354(6348):84-6 Lerner, R.A. 1982. Tapping the immunological repertoire to produce antibodies of predetermined specificity. Nature (Lond.) 299:592-596. Liu, C.-F. and Tam, J.P. 1994. Peptide segment ligation strategy without use of protecting groups. Proc. Natl. Acad. Sci. U.S.A. 91:6584-6588. Liu, C.-F. and Tam, J.P. 1997. Synthesis of a symmetric branched peptide: Assembly of a cyclic peptide on a small tetraacetate template. Chem. Commun. 1619-1620. Lu, Y.A., Clavijo, P., Galantino, M., Shen, Z.-Y., Liu, W., and Tam, J.P. 1991. Chemically unambiguous peptide immunogen: Preparation, orientation and antigenicity of purified peptide conjugated to the multiple antigen peptide system. Mol. Immunol. 28(6):623-630. Merrifield, R.B. 1963. Solid phase synthesis. I. J. Am. Chem. Soc. 85:2149-2154. Mutter, M. and Vuilleumier, S. 1989. A chemical approach to protein design-template-assembled synthetic proteins (TASP). Angew. Chem. Int. Ed. Engl. 28:535-554. Nardelli, B., Lu, Y.A., Shim, D.R., Delpierre-Defoort, C., Profy, A.T., and Tam, J.P. 1992. A chemically defined synthetic vaccine model for HIV-1. J. Immunol. 148:914-920. Pallin, D.T. and Tam, J.P. 1995. Cyclisation of totally unprotected peptides in aqueous solution by oxime formation. Chem. Commun. 2021-2022. Pallin, T.D. and Tam, J.P. 1996. Assembly of cyclic peptide dendrimers from linear building blocks in aqueous solution. Chem. Commun. 11:13451346. Pessi, A., Valmori, D., Migliorni, P., Tougne, C., Bianchi, E., Lambert, P.-H., Corradin, G., and Giudice, del G. 1991. Lack of H-2 restriction of Plasmodium falciparum (NANP) sequence as multiple antigen peptide. Eur. J. Immunol. 21:2273-2276.



Posnett, D.N., McGrath, H., and Tam, J.P. 1988. A novel method for producing antipeptide antibodies. J. Biol. Chem. 263:1719-1725. Sarin, V.K., Kent, S.B.H., Tam, J.P., and Merrifield, R.B. 1981. Quantitative monitoring of solidphase peptide synthesis by the ninhydrin reaction. Anal Chem. 117:147-157. Sasaki, T. and Kaiser, E.T. 1989. Helicrome: Synthesis and enzymatic activity of a designed heme protein. J. Am. Chem. Soc. 111:380-381. Schneider, J.P. and Kelly, J.W. 1995. Templates that induce α-helical, β-sheet, and loop conformations. Chem. Rev. 95:2169-2187. Shao, J. and Tam. J.P. 1995. Unprotected peptides as building blocks for the synthesis of peptide dendrimers with oxime, hydrazone and thiazolidine linkages. J. Am. Chem. Soc. 117:38933899. Spetzler, J.C. and Tam, J.P. 1996. Self-assembly of cyclic peptides on a dendrimer: Multiple cyclic antigen peptides. Pept. Res. 9:290-296. Stewart, J.M. and Young, J.D. 1984. Solid Phase Peptide Synthesis, 2nd ed. Pierce Chemical Co., Rockford, Ill. Tam, J.P. 1988. Synthetic peptide vaccine design: Synthesis and properties of a high density multiple antigenic peptide system. Proc. Natl. Acad. Sci. U.S.A. 85:5409-5413. Tam, J.P. 1995. Synthesis and applications of branched peptides in immunological methods and vaccines. In Peptides: Synthesis, Structures and Applications (ed. B. Gutte), pp. 455-501. Academic Press, San Diego. Tam, J.P. 1996. Recent advances in multiple antigen peptides. J. Immunol. Methods 196:17-32. Tam, J.P. and Lu, Y.-A. 1989. Vaccine engineering: Enhancement of immunogenicity of synthetic peptide vaccine related to hepatitis in chemically defined models consisting of T and B cell epitopes. Proc. Natl. Acad. Sci. U.S.A. 86:90849088. Tam, J.P. and Spetzler, J.C. 1997. Multiple antigen peptide system. Methods Enzymol. 289:612-637. Tam, J.P., Clavijo, P., Lu, Y.A., Nussenzweig, R.S., Nussenzweig, V., and Zavala, F. 1990. Incorporation of T and B epitopes of the circumsporozoite protein in a chemically defined synthetic vaccine against malaria. J. Exp. Med. 171:299306.

Tam, J.P., Wu, C.-R., Liu, W., and Zhang, J.-W. 1991. Disulfide bond formation in peptides by dimethyl sulfoxide: Scope and applications. J. Am. Chem. Soc. 113:6657-6662. Unson, C.G., Erickson, B.W., Richardson, D.G., and Richardson, J.S. 1984. Protein Engineering: Design and synthesis of a protein. Fed. Proc. 43:1837. Wrighton, N.C., Balasubramanian, P., Barbone, F.P., Kashyap, A.K., Farrell, F.X., Jolliffe, L.K., Barrett, R., and Dower, W.J. 1997. Increased potency of an erythropoietin peptide mimetic through covalent dimerization. Nat. Biotechnol. 15:1261-1265. Zhang, L. and Tam, J.P. 1997. Synthesis and application of unprotected cyclic peptides as building blocks for peptide dendrimers. J. Am. Chem. Soc. 119:2363-2370.

Key References Fields, G.B. and Noble, R. 1990. Solid phase peptide synthesis utilizing 9-fluorenylmethoxycarbonyl amino acids. Int. J. Pept. Protein Res. 35:161214. An extensive and useful literature review of Fmoc solid-phase peptide synthesis. Grant, G.A. (ed.) 1992. Synthetic Peptides—A User’s Guide. W.H. Freeman, New York. A practical book for solid-phase peptide synthesis which covers synthesis, purification, analysis, and applications of synthetic peptides. References include papers published in 1992. Stewart and Young, 1984. See above. An excellent laboratory book for solid-phase peptide synthesis. Tam and Spetzler, 1997. See above. A practical protocol for preparing MAPs and cMAP using direct and indirect approaches. Zhang and Tam, 1997. See above. A useful research paper discussing preparation and analysis of cyclic peptides and cMAP.

Contributed by James P. Tam and Jane C. Spetzler Vanderbilt University School of Medicine Nashville, Tennessee



Supplement 17

Disulfide Bond Formation in Peptides

UNIT 18.6

The formation of disulfide bridges is often a crucial final stage in peptide synthesis. There is compelling evidence that the disulfide pattern can be critical in the folding and structural stabilization of many natural peptide and protein sequences, while the artificial introduction of disulfide bridges into natural or designed peptides may often improve biological activities/specificities and stabilities. This unit will provide the experimentalist with a highly selective, albeit state-of-the-art, menu of procedures that can be tried in order to establish intramolecular or intermolecular disulfide bridges in targets of varying complexities. In all, this unit consists of 15 procedures, divided into two major classes: (1) formation of disulfides from free thiol precursors (see Basic Protocols 1 through 5 and Alternate Protocols 1 through 5), and (2) symmetrical formation of disulfides from S-protected precursors (see Basic Protocols 6 through 8 and Alternate Protocols 6 and 7). See Strategic Planning for further discussion of this breakdown. Directed methods for the unsymmetrical formation of disulfides, which offer control when two different chains are to be linked, use relatively sophisticated and specialized chemistry; examples are not provided herein. Reactions to form disulfides can be carried out with peptides either in solution or while they remain anchored to a polymeric support—in the latter case, taking advantage of the pseudo-dilution phenomenon that favors intramolecular cyclization over intermolecular side reactions. Other possible advantages of solid-phase methods relate to the circumvention of peptide solubility issues, and the chance to readily remove excess oxidizing agents as well as to exchange the solvent milieu by simple filtration and washing steps. The authors assume throughout this unit that the linear precursors (free or S-protected), which are to be converted to the corresponding disulfide species, can be prepared in a reasonable state of chemical homogeneity. When the plan is to carry out the disulfide-forming steps in solution, it is often advantageous to first purify the linear precursor. Solution and solid-phase methods of peptide synthesis, and purification of synthetic products both before or after the chemical steps to form disulfide bonds, are beyond the scope of the unit, but have been amply documented in the primary and review literature. STRATEGIC PLANNING Formation of Disulfides from Free Thiol Precursors The most straightforward approach for the preparation of disulfide-containing peptides involves initial assembly of the linear chain, using the same (invariably) acid-labile protecting group for all Cys residues. Figure 18.6.1 shows the structures of some such protecting groups that are compatible with solid-phase peptide synthesis (SPPS) methods: S-Xan, S-Tmob, S-Mmt, or S-Trt for Fmoc SPPS, and S-Mob or S-Meb for Boc SPPS. Subsequently, the polypeptide is released from the support, concomitant with removal of acid-labile protecting groups for the side chains of all (or almost all) other amino acid residues. This may be followed by treatment with reducing agents to ensure the sole presence of the linear monomeric chain, and then purification under acidic conditions that minimize the possibility of inadvertent premature oxidation. The resultant precursor, in which all Cys residues should be in the free sulfhydryl form, can then be subjected to a variety of solution oxidation procedures, keeping in mind that the precise experimental conditions, e.g., pH, ionic strength, organic cosolvent, temperature, time, and concentration, can often make a substantial difference in terms of the quality of disulfide product obtained. High dilution is recommended in order to minimize physical aggregation, and also to minimize the chemical formation of dimers, oligomers, or intractable polymers. Contributed by Lin Chen, Ioana Annis, and George Barany Current Protocols in Protein Science (2001) 18.6.1-18.6.19 Copyright © 2001 by John Wiley & Sons, Inc.



In a variation of the general approach, free sulfhydryl precursors can be converted to the corresponding disulfides while the peptide is still retained on the solid support; this requires cysteine protection that is selectively removable in an earlier on-resin step. Practical examples of this approach are given in Basic Protocols 1 to 5 and Alternate Protocols 1 to 5. Symmetrical Formation of Disulfides from S-Protected Precursors Several oxidizing reagents can be used for the formation of disulfide bridges directly from certain S-protected cysteine residues, without the intermediacy of free bis(thiol) intermediates. This approach is used for intramolecular or intermolecular pairing of two Cys residues that originally have the same protecting group, and it can be generalized in experiments aimed at the construction of multiple disulfides. Most of the S-protecting groups already shown in Figure 18.6.1 can be manipulated in this way, as can the S-Acm group, the structure of which is presented in Figure 18.6.2. Typical oxidative reagents include iodine, thallium (III) trifluoroacetate [Tl(tfa)3], and a variety of alkyltrichlorosilane-sulfoxide combinations, and both solution and on-resin deprotection/oxidation methodologies have been described. In most cases, caution is required when peptide substrates contain sensitive residues, especially Met and Trp. Practical examples of this approach are given in Basic Protocols 6 through 8 and Alternate Protocols 6 and 7.

Typically used with Boc SPPS

– S – CH2

X = CH3

X

p-methylbenzyl (Meb)

X = CH3O p-methoxybenzyl (Mob)

Typically used with Fmoc SPPS

–S

9H-xanthen-9-yl (Xan) O

H

CH3O – S – CH2

OCH3

2,4,6-trimethoxybenzyl (Tmob)

OCH3

–S –C

X

X = CH3O 4-methoxytriphenylmethyl (Mmt) X = H triphenylmethyl (trityl; Trt)


Figure 18.6.1 Acid-labile Cys protecting groups. The protecting group structure is drawn to include the sulfur (shown in bold) of the cysteine that is being protected.



Typically used with Boc SPPS

– S – CH2

9 - fluorenylmethyl

H

Typically used with either Boc or Fmoc SPPS O – S – CH2 – NH – C – CH3 acetamidomethyl (Acm)

CH3 – S – C – CH3

tert-butyl (tBu)

CH3 CH3 – S – S – C – CH3

tert-butylmercapto (StBu)

CH3

Figure 18.6.2 Essentially acid-stable Cys protecting groups that are removed by orthogonal methods. The protecting group structure is drawn to include the sulfur (shown in bold) of the cysteine that is being protected.

AIR OXIDATION Molecular oxygen can promote disulfide formation under slightly alkaline conditions, usually pH 7.5 to 8.5. Oxidation is carried out by simple aeration under gentle stirring, or by bubbling oxygen through a dilute solution of the peptide. Reactions are accelerated at higher pH, so long as materials are soluble. This widely used approach may be subject to one or more of the following limitations: (1) oligomerization, despite precautions to carry out reactions at low concentrations of substrate; (2) inadequate solubility of some basic or hydrophobic peptides; (3) long reaction times (1 to 5 days); (4) difficulty in optimizing the oxidation, because the rate depends on trace amounts of metal ions; and (5) accumulation of side products due to oxidation of Met residues.

BASIC PROTOCOL 1

Materials Poly(thiol) peptide, previously purified (see UNIT 8.7 for purification techniques) Buffer, selected from among: 0.1 to 0.2 M Tris⋅Cl, pH 7.7 to 8.7 (APPENDIX 2E) Tris⋅acetate, pH 7.7 to 8.7 0.01 M phosphate buffers, pH 7 to 8 (APPENDIX 2E) 0.01 M ammonium bicarbonate, pH 8 Air or oxygen Additional reagents and equipment for dialysis (UNIT 4.4 and APPENDIX 3B), HPLC (UNIT 8.7), and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Dissolve the peptide substrate in the appropriate buffer, at a typical concentration of 0.01 to 0.10 mg/ml (∼0.01 to 0.1 mM). 2. Stir the solution in open atmosphere, or while oxygen is bubbling through it.



Supplement 23

3. Monitor progress of the reaction by analytical HPLC (UNIT 8.7); withdraw small aliquots (e.g., 20 µl to fill an HPLC injection loop of the same size) to carry out the analysis. 4. Upon completion of the oxidation reaction, concentrate the peptide material by direct lyophilization. Alternatively, first quench the reaction by acidification, and perform dialysis (UNIT 4.4 or APPENDIX 3B) to exchange out salts. Some disulfide-containing peptides are labile when concentrated at alkaline pH, whereas the presence of acid minimizes disulfide interchange.

5. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.3), or ion-exchange (UNIT 8.2) chromatography, as appropriate. Peptide concentrations have been reported that are as as low as 1 ìM and as high as 1 mM, and the reaction time can vary from a few hours to a few days. As a starting point, try one of the buffers listed in Materials. For the oxidation of certain peptides, the use of organic cosolvents [e.g., methanol, acetonitrile, dioxane, and 2,2,2-trifluoroethanol (TFE)] and the addition of tertiary amines [e.g., N-methylmorpholine (NMM), triethylamine, or N,N-diisopropylethylamine (DIEA)], has been recommended. Addition of moderate concentrations of denaturants such as 0.5 to 3 M urea or 0.1 to 1.5 M guanidinium hydrochloride may help avoid aggregation and facilitate the renaturation process. Addition of 0.1 to 1 ìM CuCl2 has also been reported to improve certain oxidations. ALTERNATE PROTOCOL 1

ON-RESIN AIR OXIDATION In favorable cases, it is possible to use air to oxidize the resin-bound product of a linear sequence assembled by SPPS. According to this plan, the protecting groups for Cys must be removed selectively while negligible or no cleavage occurs of the anchoring linkage; whether other side-chain protecting groups are retained or removed at the same time is usually not as paramount a consideration. With acid-labile Cys protecting groups (Fig. 18.6.1) this means that the anchor should be acid-stable, or more commonly, that the Cys protecting group chosen is one that can be removed by short treatment with dilute acid. Alternatively, the usual acidolyzable anchors can be used together with acid-stable, orthogonally removable Cys protecting groups (Fig. 18.6.2). Once the on-resin deprotection step has been completed, the resin is washed, suspended in a polar solvent in the presence of a tertiary amine base, and exposed to air for the required oxidation. Subsequent cleavage from the support may provide, in a respectable yield, the wanted monomeric (if intramolecular cyclization) or homodimeric (if intermolecular co-oxidation) disulfide product; peptidic material not readily accounted for is invariably oligomeric and serves to diminish the overall yield. Materials Protected peptide-resin, prepared by optimized linear SPPS chain assembly Deprotection reagents and solvents Appropriate wash solvents 0.02 to 0.175 M triethylamine in N-methylpyrrolidone (NMP) Air or oxygen Appropriate peptide cleavage cocktail (UNIT 18.5), and materials for workup 2-ml plastic syringe fitted with a polypropylene frit (use larger syringe for larger amounts of resin)


Additional reagents and equipment for HPLC (UNIT 8.7) and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography



1. Place protected peptide-resin into a 2-ml plastic syringe fitted with a polypropylene frit, and carry out a selective deprotection step that removes protection on Cys while not affecting the anchoring linkage. Upon completion of the deprotection, wash thoroughly with appropriate solvents. For example, S-Tmob or S-Xan can be removed selectively by brief treatment with dilute acid (Munson and Barany, 1993; Munson et al., 1993; Hargittai and Barany, 1999). S-StBu can be removed by thiolysis (Eritja et al., 1987). S-Acm can be removed by treatment with Hg2+ salts (Tam and Shen, 1992). S-Fm can be removed by piperidine-promoted β-elimination (Albericio et al., 1991). Precise protocols to do this are in the aforementioned literature references.

2. Incubate the peptide-resin (containing free thiol moieties) with 0.02 to 0.175 M (2 to 10 equivalents) triethylamine in NMP at 25°C. Other polar solvents (e.g., DMF) that swell the resin can be used also.

3. Gently bubble air or oxygen through the suspension. 4. Optional: Monitor progress of the reaction by taking small aliquots of the peptideresin (e.g., 10 to 20 mg) for final cleavage, and evaluate the cleaved material by HPLC analysis. The reaction time can vary from 5 to 36 hr. This step is optional, but is particularly valuable as part of exploratory studies.

5. Cleave the peptide from the resin with appropriate (thiol-free) peptide cleavage cocktail (UNIT 18.5). Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate. These methods have been applied to make intramolecular disulfides, and also to make symmetrical intermolecular disulfides.

CHARCOAL/AIR-MEDIATED INTRAMOLECULAR DISULFIDE FORMATION


Oxygen adsorbed onto charcoal surfaces has proven efficient in mediating disulfide bond formation in a variety of peptides under basic conditions. The reactions were significantly faster than DMSO- or air-mediated cyclizations of the same substrates. Thermodynamic studies suggest that cyclization is accelerated by reduction of entropy of the peptides, upon transient adsorption to the charcoal surface, resulting in a lower activation energy (Volkmer-Engert et al., 1998). Materials Bis(thiol) peptide, previously purified 5% (v/v) aqueous NH4OH Granulated charcoal Ellman’s reagent (UNIT 18.3) Additional reagents and equipment for assaying free sulfhydryls with Ellman’s reagent (UNIT 18.3) 1. Dissolve the bis(thiol) peptide in water to a final concentration of 1 mg/ml (∼1 mM). 2. Adjust the pH to 7.5 to 8.0 with 5% aqueous NH4OH. 3. Add granulated charcoal to the peptide solution, using up to a 1:1 (w/w) ratio of charcoal to peptide.



Supplement 23

4. Gently shake the heterogeneous reaction mixture at 25°C. 5. Monitor the progress of the reaction by Ellman’s assay for disappearance of free sulfhydryls. Take 70-µl aliquots of the reaction mixture, dilute each with 0.7 ml of H2O and 70 µl of Ellman’s reagent, and measure the absorbance of 2-nitro-5-thiobenzoic acid anion (NTB−) at 420 nm. In general, the reaction is complete in 2 to 6 hr. See Figure 18.6.3 and UNIT 18.3 for details on assaying free sulfhydryls with Ellman’s reagent.

6. Upon completion of the reaction, filter the reaction mixture, and lyophilize the filtrate. BASIC PROTOCOL 2

INTRAMOLECULAR DISULFIDE FORMATION BY POTASSIUM FERRICYANIDE OXIDATION Efficient oxidation rates have been reported, especially in the oxytocin and somatostatin families, when using potassium ferricyanide, a relatively mild inorganic oxidizing reagent. Because K3Fe(CN)6 is slightly light sensitive, reactions are best conducted in the dark. The indicated order of addition, i.e., peptide to oxidizing agent, is designed to favor intramolecular cyclization because the concentration of free thiol species is kept minimal. Oxidation side products are possible when Met or Trp residues are present in the substrate. Materials Poly(thiol) peptide, previously purified 0.01 M phosphate buffer, pH ∼7 0.01 M aqueous K3Fe(CN)6 solution 10% (v/v) aqueous NH4OH 50% (v/v) aqueous acetic acid Celite resin (Aldrich) AG-3 anion-exchange resin Additional reagents and equipment for HPLC (UNIT 8.7) and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Dissolve the poly(thiol) peptide in 0.01 M phosphate buffer, pH ∼7, to achieve a final concentration of 0.1 to 1 mg/ml (∼0.1 to 1 mM). 2. At 25°C under nitrogen, add the peptide solution slowly to a 0.01 M aqueous K3Fe(CN)6 solution (the amount of oxidant used should be in 20% excess over the calculated theoretical amount). Addition times vary between 6 and 24 hr, with purer products noted upon slower addition.

3. Keep the reaction mixture constant at pH 6.8 to 7.0, by addition of 10% aqueous NH4OH. 4. After completion of the reaction, adjust the pH to 5 with 50% aqueous acetic acid. 5. Remove ferrocyanide and ferricyanide ions by filtering first through Celite resin and then, under mild suction, through a bed of AG-3 weakly basic anion-exchange resin. 6. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate.




INTERMOLECULAR OR INTRAMOLECULAR DISULFIDE FORMATION BY POTASSIUM FERRICYANIDE OXIDATION


The protocol that follows differs from the preceding one principally with respect to order of addition, and it is carried out on a shorter time frame. General considerations are the same. The extent of intermolecular homodimer formation is favored by increased peptide concentration, but intramolecular cyclization is also possible if conformationally favored. For Materials see Basic Protocol 2. 1. Dissolve the poly(thiol) peptide in 0.01 M phosphate buffer, pH ∼7, to achieve a final concentration of 0.1 to 1 mg/ml. 2. At 25°C and over a 30-min period, titrate the peptide solution with 0.01 M aqueous K3Fe(CN)6 solution, until a slight yellow color persists. 3. Follow the oxidation by Ellman analysis (optional; UNIT 18.3). 4. Adjust the pH to 5 with 50% aqueous acetic acid, and remove the oxidant with AG-3 anion-exchange resin. 5. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate. ON-RESIN DISULFIDE FORMATION BY POTASSIUM FERRICYANIDE OXIDATION


Following the same general motivation explained in Alternate Protocol 1, resin-bound thiols can, under favorable circumstances, be oxidized by reagents other than air, e.g., K3Fe(CN)6. Materials Protected peptide-resin, prepared by optimized linear SPPS chain assembly Dimethylformamide (DMF) 0.1 to 0.5 M K3Fe(CN)6 solution Dichloromethane (CH2Cl2) Appropriate (thiol-free) peptide cleavage cocktail (UNIT 18.5), and materials for workup 2-ml plastic syringe fitted with a polypropylene frit (use larger syringe for larger amounts of resin) Additional reagents and equipment for HPLC (UNIT 8.7) and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Place the protected peptide-resin into a 2-ml plastic syringe fitted with a polypropylene frit, and carry out a selective deprotection step that removes protection on Cys while not affecting the anchoring linkage. For more details, see the corresponding section of Alternate Protocol 1. At the end of the process, swell the peptide-resin in DMF, and then drain the mixture. 2. Incubate the swollen peptide-resin with a homogeneous mixture of 0.1 to 0.5 M aqueous K3Fe(CN)6 with DMF (1:1 to 1:10 v/v ratio), at 25°C in the dark. The overall amount of K3Fe(CN)6 used is calculated to be 10 to 20 equivalents with respect to the level of SH groups on the peptide-resin, and the higher concentration of oxidant in the aqueous solution requires correspondingly less of the DMF co-solvent.

3. Upon completion of the incubation (12 to 24 hr reaction), wash the peptide-resin with water, DMF, and CH2Cl2.



Supplement 23

4. Cleave the peptide from the resin with the appropriate (thiol-free) peptide cleavage cocktail (UNIT 18.5) and materials. 5. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate. BASIC PROTOCOL 3

OXIDATION UNDER SLIGHTLY ACIDIC pH CONDITIONS BY DMSO Dimethylsulfoxide (DMSO)-promoted oxidation of thiols is efficient over an extended pH range, i.e., pH 3 to 8. The oxidant is miscible with water, so concentrations of up to 20% (v/v) can be achieved and the higher concentrations in turn promote faster reactions. Additional advantages cited for this approach (Tam et al., 1991) are (1) improved peptide solubility; (2) the effect of DMSO as a denaturing co-solvent; and (3) compatibility with side-chains susceptible to oxidation (Met, Trp, or Tyr). A limiting factor, however, is the difficulty in removal of DMSO upon completion of the oxidation. Materials Poly(thiol) peptide, previously purified 5% (v/v) aqueous acetic acid Ammonium carbonate, (NH4)2CO3 Dimethylsulfoxide (DMSO) 0.05% (v/v) trifluoroacetic acid (TFA)/5% (v/v) aqueous acetonitrile Additional reagents and equipment for HPLC (UNIT 8.7) 1. Dissolve the poly(thiol) peptide in 5% aqueous acetic acid to a final concentration of 0.5 to 1.6 mg/ml (∼0.5 to 1.6 mM). 2. Adjust pH to 6 with (NH4)2CO3. 3. Add DMSO (10% to 20% by volume) to the peptide solution. 4. Monitor progress of the reaction by analytical HPLC (UNIT 8.7); withdraw small aliquots (e.g., 20 µl to fill an HPLC injection loop of the same size) to carry out the analysis. The reaction time can vary from 5 to 24 hr.

5. Upon completion of the reaction, dilute the reaction mixture two-fold with 0.05% TFA/5% aqueous acetonitrile. 6. Load onto a preparative reversed-phase HPLC column (UNIT 8.7) for DMSO removal and product purification. ALTERNATE PROTOCOL 5

OXIDATION UNDER SLIGHTLY BASIC pH CONDITIONS BY DMSO Assuming that the peptide is soluble, it is often advantageous to carry out oxidation by DMSO at higher pH, because the intrinsically higher rate means less reagent is required and the final purification becomes correspondingly simpler. Materials Poly(thiol) peptide, previously purified 0.01 M phosphate buffer, pH 7.5 Dimethylsulfoxide (DMSO) Additional reagents and equipment for HPLC (UNIT 8.7)




1. Dissolve the poly(thiol) peptide to a final concentration of ∼1 mg/ml (∼1 mM) in 0.01 M phosphate buffer, pH 7.5. 2. Add DMSO (1% by volume), at 25°C. 3. Monitor progress of the reaction by analytical HPLC (UNIT 8.7); withdraw small aliquots (e.g., 20 µl to fill an HPLC injection loop of the same size) to carry out the analysis. The reaction time can vary from 3 to 7 hr.

4. Quench the reaction by lyophilization. OXIDATION BY REDOX BUFFERS Peptides and proteins that contain multiple disulfide bonds can be oxidized in the presence of mixtures of low molecular weight disulfides and the corresponding free thiols. This approach may provide better yields and regioselectivities than straightforward air oxidations covered earlier (see Basic Protocol 1), because the mechanism changes from direct oxidation (free radical intermediates) to thiol-disulfide exchange (thiolate intermediate), which facilitates the reshuffling of incorrectly paired disulfides to the natural arrangements. The most commonly used systems are mixtures of oxidized and reduced glutathione, cysteine, cysteamine, or 2-mercaptoethanol (also see APPENDIX 3A). As with all solution oxidations, yields are improved under high dilution conditions that reduce the potential levels of oligomerization.

BASIC PROTOCOL 4

Materials Poly(thiol) peptide, previously purified Glutathione redox buffer (see recipe) EDTA Sephadex G-10 or G-25 column Additional reagents and equipment for dialysis (UNIT 4.4 and APPENDIX 3B), HPLC (UNIT 8.7), and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Dissolve the poly(thiol) peptide in glutathione redox buffer, which contains 1 mM EDTA, to a final concentration of 0.05 to 0.1 mg/ml (∼0.05 to 0.1 mM). 2. Incubate the solution at 25°C to 35°C. 3. Monitor progress of the reaction by analytical HPLC (UNIT 8.7); withdraw small aliquots (e.g., 20 µl to fill an HPLC injection loop of the same size) to carry out the analysis. The reaction time can vary from 16 hr to 2 days.

4. Upon completion of the reaction, concentrate the peptide material by direct lyophilization. Alternatively, first quench the reaction by acidification, and perform dialysis to exchange out salts.

5. Purify the product by gel filtration on a Sephadex G-10 or G-25 column or by other standard methods, e.g., preparative HPLC (UNIT 8.7) or ion-exchange (UNIT 8.3) chromatography, as appropriate. In cases where formation of physical aggregates competes with the oxidation process, addition of a nondenaturing chaotropic agent is recommended (1 to 2 M guanidine⋅HCl or urea; also see APPENDIX 3A). If intractable precipitates form during this procedure, an alternative folding/oxidation is recommended in which the peptide is oxidized against a series of redox buffers with a slow pH gradient from 2.2 to 8.0.



Supplement 23

BASIC PROTOCOL 5

OXIDATION MEDIATED BY SOLID-PHASE ELLMAN’S REAGENT Ellman’s reagent, 5,5′-dithiobis(2-nitrobenzoic acid), when bound through two sites to a suitable solid-support (PEG-PS or modified Sephadex), is an effective and mild oxidizing reagent that promotes the formation of intramolecular disulfide bridges (Annis et al., 1998; Figure 18.6.3). The likely mechanism for this process involves a key resin-bound intermediate (Fig. 18.6.3, middle structure) to which pseudo-dilution applies; dimer formation should therefore be reduced, and indeed this is borne out experimentally. Oxidations are efficient under a wide pH range, from 2.7 to 6.6, with yields and rates increasing with higher pHs. The main side reaction involves covalent adsorption of the substrate to the support through intermolecular disulfide bridges; this diminishes the yield but does not affect the purity of the desired intramolecular disulfide that is isolated. Moreover the thiol substrate can be recovered by reduction, and recycled through the oxidation process. Materials Poly(thiol) peptide, previously purified 2:1:1 (v/v/v) suitable buffer (pH 2.7-7.0)/acetonitrile/CH3OH Solid-phase Ellman’s reagent (Annis et al., 1998) Dichloromethane (CH2Cl2) Dimethylformamide (DMF) 50-ml plastic syringe fitted with a polypropylene frit (use larger syringe for larger amounts of resin) Septum or plastic lock cap Additional reagents and equipment for HPLC (UNIT 8.7) 1. Dissolve the poly(thiol) peptide to a final concentration of ∼1 mg/ml (∼1 mM) in a 2:1:1 mixture of buffer (pH 2.7 to 7.0)/acetonitrile/CH3OH. 2. Weigh the solid-phase Ellman’s reagent (e.g., ∼50 mg resin/ml peptide solution, at 0.2 mmol/g, which corresponds to a 15-fold excess of DTNB functions over thiol groups) into a 50-ml plastic syringe fitted with a polypropylene frit. 3. Swell the resin in CH2Cl2, wash with DMF, wash again with CH2Cl2, and drain. The above steps apply when the parent support is PEG-PS. In the case of Sephadex as the parent support, only DMF washes are necessary prior to draining.

4. Plug the syringe tip using a small septum or a plastic lock cap. 5. Add the solution of peptide substrate to the syringe already containing the solid-phase Ellman’s reagent. 6. Gently stir the reaction mixture magnetically, or gently agitate on a rotary shaker, at 25°C. 7. Monitor progress of the reaction by taking aliquots (e.g., 20 µl to fill an HPLC injection loop) from the liquid phase for HPLC analysis (UNIT 8.7). Reaction times vary from 0.5 to 30 hr depending on the buffer pH, solid support, and ease of oxidation, with significantly faster rates at more basic pH, and when PEG-PS is the solid support.




NO2 C NH SH

S

SH

+

S

O R O C NH

“capture” NO2

NO2

C NH SH

S

O

HS

O

S

R C NH NO2

“cyclization”

NO2 C NH

S

S

+

HS HS

O R O C NH NO2

Figure 18.6.3 Proposed mechanism for intramolecular disulfide formation mediated by solidphase Ellman’s reagents. R = solid supports. Reprinted with the permission of the American Chemical Society (Annis et al., 1998).

8. Upon completion of oxidation, drain the liquid phase into a vial, using positive air pressure. 9. Wash the resin with the 2:1:1 mixture of buffer (pH 2.7 to 7.0)/acetonitrile/CH3OH, and combine the washes with the drained liquid phase from the previous step. 10. Lyophilize the reaction mixture to obtain the oxidized peptide. No additional purification of the oxidized peptide is required, if the original poly(thiol)peptide was pure.

SIMULTANEOUS DEPROTECTION/OXIDATION WITH IODINE Iodine-mediated oxidation has been applied in conjunction with S-Xan, S-Tmob, S-Mmt, S-Trt, and S-Acm protection. The choice of solvent is critical, and optimal solvents include 80% (v/v) acetic acid, 80% (v/v) methanol, or 80% (v/v) DMF in water. It is advisable to monitor reactions (e.g., by HPLC); once oxidation is complete, quenching should be carried out to lessen the extent of overoxidation of the thiol functionality to the corresponding sulfonic acid, as well as to prevent or minimize modification of other sensitive amino acid side-chains (Tyr, Met, Trp).

BASIC PROTOCOL 6



Supplement 23

Materials S-protected peptide (protected by S-Acm, S-Xan, S-Tmob, or S-Trt), previously purified 80% (v/v) acetic acid Iodine Carbon tetrachloride (CCl4) Additional reagents and equipment for HPLC (UNIT 8.7) and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Dissolve the S-protected peptide in 80% acetic acid to a final concentration of ∼2 mg/ml (∼2 mM). 2. Add solid iodine (5 equivalents with respect to S-protected Cys) in one portion to the solution. Mix the solution vigorously at 25°C. 3. Monitor progress of the reaction by analytical HPLC (UNIT 8.7); withdraw small aliquots (e.g., 20 µl to fill an HPLC injection loop of the same size) to carry out the analysis. The reaction time can vary from 10 min to 24 hr.

4. Upon completion of oxidation, quench the reaction by diluting to two times the volume with water. 5. Extract with CCl4 (4 to 5 times, equal volume each time) to remove the iodine, and retain the upper, aqueous phase. Upon completion of the extractions, lyophilize the aqueous phase. 6. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate. ALTERNATE PROTOCOL 6

SIMULTANEOUS ON-RESIN DEPROTECTION/OXIDATION OF S-ACM WITH IODINE The direct iodine-mediated oxidation of S-protected cysteine residues can sometimes be carried out effectively while the peptide, with its essentially full complement of protecting groups, remains on the support. A wide range of anchoring linkages and non-cysteine side-chain protecting groups are compatible in the sense that they are not cleaved prematurely under the relatively mild reaction conditions. The disulfide formation chemistry is subject to the same advantages and caveats of other solid-phase procedures covered in this unit; ultimate cleavage/deprotection from the support will provide a mixture of monomeric (desired), dimeric, and oligomeric products. Materials S-protected peptide resin (protected by S-Acm, S-Xan, S-Tmob, or S-Trt, prepared by optimized linear SPPS chain assembly) Dimethylformamide (DMF) Iodine Suitable peptide cleavage cocktail Additional reagents and equipment for for dialysis (UNIT 4.4 and APPENDIX 3B), HPLC (UNIT 8.7), and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography




1. Swell the S-protected peptide-resin in DMF. 2. Add solid iodine (5 to 10 equivalents with respect to S-protected Cys). Stir gently for 1 to 4 hr at 25°C. 3. Drain, and wash the peptide-resin with DMF and CH2Cl2. 4. Cleave the peptide with suitable peptide cleavage cocktail, avoiding the use of thiol scavengers. 5. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate. SIMULTANEOUS DEPROTECTION/OXIDATION OF S-ACM WITH Tl(III) Thallium (III) trifluoroacetate [Tl(tfa)3] is a mild oxidant that can be used in conjunction with S-Acm and S-Tmob protection. For certain specific applications, this approach has provided better yields and purities of desired disulfide products with respect to methods using I2. However the high toxicity of Tl(tfa)3, and its difficult removal from sulfur-containing peptides, represent limitations. The optimal solvent for reasons both of solubilization and chemistry is TFA, but DMF is an acceptable solvent for on-resin oxidations in conjunction with TFA-labile anchoring linkages. Anisole should be added as a scavenger for the alkyl cations generated during the deprotection/oxidation. His and Tyr survive exposure to Tl(tfa)3, but Met and Trp are compatible only if protected.

BASIC PROTOCOL 7

Materials S-protected peptide (protected by S-Acm, S-Xan, S-Tmob, or S-Trt), previously purified 19:1 (v/v) trifluoroacetic acid (TFA)/anisole Thallium (III) trifluoroacetate [Tl(tfa)3] Ethyl ether 10 to 20-ml screw-cap centrifuge tube Additional reagents and equipment for HPLC (UNIT 8.7) 1. Dissolve the S-protected peptide in 19:1 TFA/anisole to a final concentration of ∼1 mg/ml (∼1 mM), in a 10 to 20-ml screw-cap centrifuge tube. Chill to 4°C. 2. Add solid Tl(tfa)3 (0.6 equivalents per S-protected Cys). Stir the reaction for 5 to 18 hr at 4°C. 3. Optional: Monitor progress of the reaction by analytical HPLC (UNIT 8.7); withdraw small aliquots (e.g., 20 µl to fill an HPLC injection loop of the same size) to carry out the analysis. 4. Evaporate as much of the TFA as possible, under positive nitrogen flow, and add ethyl ether (∼3.5 ml per 1 µmol peptide) to precipitate the peptide. 5. Triturate for 2 min. Collect the peptide by centrifuging for 2 min at 4000 rpm in a standard benchtop clinical centrifuge, room temperature. 6. Decant the ethyl ether. Repeat the trituration/centrifugation cycle two more times to ensure complete removal of the toxic Tl salts. CAUTION: Thallium is very toxic. Handle it in the hood; use of gloves is strongly recommended. Solutions must be disposed of according to institutional guidelines. Preparation and Handling of Peptides


Supplement 23


ON-RESIN SIMULTANEOUS DEPROTECTION/OXIDATION OF S-ACM WITH Tl(III) The on-resin Tl(tfa)3 procedure represents an efficient and often preferred method for creating disulfide bonds. The general considerations from the standpoint of the reagent have already been covered (see Basic Protocol 7), and relevent discussions of solid-phase methodologies are found in Alternate Protocols 1 and 6. Materials S-protected peptide-resin (protected by S-Acm, S-Tmob, prepared by optimized linear SPPS chain assembly) Thallium (III) trifluoroacetate (Tl(tfa)3) Dimethylformamide (DMF) Anisole Dichloromethane (CH2Cl2) Suitable (thiol-free) peptide cleavage cocktail (UNIT 18.5) 2-ml plastic syringe fitted with a polypropylene frit Additional reagents and equipment for for dialysis (UNIT 4.4 and APPENDIX 3B), HPLC (UNIT 8.7), and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Swell the S-protected peptide-resin in DMF in a 2-ml plastic syringe fitted with a polypropylene frit. 2. Add Tl(tfa)3 (1.5 to 2 equivalents per S-protected group) in 19:1 DMF/anisole (∼0.35 ml/25 mg resin). Gently stir the heterogeneous mixture at 4°C. 3. Optional: Monitor progress of the reaction by cleavage of small portion of the peptide-resin and subjecting to HPLC analysis (UNIT 8.7). The reaction time can vary from 1 to 18 hr.

4. Drain, and wash the peptide-resin with DMF and CH2Cl2 to remove the excess Tl reagent. CAUTION: Thallium is very toxic, handle it in the hood. Use of gloves is strongly recommended. Solutions must be disposed of according to institutional guidelines.

5. Cleave the peptide with suitable thiol-free peptide cleavage cocktail (UNIT 18.5). Avoid the use of thiol scavengers.

6. Purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7), gel filtration (UNIT 8.2), or ion-exchange (UNIT 8.3) chromatography, as appropriate. BASIC PROTOCOL 8

ALKYLTRICHLOROSILANE-SULFOXIDE OXIDATION A harsher oxidizing milieu uses mixtures of alkyltrichlorosilanes and sulfoxides to form disulfide bridges directly from the linear substrates protected with a variety of S-protecting groups (e.g., S-Acm, S-tBu, S-Mob, S-Meb). The reactions are fast, and are reported to be compatible with pre-existing disulfides formed in an orthogonal scheme. The protecting groups present in the substrate determine the optimal chlorosilane-sulfoxide combinations, but the most generally effective milieus apply diphenylsulfoxide [Ph(SO)Ph] with either CH3SiCl3 or SiCl4. The method is compatible with most amino acid side-chains, although Trp must be protected as its Nin-formyl derivative to prevent chlorination of the side-chain indole.




Materials S-protected peptide (protected by S-Acm, S-tBu, S-Mob, or S-Meb), previously purified Trifluoroacetic acid (TFA) Diphenylsulfoxide (Ph(SO)Ph) Trichloromethylsilane (CH3SiCl3) Anisole Ammonium fluoride (NH4F) Diethyl ether Sephadex G-15 column 4 N aqueous acetic acid Additional reagents and equipment for HPLC (UNIT 8.7) and gel-filtration (UNIT 8.3) or ion-exchange (UNIT 8.2) chromatography 1. Dissolve the S-protected peptide in TFA to a final concentration of 1 to 10 µg/ml (∼1 to 10 µM). 2. Add Ph(SO)Ph (10 equiv), CH3SiCl3 (100 to 250 equiv), and anisole (100 equiv) to the solution. Allow the reaction to proceed for 10 to 30 min at 25°C. 3. Quench the reaction by addition of solid NH4F (300 equiv). 4. Precipitate the crude product with a large excess of dry diethyl ether. 5. Isolate the crude product by gel filtration on Sephadex G-15 column eluting with 4 N aqueous acetic acid. 6. Lyophilize the peptide fraction, and purify the product by standard methods, e.g., preparative HPLC (UNIT 8.7) or ion-exchange (UNIT 8.3) chromatography, as appropriate. REAGENTS AND SOLUTIONS Glutathione redox buffer Prepare a 0.1 M Tris⋅Cl or Tris⋅acetate buffer at pH 7.7 to 8.7. Add both reduced (1 to 10 mM) and oxidized (0.1 to 1.0 mM) glutathione to the buffer, with the molar ratio of reduced to oxidized glutathione typically 10:1. Alternative redox systems include: cysteine/cystine, cysteamine/cystamine, 2-mercaptoethanol/2-hydroxyethyl disulfide, dithiothreitol (DTT)/1,2-dithiane-4,5-diol (oxidized DTT). The same general procedure should be used.

COMMENTARY Background Information In natural or artificial peptides and proteins, the ease of intramolecular disulfide formation evidently depends on conformational factors, and is tied to questions of what are the driving forces (kinetic versus thermodynamic) and mechanisms (direct oxidation versus disulfide exchange) of tertiary structure acquisition in vivo and in vitro. Whereas the oxidation of simple ω-dithiols to monomeric cyclic disulfides is optimal when six-membered rings can form, the heterodetic disulfide rings found in a range of biological substances are of size 3n +

8, where n is the number of residues between a given pair of half-cystines and can start at 0 (relatively rare, because of conformational strain), and continues to 1, 2 (e.g., thioredoxin active-site loop), 3, 4, and higher. For n = 4, the conformationally flexible 20-membered ring occurs in oxytocin, prothrombin, and various toxin families. The smaller “local” loops are typically associated with β- and γ-turns; assumption of these secondary structures is possible only with certain intervening residues (e.g., Gly, Pro, Ser, Asn). The proper pairing of two particular cysteines that are relatively far



Supplement 23


apart in the linear sequence becomes possible if the corresponding folded conformation is energetically favorable, somewhat flexible, and readily accessible. The environment (steric; neighboring charges; buried versus surface) about the cysteine residues affects pKa values, and consequently the ease with which they oxidize. The rates, yields, and specificities of oxidations can be influenced as well by the reagent used, the solvent for the reaction, the presence of denaturants, and trace metals that act as catalysts. There exists a substantial empirical base of knowledge on a number of these issues, but as of yet, few clear correlations with predictive power have been drawn. Intermolecular disulfides can be established by careful co-oxidation, an approach that is often limited by the formation of homodimers along with the desired heterodimers. “Directed” methods are more general, but they are also more demanding in terms of requirements for a range of selectively removable sulfhydryl protecting groups, and the need to avoid conditions that give rise to disproportionation of unsymmetrical disulfides to the symmetric species. The issues just raised increase in complexity when more than two half-cystine residues are involved, since in principle mispairing can occur. In vitro renaturation studies are clearly predicated on the assumption that the “correct” structure (as judged by proper pairing of halfcystines) will either form preferentially or be the major species after equilibration (perhaps, as promoted in thiol-disulfide “redox” buffers or by the presence of protein factors such as disulfide isomerases or thioredoxin). While this ideal can often be achieved in practice with careful attention to experimental conditions, e.g., pH, ionic strength, temperature, time, and protein concentration, it is also common (particularly with mutant materials prepared by recombinant techniques) for substantial amounts of isomers with incorrect disulfide pairing to form. Furthermore, oxidation/folding of purified synthetic materials, even carried out under high dilution, is frequently plagued by the production of significant levels of dimeric, oligomeric, and intractable polymeric byproducts. This problem is exacerbated by the difficulty in “recycling” nonmonomeric species through alternating reduction and reoxidation steps. A central conceit of this unit is that disulfide formation can be controlled to some extent by astute choices of chemical procedures. Sulfursulfur bonds are created most often by oxidation (using air or stronger reagents) of precur-

sors with free or protected sulfhydryls. Chemistries for unsymmetrical “directed” formation of sulfur-sulfur bonds can be applied as well (the latter is not specifically covered in this unit). In general, the more complex peptide targets have been obtained in relatively low yields, only after extensive optimization of experimental protocols for synthesis, purification, and oxidation/folding. The details of how to protect cysteine, and the corresponding deblocking conditions, are central issues. Unlike the case with deprotection from most other side-chains, where the parent functionality is recovered, removal of protecting groups from cysteine can yield the corresponding free thiol, a thiolate metal salt (mercaptide), or a disulfide, depending on the reagent(s) applied and the precise reaction conditions. A significant avenue for current research is to achieve some measure of regioselectivity in cysteine pairing by applying appropriate pairwise combinations of orthogonally removable Cys protecting groups.

Critical Parameters and Troubleshooting Experimentalists seeking to obtain disulfide-containing peptides are faced with a set of options that might be considered both daunting and reassuring. Certainly, the conformational forces related to the desired covalent structures need to be favorable. Beyond that, reactions can be conducted in solution (e.g., Basic Protocols 1 to 4, Basic Protocols 6 to 8, Alternate Protocol 3, and Alternate Protocol 5) or in the solidphase (e.g., Alternate Protocols 1, 4, 6, and 7) mode, or by a hybrid approach wherein the substrate is in solution but the oxidizing reagent is resin-bound (e.g., Basic Protocol 5 and Alternate Protocol 3); the reaction milieus can range from primarily aqueous to aqueous with organic cosolvent to fully organic (even anhydrous); the linear precursor can be the free poly(thiol) or have its Cys residues appropriately protected; and the strategy to generate the disulfide array can involve a single procedure or sequential steps. A minimum prerequisite for solution strategies is that the materials are indeed soluble throughout the processes (i.e., as linear substrate, as intermediates, and as final disulfide product), and the value of high dilution has been noted earlier. Solid-phase reactions are based on the pseudo-dilution principle and, when successful, have the added advantage that they are operationally much simpler than the corresponding solution reactions.



Even before attacking the disulfide-forming steps, a plan must be made for construction of the linear sequence(s). This entails selection of the repetitively removable Nα-amino protecting group, choices of all of the corresponding side-chain protecting groups with particular emphasis on Cys, consideration of how to manage certain susceptible residues (e.g., Met, Tyr, Trp) that could potentially offer compatibility problems, and (in the case of SPPS), decisions on what resin support and anchoring handle to use, and the appropriate cleavage conditions. Reasonable guidelines and precedents for all of this can either be inferred from the text in this unit, or are covered in some of the Key References as well as the Literature Cited. Some of the initial synthetic work can even be contracted out. The next phase of the work depends on whether or not any of the disulfides will be formed by solid-phase reactions. If yes, parameters that may impact the outcome include: solvent; acidity/basicity; reaction temperature and time; and the loading, morphology, and swelling of the solid support. Later, the anchoring linkage is cleaved to release the oxidized material into solution for evaluation, additional chemical steps (sometimes), and eventual purification (under all circumstances). When disulfide-forming steps are to be carried out in solution, a myriad of experimental details may be critical, as has been described throughout this unit. Starting with purified precursors and intermediates can often make a significant difference, and in the cases that rely on oxidations of the corresponding poly(thiols), it is often imperative that such species have been pretreated with suitable reducing agents in order to ensure the absence of interfering species that contain prematurely formed disulfides (UNIT 18.3). In aqueous solution, oxidations generally proceed more rapidly with increased pH (e.g., pH 7.5 to 8.5) because of the greater level of ionization to the thiolate, but these conditions can also promote disulfide scrambling. Methods to form disulfides that are effective and reliable at lower pH (e.g., pH 2.5 to 6.0) are therefore of considerable interest. Again, because the facility of disulfide formation hinges on polypeptide conformation, changes in ionic strength and addition of denaturants can sometimes have a salutary effect. Most of the general principles raised throughout here can be extrapolated to protocols that rely on oxidizing agents other than oxygen, that are conducted in media other than water, and/or that do not involve free thiol intermediates.

As the reactions to form disulfides are underway, particularly as part of exploratory studies, conversion of precursors to products should be monitored chromatographically in order to determine the endpoints. This is most straightforward for solution processes that start with poly(thiol) precursors, and the information gleaned can be reinforced by following the disappearance of sulfhydryl titer, e.g., by Ellman’s assay. Monitoring solid-phase oxidations is somewhat more difficult, since generally the peptide-resin must be sampled, then subjected to cleavage, and finally analyzed to determine the distribution of released materials (corresponding to starting, intermediates, and product that were originally resin-bound). When working with synthetic disulfidecontaining peptides, it is often helpful to provide a full accounting of the molecular species formed. This general objective covers the isolation and structural characterization of both the target and any significant byproducts (e.g., parallel or antiparallel dimers, oligomers, and variants with misaligned disulfides). At a minimum, one should quantify components of crude synthetic products by comparison of HPLC peak areas with those of purified standards (if available) of known concentration. Also, analytical gel filtration gives information about the relative levels of monomers versus species of greater size. In order to assign and/or confirm structures, standard analytical methods for characterizing peptides, e.g., amino acid analysis, analytical HPLC, and (most informative) any of a number of mild ionization mass spectrometric methods, are supplemented by techniques adapted directly to study the disulfide arrangements such as proteolysis/peptide mapping or partial reduction/peralkylation/sequencing procedures. If after following the aforementioned guidelines for synthesis, purification, and characterization, the desired disulfide-containing structures are not obtained in sufficiently satisfactory overall yields and purities, an iterative troubleshooting process can be followed. This can range from further optimization of specific reaction conditions to trying alternative protocols to a partial or complete overhaul of the essential strategy. Literature precedents suggest that most synthetic peptides with a single or two disulfide bonds can eventually be obtained by effective routes, but the outcome for more complicated targets is by no means assured and quite dependent on the experimental art of the investigator.



Supplement 23

Anticipated Results As indicated in earlier sections of this unit, there are no definitive rules that guarantee success, and several of the described protocols may need to be evaluated individually or in tandem in order to converge upon optimal results. In calculations based on 100% being a homogeneous linear precursor, either resin-bound or in solution, many of the disulfide-forming reactions under review can proceed in yields of ≥80% (turnover of starting material can generally be taken to completion, but the required disulfide will not necessarily be the exclusive product). After an efficient purification step, the overall isolated yield usually drops to the 40% to 60% range. Clearly, when multiple reactions and purifications are carried out, the overall yields will be lower.

Time Considerations


Formation of the disulfide bond(s) in synthetic peptides is integrated with other steps, starting from linear chain assembly and including one or more purification steps. The following description assumes that pilot experiments and optimizations specific to the desired target have already been completed. Corresponding to those cases where a particular chemical protocol is especially efficient, valuable time can be saved by skipping an intermediate purification. Solid-phase synthesis of the linear sequences can be completed within several days, and is accelerated somewhat when using an automated synthesizer. Cleavage and workup procedures require several hours, and half a day of effort should be budgeted for each purification. Processing of purified samples by lyophilization can take 1 to 2 days, depending on the volume. The focus of this unit is on disulfide formation, and the individual protocols should be referred to in order to learn the recommended times that often suffice to achieve complete reaction. Potentially the most time-consuming procedures are those in which a linear poly(thiol) precursor is oxidized in solution, using aeration, a redox buffer, or the solidphase Ellman’s reagent. For these cases, reaction times of 1 to 3 days are typical. Accelerated rates may be achieved when applying specific oxidizing agents such as ferricyanide or DMSO. Methods that start from S-protected precursors and involve disulfide formation concurrent with deprotection are invariably more rapid, and are often complete in ≤4 hr (sometimes, as short as 10 min). The majority of

solid-phase reactions under review are relatively fast as well.

Literature Cited Albericio, F., Hammer, R.P., García-Echeverría, C., Molins, M.A., Chang, J.L., Munson, M.C., Pons, M., Giralt, E., and Barany, G. 1991. Cyclization of disulfide-containing peptides in solid-phase synthesis. Int. J. Peptide Protein Res. 37:402413. Annis, I., Chen, L., and Barany, G. 1998. Novel solid-phase reagents for facile formation of intramolecular disulfide bridges in peptides under mild conditions. J. Am. Chem. Soc. 120:72267238. Eritja, R., Ziehler-Martin, J.P., Walker, P.A., Lee, T.D., Legesse, K., Albericio, F., and Kaplan, B.E. 1987. On the use of S-t-butylsulphenyl group for protection of cysteine in solid-phase peptide synthesis using Fmoc-amino acids. Tetrahedron 43:2675-2680. Hargittai, B. and Barany, G. 1999. Controlled syntheses of natural and disulfide-mispaired regioisomers of α-conotoxin SI. J. Peptide Res. 54:468-479. Munson, M.C. and Barany, G. 1993. Synthesis of α-conotoxin SI, a bicyclic tridecapeptide amide with two disulfide bridges: Illustration of novel protection schemes and oxidation strategies. J. Am. Chem. Soc. 115:10203-10210. Munson, M.C., Lebl, M., Slaninová, J., and Barany, G. 1993. Solid-phase synthesis and biological activity of the parallel dimer of deamino-oxytocin. Peptide Res. 6:155-159. Tam, J.P. and Shen, Z.-Y. 1992. Efficient approach to synthesis of two-chain asymmetric cysteine analogs of receptor-binding region of transforming growth factor-α. Int. J. Peptide Protein Res. 39:464-471. Tam, J.P., Wu, C.-R., Liu, W., and Zhang, J.-W. 1991. Disulfide bond formation in peptides by dimethyl sulfoxide. Scope and applications. J. Am. Chem. Soc. 113:6657-6662. Volkmer-Engert, R., Landgraf, C., and SchneiderMergener, J. 1998. Charcoal surface-assisted catalysis of intramolecular disulfide bond formation in peptides. J. Peptide Res. 51:365-369.

Key References Albericio, F., Annis, I., Royo, M., and Barany, G. 2000. Preparation and handling of peptides containing methionine and cysteine. In Fmoc Solid Phase Peptide Synthesis: A Practical Approach (W.C. Chan and P.D. White, eds.) pp. 77-114. Oxford University Press, Oxford. Relatively recent coverage of cysteine protection and disulfide formation, in the context of a 14-chapter monograph on Fmoc solid-phase methodology.



Andreu, D., Albericio, F., Solé, N.A., Munson, M.C., Ferrer, M., and Barany, G. 1994. Formation of disulfide bonds in synthetic peptides and proteins. In Methods in Molecular Biology, Vol. 35: Peptide Synthesis Protocols (M.W. Pennington and B.M. Dunn, eds.) pp. 91-169. Humana Press, Totowa, N.J. State-of-the-art review at the time that it was written. Some experimental information and 358 literature citations. Annis, I., Hargittai, B., and Barany, G. 1997. Disulfide bond formation in peptides. Methods Enzymol. 289:198-221. Another book chapter, more recent but less comprehensive than the Andreu et al. reference above. The parent volume, edited by G.B. Fields, is a highly useful 32-chapter compendium of methods of solidphase peptide synthesis.

Contributed by Lin Chen AxCell Biosciences Corporation Newtown, Pennsylvania Ioana Annis Union Carbide Corporation Bound Brook, New Jersey George Barany University of Minnesota Minneapolis, Minnesota



Supplement 23

Guide for Resin and Linker Selection in Solid-Phase Peptide Synthesis

UNIT 18.7

The past decade has seen the chemistry of solid-phase peptide synthesis (SPPS) mature considerably. What was once a technique solely accessible to specialized laboratories has now become routine practice for chemists and biologists alike who have no specific training in peptide chemistry per se. For the first ∼25 years (beginning with Merrifield, 1963), the majority of SPPS efforts employed Boc/benzyl chemistry, which is based on a system of graduated acid lability. In Boc/benzyl-based syntheses, trifluoroacetic acid (TFA) is used for iterative removal of the t-butyloxycarbonyl (Boc) protecting group from the α-amino (Nα ) functionality, while side chain protecting groups and the peptide-resin anchorage are simultaneously cleaved using a much stronger acid, anhydrous hydrogen fluoride (HF). As investigators sought milder methods for SPPS that avoided the use of HF for the final deprotection step, the Fmoc/t-butyl approach rose in popularity. This methodology differs from Boc/benzyl chemistry in that Nα protection and side-chain protection are based on a system of orthogonality—i.e., the selective removal of each type of protection occurs by a mechanism that leaves the other type of protection entirely intact. In Fmoc/t-butyl chemistry, piperidine is used for iterative removal of the Nα 9-fluorenylmethoxycarbonyl (Fmoc) protecting group, while side chain protecting groups and the peptide-resin anchorage are simultaneously cleaved using TFA. While many monographs have been written on the theory and practice of SPPS, these discussions have almost invariably splintered into camps centered around the Boc/benzyl methodology or the Fmoc/t-butyl methodology, with little to no interplay between the two. While the chemical minutiae of these two approaches are ostensibly at odds, the development of linkers and resins for Boc-based SPPS and Fmoc-based SPPS have followed the same chemical trends. It is noteworthy that some of the most intriguing applications of SPPS have bridged the gap between these two strategies. For instance, peptide nucleic acids (PNAs), which hold promise in drug delivery and antisense technologies, have brought the concept of orthogonality between side-chain protection and Nα protection, hitherto a hallmark of the Fmoc approach, to Boc chemistry. Conversely, the synthesis of glycopeptides and fluorescence resonance energy transfer (FRET) substrates using Fmoc chemistry requires the application of side-chain protection that is removable under conditions of graduated acid lability, which is the basis of Boc chemistry. To both novice and more experienced practitioners of SPPS, the vast selection of commercially available linkers and resins has become something of a babel. The goal of this unit is to clarify the situation, which is best understood by distillation to first principles, through an appreciation of chemical trends and consequences, as well as practical considerations. It must be emphasized that some of the philosophies and procedures described in this unit are naturally skewed in accordance with the author’s own experience and preferences. For example, there are many satisfactory ways to couple a linker to a resin via an amide bond (although some are better than others), but only some are presented here. As the development of SPPS has involved a long series of (usually incremental) improvements upon existing methodologies, it is hoped that this unit and others will provide food for thought that might lead to further improvements in the practice of SPPS.

Peptides Contributed by Jason A. Moss Current Protocols in Protein Science (2005) 18.7.1-18.7.19 C 2005 by John Wiley & Sons, Inc. Copyright


Figure 18.7.1 resin.

Linkage of the target peptide, linker, internal reference amino acid (IRAA), and

Linker Selection In the majority of cases, synthetic peptide targets are designed to possess C-terminal carboxylic acids or primary amides. For applications in which more exotic unnatural functionalities are desired at the C-terminus, a variety of procedures, each with different degrees of complexity, generality, and effectiveness, have been devised. Although the literature is awash with technologies that allow the preparation of most any peptide-derived target, many of these reports have been only sparingly validated and applied. Furthermore, in many cases, such techniques require significant experience in synthetic and solid-phase chemistry, as well as access to equipment not normally found in most peptide synthesis laboratories. This discussion will focus on established state-of-the-art methodologies that fulfill the needs of most peptide synthesis applications. Whenever possible, the use of preformed linkers—sometimes termed handles—prepared by solution-phase organic synthesis is preferable to the traditional strategy of functionalizing linkers onto resins via direct modification of native polystyrene. One attractive feature of preformed linkers is their compatibility with the use of an internal reference amino acid (IRAA) between the linker and the resin (Fig. 18.7.1). While perhaps unnecessary for the synthesis of smaller, more routine peptides, an IRAA allows cleavage of the scissile peptide-resin linkage to be quantitated. This is particularly useful for newcomers to Fmoc-based SPPS; if, for example, an unexplainably poor yield of isolated peptide is obtained, solid-phase amino acid analysis (AAA) of the IRAA will allow the quantitation of cleavage efficiency. In such cases, the ability to accurately troubleshoot problems may allow the cleavage to be repeated and/or may make it possible to improve cleavage protocols for use in future syntheses. A second benefit of preformed linkers is that, almost without exception, their use in conjunction with high-quality amino-functionalized resins leads to significantly improved synthetic efficiency.

Linkers yielding C-terminal carboxylic acids

Resin and Linker Selection in SPPS

Peptides bearing C-terminal carboxylic acids are almost exclusively prepared using linkers derived from substituted benzyl esters (Fig. 18.7.2A). Whereas an unsubstituted benzyl ester is quite stable to acidolysis—and therefore ideally suited to use as a peptide-resin anchorage in Boc-based SPPS—the increasing prevalence of Fmocbased SPPS has spurred the development of more acid-labile peptide-resin linkages. Benzyl esters can be decorated with various alkoxy substituents and/or aryl rings to afford increasingly acid-labile peptide-resin linkages. Because of the variation in electronic properties as a function of the location, number, and type of such substituents, a wide range of linkers with predictable, graduated acid stabilities have been developed. Although only some of these linkers have found widespread application, the elegance with which graduated acid lability can be applied to linker design has been one of the key factors leading to the widespread use of solid-phase synthesis in biomedical research involving peptides, peptide-derived natural products, proteins, and other biopolymers.



Figure 18.7.2 Acidolytic bond scission in the synthesis of peptides bearing (A) a C-terminal carboxylic acid (C-O bond scission) and (B) a C-terminal primary amide (C-N bond scission). Examples shown are the p-alkylbenzyl ester linkage (panel A; Fig. 18.7.3, entry no. 1) and the p-methylbenzhydrylamide (MBHA) linkage (panel B; Fig. 18.7.4, entry no. 1).

The classical peptide-resin linkage employed in Boc-based SPPS is the p-alkylbenzyl ester (Fig. 18.7.3, entry no. 1; Merrifield, 1964). It is important to note that this is only formally a benzyl ester; in polystyrene resins, the 4-vinyl substituent of styrene (which is saturated upon polymerization) donates electron density to the benzyl ester linkage through the σ bond, thereby increasing the acid lability of the linkage. Nonetheless, this linkage is sufficiently stable to the repetitive TFA treatments necessary for subsequent Nα Boc deprotection cycles, but quantitatively cleaved by anhydrous HF during the final deprotection/cleavage step. Still, the unacceptable cumulative loss of peptide chains through repeated TFA-mediated Nα Boc deprotection cycles motivated the development of a more acid-stable alternative, the phenylacetamidomethyl (PAM) linker (Fig. 18.7.3, entry no. 2; Mitchell et al., 1976; Mitchell et al., 1978). The PAM linker is first prepared as a derivative of the desired C-terminal residue using a two-step solution-phase synthesis, and this preformed amino acid–PAM handle is then coupled to an aminomethyl-derivatized resin. The electron-withdrawing nature of the p-acetyl substituent on the PAM benzyl ester linkage renders the PAM linker ∼100 times more stable to TFA acidolysis than a traditional p-alkylbenzyl ester (obtained by direct esterification of a chloromethylderivatized resin). The result is a peptide-resin linkage that is sufficiently stable for Boc-based synthesis of large peptides (>50 residues), with negligible chain loss in each TFA-mediated Nα deprotection step. For the routine synthesis of peptides by Boc chemistry, the PAM linker is recommended in all cases in place of the classic p-alkylbenzyl ester obtained by direct esterification of chloromethyl-derivatized polystyrene. This recommendation stems not only from the greater acid stability of the PAM linker but also from chemical considerations regarding the alternative; a variety of deleterious side reactions in Boc-based SPPS have been ascribed to incomplete esterification reactions involving chloromethyl-derivatized polystyrene resins, and all of these side reactions are avoided by using preformed PAM linkers. It must be emphasized that the peptide bond connecting the PAM linker to the resin is not cleaved during HF deprotection/cleavage; only the ester linkage between the C-terminal amino acid residue and the PAM linker is cleaved, yielding the target peptide as a C-terminal acid. Preloaded Boc-Xaa-PAM resins are available from several commercial sources (e.g., Peptides International, Midwest Biotech), eliminating the need for the user to prepare the PAM-linked derivative of the C-terminal amino acid. Preformed Boc-Xaa-PAM-COOH derivatives are also commercially available (NeoMPS, Peptides International), allowing the user to select a preferred aminomethyl-derivatized resin for use in SPPS.

Peptides


Supplement 40

Figure 18.7.3 Linkers yielding peptides that bear a C-terminal carboxylic acid. Typical minimal acidolytic deprotection conditions are noted; acidolysis should be carried out at 25◦ C whenever not otherwise specified. Entries marked with an asterisk have the greatest general utility in routine peptide synthesis applications. TFA, trifluoroacetic acid; TFE, trifluoroethanol; HFIP, hexafluoroisopropanol. Resin and Linker Selection in SPPS



Despite their incompatibility with Boc-based SPPS, TFA-labile peptide-resin linkages were found early on to be ideally suited to Fmoc-based SPPS. In Fmoc-based SPPS, the use of an orthogonal, base-labile Nα protecting group eliminates the need to use HF for the final deprotection/cleavage step. Fortunately, an appropriate linker was already in hand at the time the Fmoc approach was being developed, as a TFA-labile p-alkoxybenzyl ester had previously been developed for a somewhat sparsely applied variant of Boc chemistry that relied on greater acid lability. This substituted benzyl ester, known as the Wang linker, was prepared by a base-mediated ether coupling of 4-hydroxymethylphenol and chloromethyl-functionalized polystyrene resin (Fig. 18.7.3, entry no. 3; Wang, 1973). It is important to emphasize that the commonly used term Wang resin is a misnomer, clouding the demarcation between linker and resin. Wang-functionalized polystyrene resins are commercially available from many sources, both in free alcohol form and preloaded with Fmoc-protected amino acid residues. One significant problem inherent in the synthesis of Wang-functionalized polystyrene resins is that it is difficult to achieve clean, quantitative functionalization of the resin at all chloromethyl sites with minimal polymerization of 4-hydroxybenzyl alcohol, one of the starting materials in the process. To address this issue, the preformed 2-(4-hydroxymethyl)phenoxyacetic acid (HMPA) linker was developed (Sheppard and Williams, 1982). This linker (Fig. 18.7.3, entry no. 4) can be rapidly and quantitatively coupled to any aminomethyl-derivatized resin using standard amide coupling protocols and affords roughly the same electronic characteristics as the parent Wang linker. As the p-alkoxybenzyl alcohol is substantially more difficult to acylate than an amine, autoacylation of the unprotected benzylic alcohol during linker coupling is not a significant concern, provided that an appropriately weak combination of activation chemistry and solvent is employed (see Basic Protocol 1). An additional benefit of the HMPA linker is that TFA acidolysis of the peptide-resin anchorage cleaves only the peptide-HMPA ester linkage, such that the linker remains attached to the resin. This contrasts with the traditional Wang linker, as the Wang linker–resin ether bond, in addition to the peptide–Wang linker ester bond, is cleaved during TFA acidolysis, leading to unnecessary contamination of the cleaved peptide product. Among the TFA-labile carboxylic acid linkers, the Wang linker and HMPA handle provide the most stable peptide-resin anchorages and are therefore the most useful for chemoselective side-chain modification of the resin-bound peptide chain. Modification possibilities include the coupling of biotin, a fluorophore, or a fatty acid to a side-chain amine or alcohol, as well as the phosphorylation or glycosylation of a serine side chain. In such cases, pseudo-orthogonal trityl-based protection or truly orthogonal allyl- or reductionbased protection can be used to prevent the target functionality from reacting during chain assembly, and these protecting groups can be selectively removed in the presence of more stable t-butyl esters, ethers, and carbamates following chain assembly. (For clarification, truly orthogonal side-chain protection strategies employ deprotection mechanisms that are entirely harmless to the usual Fmoc/t-butyl side-chain protection scheme, whereas pseudo-orthogonal side chain protecting groups are removed using mild acidolytic mechanisms that often give rise to varying degrees of undesired deprotection side products.) For a peptide linked to the solid support via a p-alkoxybenzyl ester, the brief TFA (1% to 5%) treatment necessary to remove trityl protecting groups, for example, leads to negligible premature chain loss due to acidolysis of the peptide-resin anchorage. (TFA concentrations here and throughout are expressed as percent v/v in an inert solvent, usually methylene chloride.) Only at the final stage of the synthesis, after all desired side-chain modifications are incorporated, is the peptide cleaved from the resin, via treatment for 1 to 2 hr with a ∼95% TFA cleavage cocktail. Peptides


Supplement 40

In some cases, a cleaved peptide bearing a free C-terminal acid but having protected side chains and/or a protected N-terminus is desired. For example, peptide-derived natural products and proteins may be prepared by the controlled, iterative condensation of partially or fully protected peptide segments. Such reactions are referred to as fragment condensations and are usually employed when it would be difficult or impossible to prepare the entire target molecule using standard stepwise SPPS. To allow cleavage of the peptide-resin anchorage while preserving the integrity of side-chain and N-terminal protection, linkers with greater acid lability, which are labile to low concentrations of TFA or other acids, have been developed. Brief treatments (30 residues). PEGA (polyethylene glycol–polyacrylamide) resin is a cross-linked, aminefunctionalized ternary copolymer commercially available from Sigma and Polymer Laboratories (Meldal, 1992). This resin contains only secondary amide and ether bonds and is therefore highly hydrophilic and inert to most peptide synthesis conditions. While intended for automated Fmoc-based SPPS under continuous-flow conditions, this resin is applicable to batchwise synthesis as well, provided that only mild rotary shaking (so as to minimize resin fracture) is performed. The handling properties of PEGA resins leave much to be desired, as they have poor mechanical stability and must always be maintained in a moist slurry to avoid irreversible collapse of the polymer matrix. A compromise between chemical and mechanical stability can be found in a solid support known as CLEAR (cross-linked ethoxyacrylate resin), which is commercially available from Peptides International (Kempe and Barany, 1996). Like PEGA, CLEAR is a ternary copolymer of hydrophilic monomers; however, CLEAR is composed partially of alkyl ester linkages, rather than the secondary amide linkages found in PEGA. While stable to all of the standard conditions encountered in Boc- and Fmoc-based SPPS, CLEAR is not stable to harsh alkali treatments. Nevertheless, the rarity of such reaction conditions in SPPS mitigates the significance of this drawback, as the instability of many peptide side chain protecting groups to base, as well as the known base-mediated side reactions involving protected peptides, has virtually abolished the need for base treatments at any stage in SPPS. This shortcoming notwithstanding, CLEAR is a useful solid support by virtue of its good mechanical stability, which is intermediate between that of PEGA and polystyrene resins.

Peptides


Supplement 40

“Hybrid” resins (polystyrene/polyethylene glycol composites) An alternative strategy for improving resin swelling in polar solvents, as well as for maintaining consistent swelling throughout peptide chain assembly, entails the grafting of polyethylene glycol (PEG) chains onto a PS-DVB resin. These hydrophilic chains are usually terminated with pendant amino groups that are ready for functionalization with an IRAA or a linker handle. This modification provides increased swelling relative to PS-DVB in most organic solvents, and particularly in the polar aprotic solvents (e.g., dimethylformamide, 1-methyl-2-pyrrolidinone, acetonitrile), in which PS-DVB exhibits comparatively poor swelling. The length and the sterically unhindered nature of the PEG chain, which serves as a bridge between the peptide chain and the polystyrene resin backbone, are thought to result in a more “solvent-like” environment for the resinbound peptide. Both resins described below are amenable to synthesis under batchwise or continuous-flow conditions, as they exhibit good mechanical stability and consistent swelling. The first chemistry used to attach PEG grafts to polystyrene resins involved an amide linkage to an MBHA-functionalized polystyrene resin (Fig. 18.7.3, entry no. 5; Zalipsky et al., 1994). The resulting resin, appropriately termed PEG-PS, is suitable only for Fmoc-based SPPS, due to the lability of the MBHA linkage to strong acids such as HF. A key feature of the PEG-PS resin is the incorporation of a norleucine residue between the PEG graft and the MBHA linker as an IRAA for solid-phase AAA. As in any other amine-derivatized resin, the free amine–functionalized PEG terminus must be acylated with a linker handle before peptide chain assembly can be initiated; the final TFA deprotection/cleavage step cleaves this linkage without detriment to the PEG-resin MBHA linkage. PEG-PS resin is commercially available from Applied Biosystems in native, linker-functionalized, and preloaded forms. An alternative approach for the attachment of PEG chains to polystyrene resins involves the generation of PEG in situ by polymerization of ethylene oxide onto a PSDVB resin. The resulting family of resins is marketed under the trade name TentaGel and is commercially available from Rapp Polymere. TentaGel resins are available in a wider selection compared with PEG-PS resins, and they can also be obtained in native, linker-functionalized, and preloaded forms. Whereas the early variants of TentaGel were marked by exceedingly low loading (∼0.1 to 0.2 mmol/g) and acid-labile benzylic ether linkages connecting the pendant PEG chains to the polystyrene backbone, significant improvements have broadened the applicability of TentaGel resins for SPPS. Notably, the high-capacity TentaGel HL resin is commercially available, and offers satisfactory loading capacities of 0.4 to 0.5 mmol/g. In addition, TentaGel resins are now made almost exclusively with an additional methylene spacer between the PEG graft and the polystyrene backbone. This feature is of the utmost importance, as it essentially eliminates the specter of acidolytic cleavage of the PEG graft from the PS-DVB core (with resulting contamination of the crude peptide product).

Resin and Linker Selection in SPPS

The choice of resin for a given synthesis is dictated principally by the manner in which the synthesis is to be performed. For manual batchwise SPPS using either Boc/benzyl or Fmoc/t-butyl chemistry, any resin may in principle be used. However, certain practical considerations must be considered prior to commencement of any synthesis. The inferior mechanical stability of PEG-based resins—principally PEGA resin and CLEAR— contraindicates their use in manual batchwise SPPS, in which mechanical agitation is likely to lead to resin fracture and frit clogging. Conventional polystyrene resins and hybrid resins, such as TentaGel and PEG-PS, are therefore recommended, as the mechanical rigidity imparted by the polystyrene core of such resins ensures the integrity of the resin throughout the course of synthesis. For automated SPPS, the mode of



operation of the synthesizer is of prime importance. Automated synthesizers that operate in a batchwise fashion (e.g., ABI 431; ABI 433; all CSBio, Rainin, and Advanced ChemTech synthesizers) are amenable to nearly all resin types, although for longer peptides, the mechanical stability of polystyrene resins is desirable. For the few synthesizers that employ continuous-flow methodology (e.g., ABI Pioneer, MilliGen/Bioresearch Crystal), resins that exhibit minimal changes in swelling throughout the course of synthesis (i.e., PEG-based and hybrid resins, as opposed to conventional polystyrene resins) are required so that instrument failure resulting from increases in resin swelling can be avoided. While automated Fmoc-based SPPS under continuous-flow conditions is most often performed using hybrid PEG-grafted polystyrene resins, it is the author’s experience that a high-quality polystyrene resin may be used for automated or manual batchwise SPPS using Fmoc/t-butyl or Boc/benzyl chemistry. Because they can be functionalized with nearly any linker for a given application, aminomethyl-derivatized polystyrene resins are recommended above all others, being of unmatched chemical and mechanical stability, versatility, and cost-effectiveness. The last of those three qualities is a key consideration for most users, as the cost of PEG-grafted polystyrene resins and PEG-based resins is in the range of two to five times the cost of aminomethyl-derivatized polystyrene resins. Furthermore, this cost comparison is on a per-weight basis and therefore does not take into account the fact that PEG-grafted resins typically have one-fourth to one-half the amine loading capacity (per gram) of standard polystyrene resins. This disparity in resin loading is of critical economic importance, as loading capacity is the ultimate limiting factor determining the total quantity of synthetic peptide product that can be prepared in a given synthesis.

Conclusions The most important point to be taken from this unit is that the linker and resin are separate entities and can be selected independently. The selection of the linker depends primarily on three factors: (1) the synthetic chemistry to be employed (Fmoc-based, Boc-based, or a combination thereof); (2) whether any protecting groups are to be left intact on the cleaved peptide product; and (3) the desired C-terminal functionality (primary amide or carboxylic acid). While usually separable, these issues are sometimes interdependent, such as when chemoselective postassembly modification of the resin-bound peptide is desired, or when partially or fully protected peptides are desired upon cleavage of the peptide-resin anchorage. It is hoped that this distillation of the somewhat daunting selection of commercially available linkers down to a simple spectrum of carbocation stabilities will allow the newcomer to peptide chemistry to make an educated decision prior to starting a synthesis. The selection of the resin is simple by comparison; here, the overriding question is whether the user plans to perform automated Fmoc-based SPPS under continuous-flow conditions. If this is the case, only a PEG-based or PEG-grafted polystyrene resin is appropriate, with the majority of published and anecdotal practical experience dealing with the use of the latter, although the former is acceptable as well. For automated or manual batchwise SPPS using either Boc-based or Fmoc-based chemistry, polystyrene resins are the most appropriate. This stems principally from their unrivaled mechanical stability throughout repetitive vortexing, stirring, sparging, and/or inversion to aid in the diffusion of reactants throughout the polymeric matrix during coupling and/or washing. From an economic standpoint, low cost, ready availability, and the wide range of amine loading capacities also serve to elevate aminomethyl-derivatized polystyrene resins to their current status as the mainstay of modern SPPS. Peptides


Supplement 40

BASIC PROTOCOL 1

FUNCTIONALIZATION OF AN AMINE-DERIVATIZED RESIN USING PREFORMED HANDLES This protocol provides three alternatives for functionalizing an amine-derivatized resin using a hydroxyl linker as a preformed handle. Although any of these three methods will work, the preferred method is the one involving a 2,4,5-trichlorophenyl (Tcp) or pentafluorophenyl (Pfp) ester linker. Preactivated Tcp esters are commercially available from Peptides International or may be prepared according to published procedures (Atherton and Sheppard, 1989).

Materials 5% (v/v) triethylamine in dimethylformamide (DMF) or 5% (v/v) triethylamine in 1:1 DMF/toluene Linker (free carboxylic acid, Tcp ester, or Pfp ester) to be used in peptide synthesis N,N -diisopropylcarbodiimide (DIC; optional) N-hydroxybenzotriazole (HOBt; optional) Benzotriazol-l-yloxytris(dimethylamino)phosphonium hexafluorophosphate (BOP; optional) DMF Amine-derivatized resin to be used in peptide synthesis Glass or polypropylene fritted vessel with Teflon stopcock Rotary shaker Wash bottle 1. Measure out the desired amount of amine-derivatized resin and transfer to a fritted vessel. Set the vessel’s stopcock to the closed position, and then preswell the resin by adding DMF containing ∼5% triethylamine or 1:1 DMF/toluene containing ∼5% triethylamine (10 ml DMF/triethylamine or DMF/toluene/triethylamine per g resin), capping the vessel, and incubating at room temperature with rotary shaking for a minimum of 1 hr, but preferably overnight. The inclusion of a base such as triethylamine or DIEA (N,N-diisopropylethylamine) in the preswelling solution is recommended, as this will neutralize any resin-bound salts.

2. Open the stopcock to drain all liquid from the vessel. Thoroughly wash the resin twice, each time by using a wash bottle to apply a constant flow of DMF for 20 sec and then allowing all liquid to drain from the vessel. After the second wash, set the stopcock in the closed position. 3a. If using a free carboxylic acid linker and DIC/HOBt catalysis: Dissolve 3 equivalents of the linker and 3 equivalents of HOBt to a concentration of 0.5 M in DMF. Use a pipet filler and pipet to apply this solution to the resin, and then apply 3 equivalents of DIC to the resin. 3b. If using a free carboxylic acid linker and BOP catalysis: Dissolve 3 equivalents of the linker and 3 equivalents of BOP to a concentration of 0.5 M in DMF. Use a pipet filler and pipet to apply this solution to the resin, and then apply 3 equivalents of DIEA to the resin. 3c. If using a Pfp or Tcp ester linker and HOBt catalysis: Dissolve 3 equivalents of the (already activated) linker and 3 equivalents of HOBt to a concentration of 0.5 M in DMF. Use a pipet filler and pipet to apply this solution to the resin. 4. To ensure complete mobility of the resin slurry during coupling, fill the vessel to between two-thirds and three-quarters of its capacity with DMF. Cap the vessel, invert and vent by opening the stopcock, and then place on a rotary shaker for several hours. Resin and Linker Selection in SPPS


5. Perform the qualitative ninhydrin test (see Basic Protocol 3) to assess the completeness of the resin functionalization reaction. Current Protocols in Protein Science

ACYLATION OF HYDROXYL LINKERS VIA REACTION WITH A MIXED ANHYDRIDE

BASIC PROTOCOL 2

This protocol describes the preferred method for acylation of a hydroxyl linker with the C-terminal amino acid of the target peptide. While there are many methods that can be used to accomplish this acylation, most carry a risk of undesirable side reactions (e.g., racemization, premature Fmoc deprotection and resultant dipeptide formation) or are exceedingly complicated, requiring expensive reagents and rigorously anhydrous reaction conditions. The method presented here represents a compromise, providing an acceptably low level of racemization (> [B]—the concentration of A at which half of the B is found in the AB complex approximates Kd. Typical Kd values for biologically significant interactions are given in Table 19.1.1. Note that many significant interactions are weak. Alternatively, affinity may be given as an equilibrium association constant, Ka. The Ka is simply Ka =

[ AB] [ A ][ B]

In other words, it is the reciprocal of the equilibrium dissociation constant: Ka = 1/Kd. Association constants are used less often than Kd, but do appear in the literature of some subfields—for example, in descriptions of antibody-antigen interactions. The strength of an interaction is directly proportional to the change in Gibbs free energy (∆G) when A and B interact, which is given by ∆G = ∆H − T∆S

where T is the temperature in degrees Kelvin, ∆S is the change in entropy (S), and ∆H is the change in enthalpy (H). ∆G is given in units of kilocalories/mole (kcal mol−1). An increase in the free energy of an interaction by 1 kcal decreases the Kd by a factor of ∼7. Note that the temperature affects the entropy term of the equation. As the temperature decreases, reactions driven by enthalpy are less affected by losses in entropy, and are usually favored. By contrast, reactions whose free energy is derived largely from favorable changes in entropy are disfavored at lower temperatures. This means that the dependence of a protein association on temperature can often provide a clue to which term predominately contributes to the free energy change involved in that association. The relationship between Ka and ∆G is defined as follows: ∆G 0 = − RT ln

Analysis of Protein-Protein Interactions

[ AB] [ A ][ B]

∆G 0 = − RT ln K a = − RT ln

1 = − RT ln K Kd d



where ∆G0 is the free energy change under standard conditions (25oC); R is the universal gas constant (1.9872 calmole−1K–1); and T is the temperature in degrees Kelvin (25°C is 289.1 K). Therefore, ∆G 0 = 0.588 ln K d

and, since ln x = 2.303 log10 x, ∆G0 = 1.36 log10 Kd. For example:

(a) Kd = 1 × 10−14 , then ∆G0 = −19.04 kcal mol −1 (b) Kd = 1 × 10−2 , then ∆G0 = −2.72 kcal mol −1 The strepavidin-biotin reaction (a) is intrinsically more favorable in the direction of binding, [AB] formation, than the low-affinity interaction (b) involving phage repressor (see Table 19.1.1). It should be pointed out that the DG values calculated above assume that the molar ratio of reactants [A] [B] and product [AB] is 1 M (standard state). Many (but by no means all) biologically important protein interactions seem to be largely driven by ∆H, or changes in enthalpy. That is fortunate, in that changes in entropy on binding are very hard to quantitate, or even think about precisely. For example, proteins in solution are surrounded by water molecules that form hydrogen bonds with surface residues. When two proteins interact, the ordered arrangement of the water molecules that surrounded the interacting surfaces of the proteins is often disrupted, and this loss of order provides an entropically favorable term to the free energy of the interaction. Such changes in the free energy due to entropy are very hard to predict. By contrast, enthalpic changes are easier to understand. If formation of one hydrogen bond liberates about −1 kcal M−1, and formation of a particular ionic contact liberates about −2 kcal M−1, then the energies of these changes are additive such that the formation of both bonds usually liberates −3 kcal M−1—which means that the Kd is decreased >100-fold by the enthalpic changes. Kinetic Parameters The above descriptions of protein interactions make no reference to the speed at which association or dissociation occurs. These speeds are given by kinetic parameters. The dissociation rate constant, kdissoc, gives the speed at which the AB complex dissociates into A and B (AB → A + B). kdissoc is a first-order rate constant—i.e., one that is dependent on the concentration of one species, in this case the AB complex—and is given by the rate of decrease in the concentration of AB: kdissoc [A][B] =

− d[ AB] dt

Its units are those of reciprocal time (t in the equation), usually given in sec−1. For example, a dissociation rate constant of 10−4 sec−1 means that one in 104 of the AB complexes present comes apart each second. Similarly, the association rate constant, kassoc, gives the speed at which A and B associate to form an AB complex (A + B → AB). This is a second-order reaction—i.e., its speed depends on the concentrations of both A and B—and its rate constant is is given by

kassoc [A][B] =

+ d[ AB] dt

Its units are those of reciprocal concentration × reciprocal time, typically given in M−1 sec−1.

Identification of Protein Interactions


Supplement 14

For example, suppose that protein A is present at a concentration of 10−6 M in a cell, that protein B is injected to a nuclear concentration of 10−5 M, and that the rate constant for this antibody-antigen association is 10−4 M−1 sec−1. After injection, the concentration of AB will be [AB] = [A] × [B] × K assoc = (10 −6 M)(10 −5 M)(10 −4 M ) = 10 −7 M sec −1

That is, in the first second after mixing, 10−7 M of AB complex will form. Every 10-fold increase in the concentration of either reactant increases the rate of product formation 10-fold. Note that, at equilibrium, by definition,

d[AB] =0 dt and, therefore, for the AB interaction, k K d = dissoc kassoc

The fact that the strength of equilibrium interactions reflects the speed of associations and dissociations has important consequences. To understand this, imagine two pairs of proteins that interact with the same Kd. Proteins A and B come together slowly, but, once the AB complex forms, it takes a long time to come apart. By contrast, proteins C and D come together rapidly, and the CD complex dissociates rapidly. There are two common cases in which these kinetic differences in the AB and CD associations would be significant. One is in measurement. Many techniques, such as the “pulldown” and immunological coprecipitation techniques described in this chapter, rely on the the fact that proteins remain associated during some sequence of steps, while they are being separated from other proteins in a mixture and while the isolated complex is being rinsed. No matter how tightly the proteins associate, if they come apart before their complex can be separated and rinsed, the complex will not be detected. Moreover, if the AB and CD interactions have the same Kd, but CD comes apart more rapidly, then a coprecipitation experiment can falsely suggest that the CD association is weaker. The second case concerns the biological effects. Many biological phenomena, such as the transcription phenotypes resulting from protein-protein interactions in two-hybrid experiments (UNIT 19.3), seem to be well-described by consideration of equilibrium measurements. However, it is worth keeping in mind that any biological process that results from the association of two proteins requires a minimum time to occur. For some enzyme-substrate interactions, the minimum time may be on the order of microseconds, but for others, such as the initiation of DNA replication, it may be measured in seconds. If the complex dissociates faster than the minimum time, then, no matter how tight the interaction, the process will not occur. If AB and CD have the same Kd, but CD dissociates faster, the association may not produce a biological effect.

Analysis of Protein-Protein Interactions

Contributed by Roger Brent The Molecular Sciences Institute Berkeley, California



Interaction Trap/Two-Hybrid System to Identify Interacting Proteins

UNIT 19.2

To understand the function of a particular protein, it is often useful to identify other proteins with which it associates. This can be done by a selection or screen in which novel proteins that specifically interact with a target protein of interest are isolated from a library. One particularly useful approach to detect novel interacting proteins—the twohybrid system or interaction trap (see Figs. 19.2.1 and 19.2.2)—uses yeast as a “test tube” and transcriptional activation of a reporter system to identify associating proteins (see Background Information). This approach can also be used specifically to test complex formation between two proteins for which there is a prior reason to expect an interaction. In the basic version of this method (see Fig. 19.2.2), the plasmid pEG202 or a related vector (see Fig. 19.2.3 and Table 19.2.1) is used to express the probe or “bait” protein as a fusion to the heterologous DNA-binding protein LexA. Many proteins, including transcription factors, kinases, and phosphatases, have been successfully used as bait proteins. The major requirements for the bait protein are that it should not be actively excluded from the yeast nucleus, and it should not possess an intrinsic ability to strongly activate transcription. The plasmid expressing the LexA-fused bait protein (see Table 19.2.1) is used to transform yeast possessing a dual reporter system responsive to transcriptional activation through the LexA operator. In one such example, the yeast strain EGY48 (see Table 19.2.2) contains the reporter plasmid pSH18-34. In this case, binding sites for LexA are located upstream of two reporter genes. In the EGY48 strain, the upstream activating sequences of the chromosomal LEU2 gene—required in the biosynthetic pathway for leucine (Leu)—are replaced with LexA operators (DNA binding sites). pSH18-34 contains a LexA operator–lacZ fusion gene. These two reporters allow selection for transcriptional activation by permitting selection for viability when cells are plated on medium lacking Leu, and discrimination based on color when the yeast is grown on medium containing Xgal (APPENDIX 4A). In Basic Protocol 1, EGY48/pSH18-34 transformed with a bait is characterized for its ability to express protein (Support Protocol 1), growth on medium lacking Leu, and for the level of transcriptional activation of lacZ (see Fig. 19.2.2A). A number of alternative strains, plasmids, and strategies are presented which can be employed if a bait proves to have an unacceptably high level of background transcriptional activation. In an interactor hunt (Basic Protocol 2), the strain EGY48/pSH18-34 containing the bait expression plasmid is transformed (along with carrier DNA made as described in Support Protocol 2) with a conditionally expressed library made in the vector pJG4-5 (see Fig. 19.2.6 and Table 19.2.3). This library uses the inducible yeast GAL1 promoter to express proteins as fusions to an acidic domain (“acid blob”) that functions as a portable transcriptional activation motif (act) and to other useful moieties. Expression of libraryencoded proteins is induced by plating transformants on medium containing galactose (Gal), so yeast cells containing library proteins that do not interact specifically with the bait protein will fail to grow in the absence of Leu (see Fig. 19.2.2B). Yeast cells containing library proteins that interact with the bait protein will form colonies within 2 to 5 days, and the colonies will turn blue when the cells are streaked on medium containing Xgal (see Fig. 19.2.2C). The DNA from interaction trap positive colonies can be analyzed by polymerase chain reaction (PCR) to streamline screening and detect redundant clones in cases where many positives are obtained in screening (see Alternate Protocol 1). The plasmids are isolated and characterized by a series of tests to confirm specificity of the interaction with the initial bait protein (Support Protocols 3 to 5). Those found to be specific are ready for further analysis (e.g., sequencing).


Contributed by Erica A. Golemis, Ilya Serebriiskii, Russell L. Finley, Jr., Mikhail G. Kolonin, Jeno Gyuris, and Roger Brent

19.2.1

Current Protocols in Protein Science (1998) 19.2.1-19.2.40 Copyright © 1998 by John Wiley & Sons, Inc.

Supplement 14

prepare bait strain(s) for interaction mating (Alternate Protocol 2, steps 1-3)

construct bait protein plasmid and transform yeast (Basic Protocol 1, step 1)

characterize bait protein expression and activity

transform cDNA library into lexA operatorLEU2 strain to make pretransformed library strain (Alternate Protocol 2, steps 4-11)

introduce cDNA library into bait strain(s) by interaction mating (Alternate Protocol 2, steps 12-20)

obtain cDNA library in pJG4-5

transform cDNA library into lexA-operator-LEU2/lexA-operator-lacZ/pBait yeast (Basic Protocol 2, steps 1-7 )

select for library plasmid (Basic Protocol 2, step 8)

assess transcriptional activity (Basic Protocol 1, steps 4-7 ) assess repressor activity (Basic Protocol 1, steps 8 -11) test for Leu requirement (Basic Protocol 1, steps 12 -13) assess protein synthesis (Support Protocol 1)

freeze and replate transformants (Basic Protocol 2, steps 9 -15)

select for interacting proteins (Basic Protocol 2, steps 16 -19)

transform E. coli Basic Protocol 2, steps 20-22)

test for specificity (Basic Protocol 2, steps 23 -27, and Support Protocol 5 )

analyze and sequence positive isolates (Basic Protocol 2, step 28, and Support Protocol 5 )

assess whether clones are independent by restriction mapping or by filter hybridization (Support Protocol 3)

or obtain profile of independent interactors by microplate plasmid rescue (Support Protocol 4) or

warehouse clones and repeat screen with less sensitive strain analyze positive clones by PCR and restriction endonuclease digestion (Alternate Protocol 1)

Figure 19.2.1 Flow chart for performing an interaction trap.



A

B

C act act bait

bait

bait LEU2

LEU2

LEU2

act

act bait

bait lacZ

bait lacZ

lacZ

Figure 19.2.2 The interaction trap. (A) An EGY48 yeast cell containing two LexA operator–responsive reporters, one a chromosomally integrated copy of the LEU2 gene (required for growth on −Leu medium), the second a plasmid bearing a GAL1 promoter–lacZ fusion gene (causing yeast to turn blue on medium containing Xgal). The cell also contains a constitutively expressed chimeric protein, consisting of the DNA-binding domain of LexA fused to the probe or bait protein, shown as being unable to activate either of the two reporters. (B) and (C), EGY48/pSH18-34/pbait-containing yeast have been additionally transformed with an activation domain (act)–fused cDNA library in pJG4-5, and the library has been induced. In (B), the encoded protein does not interact specifically with the bait protein and the two reporters are not activated. In (C), a positive interaction is shown in which the library-encoded protein interacts with bait protein, resulting in activation of the two reporters (arrow), thus causing growth on medium lacking Leu and blue color on medium containing Xgal. Symbols: black rectangle, LexA operator sequence; open circle, LexA protein; open pentagon, bait protein; open rectangle, library protein; shaded box, activator protein (acid blob in Fig. 19.2.6).

When more than one bait will be used to screen a single library, significant time and resources can be saved by performing the interactor hunt by interaction mating (see Alternate Protocol 2). In this protocol, EGY48 is transformed with library DNA and the transformants are collected and frozen in aliquots. For each interactor hunt, an aliquot of the pretransformed EGY48/library strain is thawed and mixed with an aliquot of a bait strain transformed with the bait expression plasmid and pSH18-34. Overnight incubation of the mixture on a YPD plate results in fusion of the two strains to form diploids. The diploids are then exposed to galactose to induce expression of the library-encoded proteins, and interactors are selected in the same manner as in Basic Protocol 2. The advantage to this approach is that it requires only one high-efficiency library transformation for multiple hunts with different baits. It is also useful for bait proteins that are somewhat toxic to yeast; yeast expressing toxic baits can be difficult to transform with the library DNA. CHARACTERIZING A BAIT PROTEIN The first step in an interactor hunt is to construct a plasmid that expresses LexA fused to the protein of interest. This construct is transformed into reporter yeast strains containing LEU2 and lacZ reporter genes, and a series of control experiments is performed to establish whether the construct is suitable as is or must be modified, and whether alternative yeast reporter conditions should be used. These controls establish that the bait protein is made as a stable protein in yeast, that it is capable of entering the nucleus and binding LexA operator sites, and that it does not appreciably activate transcription of the LexA operator–based reporter genes. This last is the most important constraint on use of this system. The LexA-fused bait protein must not activate transcription of either re-

BASIC PROTOCOL 1



Supplement 14

Table 19.2.1

Interaction Trap Componentsa,b

Plasmid name/source

Selection In yeast

Comment/description

In E. coli

LexA fusion plasmids HIS3 pEG202c,d,e HIS3 pJK202

Apr Apr

pNLexAe

HIS3

Apr

pGildad

HIS3

Apr

pEE202I

HIS3

Apr

pRFHM1e,f (control)

HIS3

Apr

pSH17-4e,f (control) pMW101f

HIS3

Apr

HIS3

Cmr

pMW103f

HIS3

Kmr

pHybLex/Zeof,g Zeor

Zeor

Activation domain fusion plasmids pJG4-5c,d,e,f TRP1 Apr

pJG4-5I

TRP1

Apr

pYESTrpg

TRP1

Apr

pMW102f

TRP1

Kmr

pMW104f

TRP1

Cmr

LacZ reporter plasmids URA3 pSH18-34d,e,f

Apr

pJK103e

URA3

Apr

pRB1840e

URA3

Apr

pMW112f pMW109f

URA3 URA3

Kmr Kmr

Contains an ADH promoter that expresses LexA followed by polylinker Like pEG202, but incorporates nuclear localization sequences between LexA and polylinker; used to enhance translocation of bait to nucleus Contains an ADH promoter that expresses polylinker followed by LexA; for use with baits where amino-terminal residues must remain unblocked Contains a GAL1 promoter that expresses same LexA and polylinker cassette as pEG202; for use with baits whose continuous presence is toxic to yeast An integrating form of pEG202 that can be targeted into HIS3 following digestion with KpnI; for use where physiological screen requires lower levels of bait to be expressed Contains an ADH promoter that expresses LexA fused to the homeodomain of bicoid to produce nonactivating fusion; used as positive control for repression assay, negative control for activation and interaction assays ADH promoter expresses LexA fused to GAL4 activation domain; used as a positive control for transcriptional activation Same as pEG202, but with altered antibiotic resistance markers; basic plasmid used for cloning bait Same as pEG202, but with altered antibiotic resistance markers; basic plasmid used for cloning bait Bait cloning vector compatible with interaction trap and all other two-hybrid systems; minimal ADH promotor expresses LexA followed by extended polylinker Contains a GAL1 promoter that expresses nuclear localization domain, transcriptional activation domain, HA epitope tag, cloning sites; used to express cDNA libraries An integrating form of pJG4-5 that can be targeted into TRP1 by digestion with Bsu36I (New England Biolabs); to be used with pEE202I to study interactions that occur physiologically at low protein concentrations Contains a GAL1 promoter that expresses nuclear localization domain, transcriptional activation domain, V5 epitope tag, multiple cloning sites; contains f1 ori and T7 promoter/flanking site; used to express cDNA libraries (Invitrogen) Same as pJG4-5, but with altered antibiotic resistance markers; no libraries yet available Same as pJG4-5, but with altered antibiotic resistance markers; no libraries yet available Contains 8 LexA operators that direct transcription of the lacZ gene; one of the most sensitive indicator plasmids for transcriptional activation Contains two LexA operators that direct transcription of the lacZ gene; an intermediate reporter plasmid for transcriptional activation Contains 1 LexA operator that directs transcription of the lacZ gene; one of the most stringent reporters for transcriptional activation Same as pSH18-34, but with altered antibiotic resistance marker Same as pJK103, but with altered antibiotic resistance marker continued



Table 19.2.1

Plasmid name/source pMW111f pMW107f pMW108f pMW110f pJK101e,f (control)

Interaction Trap Componentsa,b, continued

Selection In yeast

In E. coli

URA3 URA3 URA3 URA3 URA3

Kmr Cmr Cmr Cmr Apr

Comment/description Same as pRB1840, but with altered antibiotic resistance marker Same as pSH18-34, but with altered antibiotic resistance marker Same as pJK103, but with altered antibiotic resistance marker Same as pRB1840, but with altered antibiotic resistance marker Contains a GAL1 upstream activating sequence followed by two lexA operators followed by lacZ gene; used in repression assay to assess bait binding to operator sequences

aAll plasmids contain a 2µm origin for maintenance in yeast, as well as a bacterial origin of replication, except where noted (pEE202I, pJG4.5I). bInteraction Trap reagents represent the work of many contributors: the original basic reagents were developed in the Brent laboratory (Gyuris et al.,

1993). Plasmids with altered antibiotic resistance markers (all pMW plasmids) were constructed at Glaxo in Research Triangle Park, N.C. (Watson et al., 1996). Plasmids and strains for specialized applications have been developed by the following individuals: E. Golemis, Fox Chase Cancer Center, Philadelphia, Pa. (pEG202); J. Kamens, BASF, Worcester, Mass. (pJK202); cumulative efforts of I. York, Dana-Farber Cancer Center, Boston, Mass. and M. Sainz and S. Nottwehr, U. Oregon (pNLexA); D.A. Shaywitz, MIT Center for Cancer Research, Cambridge, Mass. (pGilda); R. Buckholz, Glaxo, Research Triangle Park, N.C. (pEE2021, pJG4-51); J. Gyuris, Mitotix, Cambridge, Mass. (pJG4-5); S. Hanes, Wadsworth Institute, Albany, N.Y. (pSH17-4); R.L. Finley, Wayne State University School of Medicine, Detroit, Mich. (pRFHM1); S. Hanes, Wadsworth Institute, Albany, N.Y. (pSH18-34); J. Kamens, BASF, Worcester, Mass. (pJK101, pJK103); R. Brent, The Molecular Sciences Institute, Berkeley, California (pRB1840). Specialized plasmids not yet commercially available can be obtained by contacting the Brent laboratory at (510) 647-0690 or [email protected] or the Golemis laboratory, (215) 728-2860 or [email protected]. cSequence data are available for pEG202 (pLexA) accession number pending. dPlasmids commercially available from Clontech and OriGene; for Clontech pEG202 is listed as pLexA, pJG4-5 as pB42AD, and pSH18-34 as p8op-LacZ. ePlasmids and strains available from OriGene. fIn pMW plasmids the ampicillin resistance gene (Apr) is replaced with the chloramphenicol resistance gene (Cmr) and the kanamycin resistance gene (Kmr) from pBC SK(+) and pBK-CMV (Stratagene), respectively. The choice between Kmr and Cmr or Apr plasmids is a matter of personal taste; use of basic Apr plasmids is described in the basic protocols. Use of the more recently developed reagents would facilitate the purification of library plasmid in later steps by eliminating the need for passage through KC8 bacteria, with substantial saving of time and effort. Apr has been maintained as marker of choice for the library plasmid because of the existence of multiple libraries already possessing this marker. These plasmids are the basic set of plasmids recommended for use. gPlasmids commercially available from Invitrogen as components of a Hybrid Hunter kit; this kit also includes all necessary positive and negative controls (not listed in this table). See Background Information for further details on commercially available reagents.

porter—the EGY48 strain (or related strain EGY191) that expresses the LexA fusion protein should not grow on medium lacking Leu, and the colonies should be white on medium containing Xgal. The characterized bait protein plasmid is used for Basic Protocol 2 to screen a library for interacting proteins. Materials DNA encoding the protein of interest Plasmids (see Table 19.2.1): pEG202 (see Fig. 19.2.3), pSH18-34 (see Fig. 19.2.4), pSH17-4, pRFHM1, and pJK101 for basic characterization; other plasmids for specific circumstances as described (Clontech, Invitrogen, OriGene, or R. Brent) Yeast strain EGY48 (ura3 trp1 his3 3LexA-operator-LEU2), or EGY191 (ura3 trp1 his3 1LexA-operator-LEU2; Table 19.2.2) Complete minimal (CM) medium dropout plates (APPENDIX 4L), supplemented with 2% (w/v) of the indicated sugars (glucose or galactose), in 100-mm plates: Glu/CM, −Ura, −His Gal/CM, −Ura, −His Gal/CM, −Ura, −His, −Leu Z buffer (APPENDIX 4L) with 1 mg/ml 5-bromo-4-chloro-3-indolyl-β-D-galactosidase (Xgal)



Supplement 15

Pst I 8672

Aat II 7995 Stu I 7835 Nru I 7765 Pst I 7671 Pst I 7590

Tth 111I 10059 Nar I 180 Nae I 246 SphI 1110 Hin dIII 1514 Mlu I 1616 Pme I 2056 pBR backbone ADHpro EcoRI 2144 Bam HI r Sal I * Ap Nco I lexA Not I Xho I Sal I *2182 ADH ter Pst I 2188 pEG202 SphI 2396 10166 bp

HIS3 Hin dIII 6464 Bst XI 6380 HindIII 6277 Pst I 6089 Bss HII 5960

Sac I 5113

2 µm ori

Xba I 3487

Pst I 4780 Avr II 5004

Polylinker sequence

Sal I* Not I Sal I* EcoRI BamHI NcoI XhoI GAA TTC CCG GGG ATC CGT CGA CCA TGG CGG CCG CTC GAG TCG AC

Figure 19.2.3 LexA-fusion plasmids: pEG202. The strong constitutive ADH promoter is used to express bait proteins as fusions to the DNA-binding protein LexA. Restriction sites shown in this map are based on recently compiled pEG202 sequence data and include selected sites suitable for diagnostic restriction endonuclease digests. A number of restriction sites are available for insertion of coding sequences to produce protein fusions with LexA; the polylinker sequence and reading frame relative to LexA are shown below the map with unique sites marked in bold type. The sequence 5′-CGT CAG CAG AGC TTC ACC ATT G-3′ can be used to design a primer to confirm correct reading frame for LexA fusions. Plasmids contain the HIS3 selectable marker and the 2µm origin of replication to allow propagation in yeast, and the Apr antibiotic resistance gene and the pBR origin of replication to allow propagation in E. coli. In the recently developed LexA-expression plasmids pMW101 and pMW103, the ampicillin resistance gene (Apr) has been replaced with the chloramphenicol resistance gene (Cmr) and the kanamycin resistance gene (Kmr), respectively (see Table 19.2.1 for details).

Gal/CM dropout liquid medium (APPENDIX 4L) supplemented with 2% Gal Antibody to LexA or fusion domain: monoclonal antibody to LexA (Clontech, Invitrogen) or polyclonal antibody to LexA (available by request from R. Brent or E. Golemis) H2O, sterile 30°C incubator Nylon membrane Whatman 3MM filter paper Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

Additional reagents and equipment for subcloning DNA fragments (Struhl, 1987), lithium acetate transformation of yeast (APPENDIX 4L), liquid assay for β-galactosidase (APPENDIX 4L), preparation of protein extracts for immunoblot analysis (see Support Protocol 1), and immunoblotting and immunodetection (UNIT 10.10)



Smal 0.00 Bam HI 0.01 EcoRI 0.01 XhoI 0.28

PstI 9.42 Hind III 9.22 EcoRI 9.19

EcoRI 0.60 Hind III 0.61 Bam HI 0.62

PstI 8.95 URA3

lexA op GAL1pro

2µm

lacZ LacZ reporter 10.3 kb

SacI 2.63

Apr

Hind III 7.05 Eco RI 6.95

pBR ori

Eco RI 3.70

Pst I 6.20

Figure 19.2.4 LacZ reporter plasmid. pRB1840, pJK103, and pSH18-34 are all derivatives of LR1∆1 (West et al., 1984) containing eight, two, or one operator for LexA (LexAop) binding inserted into the unique XhoI site located in the minimal GAL1 promoter (GAL1pro; 0.28 on map). The plasmid contains the URA3 selectable marker, the 2µm origin to allow propagation in yeast, the ampicillin resistance (Apr) gene, and the pBR322 origin (ori) to allow propagation in E. coli. Numbers indicate relative map positions. In the recently developed derivatives, the ampicillin resistance gene (Apr) has been replaced with the chloramphenicol or kanamycin resistance genes (see Table 19.2.1 for details).

NOTE: All solutions and equipment coming into contact with cells must be sterile, and proper sterile technique should be used accordingly. Transform yeast with the bait protein plasmid 1. Using standard subcloning techniques (e.g., Struhl, 1987), insert the DNA encoding the protein of interest into the polylinker of pEG202 (see Fig. 19.2.3) or other LexA fusion plasmid to make an in-frame protein fusion. The LexA fusion protein is expressed from the strong alcohol dehydrogenase (ADH) promoter. pEG202 also contains a HIS3 selectable marker and a 2ìm origin for propagation in yeast. pEG202 with the DNA encoding the protein of interest inserted is designated pBait. Uses of alternative LexA fusion plasmids are described in Background Information.

2. Perform three separate lithium acetate transformations (APPENDIX 4L) of EGY48 using the following combinations of plasmids: pBait + pSH18-34 (test) pSH17-4 + pSH18-34 (positive control for activation) pRFHM1 + pSH18-34 (negative control for activation). Use of the two LexA fusions as positive and negative controls allows a rough assessment of the transcriptional activation profile of LexA bait proteins. pEG202 itself is not a good negative control because the peptide encoded by the uninterrupted polylinker sequences is itself capable of very weakly activating transcription.



Supplement 17

Table 19.2.2

Interaction Trap Yeast Selection Strainsa

Strain

Relevant genotype

EGY48b,c,d

MATα trp1, his3, ura3, lexAops-LEU2

6

EGY191

MATα trp1, his3, ura3, lexAops-LEU2

2

L40c

MATα trpl, leu2, ade2, GAL4, lexAops-HIS34, lexAops-lacZ8

Number of operators

Comments/description lexA operators direct transcription from the LEU2 gene; EGY48 is a basic strain used to select for interacting clones from a cDNA library EGY191 provides a more stringent selection than EGY48, producing lower background with baits with instrinsic ability to activate transcription Expression driven from GAL1 promoter is constitutive in L40 (inducible in EGY strains); selection is for HIS prototrophy. Integrated lacZ reporter is considerably less sensitive than pSH18-34 maintained in EGY strains

aInteraction Trap reagents represent the work of many contributors: the original basic reagents were developed in the Brent laboratory

(Gyuris et al., 1993). Strains for specialized applications have been developed by the following individuals: E. Golemis, Fox Chase Cancer Center, Philadelphia, Pa. (EGY48, EGY191); A.B. Vojtek and S.M. Hollenberg, Fred Hutchinson Cancer Research Center, Seattle, Wash. (L40). Specialized strains not yet commercially available can be obtained by contacting the Brent laboratory at The Molecular Sciences Institute, Berkeley, (510) 647-0690 or [email protected], or the Golemis laboratory, (215) 728-2860 or [email protected]. bStrains commercially available from Clontech. cStrains commercially available from Invitrogen as components of a Hybrid Hunter kit; the kit also includes all necessary positive and negative controls (not listed in this table). See Background Information for further details on commercially available reagents. dStrains commercially available from OriGene.

pSH18-34 contains a 2ìm origin and a URA3 selectable marker for maintenance in yeast, as well as a bacterial origin of replication and ampicillin-resistance gene. It is the most sensitive lacZ reporter available and will detect any potential ability to activate lacZ transcription. pSH17-4 is a HIS3 2ìm plasmid encoding LexA fused to the activation domain of the yeast activator protein GAL4. This fusion protein strongly activates transcription. pRFHM1 is a HIS3 2ìm plasmid encoding LexA fused to the N-terminus of the Drosophila protein bicoid. This fusion protein has no ability to activate transcription.

3. Plate each transformation mixture on Glu/CM −Ura, −His dropout plates. Incubate 2 days at 30°C to select for yeast that contain both plasmids. Colonies obtained can be used simultaneously in tests for the activation of lacZ (steps 4 to 7) and LEU2 (steps 12 to 13) reporters.

Assay lacZ gene activation by β-galactosidase assay 4. Streak a Glu/CM −Ura, −His master dropout plate with at least five or six independent colonies obtained from each of the three transformations in step 3 (test, positive control, and negative control) and incubate overnight at 30°C. The filter assay described in Steps 5a to 7a (based on Breeden and Nasmyth, 1985) provides a rapid assay for β-galactosidase transcription. Alternatively, a liquid assay (APPENDIX 4L) or a plate assay (described in Steps 5b to 7b) may be used.

Perform filter assay for β-galactosidase activity: 5a. Lift colonies by gently placing a nylon membrane on the yeast plate and allowing it to become wet through. Remove the membrane and air dry 5 min. Chill the membrane, colony side up, 10 min at −70°C. Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

Whatman 3MM filters can be cut to the size of the yeast plate as a more economical alternative to nylon membranes for performing lifts. In addition, two or three 5-min temperature cycles (−70°C to room temperature) can be used instead of a single cycle to promote better lysis; this may be worth doing if there is difficulty visualizing blue color.



6a. Cut a piece of Whatman 3MM filter paper slightly larger than the colony membrane and soak it in Z buffer containing 1 mg/ml Xgal. Place colony membrane, colony side up, on Whatman 3MM paper, or float it in the lid of a petri dish containing ∼2 ml Z buffer with 1 mg/ml Xgal. Acceptable results may be obtained using as little as 300 ìg/ml Xgal.

7a. Incubate at 30°C and monitor for color changes. It is generally useful to check the membrane after 20 min, and again after 2 to 3 hr. Strong activators will produce a blue color in 5 to 10 min, and a bait protein (LexA fusion protein) that does so is unsuitable for use in an interactor hunt using this lacZ reporter plasmid. Weak activators will produce a blue color in 1 to 6 hr (compare versus negative control pRFHMI which will itself produce a faint blue color with time) and may or may not be suitable. Weak activators should be tested using the repressor assay described in steps 8 to 11.

Perform Xgal plate assay for lacZ activation: 5b. Prepare Z buffer Xgal plates as described in APPENDIX 4L. For activation assays, plates should be prepared with glucose as a sugar source. For repression assays (steps 8 to 11), galactose should be used as a sugar source. In our experience, when patching from a master plate to Xgal plates, sufficient yeast are transferred that plasmid loss is not a major problem even in the absence of selection; this is balanced by the desire to assay sets of constructs on the same plate to eliminate batch variation in Xgal potency. Hence, plates should be made either with complete minimal amino acid mix, or by dropping out only uracil (−Ura), to make the plates universally useful.

A +++ endogenous

GAL4 GALUAS

ops

lacZ

plasmid JK101

B

P1

endogenous

L e x A ops

GAL4 GALUAS

+ lacZ

plasmid JK101

Figure 19.2.5 Repression assay for DNA binding. (A) The plasmid JK101 contains the upstream activating sequence (UAS) from the GAL1 gene followed by LexA operators upstream of the lacZ coding sequence. Thus, yeast containing pJK101 will have significant β-galactosidase activity when grown on medium in which galactose is the sole carbon source because of binding of endogenous yeast GAL4 to the GALUAS (B). LexA-fused proteins (P1-LexA) that are made, enter the nucleus, and bind the LexA operator sequences (ops) will block activation from the GALUAS, repressing β-galactosidase activity (+) 3- to 5-fold. On glucose/Xgal medium, yeast containing pJK101 should be white because GALUAS transcription is repressed.



Supplement 14

6b. Streak yeast from master plate to Xgal plate and incubate at 30°C. 7b. Examine plates for color development at intervals over the next 2 to 3 days. Strongly activating fusions should be visibly blue on the plate within 12 to 24 hr; moderate activators will be visibly blue after ∼2 days. When a bait protein appreciably activates transcription under these conditions, there are several recourses. The first and simplest is to switch to a less sensitive lacZ reporter plasmid; use of pJK103 and pRB1840 may be sufficient to reduce background to manageable levels. If this fails to work, it is frequently possible to generate a truncated LexA fusion that does not activate transcription.

Confirm fusion-protein synthesis by repression assay For LexA fusions that do not activate transcription, confirm by performing a repression assay (Brent and Ptashne, 1984) that the LexA fusion protein is being synthesized in yeast (some proteins are not) and that it is capable of binding LexA operator sequences (Fig. 19.2.5). The following steps can be performed concurrently with the activation assay. 8. Transform EGY48 yeast with the following combinations of plasmids (three transformations): pBait + pJK101 (test) pRFHM1 + pJK101 (positive control for repression) pJK101 alone (negative control for repression). 9. Plate each transformation mix on Glu/CM −Ura, −His dropout plates or Glu/CM −Ura dropout plates as appropriate to select yeast cells that contain the indicated plasmids. Incubate 2 to 3 days at 30°C until colonies appear. 10. Streak colonies to a Glu/CM −Ura, −His or Glu/CM −Ura dropout master plate and incubate overnight at 30°C. 11. Assay β-galactosidase activity of the three transformed strains (test, positive control, and negative control) by liquid assay (using Gal/CM dropout liquid medium), filter assay (steps 5a to 7a, first restreaking to Gal/CM plates to grow overnight), or plate assay (steps 5b to 7b, using Gal/CM −Ura XGal plates). This assay should not be run for more than 1 to 2 hr for membranes, or 36 hr for Xgal plates, as the high basal lacZ activity will make differential activation of pJK101 impossible to see with longer incubations. Use of Xgal plates, and inspection 12 to 24 hr after streaking, is generally most effective. The plasmid pJK101 contains the galactose upstream activating sequence (UAS) followed by LexA operators upstream of the lacZ coding sequence. Thus, yeast containing pJK101 will have significant β-galactosidase activity when grown on medium in which galactose is the sole carbon source because of binding of endogenous yeast GAL4 to the GALUAS. LexA-fused proteins that are made, enter the nucleus, and bind the LexA operator sequences block activation from the GALUAS, repressing β-galactosidase activity 3- to 20-fold. Note that on Glu/Xgal medium, yeast containing pJK101 should be white, because GALUAS transcription is repressed.

12. If a bait protein neither activates nor represses transcription, perform immunoblot analysis by probing an immunoblot of a crude lysate with antibodies against LexA or the fusion domain to test for protein synthesis (see Support Protocol 1).

Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

Even if a bait protein represses transcription, it is generally a good idea to assay for the production of full-length LexA fusions, as occasionally some fusion proteins will be proteolytically cleaved by endogenous yeast proteases. If the protein is made but does not repress, it may be necessary to clone the sequence into a LexA fusion vector that contains a nuclear localization motif, e.g., pJK202 (see Table 19.2.1), or to modify or truncate the fusion domain to remove motifs that target it to other cellular compartments (e.g., myristoylation signals).



Test for Leu requirement These steps can be performed concurrently with the lacZ activation and repression assays. 13. Disperse a colony of EGY48 containing pBait and pSH18-34 reporter plasmids into 500 µl sterile water. Dilute 100 µl of suspension into 1 ml sterile water. Make a series of 1/10 dilutions in sterile water to cover a 1000-fold concentration range. 14. Plate 100 µl from each tube (undiluted, 1/10, 1/100, and 1/1000) on Gal/CM −Ura, −His dropout plates and on Gal/CM −Ura, −His, −Leu dropout plates. Incubate overnight at 30°C. There will be a total of eight plates. Gal/CM −Ura, −His dropout plates should show a concentration range from 10 to 10,000 colonies and Gal/CM −Ura, −His, −Leu dropout plates should have no colonies. Actual selection in the interactor hunt is based on the ability of the bait protein and acid-fusion pair, but not the bait protein alone, to activate transcription of the LexA operator-LEU2 gene and allow growth on medium lacking Leu. Thus, the test for the Leu requirement is the most important test of whether the bait protein is likely to have an unworkably high background. The LEU2 reporter in EGY48 is more sensitive than the pSH18-34 reporter for some baits, so it is possible that a bait protein that gives little or no signal in a β-galactosidase assay would nevertheless permit some level of growth on −Leu medium. If this occurs, there are several options for proceeding, the most immediate of which is to substitute EGY191 (see Table 19.2.2), a less sensitive screening strain, and repeat the assay. As outlined in this protocol, the authors recommend the strategy of performing the initial screening using the most sensitive reporters and then, if activation is detected, screening with increasingly less sensitive reporters (see Critical Parameters for further discussion).

PERFORMING AN INTERACTOR HUNT An interactor hunt involves two successive large platings of yeast containing LexA-fused probes and reporters and libraries in pJG4-5 (Fig. 19.2.6, Table 19.2.3) with a cDNA expression cassette under control of the GAL promoter. In the first plating, yeast are plated on complete minimal (CM) medium −Ura, −His, −Trp dropout plates with glucose (Glu) as a sugar source to select for the library plasmid. In the second plating, which selects for yeast that contain interacting proteins, a slurry of primary transformants is plated on CM −Ura, −His, −Trp, −Leu dropout plates with galactose/raffinose (Gal/Raff) as the sugar source. This two-step selection is encouraged for two reasons. First, a number of interesting cDNA-encoded proteins may be deleterious to the growth of yeast that bear them; these would be competed out in an initial mass plating. Second, it seems likely that immediately after simultaneous transformation and Gal induction, yeast bearing particular interacting proteins may not be able to initially express sufficient levels of these proteins to support growth on medium lacking Leu. Library plasmids from colonies identified in the second plating are purified by bacterial transformation and used to transform yeast cells for the final specificity screen.

BASIC PROTOCOL 2

A list of libraries currently available for use with this system is provided in Table 19.2.3. The protocol outlined below describes the steps used to perform a single-step screen that should saturate a library derived from a mammalian cell. For screens with libraries derived from lower eukaryotes with less complex genomes, fewer plates will be required. Occasionally, baits that seemed well-behaved during preliminary tests produce unworkably high backgrounds of “positives” during an actual screen (see Background Information and Critical Parameters). To forestall the waste of time and materials performing a screen with such a bait would entail, an alternative approach is to perform a scaled-back



Supplement 14

screen when working with a new bait (e.g., 5 rather than 30 plates of primary transformants). The results can be assessed before doing a full screen; it is then possible to switch to lower-sensitivity reporter strains and plasmids, if appropriate. Although individual baits will vary, the authors’ current default preference is to use the lacZ reporter pJK103 in conjunction with either EGY48 or EGY191. Polymerase chain reaction (PCR) can also be used in a rapid screening approach that may be preferable if a large number of positions are obtained in a library screen (see Alternate Protocol 1).

SacI 6440 PvuII 6253 Afl III 6075

KpnI 6446

HindIII 528 EcoRI 849 XhoI 861 GAL pro Alw NI 5661 HindIII 867 fusion SphI 1191 cassette BamHI 1330 pUC backbone XbaI 1336 ADH ter SalI 1342 NotI 1350 PstI 1364 Ap HindIII 1474 pJG4-5 6449 bp ScaI 4704

2µm ori

AatII 4264

XbaI 2072

TRP1 XbaI 4002 HindIII 3573

PstI 3365

Fusion cassette NLS

HA Tag

B42 domain

EcoRI

XhoI

ATG GGT GCT CCT CCA AAA AAG AAG ... CCC GAA TTC GGC CGA CTC GAG AAG CTT ... M G A P P K K K ... P E F G R L E K L ...


Figure 19.2.6 Library plasmids: pJG4-5. Library plasmids express cDNAs or other coding sequences inserted into unique EcoRI and XhoI sites as a translational fusion to a cassette consisting of the SV40 nuclear localization sequence (NLS; PPKKKRKVA), the acid blob B42 domain (Ruden et al, 1991), and the hemagglutinin (HA) epitope tag (YPYDVPDYA). Expression of cassette sequences is under the control of the GAL1 galactose-inducible promoter. This map is based on the sequence data available for pJG4-5, and includes selected sites suitable for diagnostic restriction digests (shown in bold). The sequence 5′-CTG AGT GGA GAT GCC TCC-3′ can be used as a primer to identify inserts or to confirm correct reading frame. The pJG4-5 plasmid contains the TRP1 selectable marker and the 2µm origin to allow propagation in yeast, and the antibiotic resistance gene and the pUC origin to allow propagation in E. coli. In the recently developed pJG4-5 derivative plasmids pMW104 and pMW102, the ampicillin resistance gene (Apr) has been replaced with the chloramphenicol resistance gene (Cmr) and the kanamycin resistance gene (Kmr), respectively (see Table 19.2.2 for details). Currently existing libraries are all made in the pJG4-5 plasmid (Gyuris et al., 1993) shown on this figure. Unique sites are marked in bold type.



Materials Yeast containing appropriate combinations of plasmids (see Table 19.2.1 and Table 19.2.2): EGY48 containing LexA-operator-lacZ reporter and pBait (see Basic Protocol 1) EGY48 containing LexA-operator-lacZ reporter and pRFHM-1 EGY48 containing LexA-operator-lacZ reporter and any nonspecific bait Complete minimal (CM) dropout liquid medium (APPENDIX 4L) supplemented with sugars (glucose, galactose, and/or raffinose) as indicated [2% (w/v) Glu, or 2% (w/v) Gal + 1% (w/v) Raff]: Glu/CM −Ura, −His Glu/CM −Trp Gal/Raff/CM −Ura, −His, −Trp H2O, sterile TE buffer (pH 7.5; APPENDIX 2E)/0.1 M lithium acetate Library DNA in pJG4-5 (Table 19.2.3 and Fig. 19.2.6) High-quality sheared salmon sperm DNA (see Support Protocol 2) 40% (w/v) polyethylene glycol 4000 (PEG 4000; filter sterilized)/0.1 M lithium acetate/TE buffer (pH 7.5) Dimethyl sulfoxide (DMSO) Complete minimal (CM) medium dropout plates (APPENDIX 4L) supplemented with sugars and Xgal (20 µg/ml) as indicated [2% (w/v) Glu, and 2% (w/v) Gal + 1% (w/v) Raff]: Glu/CM −Ura, −His, −Trp, 24 × 24–cm (Nunc) and 100-mm Gal/Raff/CM −Ura, −His, −Trp, 100-mm Gal/Raff/CM −Ura, −His, −Trp, −Leu, 100-mm Glu/Xgal/CM −Ura, −His, −Trp, 100-mm Gal/Raff/Xgal/CM −Ura, −His, −Trp, 100-mm Glu/CM −Ura, −His, −Trp, −Leu, 100-mm Glu/CM −Ura, −His, 100-mm Gal/CM −Ura, −His, −Trp, −Leu, 100-mm TE buffer (pH 7.5), sterile (optional) Glycerol solution (see recipe) E. coli KC8 (pyrF leuB600 trpC hisB463; constructed by K. Struhl and available from R. Brent) LB/ampicillin plates (APPENDIX 4A) E. coli DH5α or other strain suitable for preparation of DNA for sequencing Bacterial defined minimal A medium plates: 1× A medium plates containing 0.5 µg/ml vitamin B1 (APPENDIX 4A) and supplemented with 40 µg/ml each Ura, His, and Leu 30°C incubator, with and without shaking Low-speed centrifuge and rotor 50-ml conical tubes, sterile 1.5-ml microcentrifuge tubes, sterile 42°C heating block Glass microscope slides, sterile Additional reagents and equipment for rapid miniprep isolation of yeast DNA (APPENDIX 4L), transformation of bacteria by electroporation (UNIT 5.10), miniprep isolation of bacterial DNA (APPENDIX 4C), restriction endonuclease digestion (APPENDIX 4I; optional), and agarose gel electrophoresis (APPENDIX 4F; optional) NOTE: All solutions and equipment coming into contact with cells must be sterile, and proper sterile technique should be used accordingly.



Supplement 14

Table 19.2.3 Libraries Compatible with the Interaction Trap Systema

Source of RNA/DNA Cell lines HeLa cells (human cervical carcinoma)

Independent clones

Insert size (average)b Contact information

JG

9.6 × 106

0.3-3.5 kb (1.5 kb)

Y JG

3.7 × 106 5.7 × 106

0.3-1.2 kb 0.3-3.5 kb (1.5 kb)

JG

4.0 × 106

0.7-2.8 kb (1.5 kb)

R. Brent, Clontech, Invitrogen, OriGene Invitrogen R. Brent, Clontech, OriGene R. Brent

Y Y JG JG Y

3.2 × 106 3.0 × 106 5.7 × 106 2 × 106 5.4 × 106

0.3-1.2 kb 0.5-4.0 kb (1.8 kb) (>1.3) 0.7-3.5 kb (1.2 kb) 0.3-0.8 kb

Invitrogen Clontech OriGene S. Witte Invitrogen

JG JG JG

4.0 × 106 >106 1.5 × 106

0.4-4.0 kb (2.0 kb) 0.3-2.5 kb (>0.5 kb) 0.3-3.5 kb

Clontech R. Brent R. Brent

Vector

HeLa cells (human cervical carcinoma) WI-38 cells (human lung fibroblasts), serum-starved, cDNA Jurkat cells (human T cell leukemia), exponentially growing, cDNA Jurkat cells (human T cell leukemia) Jurkat cells (human T cell leukemia) Jurkat cells (human T cell leukemia) Jurkat cells (human T cell leukemia) Be Wo cells (human fetal placental choriocarcinoma) Human lymphocyte CD4+ T cell, murine, cDNA Chinese hamster ovary (CHO) cells, exponentially growing, cDNA A20 cells (mouse B cell lymphoma) Human B cell lymphoma Human 293 adenovirus–infected (early and late stages) SKOV3 human Y ovarian cancer MDBK cell, bovine kidney MDCK cells HepG2 cell line cDNA MCF7 breast cancer cells, untreated MCF7 breast cancer cells, estrogen-treated MCF7 cells, serum-grown LNCAP prostate cell line, untreated LNCAP prostate cell line, androgen-treated Mouse pachytene spermatocytes

Y JG JG

3.11 × 106 — —

0.3-1.2 kb — —

Invitrogen H. Niu K. Gustin

Y JG JG JG JG JG JG JG JG JG

5.0 × 106 5.8 × 106 — 2 × 106 1.0 × 107 1.0 × 107 1.0 ×107 2.9 × 106 4.6 × 106 —

(>1.4 kb) (>1.2 kb) — — (>1.5 kb) (>1.1 kb) 0.4-3.5 kb (>0.8 kb) (>0.9 kb) —

OriGene OriGene D. Chen M. Melegari OriGene OriGene OriGene OriGene OriGene C. Hoog

Tissues Human breast Human breast tumor Human liver Human liver

Y Y JG Y

9 × 106 8.84 × 106 >106 2.2 × 106

Invitrogen Invitrogen R. Brent Clontech

Human liver Human liver Human lung Human lung tumor Human brain Human brain Human testis Human testis Human ovary

JG JG Y Y JG Y Y JG Y

3.2 × 106 1.1 × 107 5.9 × 106 1.9 × 106 3.5 × 106 8.9 × 106 6.4 × 106 3.5 × 106 4.6 × 106

0.4-1.2 kb 0.4-1.2 kb 0.6-4.0 kb (>1 kb) 0.5-4 kb (1.3 kb) 0.3-1.2 kb (> 1 kb) 0.4-1.2 kb 0.4-1.2 0.5-4.5 kb (1.4 kb) 0.3-1.2 kb 0.3-1.2 kb 0.4-4.5 kb (1.6 kb) 0.3-1.2 kb

Invitrogen OriGene Invitrogen Invitrogen Clontech Invitrogen Invitrogen Clontech Invitrogen continued



Table 19.2.3 Libraries Compatible with the Interaction Trap Systema, continued

Source of RNA/DNA

Vector

Independent clones


Human ovary Human ovary Human heart Human placenta Human placenta Human mammary gland Human peripheral blood leucocyte Human kidney Human fetal kidney Human spleen Human prostate Human normal prostate Human prostate Human prostate cancer Human fetal prostate Human fetal liver Human fetal liver Human fetal liver Human fetal brain

JG JG JG Y JG JG JG JG JG Y Y JG JG JG JG JG Y JG JG

4.6 × 106 3.5 × 106 3.0 × 106 4.8 × 106 3.5 × 106 3.5 × 106 1.0 × 107 3.5 × 106 3.0 × 106 1.14 × 107 5.5 × 106 1.4 × 106 1.4 × 106 1.1 × 106 — 3.5 × 106 2.37 × 106 8.6 × 106 3.5 × 106

(>1.3 kb) 0.5-4.0 kb (1.8 kb) 0.3-3.5 kb (1.3 kb) 0.3-1.2 kb 0.3-4.0 kb (1.2 kb) 0.5-5 kb (1.6 kb) (>1.3 kb) 0.4-4.5 kb (1.6 kb) (>1 kb) 0.4-1.2 kb 0.4-1.2 kb 0.4-4.5 kb (1.7 kb) (>1 kb) (>0.9 kb) — 0.3-4.5 kb (1.3 kb) 0.3-1.2 kb (>1 kb) 0.5-1.2 kb (1.5 kb)

Mouse brain Mouse brain Mouse breast, lactating Mouse breast, involuting Mouse breast, virgin Mouse breast, 12 days pregnant Mouse skeletal muscle Rat adipocyte, 9-week-old Zucker rat Rat brain Rat brain (day 18) Rat testis Rat thymus Mouse liver Mouse spleen Mouse ovary Mouse prostate Mouse embryo, whole (19-day) Mouse embryo Drosophila melanogaster, adult, cDNA D. melanogaster, embryo, cDNA D. melanogaster, 0-12 hr embryos, cDNA D. melanogaster, ovary, cDNA D. melanogaster, disc, cDNA D. melanogaster, head

JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG JG

6.1 × 106 4.5 × 106 1.0 × 107 1.0 × 107 1.0 × 107 6.3 × 106 7.2 × 106 1.0 × 107 4.5 ×106 — 8.0 × 106 8.2 × 106 9.5 × 106 1.0 × 107 4.0 × 106 — 1.0 × 105 3.6 × 106 1.8 × 106 3.0 × 106 4.2 × 106

(>1 kb) 0.4-4.5 kb (1.2 kb) 0.4-3.1 kb 0.4-7.0 kb 0.4-5.5 kb 0.4-5.3 kb 0.4-3.5 kb 0.4-5.0 kb 0.3-3.4 kb — (>1.2 kb) (>1.3 kb) (>1.4 kb) (>1 kb) (>1.2 kb) — 0.2-2.5 kb 0.5-5 kb (1.7 kb) (>1.0 kb) 0.5-3.0 kb (1.4 kb) 0.5-2.5 kb (1.0 kb)

OriGene Clontech Clontech Invitrogen Clontech Clontech OriGene Clontech OriGene Invitrogen Invitrogen Clontech OriGene OriGene OriGene Clontech Invitrogen OriGene R. Brent, Clontech, Invitrogen, OriGene OriGene Clontech OriGene OriGene OriGene OriGene OriGene OriGene OriGene H. Niu OriGene OriGene OriGene OriGene OriGene OriGene OriGene Clontech OriGene Clontech R. Brent

JG JG JG

3.2 × 106 4.0 × 106 —

0.3-1.5 kb (800 bp) 0.3-2.1 kb (900 bp) —

R. Brent R. Brent M. Rosbash continued


Supplement 15

Table 19.2.3 Libraries Compatible with the Interaction Trap Systema, continued

Source of RNA/DNA

Vector

Miscellaneous Synthetic aptamers Saccharomyces cerevisiae, S288C, genomic S. cerevisiae, S288C, genomic Sea urchin ovary Caenorhabditis elegans Agrobacterium tumefaciens Arabidopsis thaliana, 7-day-old seedlings Tomato (Lycopersicon esculentum) Xenopus laevis embryo

PJM-1 JG JG JG JG JG JG JG JG

Independent clones

>1× 109 >3 × 106 4.0 × 106 3.5 × 106 3.8 × 106 — — 8 × 106 2.2 × 106


60 bp 0.8-4.0 kb 0.5-4.0 kb (1.7 kb) (>1.2 kb) — — — 0.3-4 kb (1.0 kb)

R. Brent R. Brent OriGene Clontech OriGene — H.M. Goodman G.B. Martin Clontech

aMost libraries are constructed in either the pJG4-5 vector or the pYESTrp vector (JG or Y in the Vector column); the peptide aptamer library is made in

the pJM-1 vector. Libraries available from the public domain were constructed by the following individuals: (1) J. Gyuris; (3) C. Sardet and J. Gyuris; (4) W. Kolanus, J. Gyuris, and B. Seed; (39) D. Krainc; (50-52) R. Finley; (55) P. Watt; (54) P. Colas, B. Cohen, T. Jessen, I. Grishina, J. McCoy, and R. Brent (Colas et al., 1996). All libraries mentioned above were constructed in conjunction with and are available from the laboratory of Roger Brent, (510) 647-0690 or [email protected]. The following individual investigators must be contacted directly: (18) J. Pugh, Fox Chase Cancer Center, Philadelphia, Pa.; (8,9) Vinyaka Prasad, Albert Einstein Medical Center New York, N.Y.; (57, 58) Gregory B. Martin, [email protected]; (11) Huifeng Niu, [email protected]; (16) Christer Hoog, [email protected]; (12) Kurt Gustin, [email protected]; (6) Stephan Witte, [email protected]. bInsert size ranges for pJG4-5 based libraries originally constructed in the Brent laboratory, which are now commercially available from Clontech, were reestimated by the company.

Transform the library 1. Grow an ∼20-ml culture of EGY48 or EGY191 containing a LexA-operator-lacZ reporter plasmid and pBait in Glu/CM −Ura, −His liquid dropout medium overnight at 30°C. For best results, the pBait and lacZ reporter plasmids should have been transformed into the yeast within ∼7 to 10 days of commencing a screen.

2. In the morning, dilute culture into 300 ml Glu/CM −Ura, −His liquid dropout medium to 2 × 106 cell/ml (OD600 = ∼0.10). Incubate at 30°C until the culture contains ∼1 × 107 cells/ml (OD600 = ∼0.50). 3. Centrifuge 5 min at 1000 to 1500 × g in a low-speed centrifuge at room temperature to harvest cells. Resuspend in 30 ml sterile water and transfer to 50-ml conical tube. 4. Centrifuge 5 min at 1000 to 1500 × g. Decant supernatant and resuspend cells in 1.5 ml TE buffer/0.1 M lithium acetate. 5. Add 1 µg library DNA in pJG4-5 and 50 µg high-quality sheared salmon sperm carrier DNA to each of 30 sterile 1.5-ml microcentrifuge tubes. Add 50 µl of the resuspended yeast solution from step 4 to each tube. The total volume of library and salmon sperm DNA added should be 500/plate) within 24 to 48 hr after plating on selective medium. Some investigators omit use of a Gal/Raff/CM −Ura, −His, −Trp, −Leu master plate, restreaking directly to a Glu/CM −Ura, −His, −Trp master plate as in step 19.

Test for Gal dependence The following steps test for Gal dependence of the Leu+ insert and lacZ phenotypes to confirm that they are attributable to expression of the library-encoded proteins. The GAL1 promoter is turned off and −Leu selection eliminated before reinducing. 18. Restreak from the Gal/Raff/CM −Ura, −His, −Trp, −Leu master dropout plate to a 100-mm Glu/CM −Ura, −His, −Trp master dropout plate. Incubate overnight at 30°C until colonies form. 19. Restreak or replica plate from this plate to the following plates: Glu/Xgal/CM −Ura, −His, −Trp Gal/Raff/Xgal/CM −Ura, −His, −Trp Glu/CM −Ura, −His, −Trp, −Leu Gal/Raff/CM −Ura, −His, −Trp, −Leu. Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

At this juncture, colonies and the library plasmids they contain are tentatively considered positive if they are blue on Gal/Raff/Xgal plates but not blue or only faintly blue on Glu/Xgal plates, and if they grow on Gal/Raff/CM −Leu plates but not on Glu/CM −Leu plates. The number of positives obtained will vary drastically from bait to bait. How they are processed subsequently will depend on the number initially obtained and on the preference



of the individual investigator. If none are obtained using EGY48 as reporter strain, it may be worth attempting to screen a library from an additional tissue source. If a relatively small number (≤30) are obtained, proceed to step 20. However, sometimes searches will yield large numbers of colonies (>30 to 300, or more). In this case, there are several options. The first option is to warehouse the majority of the positives and work up the first 30 that arise; those growing fastest are frequently the strongest interactors. These can be checked for specificity, and restriction digests can be used to establish whether they are all independent cDNAs or represent multiple isolates of the same, or a small number, of cDNAs. If the former is true, it may be advisable to repeat the screen in a less sensitive strain background, as obtaining many different interactors can be a sign of low-affinity nonspecific background. Alternatively, if initial indications are that a few cDNAs are dominating the positives obtained, it may be useful to perform a filter hybridization with yeast (see Support Protocol 3) using these cDNAs as a probe to establish the frequency of their identification and exclude future reisolation of these plasmids. The second major option is to work up large numbers of positives to get a complete profile of isolated interactors (see Support Protocol 4). A third option is to temporarily warehouse the entire results of this first screen, and repeat the screen with a less sensitive strain such as EGY191, on the theory that it is most important to get stronger interactors first and a complete profile of interactors later. Finally, some investigators prefer to work up the entire set of positives initially obtained, even if such positives number in the hundreds. Particularly in this latter case, it is most effective to use Alternate Protocol 1 as a means to identify unique versus common positives.

Isolate plasmid from positive colonies by transfer into E. coli 20a. Transfer yeast plasmids directly into E. coli by following the protocol for direct electroporation (UNIT 5.10). Proceed to step 22. 20b. Isolate plasmid DNA from yeast by the rapid miniprep protocol (APPENDIX 4C) with the following alteration: after obtaining aqueous phase, precipitate by adding sodium acetate to 0.3 M final and 2 vol ethanol, incubate 20 min on ice, microcentrifuge 15 min at maximum speed, wash pellet with 70% ethanol, dry, and resuspend in 5 µl TE buffer. Cultures can be grown prior to the miniprep using Glu/CM −Trp to select only for the library plasmid; this may increase the proportion of bacterial colonies that contain the desired plasmid.

21. Use 1 µl DNA to electroporate (UNIT 5.10) into competent KC8 bacteria, and plate on LB/ampicillin plates. Incubate overnight at 37°C. Electroporation must be used to obtain transformants with KC8 because the strain is generally refractory to transformation.

22. Restreak or replica plate colonies arising on LB/ampicillin plates to bacterial defined minimal A medium plates containing vitamin B1 and supplemented with Ura, His, and Leu but lacking Trp. Incubate overnight at 37°C. Colonies that grow under these conditions contain the library plasmid. The yeast TRP1 gene can successfully complement the bacterial trpC-9830 mutation, allowing the library plasmid to be easily distinguished from the other two plasmids contained in the yeast. It is helpful to first plate transformations on LB/ampicillin plates, which provides a less stringent selection, followed by restreaking to bacterial minimal medium to maximize the number of colonies obtained (E.G., unpub. observ.).

23. Purify library-containing plasmids using a bacterial miniprep procedure (APPENDIX 4C). Some investigators are tempted to immediately sequence DNAs obtained at this stage. At this point, it is still possible that none of the isolated clones will express bona fide



Supplement 14

interactors, and it is suggested that the following specificity tests be completed before committing the effort to sequencing (also see annotation to step 28). Because multiple 2ìm plasmids with the same marker can be simultaneously tolerated in yeast, it sometimes happens that a single yeast will contain two or more different library plasmids, only one of which encodes an interacting protein. The frequency of this occurrence varies in the hands of different investigators and may in some cases account for disappearing positives if the wrong cDNA is picked. When choosing colonies to miniprep, it is generally useful to work up at least two individual bacterial transformants for each yeast positive. These minipreps can then be restriction digested (APPENDIX 4I) with EcoRI + XhoI to release cDNA inserts, and the size of inserts determined on an agarose minigel (APPENDIX 4F) to confirm that both plasmids contain the same insert. An additional benefit of analyzing insert size is that it may provide some indication as to whether repeated isolation of the same cDNA is occurring, generally a good indication concerning the biological relevance of the interactor. See Background Information for further discussion.

Assess positive colonies with specificity tests Much spurious background will have been removed by the previous series of controls. Other classes of false positives can be eliminated by retransforming purified plasmids into “virgin” LexA-operator-LEU2/LexA-operator-lacZ/pBait–containing strains that have not been subjected to Leu selection and verifying that interaction-dependent phenotypes are still observed. Such false positives could include mutations in the initial EGY48 yeast that favor growth on Gal medium, library-encoded cDNAs that interact with the LexA DNA-binding domain, or proteins that are sticky and interact with multiple biologically unrelated fusion domains. 24. In separate transformations, use purified plasmids from step 23 to transform yeast that already contain the following plasmids and are growing on Glu/CM −Ura, −His plates: EGY48 containing pSH18-34 and pBait EGY48 containing pSH18-34 and pRFHM-1 EGY48 containing pSH18-34 and a nonspecific bait (optional). 25. Plate each transformation mix on Glu/CM −Ura, −His, −Trp dropout plates and incubate 2 to 3 days at 30°C until colonies appear. 26. Create a Glu/CM −Ura, −His, −Trp master dropout plate for each library plasmid being tested. Streak adjacently five or six independent colonies derived from each of the transformation plates. Incubate overnight at 30°C. 27. Restreak or replica plate from this master dropout plate to the same series of test plates used for the actual screen: Glu/Xgal/CM −Ura, −His, −Trp Gal/Raff/Xgal/CM −Ura, −His, −Trp Glu/CM −Ura, −His, −Trp, −Leu Gal/CM −Ura, −His, −Trp, −Leu. True positive cDNAs should make cells blue on Gal/Raff/Xgal but not on Glu/Xgal plates, and should make them grow on Gal/Raff/CM −Leu but not Glu/CM −Leu dropout plates only if the cells contain LexA-bait. cDNAs that meet such criteria are ready to be sequenced (see legend to Fig. 19.2.3 for primer sequence) or otherwise characterized. Those cDNAs that also encode proteins that interact with either RFHM-1 or another nonspecific bait should be discarded. Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

It may be helpful to cross-check the isolated cDNAs with a database of cDNAs thought to be false positives. This database is available on the World Wide Web as a work in progress at http://www.fccc.edu:80/research/labs/golemis/InteractionTrapInWork.html. cDNAs reported to this database are generally those isolated only once in a screen in which obviously



true interactive partners were isolated multiple times, cDNAs that may interact with more than one bait, or cDNAs for which the interaction does not appear to make biological sense in the context of the starting bait. Although some proteins in this database may ultimately turn out in fact to associate with the bait that isolated them, they are by default unlikely to possess a unique and interesting function in the context of that bait if they are well represented in the database.

28. If appropriate, conduct additional specificity tests (see Support Protocol 5). Analyze and sequence positive isolates. The primer sequence for use with pJG4-5 is provided in the legend to Figure 19.2.4. DNA prepared from KC8 is generally unsuitable for dideoxy or automated sequencing even after use of Qiagen columns and/or cesium chloride gradients. Library plasmids to be sequenced should be retransformed from the KC8 miniprep stock (step 23) to a more amenable strain, such as DH5α, before sequencing is attempted.

RAPID SCREEN FOR INTERACTION TRAP POSITIVES Under some circumstances, it may be desirable to attempt the analysis of a large number of positives resulting from a two-hybrid screen. One such hypothetical example would be a bait with a leucine zipper or coiled coil known to dimerize with partner “A” that is highly expressed. In order to identify the rare novel partner “B”, it is necessary to work through the high background of “A” reisolates. This protocol uses the polymerase chain reaction (PCR) in a strategy to sort positives into redundant (multiple isolates) and unique classes prior to plasmid rescue from yeast, thus greatly reducing the number of plasmid isolations that must be performed. An additional benefit is that this protocol preidentifies positive clones containing one or multiple library plasmids; for those containing only one library plasmid, only a single colony needs to be prepared through KC8/DH5α.


Additional Materials (also see Basic Protocol 2) Yeast plated on Glu/CM −Ura, −His, −Trp master plate (see Basic Protocol 2, step 19) Lysis solution (see recipe) 10 µM forward primer (FP1): 5′-CGT AGT GGA GAT GCC TCC-3′ 10 µM reverse primer (FP2): 5′-CTG GCA AGG TAG ACA AGC CG-3′ Toothpicks or bacterial inoculating loops (APPENDIX 4A), sterile 96-well microtiter plate Sealing tape, e.g., wide transparent tape 150- to 212-µm glass beads, acid-washed (UNIT 5.8) Vortexer with flat plate Additional reagents and equipment for performing an interactor hunt (see Basic Protocol 2), PCR amplification of DNA (APPENDIX 4J), agarose gel electrophoresis (APPENDIX 4F), restriction endonuclease digestion (APPENDIX 4I), electroporation (UNIT 5.10), and miniprep isolation of bacterial DNA (APPENDIX 4C) 1. Perform an interactor hunt (see Basic Protocol 2, steps 1 to 19). 2. Use a sterile toothpick or bacterial inoculating loop to transfer yeast from the Glu/CM, −Ura, −His, −Trp master plate into 25 µl lysis solution in a 96-well microtiter plate. Seal the wells of the microtiter plate with sealing tape and incubate 1.5 to 3.5 hr at 37°C with shaking. The volume of yeast transferred should not exceed ∼2 to 3 ìl of packed pellet; larger quantities of yeast will reduce quality of the DNA. DNA can be efficiently recovered from master plates that have been stored up to 1 week at 4°C. If yeast have been previously



Supplement 14

gridded on master plates, transfer to microtiter plates can be facilitated by using a multicolony replicator.

3. Remove tape from the plate, add ∼25 µl acid-washed glass beads to each well, and reseal with the same tape. Firmly attach the microtiter plate to a flat-top vortexer, and vortex 5 min at medium-high power. The microtiter plate can be attached to the vortexer using 0.25-in (0.64-cm) rubber bands.

4. Remove the tape and add ∼100 µl sterile water to each well. Swirl gently to mix, then remove sample for step 5. Press the tape back firmly to seal the microtiter plate and place in the freezer at −20°C for storage. 5. Amplify 0.8 to 2.0 µl of sample by standard PCR (APPENDIX 4J) in a ∼30-µl volume using 3 µl each of the forward primer FP1 and the reverse primer FP2. Perform PCR using the following cycles: Initial step: 31 cycles:

2 min 45 sec 45 sec 45 sec

94°C 94°C 56°C 72°C.

These conditions have been used successfully to amplify fragments up to 1.8 kb in length; some modifications, such as extension of elongation time, are also effective.

6. Load 20 µl of the PCR reaction product on a 0.7% low melting temperature agarose gel (APPENDIX 4F) to resolve PCR products. Based on insert sizes, group the obtained interactors in families, i.e., potential multiple independent isolates of identical cDNAs. Reserve gel until results of step 7 are obtained. No special precautions are needed for storing the gel. Since HaeIII digests typically yield rather small DNA fragments, running the second gel does not take a lot of time. Usually, the delay does not exceed 45 to 60 min, during which time the first gel may be stored in a gel box at room temperature or wrapped in plastic wrap at 4°C.

7. While the gel is running, use the remaining 10 µl of PCR reaction product for a restriction endonuclease digestion with HaeIII in a digestion volume of ∼20 µl (APPENDIX 4I). Based on analysis of the sizes of undigested PCR products in the gel (step 6), rearrange the tubes with HaeIII digest samples so that those thought to represent a family are side by side. Resolve the digests on a 2% to 2.5% agarose gel (APPENDIX 4F). Most restriction fragments will be in the 200-bp to 1.0-kb size range so using a long gel run is advisable. This analysis should produce a distinct fingerprint of insert sizes and allow definition of library cDNAs as unique isolates or related groups. A single positive yeast will sometimes contain multiple library plasmids. An advantage of this protocol is the ready detection of multiple library plasmids in PCR reactions; thus, following subsequent bacterial transformations, only a single TRP1 colony would need to be analyzed unless multiple plasmids were already known to be present.

8. Isolate DNA fragments from the low melting temperature agarose gel (step 6). If inspection of the banding pattern on the two gels suggests that a great many reisolates of a small number of cDNAs are present, it may be worthwhile to immediately sequence PCR products representative of these clusters, but it is generally still advisable to continue through specificity tests before doing so. If the PCR products are sequenced, the FP1 forward primer works well in automated sequencing of PCR fragments, but the FP2 primer is only effective in sequencing from purified plasmid. Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

In general, priming from the AT-rich ADH terminator downstream of the polylinker/cDNA in library plasmid is less efficient than from upstream of the cDNA, and it is hard to design effective primers in this region.



9. Remove the microtiter plate of lysates from the freezer, thaw it, and remove 2 to 4 µl of lysed yeast for each desired positive. Electroporate DNA into either DH5α or KC8 E. coli as appropriate, depending on the choice of bait and reporter plasmids (see Table 19.2.1 and see Background Information for further information). Refreeze the plate as a DNA reserve in case bacteria fail to transform on the first pass. KC8 E. coli should be used for electroporation when the original reagents pEG202/pJG45/pJK101 are used for the interaction trap. An additional strength of this protocol is that it identifies redundant clones before transfer of plasmids to bacteria, thus reducing the amount of work required in cases where plasmid identity can be unambiguously assigned. However, although restriction endonuclease digestion and PCR analysis are generally highly predictive, they are not 100% certain methods for estimating cDNA identity. Thus, if there is any doubt about whether two cDNAs are the same, investigators are urged to err on the side of caution.

10. Prepare a miniprep of plasmid DNA from the transformed bacteria (APPENDIX 4C) and perform yeast transformation and specificity assessment (see Basic Protocol 2, steps 24 to 28). PERFORMING A HUNT BY INTERACTION MATING An alternative way of conducting an interactor hunt is to mate a strain that expresses the bait protein with a strain that has been pretransformed with the library DNA, and screen the resulting diploid cells for interactors (Bendixen et al., 1994; Finley and Brent, 1994). This “interaction mating” approach can be used for any interactor hunt, and is particularly useful in three special cases. The first case is when more than one bait will be used to screen a single library. Interaction mating allows several interactor hunts with different baits to be conducted using a single high-efficiency yeast transformation with library DNA. This can be a considerable savings, since the library transformation is one of the most challenging tasks in an interactor hunt. The second case is when a constitutively expressed bait interferes with yeast viability. For such baits, performing a hunt by interaction mating avoids the difficulty associated with achieving a high-efficiency library transformation of a strain expressing a toxic bait. Moreover, the actual selection for interactors will be conducted in diploid yeast, which are more vigorous than haploid yeast and can better tolerate expression of toxic proteins. The third case is when a bait cannot be used in a traditional interactor hunt using haploid yeast strains (see Basic Protocol 2) because it activates transcription of even the least sensitive reporters. In diploids the reporters are less sensitive to transcription activation than they are in haploids. Thus, the interaction mating hunt provides an additional method to reduce background from transactivating baits.


In the protocol described below, the library DNA is used to transform a strain with a LEU2 reporter (e.g., EGY48). This pretransformed library strain is then frozen in many aliquots, which can be thawed and used for individual interactor hunts. The bait is expressed in a strain of mating type opposite to that of the pretransformed library strain, and also bearing the lacZ reporter. A hunt is conducted by mixing the pretransformed library strain with the bait strain and allowing diploids to form on YPD medium overnight. The diploids are then induced for expression of the library-encoded proteins and screened for interactors as in Basic Protocol 2. NOTE: Strain combinations other than those described below can also be used in an interaction-mating hunt. The key to choosing the strains is to ensure that the bait and prey strains are of opposite mating types and that both have auxotrophies to allow selection for the appropriate plasmids and reporter genes. Also, once the bait plasmid and lacZ reporter plasmid have been introduced into the bait strain, and the library plasmids have



Supplement 14

been introduced into the library strain, the resulting bait strain and library strain must each have auxotrophies that can be complemented by the other, so that diploids can be selected. Additional Materials (also see Basic Protocols 1 and 2) Yeast strains: either RFY206 (Finley and Brent, 1994), YPH499 (Sikorski and Hieter, 1989; ATCC #6625), or an equivalent MATa strain with auxotrophic markers ura3, trp1, his3, and leu2 YPD liquid medium (APPENDIX 4L) Glu/CM –Trp plates: CM dropout plates −Trp (APPENDIX 4L) supplemented with 2% glucose pJG4-5 library vector (Fig. 19.2.6), empty 100-mm YPD plates (APPENDIX 4L) Additional reagents and equipment for lithium acetate transformation of yeast (APPENDIX 4L) Construct the bait strain The bait strain will be a MATa yeast strain (mating type opposite of EGY48) containing a lacZ reporter plasmid like pSH18-34 and the bait-expressing plasmid, pBait. 1. Perform construction of the bait plasmid (pBait; see Basic Protocol 1, step 1). 2. Cotransform the MATa yeast strain (e.g., either RFY206 or YPH499) with pBait and pSH18-34 using the lithium acetate method (APPENDIX 4L). Select transformants on Glu/CM –Ura,–His plates by incubating plates at 30°C for 3 to 4 days until colonies form. Combine 3 colonies for all future tests and for the mating hunt. The bait strain (RFY206/pSH18-34/pBait or YPH499/pSH18-34/pBait) can be tested by immunoblotting to ensure that the bait protein is expressed (see Support Protocol 1). Synthesis and nuclear localization of the bait protein can also be tested by the repression assay (see Basic Protocol 1, steps 8 to 12).

3. Optional: Assay lacZ gene activation in the bait strain (see Basic Protocol 1, steps 4 to 7). If the bait activates the lacZ reporter, a less sensitive lacZ reporter plasmid (Table 19.2.1), or an integrated version of the lacZ reporter should be tried. A bait that strongly activates the lacZ reporters usually cannot be used in a hunt based on selection of interactors with the LEU2 reporter, because the LEU2 reporters are more sensitive than the lacZ reporters. However, both reporters are less sensitive to activation by a bait in diploid cells, as compared to haploid cells. Thus, a more important test of the transactivation potential of a bait is to test the leucine requirement of diploid cells expressing it, as described in steps 6 to 20, below.

Prepare the pretransformed library strain (EGY48 + library plasmids) 4. Perform a large-scale transformation of EGY48 with library DNA using the lithium acetate method (see Basic Protocol 2, steps 1 to 8, except start with EGY48 bearing no other plasmids). To prepare for transformation, grow EGY48 in YPD liquid medium. Select library transformants on Glu/CM –Trp plates by incubating 3 days at 30°C. 5. Collect primary transformants by scraping plates, washing yeast, and resuspending in 1 pellet vol glycerol solution (see Basic Protocol 2, steps 9 to 12). Freeze 0.2 to 1.0 ml aliquots at −70° to −80°C. Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

The cells will be stable for at least 1 year. Refreezing a thawed aliquot will result in loss of viability. Thus, many frozen aliquots should be made, so that each thawed aliquot can be discarded after use.



Prepare the pretransformed control strain (EGY48 + pJG4-5) 6. Transform EGY48 grown in YPD liquid medium with the empty library vector, pJG4-5, using the lithium acetate method (APPENDIX 4L). Select transformants on Glu/CM –Trp plates by incubating 3 days at 30°C. 7. Pick and combine three transformant colonies and use them to inoculate 30 ml of Glu/CM –Trp medium. Incubate 15 to 24 hr at 30°C (to OD600 >3). 8. Centrifuge 5 min at 1000 to 1500 × g, room temperature, and remove supernatant. Resuspend in 10 ml sterile water to wash cells. 9. Centrifuge 5 min at 1000 to 1500 × g, room temperature, and remove supernatant. Resuspend in 1 pellet vol glycerol solution and freeze 100-µl aliquots at −70° to −80°C. Determine plating efficiency of pretransformed library and pretransformed control strains 10. After freezing (at least 1 hr) thaw an aliquot of each pretransformed strain (from step 5 and step 9) at room temperature. Make several serial dilutions in sterile water, including aliquots diluted 105-fold, 106-fold, and 107-fold. Plate 100 µl of each dilution on 100-mm Glu/CM –Trp plates and incubate 2 to 3 days at 30°C. 11. Count the colonies and determine the number of colony-forming units (cfu) per aliquot of transformed yeast. The plating efficiency for a typical library transformation and for the control strain will be ∼1 × 108 cfu per 100 ìl.

Mate the bait strain with the pretransformed library strain and the pretransformed control strain In steps 12 through 20, an interactor hunt is conducted concurrently with testing LEU2 reporter activation by the bait itself. For most baits, this approach will be the quickest way to isolate interactors. However, for some baits, such as those that have a high transactivation potential, or those that affect yeast mating or growth, steps 12 through 20 will serve as a pilot experiment to determine the optimal parameters for a subsequent hunt. 12. Grow a 30-ml culture of the bait strain in Glu/CM –Ura,–His liquid dropout medium to mid to late log phase (OD600 = 1.0 to 2.0, or 2 to 4 × 107cells/ml). A convenient way to grow the bait strain is to inoculate a 5-ml culture with approximately three colonies from a plate and grow it overnight at 30°C with shaking. In the morning, measure the OD600, dilute into a 30-ml culture to a final OD600 = 0.2, and grow at 30°C with shaking. The culture should reach mid to late log phase before the end of the day.

13. Centrifuge the culture 5 min at 1000 to 1500 × g, room temperature, to harvest cells. Resuspend the cell pellet in sterile water to make a final volume of 1 ml. This should correspond to ∼1 × 109 cells/ml.

14. Set up two matings. In one sterile microcentrifuge tube mix 200 µl of the bait strain with 200 µl of a thawed aliquot of the pretransformed control strain from step 9. In a second microcentrifuge tube mix 200 µl of the bait strain with ∼1 × 108 cfu (∼0.1 to 1 ml) of the pretransformed library strain from step 5. The library mating should be set up so that it contains a ∼2-fold excess of bait strain cfu over pretransformed library strain cfu. Because the bait strain was harvested in log phase, most of the cells will be viable (i.e., cells/ml = ∼cfu/ml), and the number of cfu can be sufficiently estimated from optical density (1 OD600 = ∼2 × 107 cells/ml). Under these conditions, ∼10% of the cfu in the pretransformed library strain will mate with the bait



Supplement 14

strain. Thus, a complete screen of 107 library transformants will require a single mating with at least 108 cfu of the pretransformed library strain and at least 2 × 108 cfu of the bait strain. To screen more library transformants, set up additional matings. The number of pretransformed library transformants to screen depends on the size of the library and the number of primary transformants obtained in step 5. If the size of the library is larger than the number of transformants obtained in step 5, the goal will be to screen all of the yeast transformants. In this case, complete screening of the library will require additional transformations of EGY48 and additional interactor hunts. If the size of the library is smaller than the number of transformants obtained in step 5, the goal will be to screen at least a number of transformants equivalent to the size of the library.

15. Centrifuge each cell mixture for 5 min at 1000 to 1500 × g, pour off medium, and resuspend cells in 200 µl YPD medium. Plate each suspension on a 100-mm YPD plate. Incubate 12 to 15 hr at 30°C. 16. Add ∼1 ml of Gal/Raff/CM –Ura, –His, –Trp to the lawns of mated yeast on each plate. Mix the cells into the medium using a sterile applicator stick. 17. Transfer each slurry of mated cells to a 500-ml flask containing 100 ml of Gal/Raff/CM –Ura, –His, –Trp dropout medium. Incubate with shaking 6 hr at room temperature to induce the GAL1 promoter, which drives expression of the cDNA library. 18. Centrifuge the cell suspensions 5 min at 1000 to 1500 × g, room temperature, to harvest the cells. Wash by resuspending in 30 ml of sterile water and centrifuging again. Resuspend each pellet in 5 ml sterile water. Measure OD600 and, if necessary, dilute to a final concentration of ∼1 × 108 cells/ml. This is a mixture consisting of haploid cells that have not mated and diploid cells. Under a microscope, the two cell types can be distinguished by size (diploids are ∼1.7× bigger than haploids) and shape (diploids are slightly oblong and haploids are spherical). Because diploids grow faster than haploids, this mixture will contain ∼10% to 50% diploid cells. The actual number of diploids will be determined by plating dilutions on –Ura, –His, –Trp medium, which will not support the growth of the parental haploids.

19. For each mating make a series of 1⁄10 dilutions in sterile water, at least 200 µl each, to cover a 106-fold concentration range. Plate 100 µl from each tube (undiluted, 10−1, 10−2, 10−3, 10−4, 10−5, and 10−6 dilution) on 100-mm Gal/Raff/CM –Ura, –His, –Trp, –Leu plates. Plate 100 µl from the 10−4, 10−5, and 10−6 tubes on 100-mm Gal/Raff/CM –Ura, –His, –Trp plates. Incubate plates at 30°C. Count the colonies on each plate after 2 to 5 days. 20. For the mating with the pretransformed library, prepare an additional 3 ml of a 10−1 dilution. Plate 100 µl of the 10−1 dilution on each of 20 100-mm Gal/Raff/CM –Ura, –His, –Trp, –Leu plates. Also plate 100 µl of the undiluted cells on each of 20 100-mm Gal/Raff/CM –Ura, –His, –Trp, –Leu plates. Incubate at 30°C. Pick Leu+ colonies after 2 to 5 days and characterize them beginning with step 17 of Basic Protocol 2.


The number of Leu+ colonies to pick to ensure that all of the pretransformed library has been screened depends on the transactivation potential of the bait protein itself. The transactivation potential is expressed as the number of Leu+ colonies that grow per cfu (Leu+/cfu) of the bait strain mated with the control strain, as determined in step 19 of this protocol. It can be calculated as the ratio of the number of colonies that grow on Gal/Raff/CM –Ura,–His, –Trp, –Leu to the number of colonies that grow on Gal/Raff/CM –Ura, –His, –Trp for a given dilution of the mating between the bait strain and the control strain. A bait with essentially no transactivation potential will produce less than 10−6 Leu+/cfu. For a bait to be useful in an interactor hunt it should not transactivate more than 10−4 Leu+/cfu.



To screen all of the pretransformed library, it will be necessary to pick a sufficient number of Leu+ colonies in addition to background colonies produced by the transactivation potential of the bait itself. Thus, the minimum number of Leu+ colonies that should be picked in step 20 of this protocol is given by: (transactivation potential, Leu+/cfu) × (# library transformants screened). For example, if 107 library transformants were obtained in step 2 (and at least 108 cfu of these transformants were mated with the bait strain in step 14, since only ∼10% will form diploids), and the transactivation potential of the bait is 10−4 Leu+/cfu, then at least 1000 Leu+ colonies must be picked and characterized. In other words, if the rarest interactor is present in the pretransformed library at a frequency of 10−7, to find it one needs to screen through at least 107 diploids from a mating of the library strain. However, at least 1000 of these 107 diploids would be expected to be Leu+ due to the bait background if the transactivation potential of the bait is 10−4. The true positives will be distinguished from the bait background in the next step by the galactose dependence of their Leu+ and lacZ+ phenotypes.

PREPARATION OF PROTEIN EXTRACTS FOR IMMUNOBLOT ANALYSIS To confirm that the bait fusion protein constructed in Basic Protocol 1 is synthesized properly, a crude lysate is prepared for SDS-PAGE and immunoblot analysis (UNITS 10.1 & 10.10). The presence of the target protein is detected by antibody to LexA or the fusion domain.

SUPPORT PROTOCOL 1

Materials Master plates with pBait-containing positive and control yeast on Glu/CM −Ura, −His dropout medium (see Basic Protocol 1, step 4) Glu/CM −Ura, −His dropout liquid medium: CM dropout plates −Ura, −His (APPENDIX 4L) supplemented with 2% glucose 2× Laemmli sample buffer (see recipe) Antibody to fusion domain or LexA: monoclonal antibody to LexA (Clontech, Invitrogen) or polyclonal antibody to LexA (available by request from R. Brent or E. Golemis) 30°C incubator 100°C water bath Additional reagents and equipment for SDS-PAGE (UNIT 10.1) and immunoblotting and immunodetection (UNIT 10.10) 1. From the master plates, start a 5-ml culture in Glu/CM −Ura, −His liquid medium for each bait being tested and for a positive control for protein expression (i.e., RFHMI or SH17-4). Incubate overnight at 30°C. For each construct assayed, it is a good idea to grow colonies from at least two primary transformants, as levels of bait expression are sometimes heterogenous.

2. From each overnight culture, start a fresh 5-ml culture in Glu/CM −Ura, −His at OD600 = ∼0.15. Incubate again at 30°C. 3. When the culture has reached OD600 = 0.45 to 0.7 (∼4 to 6 hr), remove 1.5 ml to a microcentrifuge tube. For some LexA fusion proteins, levels of the protein drop off rapidly in cultures approaching stationary phase. This is due to a combination of the diminishing activity of the ADH1 promoter in late growth phases and the relative instability of particular fusion domains. Thus, it is not a good idea to let cultures become saturated in the hopes of obtaining a higher yield of protein.



Supplement 14

4. Microcentrifuge cells 3 min at 13,000 × g, room temperature. When the pellet is visible, remove the supernatant. Inspection of the tube should reveal a pellet ∼1 to 3 ìl in volume. If the pellet is not visible, microcentrifuge another 3 min.

5. Working rapidly, add 50 µl of 2× Laemmli sample buffer to the visible pellet in the tube, vortex, and place the tube on dry ice. Samples may be frozen at −70°C.

6. Transfer frozen sample directly to a boiling water bath or a PCR machine set to cycle at 100°C. Boil 5 min. 7. Microcentrifuge 5 sec at maximum speed to pellet large cellular debris. 8. Perform SDS-PAGE (UNIT 10.1) using 20 to 50 µl sample per lane. 9. To detect the protein, immunoblot and analyze (UNIT 10.10) using antibody to the fusion domain or LexA. SUPPORT PROTOCOL 2

PREPARATION OF SHEARED SALMON SPERM CARRIER DNA This protocol generates high-quality sheared salmon sperm DNA (sssDNA) for use as carrier in transformation (Basic Protocol 2). This DNA is also suitable for other applications where high-quality carrier DNA is needed (e.g., hybridization). This protocol is based on Schiestl and Gietz (1989). For more details of phenol extraction or other DNA purification methods, consult APPENDIX 4E. Materials High-quality salmon sperm DNA (e.g., sodium salt from salmon testes, Sigma or Boehringer Mannheim), desiccated TE buffer, pH 7.5 (APPENDIX 2E), sterile TE-saturated buffered phenol (APPENDIX 2E) 1:1 (v/v) buffered phenol/chloroform Chloroform 3 M sodium acetate, pH 5.2 (APPENDIX 2E) 100% and 70% ethanol, ice cold Magnetic stirring apparatus and stir-bar, 4°C Sonicator with probe 50-ml conical centrifuge tube High-speed centrifuge and appropriate tube 100°C and ice-water baths 1. Dissolve desiccated high-quality salmon sperm DNA in TE buffer, pH 7.5, at a concentration of 5 to 10 mg/ml by pipetting up and down in a 10-ml glass pipet. Place in a beaker with a stir-bar and stir overnight at 4°C to obtain a homogenous viscous solution. It is important to use high-quality salmon sperm DNA. Sigma Type III sodium salt from salmon testes has worked well, as has a comparable grade from Boehringer Mannheim. Generally it is convenient to prepare 20- to 40-ml batches at a time.

2. Shear the DNA by sonicating briefly using a large probe inserted into the beaker. Interaction Trap/ Two-Hybrid System to Identify Interacting Proteins

The goal of this step is to generate sheared salmon sperm DNA (sssDNA) with an average size of 7 kb, but ranging from 2 to 15 kb. Oversonication (such that the average size is closer to 2 kb) drastically decreases the efficacy of carrier in enhancing transformation. The original version of this protocol (Schiestl and Gietz, 1989) called for two 30-sec pulses at



Table 19.2.4

System

Two-hybrid System Variantsa

DNA-binding Activation Selection domain domain

Two-hybrid Interaction trap “Improved two-hybrid” Modified two-hybrid KISS Contingent replication

GAL4 LexA GAL4 LexA GAL4 GAL4

GAL4 B42 GAL4 VP16 VP16 VP16

Activation of lacZ, HIS3 Activation of LEU2, lacZ Activation of HIS3, lacZ Activation of HIS3, lacZ Activation of CAT, hygr Activation of T-Ag, replication of plasmids

Reference Chien et al., 1991 Gyuris et al., 1993 Durfee at al., 1993 Vojtek at al., 1993 Fearon et al., 1992 Vasavada et al., 1991

aAbbreviations: CAT, chloramphenicol transferase gene; hygr, hygromycin resistance gene; T-Ag, viral large T antigen.

three-quarter power, but optimal conditions vary between sonicators. The first time this protocol is performed, it is worthwhile to sonicate briefly, then test the size of the DNA by running out a small aliquot alongside molecular weight markers on an agarose gel containing ethidium bromide. The DNA can be sonicated further if needed.

3. Once DNA of the appropriate size range has been obtained, extract the sssDNA solution with an equal volume of TE-saturated buffered phenol in a 50-ml conical tube, shaking vigorously to mix. 4. Centrifuge 5 to 10 min at 3000 × g, room temperature, or until clear separation of phases is obtained. Transfer the upper phase containing the DNA to a clean tube. 5. Repeat extraction using 1:1 (v/v) buffered phenol/chloroform, then chloroform alone. Transfer the DNA into a tube suitable for high-speed centrifugation. 6. Precipitate the DNA by adding 1⁄10 vol of 3 M sodium acetate and 2.5 vol of ice-cold 100% ethanol. Mix by inversion. Centrifuge 15 min at ∼12,000 × g, room temperature. 7. Wash the pellet with 70% ethanol. Briefly dry either by air drying, or by covering one end of the tube with Parafilm with a few holes poked in and placing the tube under vacuum. Resuspend the DNA in sterile TE buffer at 5 to 10 mg/ml. Do not overdry the pellet or it will be very difficult to resuspend.

8. Denature the DNA by boiling 20 min in a 100°C water bath. Then immediately transfer the tube to an ice-water bath. 9. Place aliquots of the DNA in microcentrifuge tubes and store frozen at −20°C. Thaw as needed. DNA should be boiled again briefly (5 min) immediately before addition to transformations. Before using a new batch of sssDNA in a large-scale library transformation, it is a good idea to perform a small-scale transformation using suitable plasmids to determine the transformation efficiency. Optimally, use of sssDNA prepared in the manner described will yield transformation frequencies of >105 colonies/ìg input plasmid DNA.



Supplement 14

SUPPORT PROTOCOL 3

YEAST COLONY HYBRIDIZATION This protocol is adapted from a modification of the classic protocol of Grunstein and Hogness (1975; Kaiser et al., 1994). It is primarily useful when a large number of putative interactors has been obtained, and initial minipreps and restriction digests have indicated that many of them derive from a small number of cDNAs; these cDNAs can then be used as probes to screen and eliminate identical cDNAs from the pool. Materials Glu/CM −Trp plates: CM dropout plates −Trp (APPENDIX 4L) supplemental with 2% glucose Master dropout plate of yeast positive for Gal dependence (see Basic Protocol 2, step 18) 1 M sorbitol/20 mM EDTA/50 mM DTT (prepare fresh) 1 M sorbitol/20 mM EDTA 0.5 M NaOH 0.5 M Tris⋅Cl (pH 7.5)/6× SSC (APPENDIX 2E) 2× SSC (APPENDIX 2E) 100,000 U/ml β-glucuronidase (type HP-2 crude solution from Helix pomatia; Sigma) 82-mm circular nylon membrane, sterile Whatman 3MM paper 80°C vacuum oven or UV cross-linker Additional reagents and equipment for bacterial filter hybridization (Strauss, 1993; Duby et al., 1988) 1. Place a sterile nylon membrane onto a Glu/CM −Trp dropout plate. From the master dropout plate of Gal-dependent positives, gently restreak positives to be screened onto the membrane and mark the membrane to facilitate future identification of hybridizing colonies. Grow overnight (∼12 hr) at 30°C. Growth for extended periods of time (i.e., 24 hr) may result in difficulty in obtaining good lysis. It is a good idea to streak positive and negative controls for the cDNAs to be hybridized on the membrane.

2. Remove membrane from plate. Air dry briefly. Incubate ∼30 min on a sheet of Whatman 3MM paper saturated with 1 M sorbitol/20 mM EDTA/50 mM DTT. Optionally, before commencing chemical lysis, membranes can be placed at −70°C for 5 min, then thawed at room temperature for one or more cycles to enhance cell wall breakage.

3. Cut a piece of Whatman 3MM paper to fit inside a 100-mm petri dish. Place the paper disc in the dish and saturate with 100,000 U/ml β-glucuronidase diluted 1:500 in 1 M sorbitol/20 mM EDTA (2 µl glucuronidase per ml of sorbitol/EDTA to give 200 U/ml final). Layer nylon membrane on dish, cover dish, and incubate up to 6 hr at 37°C until >80% of the cells lack a cell wall. The extent of cell wall removal can be determined by removing a small quantity of cells from the filter to a drop of 1 M sorbitol/20 mM EDTA on a microscope slide and observing directly with a phase-contrast microscope at ≥60× magnification. Cells lacking cell wall are nonrefractile.


4. Place membrane on Whatman 3MM paper saturated with 0.5 M NaOH for ∼8 to 10 min. 5. Place membrane on Whatman 3MM paper saturated with 0.5 M Tris⋅Cl (pH 7.5)/6× SSC for 5 min. Repeat with a second sheet of Whatman 3MM paper.



6. Place membrane on Whatman 3MM paper saturated with 2× SSC for 5 min. Then place membrane on dry Whatman paper to air dry for 10 min. 7. Bake membrane 90 min at 80°C in vacuum oven or UV cross-link. 8. Process as for bacterial filter hybridization (Strauss, 1993; Duby et al., 1988), hybridizing the membrane with probes complementary to previously isolated cDNAs. When selecting probes, either random-primed cDNAs or oligonucleotides complementary to the cDNA sequence may be used. If the cDNA is a member of a protein family, it may be advantageous to use oligonucleotides to avoid inadvertently excluding genes related but not identical to those initially obtained.

MICROPLATE PLASMID RESCUE In some cases, it is desirable to isolate plasmids from a large number of positive colonies (Basic Protocol 2, steps 18 and 19). The protocol described below is a batch DNA preparation protocol developed by Steve Kron (University of Chicago, Chicago, Ill.) as a scale-up of a basic method developed by Manuel Claros (Laboratoire de Génétique Moleculaire, Paris, France).

SUPPORT PROTOCOL 4

Materials 2× Glu/CM −Trp liquid medium: 2× CM −Trp liquid medium (APPENDIX 4L) supplemented with 4% glucose Master plate of Gal-dependent yeast colonies (see Basic Protocol 2, step 18) Rescue buffer: 50 mM Tris⋅Cl (pH 7.5)/10 mM EDTA/0.3% (v/v) 2-mercaptoethanol (prepare fresh) Lysis solution: 2 to 5 mg/ml Zymolyase 100T/rescue buffer or 100,000 U/ml β-glucuronidase (type HP-2 crude solution from Helix pomatia; Sigma) diluted 1:50 in rescue buffer 10% (w/v) SDS 7.5 M ammonium acetate Isopropanol 70% ethanol TE buffer, pH 8.0 (APPENDIX 2E) 24-well microtiter plates Centrifuge with microplate holders, refrigerated Repeating micropipettor 37°C rotary shaker Grow yeast cultures 1. Aliquot 2 ml of 2× Glu/CM −Trp medium into each well of a 24-well microtiter plate. Into each well, pick a putative positive colony. Grow overnight with shaking at 30°C. The 2× minimal medium is used to maximize the yield of yeast. Four plates can generally be handled conveniently at once, based on the number that can be centrifuged simultaneously.

2. Centrifuge 5 min at 1500 × g, 4°C. Shake off supernatant with a snap and return the plate to upright. 3. Swirl or lightly vortex the plate to resuspend cell pellets in remaining liquid. Add 1 ml water to each well and swirl lightly. Cell pellets can most easily be resuspended in residual liquid before adding new solutions. Addition of liquid can be accomplished using a repeating pipettor.



Supplement 14

4. Centrifuge 5 min at 1500 × g, 4°C. Shake off supernatant and resuspend pellet. Add 1 ml rescue buffer. 5. Centrifuge 5 min at 1500 × g, 4°C. Shake off supernatant and resuspend pellet in the small volume of liquid remaining in the plate. Lyse cells 6. To each well, add 25 µl lysis solution. Swirl or vortex to mix. Incubate (with cover on) on a rotary shaker ∼1 hr at 37°C. Lysis solution need not be completely dissolved before use. By 1 hr, lysis should be obvious as coagulation of yeast into a white precipitate. Susceptibility of yeast strains to lytic enzymes varies. If lysis occurs rapidly, then less lytic enzyme should be used. If the lysis step is allowed to go too far, too much of the partially dissolved cell wall may contaminate the final material. Lysis can be judged by examining cells with a phase-contrast microscope. Living cells are white with a dark halo and dead cells are uniformly gray. Lysis leads to release of granular cell contents into the medium. Once cells are mostly gray and many are disrupted, much of the plasmid should have been released.

7. To each well, add 25 µl of 10% SDS. Mix gently by swirling to completely disperse the precipitates. Allow plates to sit 1 min at room temperature. At this point, the wells should contain a clear, somewhat viscous solution.

Purify plasmid 8. To each well, add 100 µl of 7.5 M ammonium acetate. Swirl gently, then incubate 15 min at −70°C or −20°C until frozen. Addition of acetate should result in the formation of a massive white precipitate of cell debris and SDS. The freezing step appears to improve removal of inhibitors of E. coli transformation.

9. Remove plate from freezer. Once it begins to thaw, centrifuge 15 min at 3000 × g, 4°C. Transfer 100 to 150 µl of the resulting clear supernatants to clean 24-well plates. In general, some contamination of the supernatant with pelleted material cannot be avoided. However, it is better to sacrifice yield in order to maintain purity.

10. To each well, add ∼0.7 vol isopropanol. Mix by swirling and allow to precipitate 2 min at room temperature. A cloudy fine precipitate should form immediately after isopropanol is added.

11. Centrifuge 15 min at 3000 × g, 4°C. Shake off supernatant with a snap. 12. To each well, add ∼1 ml cold 70% ethanol. Mix by swirling, centrifuge 5 min at 3000 × g, 4°C. Shake off supernatant with a snap, invert plates and blot well onto paper towel. Allow plates to air dry. 13. To each well, add 100 µl TE buffer. Swirl well and allow to rest on bench several minutes, until the pellets appear fully dissolved. Transfer preps to microcentrifuge tubes or 96-well plates for storage at −20°C.


One to five microliters of each of the resulting preparations can be used to transform competent E. coli: for KC8, electroporation should be used (see Basic Protocol 2, step 21). Sometimes, the yield of transformants is low if E. coli carrying plasmids are not permitted time to increase the plasmid copy number above a critical threshold before the cells are placed on selective medium. Allow plenty of time for cells to express antibiotic resistance or the TRP1 gene before plating. If insufficient numbers of colonies are obtained by this approach, the final plasmid preparation can be resuspended in 20 ìl instead of 100 ìl TE buffer to concentrate the DNA stock.



ADDITIONAL SPECIFICITY SCREENING The three test plasmids outlined (pSH18-34, pRFHM1, and pEG202; see Basic Protocol 2, step 24) represent a minimal test series. If other LexA-bait proteins that are related to the bait protein used in the initial library screen are available, substantial amounts of information can be gathered by additional specificity tests. For example, if the initial bait protein was LexA fused to the leucine zipper of c-Fos, specificity screening of interactor-hunt positives against the leucine zippers of c-Jun or GCN4 in addition to that of c-Fos might allow discrimination between proteins that are specific for fos versus those that generically associate with leucine zippers.

SUPPORT PROTOCOL 5

REAGENTS AND SOLUTIONS Use deionized, distilled water in all recipes and protocol steps. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Glycerol solution 65% (v/v) glycerol, sterile 0.1 M MgSO4 25 mM Tris⋅Cl, pH 8.0 (APPENDIX 2E) Store up to 1 year at room temperature Laemmli sample buffer, 2× 10% (v/v) 2-mercaptoethanol (2-ME) 6% (w/v) SDS 20% (v/v) glycerol 0.2 mg/ml bromphenol blue 0.025× Laemmli stacking buffer (see recipe; optional) Store up to 2 months at room temperature This reagent can conveniently be prepared 10 ml at a time.

Laemmli stacking buffer, 2.5× 0.3 M Tris⋅Cl, pH 6.8 0.25% (w/v) SDS Store up to 1 month at 4°C Lysis solution 50 mM Tris⋅Cl, pH 7.5 (APPENDIX 2E) 10 mM EDTA 0.3% (v/v) 2-mercaptoethanol (2-ME), added just before use 2% (v/v) β-glucuronidase from Helix pomatia (Type HP-2; Sigma), added just before use COMMENTARY Background Information Interaction-based cloning is derived from three experimental observations. In the first, Brent and Ptashne (1985) demonstrated that it was possible to assemble a novel, functional transcriptional activator by fusing the DNAbinding domain from one protein, LexA, to the activation domain from a second protein, GAL4. This allowed the use of a single reporter system containing a single DNA-binding motif, the LexA operator, to study transcriptional ac-

tivation by any protein of interest. In the second, Ma and Ptashne (1988) built on this work to demonstrate that the activation domain could be brought to DNA by interaction with a DNAbinding domain. In the third, Fields and Song (1989), working independently of Ma and Ptashne, used two yeast proteins, SNF1 and SNF4, to make an SNF1 fusion to the DNAbinding domain of GAL4 and an SNF4 fusion to the GAL4 activation domain. They demonstrated that the strength of the SNF1-SNF4



Supplement 14


interaction was sufficient to allow activation through a GAL4 DNA-binding site. From this, they suggested the feasibility of selecting interacting proteins by performing screens of cDNA libraries made so that library-encoded proteins carried activating domains. Several groups have developed cDNA library strategies along these lines, with some systems using LexA and others using GAL4 as the DNA-binding domain (Table 19.2.4). LexA and GAL4 each have different properties that should be considered when selecting a system. LexA is derived from a heterologous organism, has no known effect on the growth of yeast, possesses no residual transcriptional activity, can be used in GAL4+ yeast, and can be used with a Gal-inducible promoter. Because GAL4 is an important yeast transcriptional activator, it has the disadvantage that experiments must be performed in gal4− yeast strains to avoid background due to activation of the reporter system by endogenous GAL4. Such gal4− strains are frequently less healthy and more difficult to transform than wild-type strains, and either libraries must be constitutively expressed or alternate inducible systems must be used. By contrast, the GAL4 DNA-binding domain may be more efficiently localized to the nucleus and may be preferred for some proteins (for a review of GAL4-based systems, see Bartel et al., 1993). Whichever system is used, it is important to remember that the bait protein constitutes a novel fusion protein whose properties may not exactly parallel those of the original unfused protein of interest. Although systems using the two-hybrid paradigm have been developed in mammalian cells (see Table 19.2.4), these have not been used effectively in library screens. It seems likely that the organism of choice for two-hybrid identification of novel partner proteins will remain yeast. cDNAs that pass specificity tests are referred to as positives, or “true positives.” In interactor hunts conducted to date, anywhere from zero to practically all isolated plasmids passed the final specificity test. If no positives are obtained, the tissue source for the library originally used may not be appropriate, and a different library may produce better results. However, there are some proteins for which no positives are found. Various explanations for this are provided below. Conversely, some library-encoded proteins are known to be isolated repeatedly using a series of unrelated baits, and these proteins demonstrate at least some specificity. One of these, heat shock protein 70, might be explained by positing that it

assists the folding of some LexA-fused bait proteins, or alternatively, that these bait proteins are not normally folded. This example illustrates the point that the physiological relevance of even quite specific interactions may sometimes be obscure. Because the screen involves plating multiple cells to Gal/CM −Ura, −His, −Trp, −Leu dropout medium for each primary transformant obtained, multiple reisolates of true positive cDNAs are frequently obtained. If a large number of specific positives are obtained, it is generally a good idea to attempt to sort them into classes—for example, digesting minipreps of positives with EcoRI, XhoI, and HaeIII will generate a fingerprint of sufficient resolution to determine whether multiple reisolates of a small number of clones or single isolates of many different clones have been obtained. The former situation is a good indication that the system is working well. An important issue that arises in an interactor hunt is the question of how biologically relevant interacting proteins that are isolated are likely to be. This leads directly to the question of what Kd of association two molecules must have to be detected by an interactor hunt. In fact, this is not at all a simple issue. For the system described here, most fusion proteins appear to be expressed at levels ranging from 50 nM to 1 µM (Golemis and Brent, 1992). Given the strength of the GAL promoter, it is likely that many library-encoded proteins are expressed at similarly high levels, ≥1 µM in the nucleus (Golemis and Brent, 1992). At this concentration, which is in considerable excess over the nuclear concentration of operatorbound bait protein, a cDNA library–encoded protein should half-maximally occupy the DNA-bound bait protein if it possesses a Kd of 10−6 M, making it theoretically possible that very-low-affinity interactions could be detected. Such interactions have been observed in some cases. In contrast, some interactions that have been previously established using other methods and are predicted by known Kd to be easily detected by these means, either are not detected or are detected only weakly (Finley and Brent, 1994; Estojak et al., 1995). Because of the conservation of many proteins between lower and higher eukaryotes, one explanation for this observation is that either one or both of the partners being tested is being sequestered from the desired interaction by fortuitous association with an endogenous yeast protein. A reasonably complete investigation of the degree of correlation between in vitro determina-



tions of interaction affinity and apparent strength of interaction in the interaction trap is included in Estojak et al. (1995). The result of this investigation suggests it is important to measure the affinity of detected interactions under different conditions, using a second assay system, rather than to draw conclusions about affinity based on detection in the interaction trap. A number of different plasmids can be used for conducting an interactor hunt. Their properties are summarized in Tables 19.2.1 and 19.2.2. Because of the generous and open scientific exchange between investigators using the system, the number of available plasmids and other components has greatly expanded since the appearance of the initial two-hybrid reagents, facilitating the study of proteins inaccessible by the original system. The original parent plasmid for generating LexA fusions, pEG202 is a derivative of 202 + PL (Ruden et al., 1991; see Fig. 19.2.3) that contains an expanded polylinker region. The available cloning sites in pEG202 include EcoRI, BamHI, SalI, NcoI, NotI, and XhoI, with the reading frame as described in the legend to Figure 19.2.3. Since the original presentation of this system, a number of groups have developed variants of this plasmid that address specialized research needs. Those currently available, as well as purposes for which they are suited, are listed in Table 19.2.1. pGilda, created by David A. Shaywitz, places the LexAfusion cassette under the control of the inducible GAL1 promoter, allowing expression of the bait protein for limited times during library screening, reducing the exposure of yeast to toxic baits. pJK202, created by Joanne Kamens, adds nuclear localization sequences to pEG202, facilitating assay of the function of proteins lacking internal nuclear localization sequences. pNLexA, created by Ian York, places LexA carboxy-terminal in the fusion domain, allowing assay of interactions that require an unblocked amino-terminus on the bait protein. pEE202I, created by Mike Watson and Rich Buckholz, allows chromosomal integration of a pEG202-like bait, thus reducing expression levels so they are more physiological for bait proteins normally present at low levels intracellularly. All of these have been extensively tested by numerous researchers. pGilda, pJK202, and pEE202I work with complete reliability. pNLexA works effectively with ∼50% of the fusion domains tried, but synthesizes only very low levels of protein (relative to expression of the same fusion domain as a

pEG202 fusion) with the remaining 50%. Attachment of fusion domains amino-terminal either to LexA or GAL4 has been generally problematic in the hands of many investigators; it may be that appending additional protein sequences to the amino termini of these proteins is destabilizing, although the problem has not been rigorously investigated. A series of lacZ reporters of differing sensitivity to transcriptional activation can be used to detect interactions of varying affinity (see Table 19.2.2). These plasmids are LexA operator–containing derivatives of the plasmid LR1∆1 (West et al., 1984). In LR1∆1, a minimal GAL1 promoter lacking the GAL1 upstream activating sequences (GALUAS) is located upstream of the bacterial lacZ gene. In pSH18-34, eight LexA operators have been cloned into an XhoI site located 167 bp upstream of the lacZ gene (S. Hanes, unpub. observ.). pJK103 and pRB1840 contain two and one operator, respectively. pJK101 is similar to pSH18-34, except that it contains the GAL1 upstream activating sequences (GALUAS) upstream of two LexA operator sites. A derivative of del20B (West et al., 1984), it is used in the repression assay (Brent and Ptashne, 1984; see Fig. 19.2.5) to assess LexA fusion binding to operator. pSH17-4 is a HIS3 2µm plasmid that encodes LexA fused to the activation domain of the yeast activator GAL4. EGY48 cells bearing this plasmid will produce colonies in overnight growth on medium lacking Leu, and yeast that additionally contain pSH18-34 will turn deep blue on plates containing Xgal. This plasmid serves as a positive control for the activation of transcription. pRFHM1 is a HIS3 2µm plasmid that encodes LexA fused to the N-terminus of the Drosophila protein bicoid. The plasmid has no ability to activate transcription, so EGY48 cells that contain pRFHM1 and pSH18-34 do not grow on −Leu medium and remain white on plates containing Xgal. pRFHM1 is a good control for specificity testing, because it has been demonstrated to be sticky—that is, to associate with a number of library-encoded proteins that are clearly nonphysiological interactors (R. Finley, Wayne State University, Detroit, Mich., unpub. observ.). This protocol uses interaction libraries (Table 19.2.3) made in pJG4-5 or its derivatives (see Fig. 19.2.6). pJG4-5 was developed to facilitate isolation and characterization of novel proteins in interactor hunts (Gyuris et al., 1993). The pJG4-5 cDNA library expression



Supplement 14


cassette is under control of the GAL1 promoter, so library proteins are expressed in the presence of galactose (Gal) but not glucose (Glu). This conditional expression has a number of advantages, the most important of which is that many false-positives obtained in screens can be easily eliminated because they do not demonstrate a Gal-dependent phenotype. The expression cassette consists of an ATG to start translation, a nuclear localization signal to extend the interaction trap’s range to include proteins that are normally predominantly localized in the cytoplasm, an activation domain (acid blob; Ma and Ptashne, 1987), the hemagglutinin epitope tag to permit rapid assessment of the size of encoded proteins, EcoRI-XhoI sites designed to receive directionally synthesized cDNAs, and the alcohol dehydrogenase (ADH) termination sequences to enhance the production of high levels of library protein. The plasmid also contains the TRP1 auxotrophy marker and 2µm origin for propagation in yeast. A derivative plasmid, pJG4-5I, was created by Mike Watson and Richard Buckholz to facilitate chromosomal integration of the activation domain fusion expression plasmid. A series of recently developed derivatives of pEG202, pJG4-5, and lacZ reporter plasmids (MW101 to MW112) alter the antibiotic resistance markers on these plasmids from ampicillin (Apr) to either kanamycin (Kmr) or chloramphenicol (Cmr; Watson et al., 1996). Judiciously mixing and matching these plasmids in conjunction with Apr libraries would considerably reduce work subsequent to library screening, because the KC8 transformation, which involves trpC complementation in bacteria, could be omitted. EGY48 and EGY191 (see Table 19.2.2) are both derivatives of the strain U457 (a gift of Rodney Rothstein, Columbia University, New York, N.Y.) in which the endogenous LEU2 gene has been replaced by homologous recombination with LEU2 reporters carrying varying numbers of LexA operators, using a procedure detailed in Estojak et al. (1995). Interaction Trap–compatible reagents have recently become commercially available; Clontech and Invitrogen were the first to market such reagents and have recently been joined by OriGene. All suppliers use systems with the most sensitive reporters (EGY48 and pSH1834), and provide their own positive and negative controls for testing activation or interaction between defined proteins. For expression of bait and library proteins, the Clontech Matchmaker LexA two-hybrid system and the

OriGene Duplex-A system use some of the basic set of plasmids described here (see Table 19.2.1 for availability). Forward sequencing primers for bait and library plasmids are included in the Clontech kit, and Insert Screening Amplimer Sets for both plasmids can be acquired separately. Additional related products from Clontech include KC8 competent cells, anti-LexA monoclonal antibodies, a yeast transformation system, a yeast plasmid isolation kit, and an EGY48 partner strain for yeast mating to facilitate the analysis of interaction specificity. OriGene has a generally similar product line to Clontech. In contrast, Invitrogen has substantially modified the Interaction Trap core reagents to develop its own bait and library plasmids. pHybLex/Zeo, a novel bait plasmid, is ∼50% smaller than the original pEG202 (making it easier to clone into), and it has an enriched polylinker. Significantly, it replaces both the Apr and HIS3 genes with a novel gene that confers resistance to the antibiotic Zeocin (supplied with the kit), which provides selection in both bacteria and yeast. This elimination of auxotrophic selection for the bait plasmid renders the LexA-fusion construct usable with libraries and strains from all existing two-hybrid systems and additionally facilitates the direct selection of library plasmid in strains other than KC8. Some changes, which are designed to make the vector easier to use, have also been introduced in the library vector pYESTrp (e.g., it uses a V5 epitope tag for protein detection). The Invitrogen kit, termed Hybrid Hunter, includes the bait/library/reporter plasmids and EGY48 yeast strain as noted, and additionally includes primer sets for bait and library plasmids and the L40 yeast strain, should an investigator wish to use a HIS3 auxotrophy selection. Additional related products from Invitrogen include antibodies for detection of bait and prey fusion proteins (antiLexA and anti-V5), pJG4-5 library vector primers, and a Transformation Kit. A significant advantage of the entry of commercial entities into the Interaction Trap field is the rapid increase in the number of compatible cDNA libraries. A list of currently available premade libraries available from these companies is presented in Table 19.2.3, and custommade libraries are also available upon request. Because new libraries and other related reagents are being constantly added to the line of two-hybrid related products, it is advisable to contact the companies or visit their Web sites (www.clontech.com, www.invitrogen.com, and www.origene.com) for the latest information.



Finally, over the last several years, a number of groups have adapted basic two-hybrid strategies to more specialized applications, and they have devised strategies to broaden their basic functionality. Interaction Mating (Finley and Brent, 1994) has been used to establish extended networks of targeted protein-protein interaction. In this approach, a panel of LexAfused proteins are transformed into a MATa haploid selective strain (such as RFY206), a panel of activation-domain fused proteins are transformed into a suitable MATα haploid (such as EG448), and the two panels are crossgridded against each other for mating. Selected diploids are then screened by replica plating to selective medium. This approach complements library screening in large-scale applications, such as proposed definition of interaction maps for entire genomes (Bartel et al., 1996). Interaction mating has also provided the basis for an alternative two-hybrid hunt protocol (see Alternate Protocol 2), useful in cases when a single library will be screened with different baits. In this approach (Bendixen et al., 1994; Finley and Brent, 1994: Kolonin and Finley, 1998), a library is introduced into a single strain, like EGY48, and aliquots are stored frozen. To conduct a hunt, an aliquot is thawed and mated with a strain expressing a bait. This allows one to avoid repeated high-efficiency transformations, since a single library transformation can provide enough pretransformed yeast to conduct dozens of interactor hunts. Moreover, some yeast strains pretransformed with libraries are becoming commercially available, which may eliminate altogether the need to conduct a high-efficiency library transformation for some researchers. Two-hybrid approaches have been shown to be effective in identifying small peptides with biological activities on selected baits (Yang et al., 1995; Colas et al., 1996), which may prove to be useful as a guide to targeted drug design. Rapid screening protocols have been devised using custom-synthesized libraries expressing sheared plasmid DNA to facilitate rapid mapping of interaction interfaces (Stagljar et al., 1996). Osborne and coworkers have demonstrated the effectiveness of a tribrid (or tri-hybrid) approach, in which an additional plasmid expresses a tyrosine kinase to specifically modify a bait protein, allowing detection of SH2domain-containing partner proteins that recognize specific phosphotyrosine residues (Osborne et al., 1995). A variety of more elaborate tribrid approaches, in which a DNA-binding domain fused protein is used to present an

intermediate nonprotein compound for interaction with a library, have been developed and proven effective. These approaches have allowed the identification of proteins binding specific drug ligands (Chiu et al., 1994; Licitra and Liu, 1996), as well as the identification of proteins binding to RNA sequences (SenGupta et al., 1996; Wang et al., 1996). It is expected that the range of utility of these systems will continue to expand.

Critical Parameters and Troubleshooting To maximize chances of a successful interactor hunt, a number of parameters should be taken into account. Before attempting a screen, bait proteins should be carefully tested to ensure that they have little or no intrinsic ability to activate transcription. Bait proteins must be expressed at reasonably high levels and must be able to enter the yeast nucleus and bind DNA (as confirmed by the repression assay). Optimally, integrity and levels of bait proteins should be confirmed by immunoblot analysis, using an antibody to either LexA or the fused domain. In particular, at this time, bait proteins that have extensive transmembrane domains or are normally excluded from the nucleus are not likely to be productively used in a library screen. Proteins that are moderate to strong activators will need to be truncated to remove activating domains before they can be used. If a protein neither activates nor represses, the most likely reason is that it is not being made. This can be determined by immunoblot analysis of a crude lysate protein extract of EGY48 (UNIT 10.10; Samson et al., 1989) containing the plasmid, using anti-LexA antibodies as primary antiserum. If the full protein is not made, it may be possible to express truncated derivatives of the protein. If the protein is made, but still does not repress, it may not enter the yeast nucleus effectively, although this appears to be a relatively rare problem. In this case, introducing the coding sequence for the fused moieity into a LexA fusion vector containing a nuclear localization motif (e.g., pJK202; J. Kamens, BASF, Worcester, Mass., unpub. observ.) may solve the problem. The test for the leucine (Leu) requirement is extremely important to determine whether the bait protein is likely to yield an unworkably high background. The LEU2 reporter in EGY48 is more sensitive than the pSH18-34 reporter for some baits (Estojak et al., 1995). Therefore, it is possible that a bait protein demonstrating little or no signal in a β-galac-



Supplement 14


tosidase assay may nevertheless permit some growth on −Leu medium. If this occurs, there are several options. First, a less sensitive strain can be used, as described in the text. Second, background can sometimes be reduced further by making the EGY strain diploid (e.g., D. Krainc, Harvard Medical School, Boston, Mass.; R. Finley and R. Brent, unpub. observ.) or by performing the hunt by interaction mating as described in Alternate Protocol 2. A third option is to attempt to truncate the bait protein to remove activating function. In general, it is useful to extrapolate from the number of cells that grow on −Leu medium to the number that would be obtained in an actual library screen, and determine if this is a background level that can be tolerated. For example, if two colonies arise from 100,000 plated cells on −Leu medium, 200 to 400 would be expected in an actual screen of 106 cDNAs. Although this is a high initial number of positives, the vast majority should be eliminated immediately through easily performed controls. This is a judgment call. Finally, very rarely it happens that a bait that appears to be well behaved and negative for transcriptional activation through all characterization steps will suddenly develop a very high background of transcriptional activation following library transformation. The reason for this is currently obscure, and no means of addressing this problem has as yet been found: such baits are hence inappropriate for use in screens. The protocols described in this unit use initial screening with the most sensitive reporters followed by substitution with less sensitive reporters if activation is detected. An obvious question is, why not start out working with extremely stringent reporters and know immediately whether the system is workable? In fact, some researchers routinely use a combination of pJK103 or pRB1840 with EGY191, and obtain proteins that to date appear to be biologically relevant partners from library screens. However, extensive comparison studies using interactors of defined in vitro affinity with different combinations of LacZ and LEU2 reporters (Estojak et al., 1995) have indicated that although the most sensitive reporters (pSH1834) may in some cases be prone to background problems, the most stringent reporters (EGY191, pRB1840) may miss some interactions that certainly are biologically relevant and occur inside cells. In the end, the choice of reporters devolves to the preference of individual investigators: the bias of the authors is to cast a broad net in the early stages of a screen,

and hence to use more sensitive reporters when practicable. It is important to move expeditiously through characterization steps and to handle yeast transformed with bait plasmids with care. In cases where yeasts have been maintained on plates for extended periods (e.g., 4 days at room temperature or >2 to 3 weeks at 4°C), unexpected problems may crop up in subsequent library screens. The transformation protocol is a version of the lithium acetate transformation protocol described by Schiestl and Gietz (1989) and Gietz et al. (1992; see APPENDIX 4L) that maximizes transformation efficiency in Saccharomyces cerevisiae and produces up to 105 colonies/µg plasmid DNA. In contrast to Escherichia coli, the maximum efficiency of transformation for S. cerevisiae is ∼104 to 105/µg input DNA. It is extremely important to optimize transformation conditions before attempting an interactor hunt. Perform small-scale pilot transformations to ensure this efficiency is attained and to avoid having to use prohibitive quantities of library DNA. In addition, as for any effort of this type, it is a good idea to obtain or construct a library from a tissue source in which the bait protein is known to be biologically relevant. In practice, the majority of proteins isolated by interaction with a LexA fusion turn out to be specific for the fused domain; a smaller number are nonspecifically sticky, and to date there appears to have been only one isolation from a eukaryotic library of a protein specific for LexA. However, it is generally informative to retest positive clones on more than one LexA bait protein; ideally, library-derived clones should be tested against the LexA fusion used for their isolation, several LexA fusions to proteins that are clearly unrelated to the original fusion, and if possible, several LexA fusions that there is reason to believe are related to the initial protein (e.g., if the initial probe was LexA-Fos, a good related set would include LexA-Jun and LexA-GCN4). Colony selection for master plate production is one of the more variable parts of the procedure. For strong interactors, colonies will grow up in 2 days. However, if plates are left at 30°C, new colonies will continue to appear every day. Those that appear rapidly are most likely to reflect interactors that are biologically relevant to the bait protein. Those that appear later may or may not be relevant. However, many parameters can delay the time of colony formation of cells that contain valid interactions, including the strength of the interaction



and the level of expression of the library-encoded protein.

Anticipated Results Depending on the protein used as bait, anywhere from zero to hundreds of specific interactors will be obtained from 106 primary transformants.

Time Considerations If all goes well, once the required constructions have been made it will take ∼1 week to perform yeast transformations, obtain colonies, and determine whether bait proteins are appropriate. It will take a second week to perform library transformations, replate to selective medium, and obtain putative positives. A third week will be required to rescue the plasmid from the yeast, passage it through E. coli, transform fresh yeast, and confirm specificity.

Literature Cited Bartel, P.L., Chien, C.-T., Sternglanz, R., and Fields, S. 1993. Using the two-hybrid system to detect protein-protein interactions. In Cellular Interactions in Development: A Practical Approach (D.A. Hartley, ed.) pp. 153-179. Oxford University Press, Oxford. Bartel, P.L., Roecklein, J.A., SenGupta, D., and Fields, S. 1996. A protein linkage map of Escherichia coli bacteriophage T7. Nature Genet. 12:72-77. Bendixen, C., Gangloff, S., and Rothstein, R. 1994. A yeast mating-selection scheme for detection of protein-protein interactions. Nucl. Acids Res. 22:1778-1779. Breeden, L. and Nasmyth, K. 1985. Regulation of the yeast HO gene. Cold Spring Harbor Symp. Quant. Biol. 50:643-650. Brent, R. and Ptashne, M. 1984. A bacterial repressor protein or a yeast transcriptional terminator can block upstream activation of a yeast gene. Nature 312:612-615. Brent, R. and Ptashne, M. 1985. A eukaryotic transcriptional activator bearing the DNA specificity of a prokaryotic repressor. Cell 43:729-736. Chien, C.-T., Bartel, P.L., Sternglanz, R., and Fields, S. 1991. The two-hybrid system: A method to identify and clone genes for proteins that interact with a protein of interest. Proc. Natl. Acad. Sci. U.S.A. 88:9578-9582. Chiu, M.I., Katz, H., and Berlin, V. 1994. RAPT1, a mammalian homolog of yeast Tor, interacts with the FKBP12/rapamycin complex. Proc. Nat. Acad. Sci. U.S.A. 91:12574-12578. Colas, P., Cohen, B., Jessen, T., Grishina, I., McCoy, J., and Brent, R. 1996. Genetic selection of peptide aptamers that recognize and inhibit cyclindependent kinase 2. Nature 380:548-550. Duby, A., Jacobs, K.A., and Celeste, A. Using synthetic oligonucleotides as probes. In Current

Protocols in Molecular Biology (F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 6.4.16.4.10. John Wiley & Sons, New York. Durfee, T., Becherer, K., Chen, P.L., Yeh, S.H., Yang, Y., Kilburn, A.E., Lee, W.H., and Elledge, S.J. 1993. The retinoblastoma protein associates with the protein phosphatase type 1 catalytic subunit. Genes & Dev. 7:555-569. Estojak, J., Brent, R., and Golemis, E.A. 1995. Correlation of two-hybrid affinity data with in vitro measurements. Mol. Cell. Biol. 15:58205829. Fearon, E.R., Finkel, T., Gillison, M.L., Kennedy, S.P., Casella, J.F., Tomaselli, G.F., Morrow, J.S., and Dang, C.V. 1992. Karyoplasmic interaction selection strategy: A general strategy to detect protein-protein interactions in mammalian cells. Proc. Nat. Acad. Sci. U.S.A. 89:7958-7962. Fields, S. and Song, O. 1989. A novel genetic system to detect protein-protein interaction. Nature 340:245-246. Finley, R.L., Jr., and Brent, R. 1994. Interaction mating reveals binary and ternary connections between Drosophila cell cycle regulators. Proc. Natl. Acad. Sci. U.S.A. 91:12980-12984. Gietz, D., St. Jean, A., Woods, R.A., and Schiestl, R.H. 1992. Improved method for high-efficiency transformation of intact yeast cells. Nucl. Acids Res. 20:1425. Golemis, E.A. and Brent, R. 1992. Fused protein domains inhibit DNA binding by LexA. Mol. Cell Biol. 12:3006-3014. Grunstein, M., and Hogness, D.S. 1975. Colony hybridization: A method for the isolation of cloned DNAs that contain a specific gene. Proc. Natl. Acad. Sci. U.S.A. 72:3961-3965. Gyuris, J., Golemis, E.A., Chertkov, H., and Brent, R. 1993. Cdi1, a human G1- and S-phase protein phosphatase that associates with Cdk2. Cell 75:791-803. Kaiser, C., Michaelis, S., and Mitchell, A. 1994. Methods in Yeast Genetics, a Cold Spring Harbor Laboratory Course Manual, pp.135-136. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. Kolonin, M.G. and Finley, R.L., Jr. 1998. Targeting cyclin-dependent kinases in Drosophilia with peptide aptamers. Proc. Natl. Acad. Sci. U.S.A. In press. Licitra, E.J. and Liu, J.O. 1996. A three-hybrid system for detecting small ligand-protein receptor interactions. Proc. Nat. Acad. Sci. U.S.A. 93:12817-12821. Ma, J. and Ptashne, M. 1987. A new class of yeast transcriptional activators. Cell 51:113-119. Ma, J. and Ptashne, M. 1988. Converting an eukaryotic transcriptional inhibitor into an activator. Cell 55:443-446. Osborne, M., Dalton, S., and Kochan, J.P. 1995. The yeast tribrid system: Genetic detection of transphosphorylated ITAM-SH2 interactions. Bio/Technology 13:1474-1478.



Supplement 14

Ruden, D.M., Ma, J., Li, Y., Wood, K., and Ptashne, M. 1991. Generating yeast transcriptional activators containing no yeast protein sequences. Nature 350:426-430. Samson, M.-L., Jackson-Grusby, L., and Brent, R. 1989. Gene activation and DNA binding by Drosophila Ubx and abd-A proteins. Cell 57:1045-1052. Schiestl, R.H. and Gietz, R.D. 1989. High-efficiency transformation of intact yeast cells using single-stranded nucleic acids as a carrier. Curr. Genet. 16:339-346. SenGupta, D.J., Zhang, B., Kraemer, B., Pochart, P., Fields, S., and Wickens, M. 1996. A three-hybrid system to detect RNA-protein interactions in vivo. Proc. Nat. Acad. Sci. U.S.A. 93:8496-8501. Sikorski, R.S. and Hieter, P. 1989. A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae. Genetics 122:19-27. Stagljar, I., Bourquin, J.-P., and Schaffner, W. 1996. Use of the two-hybrid system and random sonicated DNA to identify the interaction domain of a protein. BioTechniques 21:430-432. Strauss, W.M. 1993. Using DNA fragments as probes. In Current Protocols in Molecular Biology (F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 6.3.1-6.3.6. John Wiley & Sons, New York. Struhl, K. 1987. Subcloning of DNA fragments. In Current Protocols in Molecular Biology (F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 3.16.1-3.16.11. John Wiley & Sons, New York. Treco, D.A. and Winston, F. 1992. Growth and manipulation of yeast. In Current Protocols in Molecular Biology (F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 13.2.1-13.2.12. John Wiley & Sons, New York. Vasavada, H.A., Ganguly, S., Germino, F.J., Wang, Z.X., and Weissman, S.M. 1991. A contingent replication assay for the detection of protein-protein interactions in animal cells. Proc. Nat. Acad. Sci. U.S.A. 88:10686-10690. Vojtek, A.B., Hollenberg, S.M., and Cooper, J.A. 1993. Mammalian Ras interacts directly with the serine/threonine kinase Raf. Cell 74:205-214. Wang, Z.F., Whitfield, M.L., Ingledue, T.C.3, Dominski, A., and Marzluff, W.F. 1996. The protein that binds the 3′ end of histone mRNA: A novel RNA-binding protein required for histone pre-mRNA processing. Genes & Dev. 10:3028-3040. Watson, M.A., Buckholz, R., and Weiner, M.P. 1996. Vectors encoding alternative antibiotic resistance for use in the yeast two-hybrid system. BioTechniques 21:255-259.

West, R.W.J., Yocum, R.R., and Ptashne, M. 1984. Saccharomyces cerevisiae GAL1-GAL10 divergent promoter region: Location and function of the upstream activator sequence UASG. Mol. Cell Biol. 4:2467-2478. Yang, M., Wu, Z., and Fields, S. 1995. Protein-peptide interactions analyzed with the yeast two-hybrid system. Nucl. Acids Res. 23:1152-1156.

Key Reference Gyuris et al., 1993. See above. Initial description of interaction trap system.

Internet Resources http://www.clontech.com http://cmmg.biosci.wayne.edu/rfinley/lab.html Source of two-hybrid information, protocols, and links. http://www.invitrogen.com http://www.origene.com Commercial sources for basic plasmids, strains, and libraries for interaction trap experiments. [email protected] [email protected] Sources of interaction trap plasmids for specialized interactions. http://www.fccc.edu:80/research/labs/golemis/ InteractionTrapInWork/html Database for false postive proteins detected in interaction trap experiments; analysis of two-hybrid usage. http://xanadu.mgh.harvard.edu/brentlabhome page4.html Database of interaction trap protocols and related issues.

Contributed by Erica A. Golemis and Ilya Serebriiskii Fox Chase Cancer Center Philadelphia, Pennsylvania Russell L. Finley, Jr. and Mikhail G. Kolonin (hunt by interaction mating) Wayne State University School of Medicine Detroit, Michigan Jeno Gyuris Mitotix, Inc. Cambridge, Massachusetts Roger Brent The Molecular Sciences Institute Berkeley, California




Phage-Based Expression Cloning to Identify Interacting Proteins

UNIT 19.3

Interaction cloning (also known as expression cloning) is a technique to identify and clone genes which encode proteins that interact with a protein of interest, or “bait” protein. Phage-based interaction cloning requires a gene encoding the bait protein and an appropriate expression library constructed in a bacteriophage expression vector, such as λgt11. The gene encoding the bait protein is used to produce recombinant fusion protein in E. coli. The cDNA is radioactively labeled with 32P. A recognition site for cyclic adenosine 3′, 5′-phosphate (cAMP)–dependent protein kinase (protein kinase A; PKA) is introduced into the recombinant fusion protein to allow its enzymatic phosphorylation by PKA and [γ-32P]ATP. The procedure presented here (see Basic Protocol) involves a fusion protein consisting of bait protein and glutathione-S-transferase (GST) with a PKA site at the junction between them (the protocol can, however, be adapted to use other PKA-containing recombinant proteins). The labeled protein is subsequently used as a probe to screen a λ bacteriophage-derived cDNA expression library, which expresses β-galactosidase fusion proteins that contain in-frame gene fusions. The phages lyse cells, form plaques, and release fusion proteins that are adsorbed onto nitrocellulose membrane filters. The filters are blocked with excess nonspecific protein to eliminate nonspecific binding and probed with the radiolabeled bait protein (see Fig. 19.3.1). This procedure leads directly to the isolation of genes encoding the interacting protein, bypassing the need for purification and microsequencing or for antibody production. NOTE: Radioactive label is used in this protocol, and appropriate precautions and shielding should be used (APPENDIX 2B). STRATEGIC PLANNING There are two important choices one must make to begin this procedure: (1) how to design the bait protein and (2) how to construct or acquire an appropriate phage-derived expression library. Vectors for recombinant fusion protein expression that contain a PKA recognition site can be obtained commercially. Several companies now sell these vectors with various affinity tags such as GST (Pharmacia Biotech), histidine (Novagen), or calmodulin-binding protein (Stratagene). Alternatively, one can engineer the PKA recognition site (the five–amino acid sequence RRASV) into existing vectors by using synthetic DNA that encodes it. Lambda-derived expression libraries that direct the expression of cDNAs made from many different mRNA sources are widely available. Alternatively, a library may be constructed for a particular experimental purpose; when designing a library to be used in expression cloning, several points should be considered. Libraries made from cDNA synthesized with random primers or an oligo dT-primer during first-strand synthesis (Klickstein et al., 1995; Klickstein and Neve, 1991) can often be advantageous in that multiple clones representing different portions of the same protein can be identified in the screening. Analysis of coding regions present in these multiple clones could provide useful information about what region of the protein is responsible for the observed interaction. Full-length cDNA clones can subsequently be obtained from other available libraries. Many suitable λ vectors are available for constructing cDNA expression libraries. The most widely used cDNA expression vector has been λgt11 (Huynh et al., 1985). There are modifications of these λ vectors, however, that facilitate the recovery of the cDNA Contributed by Julie M. Stone Current Protocols in Protein Science (1999) 19.3.1-19.3.9 Copyright © 1999 by John Wiley & Sons, Inc.



inserts by avoiding the necessity of time-consuming λ phage DNA preparations. These modified vectors include those that make use of the Cre-lox recombination for in vivo conversion of recombinant phages into plasmid DNA in Cre recombinase-expressing host strains (e.g., λZipLox, Life Technologies) as well as those that employ helper phage for in vivo excision of a phagemid vector (e.g., the λZAP series of vectors, Stratagene). Most λ vectors available produce fusions of the cDNA inserts to β-galactosidase, because they are cloned into the lacZ gene. However, some λ-derived expression vectors (e.g., λSCREEN-1, Novagen) direct expression of proteins fused to the T7 DNA polymerase promoter (gene 10 under control of the T7 promoter) and may yield higher expression of the library proteins (Margolis et al., 1992). BASIC PROTOCOL

INTERACTION CLONING Phage-based interaction expression cloning is a simple, rapid, and powerful technique to identify interacting proteins. A protein of interest is expressed as a recombinant fusion protein and labeled with 32P at a serine residue in an engineered PKA recognition site to facilitate detection. β-galactosidase proteins that are fused in-frame to cDNA inserts in a bacteriophage λ-derived expression library are produced by the phage and adsorbed onto nitrocellulose filters. The filters are then screened with the radiolabeled protein probe to identify phage clones that express an interacting protein.

A G ST

PKA ba it

B probed with GST-bait probed with GST

Phage-Based Expression Cloning

Figure 19.3.1 Schematic representation of the interaction cloning technique used to identify proteins that associate with a protein of interest (bait protein), and the expected results of a successful screen. (A) Expression of β-galactosidase fusion proteins from in-frame cDNA inserts of a λgt11 Arabidopsis library is induced with IPTG-impregnated nitrocellulose membrane filters (indicated by an oval). Filters are probed with GST-bait fusion protein labeled with 32P (•) at the PKA recognition site located at the junction of the fusion. Interacting clones are detected by autoradiography. (B) Representative autoradiogram of a tertiary screen of a positive plaque after the control experiment to determine whether the interaction is specific for the bait portion of the probe. The top half of the filter is probed with GST-bait while the bottom half is probed with GST alone as a control.



Materials cAMP-dependent protein kinase (PKA; e.g., 250-U lots from Sigma) 40 mM DTT, prepared fresh 10× PKA buffer (see recipe) 10 mCi/ml [γ-32P]ATP (6000 mCi/mmol) Purified glutathione-S-transferase (GST)–bait protein fusion protein with a PKA recognition site (UNIT 6.6), at ∼0.1 to 1 µg/µl concentration Z′-KCl (see recipe), ice cold Sephadex G-50 equilibrated in Z′-KCl E. coli Y1090r− or other appropriate host strain LB medium containing appropriate selective antibiotic (APPENDIX 4B), 10 mM MgSO4, and 0.2% maltose 10 mM MgSO4 10 mM IPTG (Table 1.4.2) 150- or 100-mm LB plates (with antibiotic, if necessary; APPENDIX 4B) 0.7% top agarose (APPENDIX 4B), 47°C Tris-buffered saline with Triton X-100 (TBS-T; see recipe) India ink HEPES blocking buffer (HBB; see recipe) Binding buffer (BB; see recipe) Suspension medium (SM; see recipe) Chloroform 3-ml disposable plastic columns or disposable syringe and glass wool Scintillation counter Tabletop centrifuge or equivalent Nitrocellulose membrane filters (137- and 82-mm disks) 22-G needle Additional reagents and equipment for preparation and purification of recombinant glutathione-S-transferase fusion protein (UNIT 6.6), SDS-PAGE (optional; UNIT 10.1), autoradiography (UNIT 10.11), titering and plating λ phage to generate plaques (Lech and Brent, 1988; Quertermous, 1996), and purification of bacteriophage clones (Quertermous, 1987) Prepare the 32P-labeled protein probe 1. Resuspend 250 U PKA in 25 µl freshly prepared 40 mM DTT. Let the reconstituted enzyme stand at room temperature ∼10 min before use. It is important to use freshly prepared 40 mM DTT. PKA is extremely unstable after reconstitution; the enzyme stock solution can be stored temporarily at 4°C but retains activity for only 2 to 3 days.

2. Prepare a phosphorylation reaction mixture containing: 1 µl 10 U/µl PKA (from step 1) 3 µl 10× PKA buffer 5 µl 10 mCi/ml (6000 mCi/mmol) [γ-32P]ATP 1 to 10 µl (∼1 µg) purified GST-bait fusion protein H2O to 30 µl. Incubate 1 hr at room temperature. A fusion protein unrelated to the bait protein, but containing the PKA recognition site, should also be expressed for use as a control to determine whether the observed interaction is specific for the bait moiety. Due to the instability of PKA, it is advisable to label both the bait protein and the unrelated control fusion protein simultaneously (in separate reactions).



Supplement 15

3. Add 170 µl ice-cold Z′-KCl (to stop reaction) and store on ice until use. 4. Prepare a gel filtration column by pouring Sephadex G-50 equilibrated in Z′-KCl into a 3-ml disposable plastic column (or a 3-ml syringe with a glass wool plug) for a final bed volume of ∼3 ml. Allow the column to drain until the level of the buffer is at the top of the column bed. The Sephadex G-50 can be swelled in water or buffer and stored at 4°C for months. In this case, five to ten column volumes of Z′-KCl can be used to equilibrate the column after it is poured.

5. Load the entire phosphorylation reaction (from step 3) onto the column and collect the effluent in a 1.5-ml microcentrifuge tube. Place a second tube under the column, add a 200-µl aliquot of Z′-KCl, and collect the effluent again. Repeat the above loading and collecting steps an additional ten times to collect a total of twelve 200-µl fractions. 6. Measure the Cerenkov counts with a scintillation counter to identify the fractions with the highest specific activities, and calculate cpm/µl. Typically two peaks of radioactivity are observed. The first peak elutes in fractions five to nine and corresponds to labeled protein. The second peak, usually found in the last few fractions collected, corresponds to unincorporated ATP and should be discarded. The hottest fraction(s), which elute first, should be used. Fractions can be stored at 4°C for several weeks.

7. Optional: Analyze a small amount (1 to 2 µl) of the 32P-labeled protein probe by SDS-PAGE and autoradiography (UNITS 10.1 & 10.11). Typically two strong signals are observed: one at the predicted size of the GST-bait fusion protein and the other at the predicted molecular weight of the GST alone (28 kDa). This is due to the fusion protein’s inherent susceptibility to protease cleavage at the GST-bait protein junction. The presence of labeled GST protein in the probe should not interfere with the screen; a control experiment with labeled GST will be performed at the later stages of screening.

Prepare host strain cells and dilution of bacteriophage cDNA library 8. Using serial dilution, determine the titer of the bacteriophage cDNA library. See Lech and Brent (1988) for serial dilution procedure.

9. Grow an overnight 50-ml culture of E. coli Y1090r− (or other appropriate host strain; APPENDIX 4B) in LB medium containing an appropriate selective antibiotic, 10 mM MgSO4, and 0.2% maltose. 10. Centrifuge cells 10 min at 2000 × g, room temperature, and resuspend in 25 ml of 10 mM MgSO4. The resuspended cells can be stored up to 1 week at 4°C before use.

Prepare the filters to screen the bacteriophage cDNA expression library 11. Soak eight 137-mm nitrocellulose filters in 10 mM IPTG for 15 min at room temperature and air dry. IPTG-impregnated filters can be stored in a petri plate until use.


12. Prepare eight 1.5-ml microcentrifuge tubes each containing 0.6 ml Y1090r− cells (from step 10) and ∼40,000 pfu bacteriophage (from step 8), and incubate 15 min at 37°C.



13. Add contents of each tube to 7 ml of 0.7% top agarose at 47°C and pour onto 150-mm LB plates (with antibiotic, if necessary). Incubate plates ∼3 hr at 42°C, until small plaques are visible. 14. Overlay the plates with IPTG-impregnated filters and incubate an additional 6 to 8 hr at 37°C. Formation of plaques in the absence of induction of the lacZ gene promoter ensures that any library-encoded proteins that are deleterious to phage growth will not be expressed while the phage are forming a plaque. Incubation in the presence of IPTG-impregnated filters is usually performed for a 6- to 8-hr period, but can proceed overnight for convenience.

15. Chill plates 15 min (or overnight) at 4°C. 16. Pierce each filter in several locations with a 22-G needle dipped in India ink to mark orientation. Remove filters from plates and wash 15 min in TBS-T at room temperature with shaking. Screen bacteriophage cDNA expression library 17. Incubate filters 1 to 4 hr with rocking at 4°C in 100 ml HBB. This blocking step (to reduce nonspecific binding) can also be performed overnight for convenience.

18. Incubate overnight with rocking at 4°C in BB containing 2.5–5 × 105 cpm/ml of radiolabeled fusion protein (from step 6). All eight filters can be placed in one 150-mm plate and probed with 30 to 40 ml of solution. The plate is wrapped in Parafilm and placed in a Plexiglas box for shielding. Alternatively, the filters can be placed with solution in a heat-sealable bag. The probe solution can be stored at 4°C and used for the subsequent secondary and tertiary screenings.

19. Wash membrane filters three times, 10 min each, in 100 ml BB with shaking at room temperature. Air dry and expose to film (see UNIT 10.11). 20. Using the large end of a Pasteur pipet, take an agarose plug at the position of the positive clone. Place into a 1.5-ml microcentrifuge tube and add 1 ml SM and 1 drop of chloroform. Agarose plugs can be stored at 4°C for months. Therefore, if many putative positive clones are identified in the primary screen, they can all be stored for future analysis if necessary.

21. Determine the titer by serial dilution and perform successive screening procedures to obtain purified clones. See Lech and Brent (1988) for serial dilution procedure and Quertermous (1987) for screening methods. Subsequent screens are performed on 100-mm LB plates with ∼2000 pfu/plate for secondary screens and ∼300 to 500 pfu/plate for tertiary screens. Before purified clones are obtained (usually during the tertiary screen), a control to eliminate clones that might interact with the GST fusion portion of the probe should be performed. To do this, cut the filters in half and probe one half with the labeled GST fusion to the protein of interest and the other with labeled GST alone or an unrelated control GST fusion protein (see Fig. 19.3.1).



Supplement 15

REAGENTS AND SOLUTIONS Use deionized, distilled water in all recipes and protocol steps. All solutions should be prepared from sterile, autoclaved stock solutions, except Z′-KCl, which should be filter sterilized. For common stock solutions, see APPENDIX 2A; for suppliers, see SUPPLIERS APPENDIX.

Binding buffer (BB) 20 mM HEPES⋅OH, pH 7.4 7.5 mM KCl 0.1 mM EDTA 2.5 mM MgCl2 1% (w/v) nonfat dry milk The solution can be prepared without milk and stored indefinitely at room temperature (add milk prior to use).

HEPES blocking buffer (HBB) 20 mM HEPES⋅OH, pH 7.4 5 mM MgCl2 1 mM KCl 5% (w/v) nonfat dry milk The solution can be prepared without milk and stored indefinitely at room temperature (add milk prior to use).

PKA buffer, 10× 200 mM Tris⋅Cl, pH 7.5 (APPENDIX 2E) 10 mM DTT 1 M NaCl 120 mM MgCl2 Store up to 6 months at room temperature Suspension medium (SM) 5.8 g NaCl 2 g MgSO4⋅7H2O 50 ml 1 M Tris⋅Cl, pH 7.5 (APPENDIX 2E) 5 ml 2% (w/v) gelatin H2O to 1 liter Sterilize by autoclaving. Store up to several months at 4°C Gelatin is prepared by adding 2 g gelatin to 100 ml H2O, then autoclaving to dissolve when needed.

Tris-buffered saline with Triton X-100 (TBS-T) 10 mM Tris⋅Cl, pH 8.0 (APPENDIX 2E) 150 mM NaCl 0.05% (v/v) Triton X-100 Store up to 6 months at room temperature


Z′-KCl 25 mM HEPES⋅OH, pH 7.4 12.5 mM MgCl2 20% (w/v) glycerol 100 mM KCl 1 mg/ml BSA 1 mM DTT Filter sterilize Store up to 6 months at 4°C



COMMENTARY Background Information Historically, interacting proteins have been isolated by biochemical approaches, which required purification of the interacting protein for antibody production or microsequencing before a clone encoding the protein could be identified. Two of the molecular biological approaches described in this chapter, the yeast two-hybrid system (UNIT 19.2) and phage-based interaction expression cloning (this unit), however, directly yield a clone encoding the interacting protein. Bacteriophage cDNA expression libraries are commonly screened using antibodies (see St. John, 1990 and references therein) or radiolabeled DNA probes to identify DNA-binding proteins (see Singh, 1991 and references therein). Modifications to identify interacting proteins include screening with the protein of interest (the bait protein) and detecting that protein with antibodies (Chapline et al., 1993). However, by screening with a radiolabeled protein probe, one avoids the additional incubations and washes which are necessary for immunodetection but that increase the likelihood of disrupting weak protein-protein interactions. 125I-labeled protein probes have been used successfully to screen for interacting proteins (Hoeffler et al., 1991), although the technique described in this unit avoids the need to label the proteins with 125I, and the complications of handling this isotope. Protein probes autophosphorylated with 32P were originally used to screen cDNA expression libraries in the isolation of proteins that interact with receptor protein kinases, a technique referred to as CORT (cloning of receptor targets; Skolnik et al., 1991; Lowenstein et al., 1992; Margolis et al., 1992). By introducing a PKA recognition site into the protein probe, the technique was made suitable for proteins that were not themselves protein kinases (Blanar and Rutter, 1992; Kaelin et al., 1992). Phage-based interaction expression cloning as described in this unit has been used successfully to identify many interacting proteins, but may not be successful for all types of interactions. For example, many related proteins (known as A-kinase anchoring proteins, or AKAPs) have been identified by their ability to interact with the type II cAMP-dependent kinase regulatory subunit, RII (Carr and Scott, 1992). All the AKAPs can bind RII under denaturing conditions (Lester et al., 1996), illustrating that the success of interaction cloning is often dependent on the nature of the interactions and that interactions less dependent on three-di-

mensional structure may be favored. However, there are many examples of proteins identified by this technique that cannot interact under the denaturing conditions often used to confirm the binding (commonly referred to as overlay assays or Far Western analysis). For example, a protein domain identified by its interaction with a plant receptor–like protein kinase is capable of interacting with a denatured form of the bait protein, but is unable to interact when the interacting protein is denatured (Stone et al., 1994). This fact is consistent with the idea that immobilization on membranes denatures some proteins but not others.

Critical Parameters and Troubleshooting The success of phage-based interaction expression cloning is inherently dependent on the quality of the protein probe used and the extent of representation of the bacteriophage cDNA library. GST fusion proteins are often obtained in high quantity in a soluble form, avoiding the necessity of solubilizing and refolding during protein purification from E. coli extracts. In most cases the PKA recognition site is readily accessible, and allows production of radiolabeled protein with a high specific activity. The strength of the protein-protein interaction is also critical. Depending on the nature of the interaction, a filter-binding technique might not be suitable. This technique may not detect weak or transient interactions, such as an enzyme/substrate interaction, and techniques such as the yeast two-hybrid system (UNIT 19.2) might be more appropriate. If no positive clones are identified or if background is high, it may be helpful to add reducing agents or detergents or to alter the salt conditions of the binding and wash solutions (Vinson et al., 1988). Moreover, denaturation and renaturation of the proteins on the filters using 6 M guanidine⋅HCl (Singh, 1991) may facilitate the recovery of clones expressing interacting proteins, because adsorption of the β-galactosidase fusion proteins to nitrocellulose filters may alter the conformation of the proteins. For a discussion of common problems with this procedure and their diagnosis and possible solutions, see Table 19.3.1.

Anticipated Results Depending on the nature of the interaction being sought by this technique, many interact-



Supplement 15

Table 19.3.1

General Troubleshooting Guide for Interaction Cloning

Problem

Possible cause

Solution

High background

Insufficient washing

Use more BB wash solution

Poor choice of nitrocellulose membrane filters Insufficient blocking

Try a different supplier for membranes Block overnight in HBB

Poor-quality protein probe

Check the probe by SDS-PAGE and autoradiography

Wash conditions too stringent

Reduce wash times Vary salt and/or detergent concentration

Poorly folded library fusion proteins

Perform 6 M guanidine⋅HCl denaturation/renaturation (Singh, 1991)

Wash and binding conditions insufficiently stringent Library biased to proteins that interact with the affinity tag

Increase salt and/or detergent in BB

No positive plaques

Too many positive plaques

ing clones, none, or just a few may be identified. However, the technique is as simple as screening a library by hybridization and often well worth the effort. In any case, observed interactions should be confirmed by other means.

Include unlabeled affinity tag in blocking, binding, and wash solutions

take an additional week. There are a number of steps that can be performed either rapidly or overnight, providing a great deal of convenience and flexibility.

Literature Cited Time Considerations


The interaction cloning technique described in this unit is extremely rapid in comparison with other techniques to identify interacting proteins, such as the yeast two-hybrid system (UNIT 19.2). The technique does not require any unusual reagents that would not be readily available in any laboratory routinely engaged in molecular biology, other than the cAMP-dependent protein kinase. Once the appropriate recombinant fusion protein and cDNA expression library are obtained, the primary screen can be completed in a few days. Once the appropriate construct for expression of the GST-bait is obtained, purification of the recombinant GST-bait protein can be achieved in 1 day. 32P labeling of the bait protein requires 2 hr (several additional hours if the optional SDSPAGE and autoradiography are performed). The initial screening of the bacteriophage cDNA library takes 3 days; day 1 to prepare the filters and induce expression of library-encoded proteins, day 2 for blocking and probing overnight with 32 P-labeled bait protein, and day 3 for washing and autoradiography. Subsequent purification of the cDNA clones should only

Blanar, M.A. and Rutter, W.J. 1992. Interaction cloning: Identification of a helix-loop-helix zipper protein that interacts with c-Fos. Science 256:1014-1018. Carr, D.W. and Scott, J.D. 1992. Blotting and bandshifting: Techniques for studying protein-protein interactions. Trends Biochem. Sci. 17:246-249. Chapline, C., Ramsay, K., Klauck, T., and Jaken, S. 1993. Interaction cloning of protein kinase C substrates. J. Biol. Chem. 268:6858-6861. Hoeffler, J.B., Lustbader, J.W., and Chen, C.Y. 1991. Identification of multiple nuclear factors that interact with cyclic AMP response element-binding protein and activation transcription factor-2 by protein interactions. Mol. Endocrinol. 5:256266. Huynh, T.V., Young, R.A., and Davis, R.W. 1985. Constructing and screening cDNA libraries in λgt10 and λgt11. In DNA Cloning: A Practical Approach (D.M. Glover, ed.) pp. 49-78. IRL Press, Oxford. Kaelin, W.G.J., Krek, W., Sellers, W.R., DeCaprio, J.A., Ajchenbaum, F., Fuchs, C.S., Chittenden, T., Li, Y., Farnham, P.J., Blanar, M.A., Livingston, D.M., and Flemington, E.K. 1992. Expression cloning of a cDNA encoding a retinoblastoma-binding protein with E2F-like properties. Cell 70:351-364.



Klickstein, L.B. and Neve, R. L. 1991. Ligation of linkers or adapters to double-stranded cDNA. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 5.6.1-5.6.10. John Wiley & Sons, New York.

St. John, T. P. 1990. Immunoscreening of fusion proteins produced in lambda plaques. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 6.7.16.7.6. John Wiley & Sons, New York.

Klickstein, L.B., Neve, R.L., Golemis, E.A., and Gyuris, J. 1995. Conversion of mRNA into double-stranded cDNA. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 5.5.1-5.5.14. John Wiley & Sons, New York.

Singh, H. 1991. Detection, purification, and characterization of cDNA clones encoding DNA-binding proteins. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 12.7.1-12.7.10. John Wiley & Sons, New York.

Lech, K. and Brent, R. 1988. Plating lambda phage to generate plaques. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 1.11.1-1.11.4 John Wiley & Sons, New York.

Skolnik, E.Y., Margolis, B., Mohammadi, M., Lowenstein, E., Fischer, R., Drepps, A., Ullrich, A., and Schlessinger, J. 1991. Cloning of PI3 kinase–associated p85 utilizing a novel method of expression/cloning of target proteins for receptor tyrosine kinases. Cell 65:83-90.

Lester, L.B., Coghlan, V.M., Nauert, B., and Scott, J.D. 1996. Cloning and characterization of a novel A-kinase anchoring protein: AKAP220, association with testicular peroxisomes. J. Biol. Chem. 271:9460-9465.

Stone, J.M., Collinge, M.A., Smith, R.D., Horn, M.A. and Walker, J.C. 1994. Interaction of a protein phosphatase with an Arabidopsis serinethreonine receptor kinase. Science 266:793-795.

Lowenstein, E.J., Daly, R.J., Batzer, A.G., Li, W., Margolis, B., Lammers, R., Ullrich, A., Skolnik, E.Y., Bar-Sagi, D., and Schlessinger, J. 1992. The SH2 and SH3 domain-containing protein GRB2 links receptor tyrosine kinases to ras signaling. Cell 70:431-442. Margolis, B., Silvennoinen, O., Comoglio, F., Roonprapunt, C., Skolnik, E., Ullrich, A., and Schlessinger, J. 1992. High-efficiency expression/cloning of epidermal growth factor–receptor-binding proteins with src homology 2 domains. Proc. Natl. Acad. Sci. U.S.A 89:8894-8898. Quertermous, T. 1987. Purification of bacteriophage clones. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 6.5.1-6.5.2. John Wiley & Sons, New York. Quertermous, T. 1996. Plating and transferring bacteriophage libraries. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 6.1.1-6.1.4. John Wiley & Sons, New York.

Vinson, C.R., LaMarco, K.L., Johnson, P.F., Landschulz, W.H., and McKnight, S.L. 1988. In situ detection of sequence-specific DNA binding activity specified by a recombinant bacteriophage. Genes & Dev. 2:801-806.

Key References Blanar and Rutter, 1992. See above. The basic protocol described in this unit is modified directly from the Blanar and Rutter protocol. Huynh et al., 1985. See above. Provides an excellent description of constructing and screening λgt11 cDNA expression libraries.

Contributed by Julie M. Stone Massachusetts General Hospital Boston, Massachusetts



Supplement 15

Detection of Protein-Protein Interactions by Coprecipitation

UNIT 19.4

Coprecipitation of proteins from whole-cell extracts is a valuable approach to test for physical interactions between proteins of interest. When a precipitating antibody is used, this method is referred to as co-immunoprecipitation. Coprecipitation can be used to study interactions between known proteins under a variety of conditions and as a means of identifying components of a complex. Coprecipitation may be the single method of choice, or may be used in combination with other methods that detect protein-protein interactions, such as two-hybrid analysis and copurification schemes (UNIT 19.2), and tests of physical associations using purified proteins. This unit describes basic approaches to immunoprecipitating tagged proteins from whole-cell extracts. The approaches described can be adapted for other systems. In a typical experiment, as described here, cells are lysed and a whole-cell extract is prepared under nondenaturing conditions (see Strategic Planning). The protein is precipitated from the lysate with a solid-phase affinity matrix and the precipitate is tested for the presence of a second specifically associated protein (see Basic Protocol and Alternate Protocol). The approach can be used for native or epitope-tagged proteins for which antibodies are available, or for recombinant proteins that have been engineered to bind with high affinity to a molecule that can be coupled to a solid-phase matrix (see Strategic Planning). The presence of an associated protein is detected by separating the precipitated proteins by SDS-PAGE (UNIT 10.1) and then immunoblotting (UNIT 10.10) with a second antibody that recognizes the putative associated protein. Controls to test specificity of interaction are crucial (see Strategic Planning). For additional background reading, the user should consult UNIT 3.8 for a theoretical discussion of immunoprecipitation and Coligan et al. (1999) for principles of antibody production and immunoassays; see UNIT 19.3 for approaches to tagging proteins; and see APPENDIX 4L and UNITS 5.6-5.8 for transformation and propagation of yeast. For an in-depth review of immunoprecipitation techniques, see Chapter 11 of Harlow and Lane (1988). For an in-depth review of coprecipitation and other approaches to detect protein-protein interactions, see Phizicky and Fields (1995). STRATEGIC PLANNING Detecting the Proteins in Question The first step is to generate reagents that detect the two proteins in the coprecipitate under nondenaturing conditions. If antibodies are available that can immunoprecipitate the proteins under nondenaturing conditions, then these can be used. Alternatively, the proteins can be differentially tagged in a variety of ways to allow their detection with commercially available antibodies or other affinity reagents. The tagged proteins are then introduced into the host organism using expression vectors. All tagged proteins must be assessed for function in vivo. A frequently used option is to add a short peptide or epitope that is recognized by a commercially available high-affinity monoclonal antibody (mAb). The epitope is typically added at the amino or carboxyl terminus, although internal positions that do not disrupt function can also be used. Two frequently used epitopes are derived from influenza hemagglutinin protein (HA) and human c-Myc and are recognized by high-affinity mAbs (12CA5 and 9E10, respectively; Kolodziej and Young, 1991). Others, such as FLAG, are Contributed by Elaine A. Elion Current Protocols in Protein Science (1999) 19.4.1-19.4.9 Copyright © 1999 by John Wiley & Sons, Inc.



also available (BioSupplyNet Source Book, 1999). The choice of epitope may be dictated by its amino acid composition. It is often useful to insert tandem copies of the epitope in order to increase sensitivity. The number of additional tandem copies can range widely, from one (Field et al., 1988) to several (e.g., three; Tyers et al., 1993) to many (e.g., nine; Feng et al., 1998). Proteins can also be fused to small proteins or peptides that have high affinity to small molecules that can be attached to a solid support. This is a particularly valuable approach when the protein to be precipitated comigrates with immunoglobulin heavy or light chains in an SDS-polyacrylamide gel. Such alternative tagging methods include fusion to glutathione-S-transferase (to allow purification by a glutathione affinity matrix; UNIT 6.6) or maltose-binding protein (to allow purification by a maltose affinity matrix; UNIT 5.1). An excellent reference for identifying sources of commercially available antibodies and approaches to tagging proteins is the BioSupplyNet Source Book (1999). See Coligan et al. (1999) and Harlow and Lane (1988) for the generation and purification of specific antibodies. See Chapters 5, 6, and 7 of this manual for a discussion of tagging and expressing proteins. Preparing Whole-Cell Extracts The second step in a successful coprecipitation is generating whole-cell extracts that optimize the yield and activity of the proteins to be analyzed, using lysis buffer conditions that permit recognition of the proteins by the affinity matrix. The yield of total protein in a whole-cell extract is not always a reliable indicator of the relative yield and activity of specific proteins, so it is wise to verify both parameters at the onset of an experiment before proceeding with the coprecipitation. Yield and activity can be affected by a number of factors (see Chapter 10; see Harlow and Lane, 1988). Small variations in the relative amounts of salt and detergents in the lysis buffer can have large effects on yield and activity, as can the speed and efficiency of cell breakage. Both factors are particularly important for less soluble proteins that associate with macromolecular structures such as membranes or cytoskeleton. In addition, global inhibition of proteolysis through the inclusion of multiple classes of protease inhibitors may be essential. Methods for preparing whole-cell extracts from yeast (UNITS 5.6-5.8), Escherichia coli (UNIT insect cells (UNIT 5.12), and mammalian cells (UNITS 5.9 & 5.10) can be found elsewhere in this manual, and specifics will not be discussed here. In general, the lysis buffer conditions are not very different from the coprecipitation conditions. It is recommended that the investigator begin by comparing small-scale extract preparations that vary the amount of salt and nonionic detergent. As a starting point, a basic lysis buffer might contain the following components.

6.2),

Basic components. Basic components include a buffering agent (such as 50 mM Tris⋅Cl, pH 7.5), a small amount of nonionic detergent (such as 0.1% [v/v] Triton X-100), salt (such as 100 mM NaCl), a reducing agent (such as 1 mM DTT), and 10% (v/v) glycerol as stabilizer. Protease inhibitors. Protease inhibitor cocktails are described here in UNIT 5.8 and are also commercially available. A reasonable starting point would be to include 5 µg/ml each of chymostatin, pepstatin A, leupeptin, and antipain, as well as 1 mM phenylmethysulfonyl fluoride and 1 mM benzamidine. Detection of Protein-Protein Interactions by Coprecipitation

Chelating agents. EGTA (∼15 mM) is commonly included to chelate divalent metal ions that are essential for metalloprotease activity. Because EGTA also inhibits other metaldependent enzymes, it may be omitted, combined with the addition of a needed metal ion, and/or substituted with EDTA.



Phosphatase inhibitors. If the phosphorylation state of the proteins in question is important, a mixture of phosphatase inhibitors should also be included in the lysis buffer. A starting mixture could contain 2.5 mM each meta- and ortho-vanadate, 10 mM NaF, and 10 mM β-glycerol phosphate. Simple modifications of this initial buffer include varying the amount of NaCl (from 0 to 500 mM) and of Triton X-100 (from 0% to 1%). The investigator may choose to compare different means of breaking the cells (for example, glass-bead breakage versus liquid nitrogen/grinding methods for yeast cells; UNITS 5.6-5.8).

1. Generate antibodies to protein 1 and protein 2; or differentially tag them ( ) and introduce genes encoding tagged proteins into host cell

protein 1

protein 2

2. Prepare whole-cell extract

3. Incubate extract with antibody to protein 1 4. Incubate extract with protein A–Sepharose, which binds antibody

5. Collect Sepharose beads by centrifugation

discard

P2 P1

supernatant

pellet

P2-NT P2 P1 P1-NT

protein 2 protein 1 Ig h Ig I

6. Wash pellet several times to remove proteins not bound to protein A–Sepharose. 7. Dissociate proteins from protein A – Sepharose. Separate proteins by SDS - PAGE. Immunoblot with antibody to protein 2 (P2). Reprobe with antibody to protein 1 (P1).

Figure 19.4.1 Flow chart for the coprecipitation of two proteins that have been differentially tagged and introduced into the host organism. Ig h and Ig l, immunoglobulin heavy and light chains; NT, no tag.



Supplement 17

Total protein concentration in the whole-cell extract is generally assayed by using the Bio-Rad protein assay and calculating protein concentration (UNIT 3.4). Extracts should be tested for the amount of each specific protein by immunoblot analysis (UNIT 10.10), analyzing 25 to 75 µg of total protein. In general, it is best to test for the presence of a second established protein (such as a housekeeping enzyme, cytoskeletal or ribosomal protein, or a previously defined component in the pathway being studied) as an internal control for normalization and as a positive control for the immunoblot. The amount of specific protein in the whole-cell extract is compared to the amount that is recovered by precipitation with an affinity matrix. Control Tests for Specificity of Interaction Controls are essential to verify that the antibodies and protein-protein interactions are specific. Proper controls are simplest to set up when the proteins are differentially tagged. In this instance, two parallel extracts are prepared from strains that contain each protein lacking the tag in the presence of the second tagged protein. An example is shown in the idealized gel in Figure 19.4.1, which includes lanes containing untagged protein 1 + tagged protein 2 and tagged protein 1 + untagged protein 2. If the antibodies are specific, untagged protein 1 will not immunoprecipitate. The presence of untagged protein 1 in the immunoprecipitate will indicate that it binds the affinity matrix nonspecifically. If the interaction between proteins 1 and 2 is specific, then tagged protein 2 will be present in the immunoprecipitate of tagged protein 1, but not in its absence. If antibodies to native proteins are used, it is necessary to compare extracts made from strains harboring deletions of the proteins in question to test for the specificity of the antibody and the interaction. However, this is obviously possible only if the deletions do not cause inviability. If deletion mutations cannot be used, a common approach is to show that the preimmune serum or an antibody not known to be specific to either of the proteins in question does not coprecipitate them in a parallel experiment. However, the latter two controls do not rule out the possibility that the antibody is precipitating the protein in question through an indirect association. It is also essential to compare the amount of coprecipitated protein with the amounts of the two proteins in question in the whole-cell extract. This allows one to determine whether apparent differences in the ability of the two proteins to coprecipitate are a secondary consequence of the relative abundance of the proteins. This control is particularly important when an interaction has been established and the investigator wishes to search for regulatory changes in association apart from changes in abundance. BASIC PROTOCOL


COPRECIPITATING PROTEINS WITH PROTEIN A/G–SEPHAROSE Once the conditions of extract preparation have been established (see Strategic Planning), the next step is to test for coprecipitation of the specific proteins. This protocol describes a standard coprecipitation procedure that uses an antibody coupled to protein A– Sepharose or protein G–Sepharose. An alternative coprecipitation method that uses GST coupled to glutathione-agarose is also provided (see Alternate Protocol). It is essential to keep all buffers and tubes cold by using an ice bath and a refrigerated centrifuge. The conditions of coprecipitation match the conditions of the lysis buffer described above. Materials Whole-cell extract (see Strategic Planning) Antibody Co-immunoprecipitation buffer (see recipe) 5 M NaCl (APPENDIX 2E)



Protein A/G–Sepharose slurry (see recipe) 2× sample buffer for SDS-PAGE (UNIT 10.1) Test tube rotator 20-ml syringe and 18-G needle Hamilton syringe Additional reagents and equipment for SDS-PAGE (UNIT 10.1) and immunoblotting (UNIT 10.10) 1. Prepare duplicate samples in microcentrifuge tubes on ice: 0.5 to 1 mg whole-cell extract 1 µg antibody 5 M NaCl to equalize at 100 mM NaCl Co-immunoprecipitation buffer to 0.5 ml final volume. Adjust buffer if necessary for activity of the protein in question (see Strategic Planning, discussion of chelating agents).

2. Invert tube gently several times and incubate on ice for 90 min with occasional tube inversion. It is recommended that the investigator begin with a 90-min incubation. However, this incubation step can be shortened or lengthened.

3. Microcentrifuge 10 min at maximum speed, 4°C, to pellet nonspecific aggregates. Transfer supernatant to a new microcentrifuge tube. 4. Add 50 µl of protein A– or protein G–Sepharose slurry (25 to 30 µl bead volume). Be sure to evenly suspend the slurry before distributing it to the samples. Protein A has been used more frequently for historical reasons; however, protein G binds a broader range of Ig subtypes at higher efficiency.

5. Rotate tube gently at 4°C for 30 to 60 min. Rocking is much less efficient and should be avoided.

6. Gently pellet protein A/G–Sepharose by centrifuging 30 sec at 1000 rpm in a tabletop centrifuge, 4°C. 7. Wash pellet three times with 1 ml co-immunoprecipitation buffer. For each wash, gently invert tube three times before pelleting. After each pelleting, use a 20-ml syringe with an 18-G needle to aspirate and remove supernatant. It may be possible to omit the costly protease inhibitors from the buffer at this stage, but this has not been attempted to date.

8. Aspirate as much liquid as possible from the final without touching the beads and add 25 µl of 2× sample buffer. If desired, samples containing sample buffer can be frozen up to several months at −80°C prior to SDS-PAGE. In this case the buffer should be prepared with sterile stock solutions and made with 1 mM sodium azide included.

9. Prepare for SDS-PAGE analysis by boiling for 5 min, vortexing, and microcentrifuging briefly to pellet beads. Use a Hamilton syringe to load eluates onto an SDS-polyacrylamide gel, arranging duplicate samples to allow preparation of duplicate blots. Separate by electrophoresis (UNIT 10.1). A Hamilton syringe works well to remove the eluate from the beads during loading.



Supplement 17

10. Immunoblot duplicate samples separately with antibodies for each of the two proteins (UNIT 10.10). Be sure to include aliquots of the whole-cell extract for comparison and as a positive control for the immunoblot. Each immunoblot can be reprobed with the antibody to the other protein. ALTERNATE PROTOCOL

COPRECIPITATING A GST FUSION PROTEIN GST fusion proteins may be coprecipitated by following the co-immunoprecipitation procedure (see Basic Protocol) with the modifications outlined below. This procedure might be used when the protein in question comigrates with immunoglobulin heavy or light chain in an SDS-PAGE gel or if the antibodies being used precipitate too many cross-reacting proteins, such as the protein being tested for association. Furthermore, it is possible to dissociate the purified GST fusion from the solid-state glutathione resin under gentle conditions through the use of imidazole. The approach is also useful in that it will increase the size of the protein sufficiently that this size increase can be used as a diagnostic feature in analyzing complexes. Additional Materials (also see Basic Protocol) Glutathione-agarose or glutathione-Sepharose slurry (see recipe) 1. Prepare duplicate samples as described for protein A/G–Sepharose (see Basic Protocol, step 1), omitting antibody. 2. Microcentrifuge 10 min at maximum speed, 4°C, to pellet nonspecific aggregates. Transfer supernatant to new microcentrifuge tube. 3. Add 30 µl glutathione-agarose or glutathione-Sepharose slurry (25- to 30-µl bead volume). Be sure to evenly suspend the slurry before distributing it to the samples. 4. Rotate the sample, pellet and wash glutathione-agarose/Sepharose, and perform SDS-PAGE and immunoblot analysis (see Basic Protocol, steps 5 to 10). REAGENTS AND SOLUTIONS Use deionized, distilled water in all recipes and protocol steps. For common stock solutions, see APPENDIX 2E; for suppliers, see SUPPLIERS APPENDIX.

Co-immunoprecipitation buffer 50 mM Tris⋅Cl, pH 7.5 (APPENDIX 2E) 15 mM EGTA 100 mM NaCl 0.1% (w/v) Triton X-100 Store at 4°C Immediately before use add: 1× protease inhibitor mix (see recipe) 1 mM dithiothreitol (DTT) 1 mM phenylmethylsulfonyl fluoride (PMSF; from fresh 250 mM solution in 95% ethanol) If necessary for activity of the protein of interest, a divalent cation may need to be included in the buffer and EGTA omitted (as for extraction buffer; see Strategic Planning, discussion of chelating agents). Detection of Protein-Protein Interactions by Coprecipitation

The protease inhibitor mix, PMSF, and DTT should be added fresh at the time of experimentation. The mixture without those components can be stored for months at 4°C with the addition of 1 mM sodium azide. PMSF is labile in aqueous buffer and should be added at the last minute.



Glutathione-agarose or glutathione-Sepharose slurry Swell 1.5 g glutathione-agarose or glutathione-Sepharose beads (e.g., Pierce, Sigma) in 30 ml of 50 mM Tris⋅Cl, pH 7.5 (APPENDIX 2E), for 1 to 2 hr on ice. Pellet beads by gravity or very gentle centrifugation (1 min at 1000 rpm in a tabletop centrifuge) and then wash four times with co-immunoprecipitation buffer (see recipe) that lacks protease inhibitor mix and contains 1 mM sodium azide. Resuspend beads in 15 ml of this buffer to yield a final slurry concentration of ∼100 mg/ml. Store at 4°C (stable for months). This recipe can be scaled up or down.

Protease inhibitor mix, 1000× Dissolve in DMSO: 5 mg/ml chymostatin 5 mg/ml pepstatin A 5 mg/ml leupeptin 5 mg/ml antipain Store in aliquots up to 1 year at −20°C Protein A/G–Sepharose slurry Swell 1.5 g protein A– or protein G–Sepharose beads (e.g., Pierce, Sigma) in 30 ml of 50 mM Tris⋅Cl, pH 7.5 (APPENDIX 2E), for 1 to 2 hr on ice. Pellet beads by gravity or very gentle centrifugation (1 min at 1000 rpm in a tabletop centrifuge) and then wash four times with co-immunoprecipitation buffer (see recipe) that lacks protease inhibitor mix and contains 1 mM sodium azide. Resuspend beads in 15 ml of this buffer to yield a final slurry concentration of ∼100 mg/ml. Store at 4°C (stable for months). This recipe can be scaled up or down.

COMMENTARY Background Information Coprecipitation is a powerful and simple approach to test for a physical interaction between proteins. There are many reasons to incorporate coprecipitation into a study. First, as a form of protein affinity chromatography, the method may be sensitive enough to detect weak associations that do not withstand the rigors of standard purification methods involving substantial dilution of the initial cell extract. Second, coprecipitation tests for associations between proteins within the milieu of a whole-cell extract, where the proteins are present at native concentration in a complex mixture of other cellular components. This feature makes it an important partner to two-hybrid methods and direct tests of interactions using purified proteins, because it provides a way to verify that a positive interaction reflects a true in vivo association. For example, nonphysiological interactions can be detected when purified proteins are present at too elevated a concentration. A falsely positive interaction between two proteins can also arise in a two-hybrid test when protein domains are inappropriately exposed

due to altered folding. In addition, not all proteins are amenable to two-hybrid analysis; a negative result may mask a true association. Nevertheless, a word of caution is in order. The ability to coprecipitate two proteins from a cellular extract is not proof that a particular interaction normally takes place in vivo. Additional experiments are needed to argue that a given interaction is not the result of mixing cell contents during extract preparation. Such evidence could include colocalization of the proteins or demonstration of functional relatedness. When performing coprecipitation, it is important to precipitate from both directions (i.e., individually precipitating protein 1 and protein 2, and testing for the presence of protein 2 and protein 1, respectively). This is important in that it can provide further verification of an interaction between the proteins. It is also important because it is possible the interaction will only be detected in one direction. An inability to detect an interaction in one direction could be due to a variety of factors including obstruction of an interaction by the binding of



Supplement 17

the antibody or other affinity agent, or differences in pool size representation of each protein. For example, protein 1 may bind to many proteins besides protein 2, while most of protein 2 binds to protein 1. In this scenario, it would be anticipated that detection of their association will be most efficient when protein 2 is precipitated.



It is important to vary conditions of both the extract preparation and the coprecipitation to determine what is optimal. When starting from scratch, it is most prudent to use a range of lysis and precipitation conditions from less to more stringent in terms of the amount of salt and nonionic detergent. When no interaction is detected, it is worthwhile to use less stringent conditions (reduced salt with little or no nonionic detergent). In addition, it may be necessary to avoid any dilution of the whole-cell extract. This can be done by adding the protein A/G–Sepharose directly to the extract after an initial clarification centrifugation and by using smaller wash volumes. Depending on the strength and nature of the interaction, the precipitation can be done in the presence of a mixture of detergents that includes ionic detergents (for example, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS), similar to that described in RIPA buffer (UNIT 3.8). In addition, it may be necessary to increase the expression levels of the proteins in question to be able to readily detect them by coprecipitation. A range of expression levels is recommended, because a level that is too high can lead to unregulated interactions (Feng et al., 1998). Alternatively, one can scale up the coprecipitation and use more than 0.5 to 1 mg of whole-cell extract (Feng et al., 1998). Here, the limiting factor is the concentration of the extracts, which must be high enough to allow the volume of the coprecipitation mixture to remain low. Larger-scale extract preparations may be necessary to generate more concentrated extracts. Finally, in cases of failure due to low abundance of the proteins in the host organism, one can overexpress a tagged version of one of the two proteins in the same or another host (such as E. coli), concentrate this protein by preimmobilization on an appropriate affinity matrix, and then incubate the affixed protein with extracts from the host organism. The most important objective in these experiments is to generate as great a signal-tonoise ratio as possible and avoid problems of

background. A variety of parameters can be changed to enhance the co-immunoprecipitation. Optimization of the precipitating antibody is one possibility. Protein A–Sepharose and protein G–Sepharose should give results comparable to anti-Ig serum. However, direct coupling of the antibody to Sepharose may lead to lower background and more quantitative precipitation. In addition, varying the ratio of antibody to whole-cell extract and the total amount of whole-cell extract is strongly suggested to determine the optimal amount of antibody that gives the most precipitation with the least amount of background. Affinity purification of the antibody may be necessary if the antibody immunoprecipitates additional crossreacting proteins. Additional approaches can be taken to minimize background. First, better clarification of the cell extract can be achieved by precentrifugation at 100,000 × g. These extracts can be directly used for coprecipitation without an intervening freezing step, which can increase the amount of protein precipitation. Second, both the lysis buffer and the coprecipitation buffer can be supplemented with 1% BSA to reduce nonspecific binding to the affinity matrix. Third, the whole-cell extract can be preincubated with protein A/G–Sepharose to remove nonspecific proteins that bind to the solid support. Fourth, the amount of salt and detergent can be increased in both the coprecipitation and the washes to reduce nonspecific binding. Fifth, increasing the number of washes may also help, although it may reduce the amount of specific protein that remains associated. Sixth, the expression levels of the proteins in question can be increased to generate a stronger signal that is above background binding. Alternatively, it may be possible to produce a whole-cell extract that is enriched for the proteins in question (e.g., by preparing a nuclear extract if the proteins are known to be in the nucleus). In instances where one of the proteins binds nonspecifically to Sepharose, the substitution of an agarose-based affinity matrix may help solve the problem. Finally, it may be necessary to generate a different set of reagents to precipitate the proteins in question (i.e., different antibodies and/or protein tag).

Anticipated Results Provided suitable antibodies are available to the proteins in question and the physical interaction is stable under the coprecipitation conditions, it should be possible to detect an interaction between two proteins.



Time Considerations Once the extracts are prepared, coprecipitation can be done within 3 to 4 hr, yielding samples ready to load on a gel for SDS-PAGE and immunoblot analysis.

Literature Cited BioSupplyNet Source Book. 1999. BioSupplyNet, Plainview, N.Y., and Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. Coligan, J.E., Kruisbeck, A.M., Margulies, D.H., Shevach, E.H., and Strober, W. 1999. Current Protocols in Immunology. John Wiley & Sons, New York. Feng, Y., Song, L.-Y., Kincaid, E., Mahanty, S.K., and Elion, E.A. 1998. Functional binding between Gβ and the LIM domain of Ste5 is required to activate the MEKK Ste11. Curr. Biol. 8:267-278. Field, J., Nikawa, J., Broek, D., MacDonald, B., Rodgers, L., Wilson, I.A., Lerner, R.A., and Wigler, M. 1988. Purification of RAS-responsive adenylyl cyclase complex from Saccharomyces cerevisiae by use of an epitope addition method. Mol. Cell. Biol. 8:2159-2165. Harlow, E. and Lane, D. 1988. Antibodies: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

Kolodziej, P.A. and Young, R.A. 1991. Epitope tagging and protein surveillance. Methods Enzymol. 194:508-519. Phizicky, E.M. and Fields, S. 1995. Protein-protein interactions: Methods for detection and analysis. Microbiol. Rev. 59:94-123. Tyers, M., Tokiwa, G., and Futcher, A.B. 1993. Comparison of the Saccharomyces cerevisiae G1 cyclins: Cln3 may be an upstream activator of Cln1, Cln2, and other cyclins. EMBO J. 11:17731784.

Key References BioSupplyNet Source Book. 1999. See above. Published yearly. Instant access is available on the WWW, at http://www.biosupplynet.com; hard copy may be requested by fax at (609) 786-4415. Information on its contents may be obtained by telephone at (516) 349-5595, fax at (516) 349-5598, or email at [email protected]. Phizicky and Fields, 1995. See above. General discussion of methodologies for detecting protein-protein interactions as well as their merits and drawbacks.

Contributed by Elaine A. Elion Harvard Medical School Boston, Massachusetts



Supplement 17

Imaging Protein-Protein Interactions by Fluorescence Resonance Energy Transfer (FRET) Microscopy

UNIT 19.5

Specific protein-protein interactions, whether induced by covalent protein modifications or not, are generally considered to mediate cellular signaling and function. Detection of these processes has long been restricted to bulk biochemical methods such as immunoprecipitation (UNIT 9.8) and immunoblotting (UNIT 10.10). These approaches have proven invaluable, e.g., in uncovering the major signal transduction pathways by delineating the hierarchy of protein-protein interactions and kinase-substrate relationships in the different phosphorylation cascades. However, immunoblotting techniques lack spatial information and the interactions that are detected depend on the stability of the complex during the experimental conditions that exist outside the cell in homogenates. Detection of proteins using immunofluorescence provides spatial information on the micrometer scale, and it is therefore not possible to infer protein-protein interactions. Limited information on the phosphorylation status of proteins can be obtained by the application of phospho-specific antibodies against specific phosphorylated residues in a given protein (see UNIT 13.4). The use of these antibodies is, however, restricted by availability and specificity. Only a small number of antibodies against phosphoproteins exist and phosphotyrosine residues have proven to produce the most specific antibodies thus far, leaving a large number of modifications undetectable by microscopy. This unit describes the preparation and execution of a typical fluorescence resonance energy transfer (FRET) experiment. The following procedures will specifically refer to the use of a GFP-tagged protein as donor and Cy3-labeled antibodies as acceptor. As an example, covalent modification by tyrosine phosphorylation of the epidermal growth factor receptor (EGFR) is determined by FRET between C-terminal GFP-tagged EGFR and Cy3-labeled anti-phosphotyrosine antibodies (Wouters and Bastiaens, 1999). FRET is measured by release of donor quenching through acceptor photobleaching. An advantage of this method is the convenient setup; it does not require specialized equipment but can be performed using common microscopy equipment, preferably a confocal laser scanning microscope (Bastiaens and Jovin, 1996; Bastiaens et al., 1996; Wouters et al., 1998). The method is quantitative, and by simple relationships, a FRET efficiency is obtained. The sample preparation described in this unit is identical for other FRET microscopy techniques or other donor-acceptor pairs. As an illustration, the result of a more advanced FRET measurement using fluorescence lifetime imaging microscopy (FLIM), which is based on the change in GFP fluorescence lifetime (see Background Information), is also included (Bastiaens and Squire, 1999; Ng et al., 1999; Wouters and Bastiaens, 1999). This unit describes FRET microscopy based on release of quenched donor fluorescence after acceptor photobleaching (see Basic Protocol), microinjection of reagents into the nucleus or cytosol (see Support Protocol 1), and labeling antibodies with Cy3 (see Support Protocol 2). FRET MICROSCOPY OF FIXED CELLS A number of quantitative FRET microscopy techniques are available at this time. A technique based on the release of quenched donor fluorescence after acceptor photobleaching will be described in this unit. This technique requires a pixel-by-pixel reference image to be created by photobleaching of the acceptor through continuous illumination at the absorption maximum of the acceptor. The time scale of the phoContributed by Fred S. Wouters and Philippe I.H. Bastiaens Current Protocols in Protein Science (2001) 19.5.1-19.5.15 Copyright © 2001 by John Wiley & Sons, Inc.

BASIC PROTOCOL



tobleaching process is on the order of minutes. In a lifetime-based FRET approach, the photobleaching step is used at the end of a time-lapse sequence of measurements to introduce an intracellular reference in live cells. In the donor intensity approach (unquenching), described here in detail, the time scale of photobleaching restricts the technique to fixed cells but has the advantages of being relatively simple and quantitative. Materials Cells of interest Plasmid for GFP-tagged protein Transfection reagent (e.g., Fugene 5 from Boehringer Mannheim, Lipofectin from Life Technologies, Effectene from Qiagen, or Superfect from Qiagen) Serum-free medium Low-background fluorescence CO2-independent medium (Life Technologies, or see recipe) Phosphate-buffered saline (PBS; see recipe), pH 7.4 4% (w/v) formaldehyde fixative solution (see recipe) Quench solution: 50 mM Tris⋅Cl (pH 8.0)/100 mM NaCl 0.1% (v/v) Triton X-100 in PBS Antibody (e.g., PY72 monoclonal anti-phosphotyrosine antibody) labeled with Cy3 (see Support Protocol 2) 1% (w/v) bovine serum albumin (BSA, fraction V) in PBS Mowiol mounting medium (see recipe) 6- and 12-well tissue culture plates Coverslips Microscope slides Confocal laser scanning microscope (e.g., Zeiss LSM 510), equipped with argon (488 nm) and He/Ne (543 nm) lasers selected by the HFT 488/543 double dichroic filter, GFP fluorescence selected by the NFT 545 dichroic and BP 505-530 emission filter, and Cy3 fluorescence selected by the LP560 emission filter Imaging software package (e.g., NIH-image or IPLab Spectrum from Scanalytics) Additional reagents and equipment for transfection of mammalian cells (APPENDIX 3C) NOTE: All solutions and equipment coming into contact with cells must be sterile, and aseptic technique should be used accordingly. NOTE: All incubations are performed in a humidified 37°C, 5% CO2 incubator unless otherwise specified. Some media (e.g., DMEM) may require altered levels of CO2 to maintain pH 7.4. Prepare cells 1. Seed the cells onto culture dishes containing coverslips. For live-cell FRET experiments, seed the cells onto glass-bottom MatTek culture dishes. The cells should be adherent and grow in a monolayer.

2. Transfect cells with the plasmid for GFP-tagged protein of interest using calcium phosphate precipitation, lipofection, or the activated dendrimer reagent, Superfect.

Imaging Protein-Protein Interactions by FRET Microscopy

For cells that are difficult to transfect, due to resistance to transfection (e.g., primary cell lines), or due to cytotoxicity of the transfection reagent, the DNA for the GFP-tagged protein can be introduced by nuclear microinjection (see Support Protocol 1). The latter method is also particularly useful when the expression levels of the protein of interest have to be controlled accurately.



3. Grow the cells, typically 1 to 2 days post transfection, until expression levels of the GFP-tagged protein are high enough to detect with fluorescence microscopy. When needed, starve the cells overnight with serum-free medium to make them semi-quiescent. For cells that undergo apoptosis upon serum deprivation, lower the serum concentration in the medium to 0.5%.

4. On the day of the experiment, transfer the coverslips to a 12-well tissue culture plate. Subject the cells to the desired experimental conditions (e.g., growth factor/hormone stimulation, or incubation with drugs or inhibitors).

Fix cells 5. Wash the cells with ice-cold PBS, aspirate, and add 1 ml 4% formaldehyde fixative solution. Allow cells to fix at room temperature for 10 min. The common alternative fixation protocol, where cells are incubated in −20°C methanol for 5 min, is not advised. Strictly speaking, this treatment does not fix the tissue, but precipitates the cellular proteins. Protein-protein interactions are therefore likely to be affected.

6. Quench excess fixative with quench solution by briefly washing the cells, then changing to fresh quench solution and incubating 5 min at room temperature. In a first short wash, the excess fixative is removed. In a second 5-min incubation, the remaining aldehyde groups are allowed to react with the primary amino group on the Tris molecule. Alternative, equally effective, quench solutions are 0.1 M hydroxylamine or 0.1 M glycine.

7. Permeabilize cellular membranes by incubating 5 min with 0.1% Triton X-100 to allow penetration of acceptor-labeled molecules into fixed cells. In this treatment, a compromise is made between optimal morphology and permeabilization. A milder detergent treatment with 0.1% saponin in PBS or a harsher permeabilization with −20°C methanol for 5 min can be considered if morphology or permeabilization, respectively, is compromised.

8. Wash the cells with PBS to remove the permeabilization solution. Add antibody 9. Dilute Cy3-labeled antibodies appropriately in 1% BSA/PBS. Incubate the coverslips with appropriately diluted Cy3-labeled antibodies for 1 hr, at room temperature. Typical concentrations are 0.1 to 10 ìg/ml, but it is advised that a titration series be performed to determine the antibody concentration where epitope binding saturates. To minimize the amount of antibody needed, press a sheet of Parafilm to the bench with some water, place 25-ìl drops of antibody solution on the Parafilm, place the coverslips cells-down on these drops, and incubate 1 hr. To facilitate handling of the coverslips at the end of the 1-hr incubation, place a pipet to the edge of a coverslip and add ∼100 ìl PBS underneath it. This will lift the coverslip so that it can be easily picked up with jeweler’s forceps.

10. Transfer the coverslips to 6-well tissue culture plates and wash four times, each time with 3 ml PBS, to remove excess antibody. 11. Blot off excess PBS with tissue and mount the coverslips on slides with ∼10 µl Mowiol mounting solution. Allow Mowiol to harden overnight at 4°C before imaging. Identification of Protein Interactions


Supplement 23

If necessary, cells can be imaged after a short drying period of 1 hr if the coverslip is attached to the slide by painting the edge of the coverslip with molten agarose or rubber cement at four points. Do not use nail polish since this has been shown to quench GFP fluorescence.

Examine cells and calculate results 12. View the specimen on a confocal microscope using a 63× or 100× oil-immersion objective. Acquire an image in the GFP channel (excitation, 488 nm; emission, NFT 545, BP505-530). Either take an image of the entire field of vision or select a region of interest containing the cell that is to be imaged. GFP fluorescence is selected by the NFT 545 dichroic and BP 505-530 emission filter. Do not use the full dynamic range of the detector, since unquenching will result in added fluorescence. Depending on the make of the confocal microscope, make sure that a second image can be made at exactly the same location. Do not adjust the settings (i.e., pinhole size, contrast, brightness, laser power, or averaging) for the GFP channel. This acquisition provides the FRET-quenched donor image (FDA), since the acceptor is present throughout the cell, causing FRET with the GFP donor. Minimize GFP photobleaching at this point by limiting the illumination.

13. Change to the Cy3 channel (excitation, 543 nm; emission, LP560) and take an image of the cell, minimizing Cy3 photobleaching. Select a portion of the cell in which the FRET efficiency is to be determined. Cy3 fluorescence is selected by the LP560 emission filter. This region is where the acceptor will be photobleached, thereby revealing the unquenched donor intensities enabling FRET calculation. Consequently, any portion of the cell where the acceptor is not photobleached serves as control [i.e., 1 − FDA/FDA = 0]. This area in the same cell will provide an essential control to judge the effects of photobleaching GFP, lateral movement, or focal mismatch between the two consecutive donor images.

14. Photobleach the selected acceptor region by repeated scanning with the 543-nm He/Ne laser line at full power. Follow the progress of photobleaching by monitoring the intensities of the respective images (emission: NFT 545, BP505-530). Continue until there is no more discernible Cy3 intensity. Depending on the staining intensity and the area that is bleached, this will typically take 1 to 20 min. When selecting the photobleaching region on the Zeiss LSM 510 confocal microscope, crop the acceptor image to this region (and note the zoom settings to return to the original image after bleaching) instead of selecting a region of interest (ROI). In the latter option, the entire area will be scanned but the laser will only switch on in the ROI. Consequently, the duty cycle is low and bleaching will take considerably longer.

15. Return to the GFP channel and make the second acquisition, using identical settings and location of the prebleached image. This provides the FD reference in the region where the acceptor was photobleached.


16. To calculate the FRET efficiency in the bleached area, use an appropriate imaging software package. Subtract the prebleach donor image from the post-bleach donor image (i.e., FD − FDA). Divide this image by the postbleach donor image: (FD − FDA)/FD; this is identical to 1 − (FDA/FD) = E, the image of FRET efficiencies. A software package capable of performing the abovementioned image processing is NIH Image. This package is freely available from http://rsb.info.nih.gov/nih-image/ and ver-



sions are available for Apple Macintosh, Windows (Scion image) and even a Java version is available that runs on any platform (Image J). A more versatile commercial package is IPLab Spectrum. Image arithmetic for calculating the FRET efficiency should be performed on a region of interest obtained by thresholding the images in order to prevent the amplification of background noise. Therefore, thresholds should be chosen in such a way that the background is omitted in the calculation. Accuracy can be improved by subtracting the average background intensity from the images before calculating the FRET efficiency.

NUCLEAR AND CYTOSOLIC MICROINJECTION The following is a generic protocol for introduction of DNA or labeled proteins into a cell. DNA is microinjected directly into the nucleus to achieve controlled and high expression of the protein of interest. Labeled proteins are injected into the cytosolic region next to the nucleus since the cell is at its highest here, facilitating injection.

SUPPORT PROTOCOL 1

Materials Cells of interest, cultured in MatTek glass-bottom 35-mm dishes (MatTek) DNA (e.g., human EGFR cDNA in the Clontech pEGFP-N1 expression vector; APPENDICES 4C & 4D) HPLC-grade water Millex-GV4 0.22-µm filtration unit GELloader tips (Eppendorf) Needles for microinjection (e.g., Femtotip from Eppendorf) Microinjector (e.g., Eppendorf model 5244) Micromanipulator (e.g., Eppendorf model 5170) Inverted microscope with 10× and 40× air objectives Additional reagents and equipment for preparation of DNA (APPENDIX 4C) 1. Prepare DNA, to be microinjected, of the highest possible quality. In the authors’ experience, double cesium chloride–banded DNA (Wilson, 1994) and DNA purified by Qiagen ion-exchange resin (Qiaprep midi/maxiprep columns; Budelier and Schorr, 1998) perform equally well. Cy3-labeled protein (e.g., antibody against protein of interest; see Support Protocol 2) may also be microinjected.

2a. For optimal expression a few hours after microinjection: Dilute DNA in HPLC-grade water to 100 µg/ml. 2b. For expression overnight: Dilute DNA in HPLC-grade water to 1 µg/ml. 3. Clear the DNA solution to prevent blocking of the glass needle during microinjection as follows. Place a 0.22 µm Millex filtration unit in a 0.5-ml microcentrifuge tube and place the entire unit in a 1.5-ml microcentrifugation tube to enable centrifugation. Filter 10 µl of the DNA solution by microcentrifugating 1 min at maximum speed, room temperature. Since these membranes have low-protein-binding characteristics, they can also be used for clearing Cy3-labeled proteins (see Support Protocol 2). The recovery of Cy3-labeled antibodies when cleared this way is also generally very high. Alternatively, microcentrifuge the Cy3-labeled protein for 20 min at maximum speed. Sacrifice a small amount of solution to prevent disturbing the pellet of aggregated protein. Identification of Protein Interactions


Supplement 23

4. Load 2 µl of DNA solution (or Cy3-protein solution) using GELloader tips into the capillary glass needle of the microinjector. Commercially available needles (Femtotip from Eppendorf) can be used. These needles fit directly into the needle holder of the Eppendorf microinjection device. Carefully remove the pipet tip that protects the needle. In the authors’ experience, this is most easily performed by holding the needle pointing downwards and loosening the tip by rotation until it falls to the ground. If access is available to a needle-pulling device, make sure that the diameter of the needle opening is ∼0.25 ìm. The diameter of the needle opening can be estimated using a simple syringe-operated micropipet bubble meter (Clark Electronic Instruments) by measuring the air pressure required to expel air bubbles from the pipet into a liquid (Mittman et al., 1987).

5. On an inverted microscope, microinject the DNA solution (or Cy3-protein solution) into the nucleus (or perinuclear cytosol) of cells grown in MatTek culture dishes. Typical settings are: 0.3 sec, 150 to 400 hPa injection pressure with 20-hPa back-pressure to prevent medium from entering the needle. The injection pressure may be varied according to the needle opening and cell type. No major movement of the nucleus (or cell organelles) should be observed. A visual indication for excessive pressure is the separation of the nucleus from the surrounding cellular material (i.e., light ring around the nucleus) and leakage into the cytosol, visible by movement of the cellular organelles. Restrict the microinjection procedure to a maximum of 10 min. In the authors’ laboratory, microinjection is performed at room temperature in normal CO2-dependent medium. After 10 min, the medium starts to acidify significantly (i.e., purple medium). CO2-independent (or HEPES-buffered) media can be used for longer periods. SUPPORT PROTOCOL 2


PROTEIN LABELING WITH Cy3 Proteins are labeled on unprotonated free amino groups (i.e., α-amino terminus or ε-amino groups on lysine side chains) by the succinimide esters of the fluorescent sulfoindocyanine (Cy) dyes (also see Haugland, 2000). Alkaline labeling conditions ensure deprotonation of amino groups. Cy-dyes are water soluble and have high extinction coefficients, making them particularly useful for sensitive detection of proteins with minimal disturbance of protein function. The following protocol describes the labeling of antibodies with Cy3, a suitable donor for GFP in a FRET experiment, and also explains how to remove stabilizing compounds (e.g., gelatin, BSA), which are often added to prolong shelf life, as these contain amino groups that would compete with the labeling reaction. Materials Antibody (PY72 monoclonal anti-phosphotyrosine antibody) 1 M Tris⋅Cl, pH 8.0 (APPENDIX 2E) 10 mM and 100 mM Bicine/NaOH, pH 8.0 100 ml citric acid/NaOH, pH 2.8 1 M Bicine/NaOH, pH 9.0 1 M NaCl (APPENDIX 2E) Labeling buffer: 100 mM Bicine/NaOH (pH 8.0)/100 mM NaCl Cy3.29–OSu monofunctional sulfoindocyanine succinimide ester (Amersham Pharmacia Biotech) Dimethylformamide (DMF) dried by addition of 10 to 20 mesh 3-Å pore diameter molecular sieve dehydrate (Fluka) 1-ml Protein G HiTrap columns (Amersham Pharmacia Biotech) Centricon YM30 concentrators (Amicon)



Biogel P6DG Econopac prepacked size-exclusion columns (5.5 × 1.5–cm, ∼10 ml; Bio-Rad) 1-ml and 10-ml syringes with HPLC Luer-Lok fitted tubing Additional reagents and equipment for spectrophotometric protein determination (UNIT 3.1) and SDS-PAGE (UNIT 10.1) Prepare antibody solution 1. Resuspend antibody to 1 mg/ml in PBS. Excess stabilizing agents containing free amino groups (e.g., BSA and gelatin) in commercially available antibody preparations compete for the labeling reagent and have to be removed by Protein A or Protein G affinity chromatography (UNIT 9.8). If the protein solution to be labeled is free of additional amino groups, proceed directly to step 10. A number of suppliers of antibodies, e.g., Transduction Laboratories and New England Biolabs, can provide their products free of these compounds. Request that the antibodies be provided at 1 mg/ml concentration in PBS.

2. Prepare a syringe-operated 1-ml protein G HiTrap column. Equilibrate column with 10 ml PBS at a maximum flow of 4 ml/min. All fluid handling is performed manually using appropriately sized syringes. These columns are easy to use and result in minimal loss of protein. A number of subclasses of IgG molecules do not bind to protein A. When using protein G or A chromatography, make sure that the antibodies are compatible (see Bonifacino and Dell’Angelica, 1998).

3. Add 0.1 vol 1 M Tris⋅Cl, pH 8.0, to the antibody solution. Commercial antibody solution is typically 0.5 ml at 0.1 mg/ml IgG.

4. Load antibody solution onto the column and wash column with 10 ml of 100 mM Bicine/NaOH, pH 8.0. Collect run-through. 5. Wash column with 10 ml of 10 mM Bicine/NaOH, pH 8.0. Collect run-through. 6. Elute column with 5 ml of 100 mM citric acid/NaOH, pH 2.8, and collect 0.5-ml fractions (i.e., 8 drops) in 1.5-ml microcentrifuge tubes containing 100 µl of 1 M Bicine/NaOH, pH 9.0, to neutralize the pH. Mix immediately and store on ice. 7. Determine the A280 of the eluted fractions using a spectrophotometer (UNIT 3.1) and pool the fractions that contain protein. Under the given conditions, the protein will typically elute in the first four fractions.

8. Add 0.1 vol of 1 M NaCl. Concentrate the solution in a YM30 Centricon to ∼200 µl by centrifuging at 5000 × g, 4°C. 9. Redilute to 2 ml with labeling buffer and repeat the concentration and redilution steps (steps 8 and 9). 10. Concentrate to ∼50 to 100 µl as in step 8 and collect the concentrated protein solution. When labeling proteins from solutions containing relatively high concentrations (i.e., millimolar) of compounds that contain free amino groups (e.g., Tris, glycine, glutathione), repeat the concentration-redilution cycle more often. Each cycle dilutes the compound ∼10 fold. Allow a maximum of 10% of contaminating primary amino groups, compared to the protein to be labeled. Identification of Protein Interactions


Supplement 23

For example, the protein concentration in a 0.5 mg/ml antibody solution is 3.3 ìM, assuming a molecular mass of 150 kDa. If this solution contains 50 mM Tris buffer (a 1.5 × 104 fold excess of free amino groups), a 1.5 × 105-fold dilution, corresponding to 5.2 concentration cycles, is needed to reach the 10% contamination level. When labeling proteins other than antibodies, a Bicine concentration of 50 mM is recommended in the labeling buffer. Additional compounds that are needed to maintain the function of the protein to be labeled should be included. At this point, no compounds containing free amino groups should be added. High concentrations of reducing agents are also known to inhibit the labeling reaction. It is recommended that these be included in the chromatography step following labeling to prevent these compounds from interfering with the labeling reaction. Choose the cut-off value of the Centricon carefully to ensure retention of the protein of interest.

Prepare dye 11. Reconstitute Cy3.29–OSu in 20 µl dry DMF to give a ∼10 mM Cy3 solution. Determine the exact concentration by measuring the absorption of a 104-fold diluted solution in PBS. From the ε550 of 150 mM−1cm−1, calculate the concentration. Cy3.29–OSu is supplied as a desiccated pellet in microcentrifuge tubes. DMF is dried by addition of hygroscopic beads to the container. The ε650 of Cy5 is 250 mM−1cm−1.

Perform labeling reaction 12. Determine the protein concentration of the antibody (or protein) solution to be labeled based on A280 reading (UNIT 3.1). Antibody concentration at A280 = 1.0 is typically 1 mg/ml. At low protein concentrations (i.e., 50 kDa) and correct for the 5% absorption of Cy3 at 280 nm. For smaller proteins or lower protein concentrations (i.e., 60°C) until the paraformaldehyde is dissolved. Add 10 ml of 10× PBS (see recipe) and allow to cool to room temperature. Adjust the pH to 7.4 using 1 M HCl (∼1 ml). Adjust to 100 ml with water and filter through a Millipore 0.45-µM filter using a syringe to remove traces of undissolved paraformaldehyde. Store up to several months at −20°C. Low-background-fluorescence CO2-independent medium Adjust the formulation of the standard medium by omitting the pH indicator phenol red, the antibiotics penicillin and streptomycin, folic acid, and riboflavin. Before use, supplement the medium with 50 mM HEPES/NaOH, pH 7.4. Store 1 to 2 months at 4°C.



Supplement 23

Mowiol mounting medium Mix 6 ml glycerol, 2.4 g Mowiol 4-88 (Calbiochem), and 6 ml water. Shake for 2 hr. Add 12 ml of 200 mM Tris⋅Cl, pH 8.5 (APPENDIX 2E), and incubate at 50°C with occasional mixing until the Mowiol dissolves (i.e., ∼3 hr). Filter through 0.45-µM Millipore filtration unit and store in aliquots up to several weeks at 4°C or up to several months at −20°C. Phosphate-buffered saline, 10× 68 g NaCl 18.8 g Na2HPO4 2 g KH2PO4 H2O to 1 liter COMMENTARY Background Information


Fluorescence resonance energy transfer (FRET) is a photophysical process that can be exploited to obtain information on protein-protein interactions and protein modification in addition to location (Clegg, 1996). Its usefulness lies in the ability to sense the presence of acceptor fluorophores in the extreme vicinity of a donor fluorophore with a maximum separation distance that is in the order of magnitude of single protein molecules. FRET is a radiationless process whereby an excited donor fluorophore transfers energy to an acceptor by dipole-dipole coupling (Förster, 1948). Due to the 6th-order distance dependence of the FRET efficiency, two discrete states of the fluorophores can be discriminated: exhibition of efficient FRET when donor and acceptor are in close proximity or no occurrence of FRET due to distance. The transfer of energy (FRET) from an excited donor fluorophore to a nearby acceptor fluorophore has a number of consequences with respect to the fluorescent properties of both fluorophores and this can be exploited to measure the efficiency of this process. In the methods presented in this unit, FRET is determined by measuring the donor fluorescence emission exclusively (Bastiaens and Squire, 1999). The acceptor fluorophore is excluded from the measurement by the choice of optical filters. As a consequence, a large excess of acceptor can be used and, when using antibodies, specificity can be sacrificed to gain higher occupancy. In photophysical terms, FRET provides an extra channel of nonradiative decay by which the excited state of the donor fluorophore is depopulated. This results in a reduced quantum yield (Q), the ratio of emitted photons over

absorbed photons. The reduction in quantum yield can be determined in two ways: (1) by the decrease in steady-state fluorescence emission; or (2) the decrease in fluorescence lifetime (τ), which characterizes the duration of the excited state of the fluorophore. The measurement of the change in quantum yield by steady-state emission has to be calibrated due to its dependence on concentration and light path. In contrast, the fluorescence lifetime is proportional to Q and is independent of concentration and light path. Fluorescence lifetimes are measured by fluorescence lifetime imaging microscopy (FLIM; Lakowicz and Berndt, 1991; Gadella et al., 1993; Gadella and Jovin, 1995) and can be performed sufficiently fast to enable realtime live cell experiments. FLIM requires a specialized microscopy setup enabling highfrequency modulation of the excitation light and the gain on the detector. This technique can be combined with acceptor photobleaching to provide an internal lifetime reference, e.g., at the end of a live-cell time-lapse sequence. A detailed description of a frequency domain FLIM set-up is given by Squire and Bastiaens (1999). This type of imaging system is gradually being made commercially available by companies such as LaVision or Lambert Instruments (see SUPPLIERS APPENDIX). Calibration in the intensity-based FRET method is achieved by photobleaching the acceptor to provide the unquenched donor (Bastiaens et al., 1996; Bastiaens and Jovin, 1998; Wouters et al., 1998). In this method, an exposure of the donor fluorescence emission is made (FDA, fluorescence emission of donor in presence of the acceptor). Removal of the acceptor fluorophore from the sample by photobleaching enables the acquisition of an unquenched donor emission image when the ex-



posure is taken with settings identical to those used for the original image (FD, fluorescence emission of the donor in absence of the acceptor). This essentially recovers that part of the fluorescence intensity that is lost to FRET. A pixel-by-pixel FRET efficiency map is obtained by simple image arithmetic: E = 1 − (FDA/FD). The specificity of the acceptor photobleaching techniques lies in the steep edge at the long-wavelength (i.e., red edge) of the absorption spectrum of the donor, enabling exclusive photobleaching of the acceptor. Essential to the success of this technique is that the photoproduct of the acceptor does not exhibit residual absorption and does not fluoresce at donor emission wavelengths. Cy3 meets both criteria to act as a proper acceptor for GFP; the same is the case for Cy5 acting as acceptor for Cy3. Since photobleaching of the acceptor occurs on a minute time scale, this type of FRET determination is restricted to fixed cells. FRET microscopy can be used to detect protein-protein interactions in single cells as if one were performing an immunoprecipitation experiment. There is, however, no need to isolate the complex or remove it from its physiological environment prior to investigation, and the experiment can be performed in living cells. A number of approaches can be followed: (1) the donor molecule can be purified, labeled with Cy3, and introduced into cells by microinjection (Bastiaens and Jovin, 1996); (2) the donor molecule can be detected by a Cy3-labeled antibody or Fab fragment (provided the antibody is highly specific) on fixed cells (Ng et al., 1999); (3) the donor protein can be fused to one of the mutants of the intrinsically fluorescent green fluorescent protein (GFP) from the jellyfish Aequoria victoria (Tsien, 1998), genetically encoded, and expressed; (4) acceptor proteins (mostly antibodies or Fab fragments) can be labeled with Cy5 (for Cy3 donors), or labeled with Cy3 (for GFP donors) and introduced by microinjection or incubation in the case of antibodies; and (5) the recently discovered and commercial available red fluorescent protein from the Anthozoa sp. (Matz et al., 1999) can be employed. The latter opens up a wide array of possibilities since an ideal acceptor for GFP can be coexpressed in the same cell, obviating the need for exogenous labeling and microinjection. Alternative methods for measuring FRET on the basis of fluorescence intensities are available. In these methods FRET is not determined from the donor fluorescence exclusively, but also from the sensitized emission of the

acceptor, excited by receiving energy from the donor. These approaches fall into two categories (1) those based on changes in donor/acceptor emission ratio (Adams et al., 1991; Miyawaki et al., 1997); and (2) those based on acceptor sensitized emission (Day, 1998; Gordon et al., 1998; Mahajan et al., 1998). These approaches are attractively simple to perform but may suffer from a number of drawbacks that cannot always be corrected for. In the ratiometric approach, the decrease in donor emission and concomitant increase in acceptor emission are imaged by calculating the ratio of these intensities at each pixel. Since this is a relative measure, a change in FRET can only be observed in live cells where the ratio changes over time. A decrease in the donor/acceptor emission ratio can be taken as an indication for the occurrence of FRET. However, this ratio is also dependent on the local concentrations of the donor and acceptor molecules measured in each pixel, which can complicate the interpretation by differential translocation of the biomolecules conjugated to donor and/or acceptor fluorophores. This problem does not occur when the donor and acceptor fluorophores are present on the same molecule (e.g., the “chameleon” Ca2+ biosensors; Miyawaki et al., 1997), since the relative concentrations are identical at each location in the cell. Measuring FRET from the sensitized emission intensity alone (by using a “FRET filter set”) is the most widely used method. Emission from the acceptor would be an exclusive readout for FRET when contamination with donor fluorescence (i.e., donor bleed-through) and direct excitation of the acceptor did not complicate the approach. Correction for these effects by extensive control measurements, some of which have to be performed in different samples where the acceptor is absent, is described in Gordon et al. (1998). Here, both donor-quenching and acceptor-emission information are used to derive the sensitized emission contribution, which is proportional to the FRET efficiency. However, sensitized emission is also proportional to the concentration of the acceptor. Therefore, relative populations of associated molecules cannot be determined. Molecules that are not participating in FRET are not detected. In contrast to donor-based measurements, no estimation can be made of the bound/unbound fractions from the sensitized emission alone. Since excitation spectra generally exhibit an extensive tail at the lower wavelength (blue-edge) the acceptor can easily be excited at the wavelength used for excitation



Supplement 23

of the donor. Direct excitation will be especially problematic when the recently discovered red fluorescent protein from the Anthozoa sp. is used as an acceptor, since its absorption spectrum shows multiple absorption peaks at the excitation optima for the green and yellow fluorescent protein.



FRET measurements are prone to false negative results. While the finding of FRET by unquenched fluorescence is highly unlikely to be caused by artifactual processes or nonspecific proximity of donor and acceptor, the absence of FRET does not provide proof for the absence of an interaction for a number of reasons. The presence of a GFP moiety on the protein of interest might interfere with its function or targeting, thus affecting the interaction with the target protein. Cloning vectors for the construction of GFP fusion proteins routinely contain a multiple cloning site that has been optimized for a maximum number of restriction sites to facilitate cloning. As a consequence, the random amino acid residues that they translate into can adopt an unfortunate secondary structure that hinders the proper behavior of the fused protein. In the case of the EGFR-GFP, the amino acid residues encoded by the cloning site were replaced by a flexible six-glycine linker between the GFP moiety and the EGFR to prevent the GFP from affecting EGFR function (Wouters and Bastiaens, 1999). Another problem might arise when the separation distance between the GFP moiety on the donor protein and the Cy3 groups on the antibody is too large (i.e., >10 nm) for energy to be transferred efficiently. This might be overcome by using antibodies raised against a different epitope, or the entire protein, thus increasing the chance of a favorable orientation of the antibody to the donor fluorophore. The labeling ratio of the antibodies is a common reason for failure to detect FRET. The labeling ratio should exceed 1 to prevent unlabeled antibody from competing with the labeled antibodies. A higher labeling ratio is beneficial for FRET detection since the R0 distance (i.e., characteristic value for a given donor-acceptor pair, the distance at which 50% of the excited state energy is transferred by FRET) is effectively increased by a larger number of acceptor molecules per donor molecule. In the authors’ experience, labeling ratios up to 5 do not significantly affect the specificity of the

antibody. However, this should be verified by using the Cy3-labeled antibody in an immunofluorescence experiment and judging the staining pattern as compared to unlabeled antibody. Remember that absolute specificity is not a requirement in the FRET assay, because only the donor photophysical properties are used. However, in extreme cases, the highly labeled antibody could become nonspecific to a point where significant amounts bind to the donor. It is essential that the donor images before and after acceptor photobleaching be taken under identical conditions. Any change in parameters that influence the collection efficiency of donor fluorescence (i.e., brightness, contrast, laser power, and averaging) will influence the calculation of the FRET efficiency. This will affect the entire image, and the control region in the E map will no longer be distributed around 0 but will be uniformly shifted. This also occurs with donor photobleaching. When the laser is used at high intensity, or when too many scans are made for averaging, a substantial amount of the GFP becomes photobleached. This will lead to a shift to negative E values in the control region. Furthermore, it is essential to verify that the sample has not moved and that the same focal plane is imaged in both donor images. Movement causes structures in the control region in the E map with positive values at the leading edge and negative values at trailing edges. Movement is most often caused by temperature differences between the slide and the objective but can also be caused by drift in the slide holder. To correct for translation of the image after acceptor photobleaching, the maximum in the correlation function between pre- and post-bleaching images gives the shift between the two images (Bastiaens and Jovin, 1998). The correlation function of the two images is obtained by Fourier transformation on both images, followed by multiplication of the conjugate of the Fourier transform of one image by the Fourier transform of the other, and performing an inverse Fourier transform on the resulting image. Another source of structured contrast in the control region in the E map can be caused by focal mismatch between the two images, originating from the apparent appearance (positive E) or disappearance (negative E) of structures. This problem can be prevented by observing the z position of the stage between exposures. A larger pinhole can reduce these problems by decreasing the depth of focus but at the cost of z resolution. When using pre- and post-z sections rather than single-focus plane images, the



Figure 19.5.1 (A) Cy3-PY72 photobleaching releases FRET-quenched EGFR-GFP emission. The histogram shows the distribution of calculated FRET efficiencies in the cell. (B) FRET measured by fluorescence lifetime imaging microscopy. The histogram shows the distribution of measured lifetime (nsec) and corresponding calculated FRET efficiencies prior and after photobleaching of the acceptor. This black and white facsimile of the figure is intended only as a placeholder; for full-color version of figure go to http://www.currentprotocols.com/colorfigures.

maximum in a three-dimensional correlation function can be used for registration of the two data sets and a correct three-dimensional E map can be obtained (Bastiaens et al., 1996).

Anticipated Results After photobleaching of the acceptor, an increase in donor emission intensity can be observed when substantial FRET is present.

Image arithmetic should produce FRET efficiencies that are closely distributed around zero in the portion of the cell that was not photobleached. Deviations from this condition indicate donor bleaching, lateral movement, or focal mismatch. With the occurrence of FRET, a separate efficiency distribution is expected in the photobleached area. The heterogeneity in this distribution contains information about



Supplement 28

variability in the formation of complexes of the proteins carrying the donor and acceptor fluorophores. Figure 19.5.1 shows an MCF-7 mammary carcinoma cell expressing EGFR-GFP that was stimulated with 100 ng/ml EGF for 5 min. This cell was incubated with Cy3-labeled anti-phosphotyrosine antibodies (e.g., 10 µg/ml PY72 monoclonal) and EGFR-GFP tyrosine phosphorylation was measured by FRET using the unquenching of GFP fluorescence by acceptor photobleaching (Fig. 19.5.1A) and FLIM (Fig. 19.5.1B). The EGFR-GFP distribution is shown in the D1 image. Most EGFR-GFP is localized at the plasma membrane in addition to internalized (punctate) EGFR-GFP. The corresponding antibody-staining A1 shows phosphorylation at the plasma membrane. A rectangular region was subjected to acceptor photobleaching (white box). The GFP fluorescence in this region increased after photobleaching as shown in the difference image (D2 − D1), demonstrating positive FRET efficiencies [(D2 − D1)/D2]. Highest FRET efficiencies can be observed in the plasma membrane corresponding to fully phosphorylated receptor at this location. The large peak in the energy-transfer efficiency histogram that is distributed around zero originates from the area outside the photobleached region and indicates proper image registration. The additional population, ranging from 15% to 35% efficiency, corresponds to the photobleached region. Figure 19.5.1B shows the analogous result of a fluorescence lifetime measurement. The steady-state fluorescence distribution of EGFR-GFP is shown in (D) and the corresponding lifetime map in (τ1). The anti-phosphotyrosine Cy3 immunofluorescence (A) is photobleached to obtain an intracellular reference lifetime in absence of acceptor (τ2). As can be seen in the fluorescencelif etim e h istog ram, these values are homogeneously distributed around an average of ∼2.0 nsec. The decrease in lifetime due to FRET is shown in the difference image (τ2 − τ1). The energy-transfer efficiency is given by the normalization of this difference to the reference EGFR-GFP lifetime [E = (τ2 − τ1)/τ2] and again shows highly phosphorylated receptor in the plasma membrane.


incubation of cells; 1/2 day of data acquisition; and 1/2 day for data analysis. Removal of gelatin/BSA from commercial antibody takes ∼3 hr. The most time-consuming step in the procedure for labeling antibodies is the buffer exchange using a Microcon concentration device. Depending on the amount of contaminating free amino groups in the original buffer, this can take between 2 and 5 hr. The labeling reaction and subsequent gel filtration chromatography takes ∼1 hr. There are a number of points where the procedure can be interrupted: (1) optimal expression after overnight incubation rather than a 3- to 4-hr incubation post microinjection can be achieved by lowering the concentration of DNA; (2) after fixation and permeabilization, the cells can be stored overnight in PBS at 4°C before antibody incubation; (3) after mounting, the cells can be viewed immediately rather than the following day, by application of rubber cement instead of Mowiol; and (4) the samples can be stored at −20°C for weeks without appreciable loss of antibody staining or FRET.

Literature Cited Adams, S.R., Harootunian, A.T., Buechler, J., Taylor S.S., and Tsien, R.Y. 1991. Fluorescence ratio imaging of cyclic AMP in single cells. Nature 349:694-697. Bastiaens, P.I.H. and Jovin, T.M. 1996. Microspectroscopic imaging tracks the intracellular processing of a signal-transduction protein: Fluorescent labeled protein kinase C beta I. Proc. Natl. Acad. Sci. U.S.A. 93:8407-8412. Bastiaens, P.I.H. and Jovin, T.M. 1998. Fluorescence resonance energy transfer microscopy. In Cell Biology a Laboratory Handbook, Vol. 3 (J.E. Celis, ed.), pp. 136-146. Academic Press, New York. Bastiaens, P.I.H. and Squire, A. 1999. Fluorescence lifetime imaging microscopy: Spatial resolution of biochemical processes in the cell. Trends Cell Biol. 9:48-52. Bastiaens, P.I.H., Majoul, I.V., Verveer, P.J., Soling, H.D., and Jovin, T.M. 1996. Imaging the intracellular trafficking and state of the AB(5) quaternary structure of cholera-toxin. EMBO J. 15:4246-4253.

Time Considerations

Bonifacino, J.S. and Dell’Angelica, E.S. 1998. Immunoprecipitation. In Current Protocols in Cell Biology (J.S. Bonifacino, M. Dasso, J.B. Hartfort, J. Lippincott-Schwartz, and K.M. Yamada, eds.), pp. 7.2.1-7.2.21. John Wiley & Sons, New York.

From cell seeding, the entire procedure can be performed in 3 to 4 days. This period includes 1 to 2 days for expression of the GFP construct after transfection or microinjection; 1/2 day for treatment, fixation, and antibody

Budelier, K. and Schorr, J. 1998. Purification of DNA by anion-exchange chromatography. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.), pp. 2.1.11-2.1.18. John Wiley & Sons, New York.



Clegg, R.M. 1996. Fluorescence resonance energy transfer spectroscopy and microscopy. In Fluorescence Imaging Spectroscopy and Microscopy (X.F. Wang and B. Herman eds.), pp. 179-251. John Wiley & Sons, New York. Day, R.N. 1998. Visualization of Pit-1 transcription factor interactions in the living cell nucleus by fluorescence resonance energy transfer microscopy. Mol. Endocrinol. 12:1410-1419. Förster, T. 1948. Zwischenmolekulare Energiewanderung und Fluoreszenz. Ann. Phys. 2:55-75. Gadella, T.W.J. and Jovin, T.M. 1995. Oligomerization of epidermal growth-factor receptors on A431 cells studied by time-resolved fluorescence imaging microscopy—a stereochemical model for tyrosine kinase receptor activation. J. Cell Biol. 129:1543-1558. Gadella, T.W.J., Jovin, T.M., and Clegg, R.M. 1993. Fluorescence lifetime imaging microscopy (FLIM)—spatial resolution of microstructures on the nanosecond time-scale. Biophys. Chem. 48:221-239. Gordon, G.W., Berry, G., Huan Liang, X., Levine, B., and Herman, B. 1998. Quantitative fluorescence resonance energy transfer measurements using fluorescence microscopy. Biophys. J. 74:2702-2713. Haugland, R.P. 2000. Antibody conjugates for cell biology. In Current Protocols in Cell Biology (J.S. Bonifacino, M. Dasso, J.B. Hartford, J. Lippincott-Schwartz, and K.M. Yamada, eds.), pp. 16.5.1-16.5.22. John Wiley & Sons, New York. Lakowicz, J.R. and Berndt, K. 1991. Lifetime-selective fluorescence imaging using an rf phase-sensitive camera. Rev. Sci. Instrum. 62:1727-1734. Mahajan, N.P., Linder, K., Berry, G., Gordon, G.W., Heim, R., and Herman, B. 1998. Bcl-2 and bax interactions in mitochondria probed with green fluorescent protein and fluorescence energy transfer. Nature Biotechnol. 16:547-552. Matz, M.V., Fradkov, A.F., Labas, Y.A., Savitsky, A.P., Zaraisky, A.G., Markelov, M.L., and Lukyanov, S.A. 1999. Fluorescent proteins from nonbioluminescent Anthozoa species. Nature Biotechnol. 17:969-973.

Mittman, S., Flaming, D.G., Copenhagen, D.R., and Belgum, J.H. 1987. Bubble pressure measurement of micropipette tip outer diameter. J. Neurosci. Methods 22:161-166. Miyawaki, A., Llopis, J., Heim, R., McCaffery, J.M., Adams, J.A., Ikura, M., and Tsien, R.Y. 1997. Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388:882-887. Ng, T., Squire, A., Hansra, G., Bornancin, F., Prevostel, C., Hanby, A., Harris, W., Barnes, D., Schmidt, S., Mellor, H., Bastiaens, P.I.H., and Parker, P.J. 1999. Imaging protein kinase C alpha activation in cells. Science 283:2085-2089. Squire, A. and Bastiaens, P.I.H. 1999. Three dimensional image restoration in fluorescence lifetime imaging microscopy. J. Microsc. 193:36-49. Tsien, R.Y. 1998. The green fluorescent protein. Annu. Rev. Biochem. 76:509-538. Wilson, K. 1994. Preparation of genomic DNA from bacteria. In Current Protocols in Molecular Biology (F.A. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds), pp. 2.4.1-2.4.5. John Wiley & Sons, Nwe York. Wouters, F.S. and Bastiaens, P.I.H. 1999. Fluorescence lifetime imaging of receptor tyrosine kinase activity in cells. Curr. Biol. 9:1127-1130. Wouters, F.S., Bastiaens, P.I.H., Wirtz, K.W.A., and Jovin, T.M. 1998. FRET microscopy demonstrates molecular association of non-specific lipid transfer protein (nsL-TP) with fatty acid oxidation enzymes in peroxisomes. EMBO J. 17:7179-7189.

Contributed by Fred S. Wouters Imperial Cancer Research Fund London, United Kingdom Philippe I.H. Bastiaens European Molecular Biology Laboratory Heidelberg, Germany



Supplement 23

High-Throughput Screening for Protein-Protein Interactions Using Yeast Two-Hybrid Arrays

UNIT 19.6

Arrays are used for parallel assays of large numbers of proteins in a single experiment. They provide an alternative to libraries for protein screening purposes. The use of arrays was fostered by genomics and proteomics with the sequencing of complete genomes, which can now be represented completely on an array. At the end of an array experiment, the positives can be identified immediately because each element, or protein, has a unique address or position in the array. Although many important studies have been carried out using DNA arrays, proteins are more chemically heterogeneous than oligonucleotides, so the construction of large-scale protein arrays that can be screened under uniform conditions is a challenge. Two types of protein arrays have been constructed: living arrays, where the proteins are expressed in cells, and nonliving arrays, where the proteins are purified and tested in vitro. Here the authors describe a protein array composed of living yeast (Saccharomyces cerevisiae) cells that can be used in a functional screen for protein interactions, the two-hybrid assay. In this assay (also see UNIT 19.2), two proteins, A and B, are expressed as fusions with, respectively, a DNA-binding domain (DBD) and a transcriptional activating domain (AD). If A and B interact in the yeast nucleus, the resulting complex will reconstitute a transcription factor, Gal4, that activates one or more reporter genes. Transcription of the reporter gene allows the cells to grow on selective media so that they can be easily scored on a selective plate. Yeast colonies expressing proteins for two-hybrid screening are arrayed in a high-density format, and proteins are screened using automated robotic procedures. The procedure can be modified for manual use or for use with alternative screening strategies such as synthetic lethal screens. With minor modifications, the array can be used to screen for protein interactions with DNA, RNA, or even small-molecule inhibitors of the two-hybrid interaction (see Background Information). The protocols described here use the yeast proteome as an example, but they can be applied to any other genome or subset thereof. The steps of the process involve the construction of the array (see Basic Protocol 1), including the cloning of the arrayed proteins (AD fusions or preys) and the DBD fusions (baits) in yeast; and the screening of the array by either manual (see Basic Protocol 2) or robotic (see Alternate Protocol) manipulation, including the selection of positives and scoring of results. NOTE: To prevent contamination, plate pouring and the transfer steps should be carried out in a sterile hood, or the work surface should be treated regularly with UV light. STRATEGIC PLANNING Many projects involving protein arrays are of a large scale and may be difficult to modify at later stages. Specifically, the size and character of the array must be designed while bearing in mind the initial construction of the array as well as the ultimate aims of the experiment. Many types of arrays are possible. Factors that may be varied include the form of protein arrayed (e.g., wild type or mutant, full length or single domain, ligand or drug bound, epitope tagged). Similarly, the arrayed proteins may be related (e.g., splice variants or alanine-substituted variants of a single protein, a family or pathway of related proteins, orthologs of a protein from different species, the entire protein complement of a model organism). Artificial peptides generated using combinatorial chemical synthesis Contributed by Gerard Cagney and Peter Uetz Current Protocols in Protein Science (2001) 19.6.1-19.6.12 Copyright © 2001 by John Wiley & Sons, Inc.



may even comprise an array. Different types of arrays will require different construction strategies, and it may be best to carry out a small-scale pilot study, incorporating positive and negative controls, before committing to a full-scale project. Although high-throughput screening projects can be performed manually, automation is strongly recommended. Highly repetitive tasks are not only boring and straining but also error prone when done manually. High-throughput screening therefore should utilize a robotic device. Robots are expensive and require programming and maintenance. Before buying a robot, one should consider whether the robot can also be used for other procedures that are not directly related to the actual screen (e.g., for follow-up studies). BASIC PROTOCOL 1

CONSTRUCTION OF A PROTEOME-SCALE PROTEIN ARRAY FOR YEAST This procedure describes the cloning and expression of yeast proteins for a high-density living array. Because the application involves a large number of proteins (>6000), polymerase chain reaction (PCR) products encoding full-length yeast proteins (as predicted by the published genome sequence) are directly cloned into yeast hosts by transformation and subsequent homologous recombination into the vectors. The recombination event takes place inside the cell following the cotransformation of linearized vector and insert DNA using a modification of the lithium acetate/polyethylene glycol technique (Orr-Weaver et al., 1981). An aliquot of each transformed yeast clone is stored individually, and the entire set of clones is grown on agar so that it can be handled conveniently for screening purposes. To perform the interaction screen, the DNAs encoding the proteins of interest are inserted into two different vectors in two different yeast strains that are identical except at the MAT locus. To clone the activating domain (AD) fusion constructs, the pOAD vector is used in a strain of mating type a (PJ69-4A). These transformants are used to create the array. To clone the DNA-binding domain (DBD) fusion constructs, the pOBD2 vector is used in a strain of mating type α (PJ69-4α). The interactions occur when the DBD and AD strains are allowed to mate (see Basic Protocol 2). The array is made from an ordered set of AD-containing strains, rather than BD-containing strains, because the former do not generally result in self-activation of transcription.

Two-Hybrid Screening with Protein Arrays

Materials DNAs encoding proteins of interest (e.g., genomic DNA, cDNA library) Appropriate primers for amplification of the protein-coding sequences Two-hybrid plasmids: pOAD and pOBD2 (Hudson et al., 1997; Cagney et al., 2000; also see Internet Resources) NcoI and PvuII or other appropriate restriction enzymes, with buffers Liquid YEPD medium (see recipe) S. cerevisiae host strains: PJ69-4A (MATa trp1-901 leu2-3,112 ura3-52 his3-200 gal4∆ gal80∆ LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ; James et al., 1996) PJ69-4α (MATα trp1-901 leu2-3,112 ura3-52 his3-200 gal4∆ gal80∆ LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ; Uetz et al., 2000) 0.1 M lithium acetate 7.75 mg/ml salmon sperm DNA (Sigma), autoclaved (15 min at 121°C) and stored at −20°C 96PEG solution (see recipe) Dimethyl sulfoxide (DMSO) 35-mm plates containing solid −Leu (for pOAD) and −Trp (for pOBD2) dropout medium (see recipe) Liquid −Leu dropout medium (see recipe)



Single-well microtiter plates (e.g., OmniTray, Nalge Nunc International) containing solid −Leu dropout medium (see recipe) 40% (v/v) glycerol 250-ml conical flasks 30°C shaking incubator 8-channel pipettor (to deliver 200 µl) and sterile pipetting troughs 96- and 384-well microtiter plates (e.g., Nalge Nunc International) Plastic adhesive tape for sealing plates 42°C incubator Centrifuge with insert for 96-well plates Toothpicks, sterile 384-Pin Replicator (Nalge Nunc International) 2-ml cryotubes (e.g., Nalge Nunc International) Additional reagents and equipment for PCR amplification (APPENDIX 4J), agarose gel electrophoresis (APPENDIX 4F), digestion of DNA with restriction endonucleases (APPENDIX 4I), and gel purification (APPENDIX 4F) and quantification (APPENDIX 4K) of DNA Amplify protein-coding DNA 1. Individually amplify the DNAs encoding the proteins of interest using appropriate primers and standard PCR conditions (APPENDIX 4J). A two-step PCR protocol may be used, where the first-round reaction uses primers specifying the 5′ and 3′ terminal ends of the DNA encoding the protein to be cloned. These primers are each flanked with ∼20-nucleotide tails that are homologous to sequences in the two-hybrid vectors. In order to increase the efficiency of homologous recombination, a second-round PCR step using ∼70-mer primers can be carried out, where the primers have ∼50 nucleotides that are homologous to the vectors (see Internet Resources; Hudson et al., 1997; Cagney et al., 2000). For the first round of PCR, specific primers must be used for each reaction; for the second round, a common set of primers is used. Appropriate primers for the cloning vector used in this protocol are: First-round forward primer: 5′-AATTCCAGCTGACCACCATGXXX20-30-3′, where ATG represents the start codon of the yeast ORF and XXX20-30 represents ORF-specific sequences of 20 to 30 bases. First-round reverse primer: 5′-GATCCCCGGGAATTGCCATG***XXX20-30-3′, where *** represents the reverse complement of one of the three stop codons and XXX20-30 represents 20 to 30 bases of the reverse complement of ORF-specific sequences at the terminus of the reading frame. Second-round forward primer: 5′-CTATCTATTCGATGATGAA GATACCCCACCAAACCCAAAAAAAGAGATCGAATTCCAGCTGACCACCATG-3′. Seco nd -ro un d reverse p rimer: 5′-CTTGCGGGGTTTTTCAGTATCTACGATTCATA GATCTCTGCAGGTCGACGGATCCCCGGGAATTGCCATG-3′. Other primers can be used when the cloning vectors are modified accordingly. It is recommended that primers be purified to the highest level that is practical (i.e., PAGE, or HPLC, UNIT 8.7). Insertions or deletions in the primers can result in the expression of fusion proteins that are out of frame. UNIT 10.1,

Suggested PCR conditions are: (first round) 30 ng template, 1.5 mM MgCl2, 5 U Taq polymerase (PE Biosystems), 0.02 U Pfu polymerase (Stratagene), 20 pmol each primer; (second round) 5 ng template, 1.5 mM MgCl2, 0.6 U Taq polymerase, 0.003 U Pfu polymerase, 2.5 pmol each primer. It is helpful to amplify the DNAs with approximately equal product sizes grouped together so that the extension times are similar and the products can be checked on the same agarose gel. Many full-length ORFs can now be purchased from commercial suppliers as PCR products or plasmids. Identification of Protein Interactions


Supplement 24

2. Examine the amplified DNAs by agarose gel electrophoresis (APPENDIX 4F) to confirm that they are the correct size. If resources permit, the DNAs can be partially or completely sequenced.

Prepare plasmids and yeast strains 3. Linearize the two-hybrid plasmids (pOAD and pOBD2 vector DNA) by digesting with NcoI and PvuII or other appropriate restriction enzymes (APPENDIX 4I). Gel purify (APPENDIX 4F) and quantify (APPENDIX 4K) the linearized plasmids. pOAD is used for the AD fusion constructs to create the array. pOBD2 is used for the DBD constructs, which will be used to screen the array. UNIT 19.2 describes alternative vector systems. The following steps can be scaled down to perform a smaller number of transformations.

4. For each multiple of 96 PCR products to be cloned, add 50 ml liquid YEPD medium to each of two 250-ml conical flasks. Inoculate each flask with one S. cerevisiae host strain (PJ69-4A and PJ69-4α) from an isolated colony using an inoculating loop and grow these cultures overnight in a 30°C shaking incubator. PJ69-4A is used to clone the AD fusion constructs, and PJ46-4α is used to clone the DBD fusion constructs.

5. Transfer each culture to a 50-ml conical centrifuge tube and centrifuge 3 min at 3000 × g, room temperature, to harvest the yeast cells. Remove the supernatants and resuspend each pellet in 2 ml of 0.1 M lithium acetate. Clone PCR products 6. Boil 1 ml of 7.75 mg/ml salmon sperm DNA in a microcentrifuge tube for 5 min and plunge it into ice water. 7. Add the following reagents (in order) to each of two 50-ml conical tubes: 20 ml 96PEG solution 0.5 ml salmon sperm DNA (step 6) 200 ng linearized pOAD or pOBD2 vector DNA (step 3) 2.5 ml DMSO. Shake vigorously for 30 sec followed by 1 min of vortexing. Add 2 ml yeast suspension (step 5) and mix well by hand for 1 min. 8. Pour each mixture into a sterile pipetting trough and use an 8-channel pipettor to deliver 200 µl into each well of a 96-well microtiter plate. Set up one plate for the pOAD vector and a second for the pOBD2 vector. Wells for control transformations, where no PCR product will be added, should be included.

9. Add 3 µl of each amplified DNA (insert DNA; step 2) to a separate vector-containing well and seal the wells with a secure plastic adhesive tape. Vortex 4 min and incubate 30 min in a 42°C incubator. A vector DNA to which no PCR product was added should be used as a negative control for transformation.

10. Centrifuge the 96-well plates 7 min at 3000 × g, room temperature.


11. Remove the supernatant by aspiration with the 8-channel pipettor. Add 200 µl sterile water to each well and resuspend the yeast with the pipettor.



12. Spread the pOAD and pOBD2 yeast suspensions onto 35-mm −Leu and −Trp plates, respectively, and incubate 2 to 3 days at 30°C. The authors suggest testing the clones by colony PCR or sequencing before assembling the array. Ideally, each colony or clone should be tested for expression of its cognate fusion protein using immunoblotting (UNIT 10.10). This may be prohibitively difficult or expensive to carry out for thousands of proteins, especially with the low-copy plasmids described here.

Construct array 13. Pick individual AD colonies from the −Leu plates using sterile toothpicks and transfer to the wells of 384-well microtiter plates containing liquid −Leu dropout medium for inclusion in the array. Note the address of each colony, preferably in a searchable database. One colony is picked from each plate if it can be tested by PCR or sequencing. If there are too many clones to test individually, two or more clones from a single AD plate may be pooled at this stage in order to avoid occasional PCR mutants or empty vectors. A larger number of pooled clones may be used, but this increases the risk of fast-growing variants out-competing the desired ones. The authors have found that recovery of array clones is best following storage in YEPD. Therefore, a duplicate 384-well microtiter plate containing YEPD should be set up with the array clones, grown overnight at 30°C, and frozen indefinitely in 20% (v/v) glycerol (i.e., after adding an equal volume of 40% glycerol) at −80°C.

14. Incubate overnight at 30°C without shaking. 15. Use a 384-Pin Replicator to transfer the clones to single-well microtiter plates containing solid −Leu dropout medium, according to manufacturer’s instructions. Duplicate each element on the array (i.e., place two colonies that express the same protein in adjacent positions on the array). Alternatively, duplicate copies of the array can be made. Using duplicate elements or arrays assures reproducibility (see Anticipated Results, discussion of false positives). When arranging the colonies on the array, standard formats should be used so that the colonies can be manipulated by robots and other standardized labware. When the array is used for screening, this format can be condensed 4-, 8-, or 16-fold to include 384, 768, or 1536 colonies per microtiter-plate footprint. However, scoring positives becomes more difficult at higher densities. For day-to-day use, the array can be stored on solid medium lacking leucine (−Leu) for up to 3 months at 4°C. Working copies of the array grown on complete medium (YEPD) can be generated from the −Leu copy every 4 weeks (or as often as needed) for use in screens (see Basic Protocol 2).

Prepare bait 16. Pick individual DBD colonies from −Trp plates and grow overnight in 20 ml YEPD at 30°C. Dilute culture 1:1 with 40% glycerol and transfer aliquots to 2-ml cryotubes. Freeze indefinitely at −80°C. MANUAL SCREENING FOR PROTEIN INTERACTIONS USING A YEAST PROTEIN ARRAY The living array (see Basic Protocol 1) can be screened for protein interactions by a mating procedure that can be carried out manually (described here) or using a robot (see Alternate Protocol). A strain expressing a single candidate protein as a DBD fusion is mated to all the colonies in the array, which contains yeast colonies expressing individual AD fusion

BASIC PROTOCOL 2



Supplement 24

proteins. After mating, the colonies are transferred to diploid-specific medium, and then to two-hybrid selective medium. To manually screen with more than one bait, replicate copies of the array are used. For large numbers of baits, robotic screening is recommended. Materials 20% (v/v) bleach (∼1% sodium hypochlorite) 95% (v/v) ethanol Yeast protein array (see Basic Protocol 1) Single-well microtiter plates (e.g., OmniTray; Nalge Nunc International) containing solid YEPD +Ade medium (see recipe), YEPD medium (see recipe), −Leu −Trp dropout medium (see recipe), and −His −Leu −Trp +3AT dropout medium (see recipe) Liquid YEPD medium (see recipe) DBD fusion–expressing yeast strain (see Basic Protocol 1) 384-Pin Replicator (Nalge Nunc International) 30°C incubator 250-ml conical flask 1. Sterilize a 384-Pin Replicator by dipping the pins into 20% bleach for 20 sec, sterile water for 1 sec, 95% ethanol for 20 sec, and sterile water again for 1 sec. Repeat this sterilization before each transfer. 2. Use the sterile pin replicator to transfer a yeast protein array to single-well microtiter plates containing solid YEPD +Ade medium and grow the array overnight in a 30°C incubator. If duplicate colonies were not used to construct the array (see Basic Protocol 1, step 15), the entire experiment should be done using duplicate arrays. YEPD medium with supplemental adenine increases the efficiency of the mating step.

3. Inoculate 20 ml liquid YEPD medium in a 250-ml conical flask with a DBD fusion–expressing yeast strain and grow overnight at 30°C. Ideally, the frozen strain (see Basic Protocol 1, step 16) is streaked out on a plate containing solid YEPD medium and grown overnight at 30°C. A colony from this plate is then used to inoculate the liquid medium. This volume is sufficient for mating to the entire array.

4. Dip the pins of the pin replicator into the DBD fusion–expressing culture and place directly onto a fresh single-well microtiter plate containing solid YEPD medium. Repeat with the required number of plates. 5. Pick up the array (i.e., AD) yeast colonies with sterilized pins and transfer them directly onto the DBD strain, so that each of the 384 DBD yeast spots per plate receives different AD yeast cells (i.e., a different AD fusion). Incubate 1 to 2 days at 30°C to allow mating. A scaffold of some kind should be used to ensure that the DBD and array strains contact each other. The use of a robot for this procedure is discussed (see Alternate Protocol). Mating will take place in 1 mm in diameter.



This is an essential control step because only diploid cells that contain Leu and Trp markers on pOAD and pOBD2, respectively, will grow on this medium. This step also helps recovery of the colonies and increases the efficiency of the next selection step.

7. Transfer the colonies to a single-well microtiter plate containing solid −His −Leu −Trp +3AT dropout medium using the sterilized pinning tool and grow at 30°C for up to 10 days (or longer if there is little or no background growth). The stringency of the screen can be varied by adding different amounts of 3AT, an inhibitor of the His3 gene product (imidazoleglycerolphosphate dehydratase). For standard screening purposes, 3 mM is suggested. In many cases (10% to 20% of full-length yeast proteins tested by the authors), the haploid strain expressing the DBD fusion has transcriptional self-activation properties. These haploid strains can be titrated on −His plates containing increasing amounts of 3AT. The highest level of 3AT tolerated should be added to the −His −Leu −Trp plates for selection of two-hybrid positive diploids. In many cases, however, the transcriptional activity is very strong (>200 mM 3AT) and alternative approaches must be considered, such as reengineering the construct to express smaller protein fragments. The vectors and strains described here are also suitable for two-hybrid screening on −Ade −Leu −Trp plates. The use of two alternative two-hybrid reporters (His and Ade) can be useful for reducing the levels of false positives, but the Ade selection is very stringent (equivalent to ∼200 mM 3AT) and the interactions of many proteins may be of insufficient affinity to permit growth on these plates.

8. Score the interactions by looking for signals (i.e., growing colonies) that are significantly above background (by size) and that are present for both duplicate colonies (see Basic Protocol 1, step 15). The plates should be examined every day. Most two-hybrid positive colonies appear within 3 to 4 days, but occasionally positive interactions can be observed later. Very small colonies are usually designated as background; however, there is no absolute measure to distinguish between background and real positives. When there are many (i.e., >30) large colonies per array of 6000 positions, the authors consider the baits used as random activators. In this case the screen should be repeated or should not be analyzed. Scoring can be done manually or using automated image analysis procedures. When using image analysis, care must be taken not to score contaminant colonies as positives. Aspects of the scoring process are discussed more fully (see Anticipated Results, discussion of analysis of screens).

ROBOTIC SCREENING FOR PROTEIN INTERACTIONS USING A YEAST PROTEIN ARRAY

ALTERNATE PROTOCOL

In many cases a hand-held 384-pin replicating tool can be used for routine transfer of colonies for screening. For larger projects, however, a robotic workstation (e.g., Biomek 2000; Beckman Coulter) may be used to speed up the screening procedures and to maximize reproducibility. A 384- or 768-pin stainless steel replicating tool (e.g., HighDensity Replication Tool; Beckman Coulter) can be used to transfer the colonies from one plate to another. Between transfer steps, the tool should be sterilized by sequential immersion into a 20% (v/v) bleach solution (20 sec), sterile water (1 sec), 95% (v/v) ethanol (20 sec), and sterile water (1 sec). The level of these liquids should be 2 to 4 mm from the base of the pin and care must be taken that the ethanol does not evaporate. It is important to ensure that plasticware is compatible with the movements of the robot. In the procedure described above, the array is gridded on eight 86 × 128–mm single-well microtiter plates (e.g., OmniTray, Nalge Nunc International) in 768-colony format. Identification of Protein Interactions


Supplement 24


C dropout powder 1 g histidine 1 g methionine 1 g arginine 2.5 g phenylalanine 3 g lysine 3 g tyrosine 4 g tryptophan 4 g leucine 4 g isoleucine 5 g glutamic acid 5 g aspartic acid 7.5 g valine 10 g threonine 20 g serine 1 g adenine 1 g uracil Store up to 2 years at room temperature For leucine-, tryptophan-, and histidine-free media (−Leu, −Trp, and −His), leave out these amino acids.

Dropout medium For liquid medium: To 800 ml H2O add: 1.7 g yeast nitrogen base without amino acids 5 g ammonium sulfate 20 g dextrose 1.4 g C dropout powder (see recipe) Add H2O to 1 liter Autoclave and store up to 10 weeks at 4°C For +3AT dropout medium, add 3 mM 3-aminotriazole (final concentration).

For solid medium: Add 16 g agar (e.g., Difco, Becton Dickinson) before autoclaving. Cool to 45°C and pour into appropriate plates. Wrap plates and store up to 6 months at 4°C. 96PEG solution 45.6 g polyethylene glycol (avg. mol. wt. 3350; e.g., Sigma) 6.1 ml 2 M lithium acetate 1.14 ml 1 M Tris⋅Cl, pH 7.5 (APPENDIX 2E) 232 µl 0.5 M EDTA, pH 8.0 (APPENDIX 2E) H2O to 100 ml Store up to 1 year at room temperature


YEPD medium For liquid medium: To 800 ml H2O add: 10 g yeast extract (e.g., Sigma; Difco, Becton Dickinson) 20 g peptone (e.g., USBiological) 20 g dextrose (e.g., Fisher) continued



Add H2O to 1 liter Autoclave and store up to 3 months at room temperature For solid medium: Add 14 g agar (e.g., Difco, Becton Dickinson) before autoclaving. Cool to 45°C and pour into appropriate plates. Wrap plates and store up to 6 months at 4°C. YEPD +Ade medium Prepare solid YEPD medium (see recipe) but add 10 ml 0.2% (w/v) adenine before autoclaving. COMMENTARY Background Information The two-hybrid system was originally invented to detect interactions between characterized proteins (Fields and Song, 1989). It quickly became clear that it could be used to identify new interactions when expression libraries were screened. With the advent of complete genome sequences, the original idea of testing known protein pairs was revived again: instead of using random libraries, all possible pairwise protein combinations were tested in order to identify interacting partners. Such systematic, genome-wide screens were first reported in studies using DNA chips but could also be applied to proteins. Arrays have the advantage that many tests can be performed under identical conditions at the same time. The results of these individual tests can be compared directly. This is important because baits vary in background levels. Preys with a tendency to cause false positives therefore can also be easily identified. In addition, because arrays allow rapid identification of positives by their position, it is not necessary to sequence positive clones. However, two-hybrid arrays as described here have disadvantages. Because all elements in the array are generated individually, the production of arrays is time and labor intensive. Furthermore, full-length proteins may not interact because steric factors prevent productive reconstitution of the transcription factor Gal4 on the promoter of the reporter gene. Other groups have recommended the use of random libraries that are screened conventionally (Fromont-Racine et al., 1997; Rain et al., 2001). Indeed, random libraries can detect interactions that have not been found using full-length proteins (Flajolet et al., 2000). On the other hand, large-scale screens of random libraries require significant sequencing capacities. For instance, the screening of 261 Helicobacter baits required the sequencing of >13,000 PCR products (Rain et al., 2001). Screens of random

libraries yield more interactions than screens of full-length libraries, but they also require highquality libraries that are representative of weakly expressed transcripts (when cDNA libraries are used). In addition to screening for protein-protein interactions, living protein arrays can be used to study DNA- and RNA-protein interactions (Kraemer et al., 2000) or genetic interactions. These might include screens of mutant proteins against arrays of strains deleted for nonessential proteins, or strains where expression of an essential gene is disrupted using promoter manipulation or drugs. In the reverse-two-hybrid screen, candidate compounds are examined for their ability to disrupt a positive two-hybrid interaction (Vidal and Legrain, 1999).

Critical Parameters Following successful cloning, the mating reaction between the bait and prey yeast strains is critical. Strains vary considerably in their mating efficiency. The strains must be in contact for ≥24 hr. Robots used for transfer functions should be calibrated regularly, in order to prevent misalignment and inefficient transfer of yeast cells. Mold contamination can be a problem, especially for plates growing for >1 week. To prevent contamination of plates, plate pouring and the transfer steps should be carried out in a sterile hood, or the work surface should be treated with UV light regularly.

Troubleshooting About 15% of all baits activate transcription on their own and show strong background growth. Weak activators can usually be screened by elevating the level of 3-aminotriazole up to 150 mM. Another approach is to use additional reporter genes, although multiple reporter genes, and therefore higher stringency, may result in the loss of weak interactions.



Supplement 24

A

B


Figure 19.6.1 Two subsequent two-hybrid screens of a full-length activation domain–open reading frame library with full-length PHO85 protein (a cyclin-dependent kinase) as bait. (A) First screen. (B) Second screen. Reproducible positives are indicated by arrows. Some reproducible positives are either nonyeast contaminants (c) or false positives (f). Nonyeast contaminants can be identified by their unusual shape, size, or color. False positives are either nonreproducible or reproducible positives, but the latter are usually found in many independent screens with unrelated baits (hence they are nonspecific). Given these criteria, the following preys were identified as reproducible positives (plates are numbered in columns): PCL10, PCL6, CLG1 (plate 4), SOR1 (plate 5), CDC36 (plate 11), PCL9, PCL2, YPL229W (plate 12), YDL246C (plate 13), PCL8 (plate 14). PCLs are cyclins, and most of them are known PHO85 interactors; CLG1 is a cyclin-like protein; CDC36 is a transcription factor; YPL229W is a protein of unknown function; SOR1 is sorbitol dehydrogenase; and YDL246C is a SOR1-related protein. Most of these preys are either known or highly plausible interactors of PHO85. The biological role of an interaction between SOR1 and PHO85 remains unclear but is supported by YDL246C, which is a relative of SOR1. Note that these two screens have an unusually high number of contaminants on plates 9 and 10. The false positives have been found in more than 10% of all screens.



When arrays are kept over months or even years, it is important to avoid cross-contamination. These events can accumulate over time and lead to misidentification of colonies. Yeast arrays can also be maintained as frozen stocks in 20% glycerol at −80°C that are used to reestablish the working arrays every couple of months.

Anticipated Results A fairly typical screen is shown in Figure 19.6.1. Ideally it should have no background growth and only a few positives. In most cases a small number of reproducible positives is found (i.e., 100 Da therefore it can be used in drug discovery applications to screen or characterize the interaction between small molecules and a panel of protein targets. Biacore A100 is well suited for information-rich, high-throughput kinetic characterization enabling selection of antibodies at early stages of development (e.g., hybridoma supernatants) and large-scale proteomics studies. Biacore A100 is also suitable for use in regulated environments.

SURFACE PREPARATION Sensor Chip Options Although SPR can be generated in thin films made from conducting metals, all Biacore sensor chips are coated with a thin,

Identification of Protein Interactions


Supplement 45

uniform gold layer. Gold has a number of advantages in that it results in a well-defined reflectance minimum when a visible light source is used to generate the SPR signal, gold is amenable to covalent attachment of surface matrix layers and, in physiological buffer conditions, gold is mostly inert. Sensor chip surfaces, except Sensor Chip Au, are covered with a covalently bound alkanethiol monolayer to prevent nonspecific adsorption of proteins to the gold surface. Sensor chips are reusable since noncovalently bound material can be removed from the Sensor Chip surface with an injection of a suitable regeneration solution.

Nonderivatized dextran-containing sensor chips The surface of Sensor Chips CM5, CM4, and CM3 is covered with a hydrogel matrix of flexible, unbranched carboxymethylated dextran covalently attached to the surface alkanethiol monolayer. The dextran layer ranges in thickness from ∼25 to 100 nm, depending on the sensor chip type. The dextran layer provides a hydrophilic, solution-like environment favorable for most interactions between proteins or other biomolecules. The carboxyl moieties can be derivatized using a wide range of well-defined chemistries for the covalent attachment of biomolecules. The flexibility of the unbranched dextran polymer allows covalently attached molecules to move with relative freedom within the surface layer. The dextran layer also increases the surface capacity in comparison with a two-dimensional surface.

Derivatized dextran-containing sensor chips

Overview of Biacore Systems

Three pre-derivatized chips are currently available for Biacore instruments, namely Sensor Chips SA, NTA, and L1. Sensor Chip SA is pre-immobilized with streptavidin and is suitable for the capture of biotinylated molecules (e.g., nucleic acids, peptides, proteins, liposomes, glycosaminoglycans). The capture of biotinylated molecules is essentially irreversible due to the high affinity of the biotinstreptavidin interaction (KD = 10−15 M). Histidine (His)-tagged recombinant proteins can be reversibly captured on NiCl2 -activated nitrilotriacetic acid (NTA) groups that are covalently coupled to the dextran matrix of Sensor Chip NTA. Captured molecules can easily be removed from the sensor chip surface with a pulse of EDTA. The affinity of the His-tag for the activated NTA surface is dependent on the microenvironment of the tag and, therefore,

the baseline stability following the capture step can vary from protein to protein. Sensor Chip L1 is derivatized with lipophilic anchors (alkane chains) designed to intercalate into lipid bilayers enabling the stable capture of intact membrane structures and liposomes.

Non-dextran sensor chips Sensor Chip C1 has a flat carboxylated surface without a dextran matrix. The carboxyl groups can be derivatized using the same immobilization chemistries used for covalent coupling to the dextran surface of Sensor Chips CM5, CM4, and CM3; however, the reduction in available surface area results in a significant decrease in the immobilization capacity (typically ∼10% of that on Sensor Chip CM5 under comparable conditions). Sensor Chip C1 is useful for monitoring interactions with large interactants such as cells and viruses. It also provides a suitable surface for studies where the flexibility of the dextran layer is not wanted (e.g., homodimerization studies, reducing avidity effects). Sensor Chip HPA lacks a carboxymethylated dextran matrix so that lipid monolayers can be formed on the hydrophobic thioalkane surface layer covering the gold film. Sensor Chip HPA is suitable for measuring interactions with a lipid monolayer or interactions with ligands associated with a membrane surface or partially inserted into the membrane. Sensor Chip L1 is more suitable for deeply inserted or transmembrane proteins. Sensor Chips Au and SIA Kit Au consist of a bare gold surface, which lends itself to the design of customized surface chemistries using self-assembled monolayers (SAM) or to study interactions between surface materials and biomolecules. Sensor Chip SIA Kit Au contains unmounted gold surfaces and a separate chip carrier for easy assembly after surface coating. This allows for the use of a wide variety of coating techniques, including those using harsh conditions (such as organic solvents) that the chip carrier would not usually withstand.

Flexchip sensor chips A number of sensor surfaces are currently available for Flexchip including bare gold, streptavidin, neutravidin, and protein A/G– coated chips. Proteins can be spotted on the bare gold surface that can subsequently be blocked with blocking solution to minimize non-specific interactions. Antibodies or Fc



fusion proteins can be spotted onto protein A/G chips, thereby resulting in a more oriented surface than that created by spotting on a bare gold surface. Streptavidin and neutravidin surfaces can be used with biotinylated peptides, nucleic acids, and proteins.

Immobilization Chemistries A critical step in the development of reliable SPR assays is the selection of the most suitable immobilization technique such that ligand activity is maintained and binding sites are available to interacting partners. Commonly utilized immobilization strategies are outlined here. Guidelines and a detailed review of immobilization chemistries have been published recently (Karlsson and Larsson, 2004).

Direct immobilization chemistries A number of covalent immobilization techniques are available to covalently attach proteins or other biomolecules to the dextran on the sensor chip surface. All types of immobilization can be performed directly in the Biacore system, and a typical immobilization reaction usually takes 9.5) as well as reducing agents such as



Supplement 45

β-mercaptoethanol and dithiothreitol. Several heterobifunctional reagents are available for introduction of reactive maleimido groups to the sensor surface, including sulfoMBS (m-maleimidobenzoyl-N-hydroxysulfosuccinimide ester), sulfo-SMCC (sulfosuccinimidyl-4-(N-maleimidomethyl)cyclohexane-l-carboxylate), and GMBS (N(g-maleimidobutyryloxy)sulfosuccinimide ester).

Aldehyde coupling Aldehyde coupling involves the formation of a hydrazone bond via condensation of hydrazide groups on the sensor surface with aldehyde groups on the ligand molecule. These aldehyde moieties may be native to the protein or introduced through mild oxidation of cis-diols present in the ligand molecule. Aldehyde coupling is particularly useful for sitedirected immobilization of glyco-conjugates, glyco-proteins, and polysaccharides, and may also be useful for orientation-specific immobilization of proteins containing functional groups that may be converted to aldehyde moieties.

Indirect (capture) immobilization General capture methods Capture approaches provide an alternative to covalent immobilization and take advantage of tags commonly used for ligand purification. This technique involves high-affinity capture of the ligand onto a capturing molecule that has been covalently immobilized using one of the techniques described earlier. The requirement for ligand purity is less stringent for capture approaches than for covalent immobilization since the capture step can also provide a ligand purification step. Another benefit of capture approaches is the creation of a homogenous surface since all ligands are similarly oriented through a common site on the ligand (the tag). The affinity of the ligand for the capture molecule should be sufficiently high to ensure that little or no ligand dissociates from the surface for the duration of an analysis cycle. Monoclonal antibodies are frequently used as capture molecules, e.g., antiGST antibodies can be immobilized and used to capture GST-tagged molecules. In general, regeneration of the surface removes both the ligand and the analyte at the end of an assay cycle such that fresh ligand must be captured for a new cycle.

Specialized capture methods The high-affinity interaction between streptavidin and biotin (KD 10−15 M) makes it an extremely useful system for the capture of biotinylated ligands. Sensor Chip SA provides a robust, user-friendly streptavidin capture surface. Owing to the high affinity of the interaction, biotinylated ligands are immobilized irreversibly and, unlike most other capture approaches, cannot be removed from the sensor chip surface. Histidine (His)-tagged recombinant proteins can be captured on Sensor Chip NTA following activation of the surface with a pulse of NiCl2 . Ligands can be removed easily from the surface by stripping the nickel ions with a pulse of EDTA. An alternative to the NTA surface is to capture His-tagged ligands onto covalently immobilized anti-His antibodies. Large baseline drifts caused by unstable ligands or poor capture may be overcome by using EDC/NHS as a post-capture cross-linking step, i.e., crosslinking the tagged protein to the anti-tag antibody. This step may compromise ligand activity if active sites of the ligand are involved in the cross-linking; therefore, it must be empirically tested for each ligand-analyte system. In general, crosslinking should be as brief as possible; 15 sec is often sufficient to achieve acceptable baseline stability without compromising ligand activity. Lipid monolayers can be adsorbed onto the hydrophobic surface of Sensor Chip HPA while liposomes can be maintained in a membrane-like lipid bilayer structure on the surface of Sensor Chip L1. On-surface reconstitution (OSR) is a recently developed methodology for handling membrane proteins on Sensor Chip L1. In this process, the membrane protein is solubilized initially in detergent and then either captured through a tag or immobilized on the chip surface using amine coupling. The next step is to inject mixed micelles, which bind to the lipophilic tails on the sensor chip surface and to the ligand. Finally, the detergent is removed, inducing the lipids to form a plasma membrane-like bilayer, linked to the immobilized membrane protein via the natural affinity of lipids for the hydrophobic regions of the proteins. In this way, it is possible to present a uniform, oriented field of plasma membrane proteins within a lipid bilayer on a sensor chip and to study how they interact with binding partners (Karlsson and Löf˚as, 2002; Stenlund et al., 2003; Navratilova et al., 2005).




ASSAY TYPES Binding Specificity Biacore is well suited to carry out qualitative studies to confirm the specificity of interactions as well as quantitative measurements for affinity, kinetics, and concentration determination. A small volume of analyte can be tested easily for selective binding to 2 to 400 targets simultaneously, depending on the instrument platform chosen. Furthermore, analyte activity is not compromised since interactants do not need to be labeled. It is possible to monitor a number of sequential binding events since each yields a concomitant increase in mass on the sensor chip surface and all stages in the binding process can thus be monitored. This approach has been used to confirm the results of conventional immunoprecipitation assays and reduce assay time (Young et al., 2002). Epitope mapping of monoclonal antibodies can be carried out at much earlier stages in development using Biacore since little sample is required and samples do not need purification or labeling. Unfractionated hybridoma culture supernatants have been used successfully for epitope mapping studies (Fägerstam et al., 1990). The technique is well suited to unattended instrument operation, and many samples can be processed without user intervention. This feature is valuable in epitope specificity determination of a large panel of MAbs, where the pair-wise combination matrix requires a large number of assay cycles. Other examples of specificity assays include identification of binding sites (Jokiranta et al., 2000), monitoring steps involved in complex formation (Schuster et al., 1993; Clark et al., 2001; Thai and Ogata, 2005), and assessing cofactor requirements for an interaction to occur (e.g., Ca2+ ; Schlattner et al., 2001). SPR technology is increasingly being used to monitor immune responses either to an immunotherapeutic protein, vaccine, or even whole virus in research and preclinical environments (Alaedini and Latov, 2001; Abad et al., 2002; Swanson, 2003; Rini et al., 2005; Thorpe and Swanson, 2005). One simple assay requiring small sample volumes can provide information regarding antibody isotype, and active or relative concentration in serum. Another advantage of using Biacore for immunogenicity studies is the ability of the technique to detect both high- and lowaffinity antibodies, whereas traditional endpoint assays (e.g., ELISA) often fail to detect

fast-dissociating antibodies (Swanson, 2003; Thorpe and Swanson, 2005).

Concentration Analysis Most chemical and spectroscopic methods used to quantify proteins measure total protein content, do not distinguish active from inactive molecules, and cannot be used with unpurified samples. Since SPR is a noninvasive technology (i.e., no light penetrates the sample), it is possible to measure sub-femtomole amounts of analyte bound to the sensor chip surface from complex matrices such as food products, serum, and cell extracts, to name a few (Nelson et al., 2000). Instrument automation decreases operator involvement thereby leading to highly reproducible measurements. Various assay formats are possible. For analytes >5000 Da, a direct binding assay format can be used with the optional response enhancement from a secondary detecting molecule. Enhancement not only increases the dynamic range of the assay but can also improve assay specificity. An enhancement step can also be used to determine the isotype of antibodies in serum that are generated in response to a protein therapeutic or vaccine. Unlike many other immunoassays, concentration analysis with Biacore requires no separation and washing steps and, since binding responses are monitored continuously, it is possible to quantify fast-dissociating, low-affinity interactants. The point at which analyte concentration is measured can be chosen, giving flexibility in the assay design, which is not available with standard end-point assays. Inhibition or competition assay formats are well suited for quantification of low-molecular-weight molecules (30 µl/min. It is important to note that mass

transport is a well-understood physical property of the system and partial mass transport limitations can be accounted for during data analysis (Myszka et al., 1998; Karlsson, 1999). Reliable detailed kinetic analysis requires data from four to six analyte concentrations, spanning the range of 0.1 to 10 times KD . Analyte concentrations must be accurately known to determine correct association rate constants. Analytes should be in the same buffer as the continuous flow buffer to minimize bulk refractive index differences that can lead to low signal-to-noise ratios. This is often most easily achieved through dilution of a concentrated analyte stock into running buffer. Samples containing high refractive index solutions, such as high salt, glycerol, or DMSO, should either be exchanged into the running buffer or the concentration of the high refractive index component should be matched precisely in the continuous flow running buffer. Kinetic assays should include a series of start-up cycles using buffer as analyte to equilibrate the surface as well as cycles with zero concentration of analytes as part of the concentration series for the purposes of double-referencing (Myszka, 1999) during data analysis. Although it is not necessary to reach equilibrium, it is recommended that the association times used be sufficient for at least one analyte concentration to reach steady state. To accurately determine dissociation rate constants, a measurable decrease in signal should occur during the dissociation period. If possible, kinetic experiments should be designed such that the data are described by the simplest interaction model. For example, in the case of an antibody-antigen interaction, the antibody should be immobilized or captured on the surface and the antigen used as analyte to avoid avidity effects resulting from the bivalent nature of the antibody. Avidity refers to the ability of an antibody to bind to two antigen molecules simultaneously, thus, the antibody may not dissociate from the antigen immobilized on the chip surface before binding another antigen molecule. Avidity effects will slow down the dissociation rate yielding enhanced affinity values compared to those measured from a 1:1 interaction. It is also important that both the ligand and analyte be as homogeneous as possible. Impurities from partially purified material can complicate the results by affecting the accurate determination of analyte concentration or introducing nonspecific binding. Lastly, analyte should be injected over both a reference surface and an active ligand surface. Reference surfaces are necessary to subtract bulk



refractive index responses from the specific binding signal as well as to ensure that there is no nonspecific interaction with the sensor chip surface. Several excellent reviews on the topic provide detailed guidelines on experimental setup and interpretation of results (Karlsson and Fält, 1997; Myszka, 1999; Myszka, 2000; Rich and Myszka, 2001; Van Regenmortel, 2003; Karlsson and Larsson, 2004).

Thermodynamics By studying temperature dependence of rate and affinity constants it is possible to determine thermodynamic parameters for a binding interaction. Not only can the equilibrium values for changes in enthalpy (H) and entropy (S) associated with complex formation be determined, but transition state energetics can also be evaluated (Roos et al., 1998). The van’t Hoff equation relates affinity (KD ) to the free energy difference (G◦ ) between the associated and unassociated states of the interacting molecules where G◦ = RT ln KD , where R is the gas constant and T is the absolute temperature. Therefore, thermodynamic parameters can be determined by measuring KD at several temperatures and plotting ln KD against inverse temperature. H◦ and S◦ may then be calculated from the slope and y-axis intercept of the resulting straight line, respectively, since ln KD = H◦ /RT – S◦ /R. In some cases, entropy may be determined more accurately by subtracting H◦ from G◦ . This approach provides complementary information to thermodynamic parameters determined through microcalorimetry since in Biacore only the direct binding is measured, whereas microcalorimetry measures all components, including heats of mixing, dilution, and hydration effects. ln KD will be a linear function of (1/T) if both H and S are independent of temperature (T); however, H and S may not always be independent of temperature. In such cases, taking changes in heat capacity (Cp◦ ) into account and nonlinear fitting of the data using the integrated van’t Hoff equation may be used to determine thermodynamic parameters (Shuman et al., 2004; Dhalluin et al., 2005). The Eyring equation can be used with either the association rate constant (ka ) or the dissociation rate constant (kd ) to calculate the transition state energetics (H◦‡ and S◦‡ ) for a given complex. Thus, a plot of ln ka or ln kd against (1/T) will have a slope of –H◦‡ /R and an intercept of S◦‡ /R – ln(h/kB ), where h is the Planck constant and kB is the Boltzmann constant.

Biacore has been used to investigate the thermodynamics of proteins interacting with small molecules (Deinum et al., 2002; Day and Myszka, 2003; Shuman et al., 2004) and with other proteins (Zeder-Lutz et al., 1997; McFarland and Strong, 2003; Baerga-Ortiz et al., 2004; Dhalluin et al., 2005). Some benefits of using Biacore to determine thermodynamic parameters include: low sample concentration and consumption; affinities up to 1011 M−1 can be measured; there is no need to wait for equilibrium since KD can be determined from kd ka . Transition state analysis can provide additional mechanistic information about an interaction, revealing if the interaction is driven by entropic or enthalpic forces.

PROTEIN INTERACTION ANALYSIS This section highlights examples of studies where Biacore technology has been used to investigate interactions between proteins and a variety of biomolecules, whole cells, and viruses. This summary represents a small sample of the over 4000 peer-reviewed publications that have made use of Biacore for investigating many disciplines within the life sciences. A searchable database containing the abstracts of these publications is available on the Biacore Website (http://www.biacore. com/lifesciences/index.html).

Protein-Protein Interactions The largest number of publications citing Biacore involve characterization of proteinprotein interactions. An example of the versatility of Biacore in protein studies is illustrated in the work of Seet and co-workers (2003). In this study, the functionality and selectivity of a putative chemokine binding protein (CBP) from a poxvirus was assayed by immobilizing the recombinant protein and screening a panel of cytokines and chemokines. The affinity and kinetic rate constants for the interaction between the CBP and several chemokines were measured. Lastly, residues important for the interaction between the chemokine MCP1 and the CBP were identified by comparing the binding profiles of MCP-1 substitution mutants to that of the wild-type protein (Seet et al., 2003). Modifications of proteins, e.g., with biotin, fluorescent labels, or polyethyleneglycol, may introduce steric hindrance and alter their binding characteristics toward their interaction partners (Peter et al., 2003). This can



Supplement 45

be especially detrimental in the case of therapeutic proteins. Since detection in Biacore is based on mass changes at the sensor chip surface, it is often used to compare the binding of native and modified proteins. The kinetic rate and affinity constants for pegylated forms of interferonα 2a to the extracellular domain of a cell membrane receptor were compared to those of the unmodified protein (Dhalluin et al., 2005). It was found that increased degrees of pegylation resulted in lower affinities mainly due to reduction in association rate. In vivo, the decreased affinity was compensated for by an increased half-life of the protein due to increased protection from proteolytic degradation and a reduction in renal clearance. A recent development is the combination of SPR with mass spectrometry (SPR-MS). While Biacore systems provide information regarding the kinetics of an interaction and the analyte concentration, they do not permit direct identification of an unknown binding partner. Thus, the possibility of combining both technologies to provide protein identification on the basis of functional binding criteria is quite powerful. This method has been implemented to identify proteins captured from cytosolic extract (Lopez et al., 2003), cell lysates (Gilligan et al., 2002; Natsume et al., 2002), tissue extracts (Kikuchi et al., 2003; Zhukov et al., 2004), and plant extracts (Borch and Roepstorff, 2004). The small sample volume required and the high sensitivity of SPR results in a small amount of material that is recovered from a single SPR experiment for MS analysis. This can be addressed through the use of MS systems with high sensitivity, multiple binding and recovery cycles or by the use of an external flow cell with an expanded area available as part of the Biacore 3000 analyte recovery package. Biacore 3000 has been updated to recover bound analyte automatically in a 2-µl volume that can be deposited directly onto MALDI targets or delivered to a recovery vial for enzymatic digestion.

Protein Interactions with Peptides and Small Molecules


Improvements in instrument sensitivity, experimental protocols, and data analysis software have allowed for routine study of the affinity and kinetics of interactions between proteins and small analytes (100 Da) in

direct binding assays. Not only have such characterizations provided valuable information regarding the interaction of drug candidates with protein targets, but also details on their interaction with circulating plasma proteins such as human serum albumin (HSA) and α1-acid glycoprotein (AGP) (Frostell-Karlsson et al., 2000; Day et al., 2002). In the case of enzymes, the data for a direct binding assay of a potential inhibitor to the enzyme may be easier to interpret than that from activity assays since interactions can be studied in the absence of substrate and without restricting the assay conditions to those best suited for catalysis of a specific substrate (Backman and Danielson, 2003). Different experimental conditions can be used to monitor their effect on the kinetics and affinity of binding interactions, e.g., the influence of pH or temperature. Binding interactions between small molecules or peptides and targets ranging from kinases (Casper et al., 2004; Nordin et al., 2005) to proteases (Karlsson et al., 2000; Backman and Danielson, 2003; Gossas and Danielson, 2003) and other enzymes (Boehm et al., 2000; Day et al., 2002), hormone receptors (Rich et al., 2002), antibiotics (Tseng and Chu, 2005), and proteins (Fisher et al., 2003; Huber, 2005; Yoshitani et al., 2005) have been characterized.

Protein Interactions with Lipids, Membranes, and Membrane-Bound Proteins The impact of Biacore technology on studies with lipids, membranes, and integral membrane proteins has been growing steadily since the introduction of Sensor Chips HPA and L1, which were designed for the creation of lipid monolayers and bilayers, respectively, on the sensor surface. In a well-designed study, de Haro et al. (2004) investigated the mechanism whereby the v-SNARE synaptobrevin, VAMP, interacts with lipids and Ca2+ / calmodulin as part of the fusion of secretory granules in the plasma membrane of neurosecretory cells. The binding of recombinant VAMP to a bilayer of acidic phopholipids created on the surface of Sensor Chip L1 was abrogated in the presence of Ca2+ / calmodulin or mutations in the calmodulin binding domain, suggesting that this region of the protein is involved in interactions with lipids and calmodulin. Antibodies raised against the putative lipid/calmodulin binding domain were able to bind to oriented VAMP proteoliposomes reconstituted with VAMP containing mutations in this domain but not to proteoliposomes reconstituted with wild-type VAMP.



These results, together with those of control experiments, demonstrated that membraneanchored VAMP is able to undergo cis-lipid interactions with acidic lipid headgroups in the outer leaflet of the vesicle membrane that may be altered by Ca2+ / calmodulin (de Haro et al., 2004). An area currently in development is the design of experimental conditions to study interactions with G-protein-coupled receptors and integral membrane proteins. The concentration of these proteins within membranes may be too low to be able to detect interactions between low-molecular-weight analytes and immobilized membrane preparations. Enrichment of membrane proteins through appropriate detergent solubilization followed by immobilization on the sensor chip surface and by on-surface reconstitution of lipid bilayers is a promising approach that has been reported by several authors (Karlsson and Löf˚as, 2002; Stenlund et al., 2003; Navratilova et al., 2005).

eric repeats was developed by Maesawa et al. (2003). Addition of nucleotides to the 3 end of biotinylated oligonucleotides captured on sensor chip SA by telomerase, present in extracts from cancer cell lines and tumor samples, results in a post-injection baseline increase due to the increased mass on the surface that is proportional to the amount of telomerase present in the extract (Maesawa et al., 2003). This Biacore assay has significant advantages over current PCR-based methods in that the direct on-surface measurement of elongation rates avoids post-PCR procedures and potential PCR artifacts. Additionally, the method is label-free and automated, with a one-step analysis that provides a rapid, reproducible readout with real-time monitoring of the process. In contrast to PCR-based methods, the telomeric repeat elongation assay enables quantitative evaluation of telomerase activity from a range of sources, including both cancer and normal somatic cells.

Protein Interactions with Nucleic Acids Interactions between proteins and nucleic acids play an extensive role in biological regulation. Although affinity data obtained under equilibrium conditions (such as electrophoretic mobility shift assays) is valuable in many instances, it fails to address the dynamic nature of protein-nucleic acid interactions. The most common approach utilized to study these interactions with Biacore is to analyze the interaction between proteins in solution and biotinylated nucleic acids captured on the streptavidin surface of sensor chip SA. DNA and RNA oligonucleotides can be synthesized with a terminal biotin group with relative ease, thus enabling the creation of a homogeneous oriented surface. Katsamba et al. (2002) used Biacore technology to elucidate the binding mechanisms of the neuronspecific RNA-binding protein, HuD, and the spliceosomal protein, U1A, with their target RNAs. A review of this work, which also provides recommendations for RNA-protein interactions, was recently published (Katsamba et al., 2002). Tsoi and Yang (2002) used Biacore to investigate the kinetics of interaction between DNA polymerase β and a series of DNA substrates. Results indicated that the polymerase-bound, single-stranded and fully matched DNA-primer templates with similar affinity, but with different rates of association and dissociation, while mismatches in the primer sequence, significantly lowered the affinity. A novel assay to monitor telomerase activity by detecting on-surface elongation of oligonucleotides containing tandem telom-

Protein Interactions with Whole Cells and Viruses On the high end of the analyte molecular weight spectrum are whole cells and viruses. A number of SPR assays to detect and identify bacteria in a wide range of contaminated samples have been developed as an alternative to conventional microbiological and biochemical methods that are much lengthier. These assays have shown very good reproducibility and equal or better sensitivity than ELISA and PCR (Bokken et al., 2003; Leonard et al., 2005). Bacterial cells and viruses, unlike eukaryotic cells, are generally robust enough to withstand immobilization and the shear forces within the microfluidic system (Dubs et al., 1991; Abad et al., 2002; Medina, 2004). By measuring the ability of various compounds to reduce the binding of cell membrane proteins to an immobilized Salmonella typhimurium surface, Medina developed an SPR-based method for the rapid evaluation of compounds that may prevent bacterial attachment to poultry skin and carcasses with a vision to develop treatments to reduce pathogen contamination of poultry foods. A number of publications have reported the use of Biacore to detect interactions between eukaryotic cells and proteins or peptides (Quinn et al., 1997; Aggarwal et al., 2005). For example, Aggarwal and co-workers designed peptides to efficiently and selectively capture metastatic epithelial cancer cells from flowing blood. More detailed kinetic information provided by Biacore analysis of phage-displayed



Supplement 45

antibodies binding to the target antigen has also proved to be useful in guiding the selection of the highest affinity antibodies (Schier and Marks, 1996). For more than 15 years, Biacore systems have been helping scientists unravel the intricacies of protein interactions. Herein, only a small portion of this exciting technology has been highlighted and the breadth of applications it addresses. The release of the new SPR array platforms (Flexchip and Biacore A100) will add new dimensions to future developments within label-free protein interaction analysis research.

LITERATURE CITED Abad, L.W., Neumann, M., Tobias, L., ObenauerKutner, L., Jacobs, S., and Cullen, C. 2002. Development of a biosensor-based method for detection and isotyping of antibody responses to adenoviral-based gene therapy vectors. Anal. Biochem. 310:107-113. Aggarwal, S., Janssen, S., Wadkins, R.M., Harden, J.L., and Denmeade, S.R. 2005. A combinatorial approach to the selective capture of circulating malignant epithelial cells by peptide ligands. Biomaterials 26:6077-6086. Alaedini, A. and Latov, N. 2001. A surface plasmon resonance biosensor assay for measurement of anti-GM(1) antibodies in neuropathy. Neurology 56:855-860. Backman, D. and Danielson, U.H. 2003. Kinetic and mechanistic analysis of the association and dissociation of inhibitors interacting with secreted aspartic acid proteases 1 and 2 from Candida albicans. Biochim. Biophys. Acta 1646:184-195. Baerga-Ortiz, A., Bergqvist, S., Mandell, J.G., and Komives, E.A. 2004. Two different proteins that compete for binding to thrombin have opposite kinetic and thermodynamic profiles. Protein Sci. 13:166-176. Boehm, H., Boehringer, M., Bur, D., Gmuender, H., Huber, W., Klaus, W., Kostrewa, D., Kuehne, H., Luebbers, T., Meunier-Keller, N., and Mueller, F. 2000. Novel inhibitors of DNA gyrase: 3D structure based biased needle screening, hit validation by biophysical methods, and 3D guided optimization. A promising alternative to random screening. J. Med. Chem. 43:2664-2674. Bokken, G.C., Corbee, R.J., van Knapen, F., and Bergwerff, A.A. 2003. Immunochemical detection of Salmonella group B, D and E using an optical surface plasmon resonance biosensor. FEMS Microbiol. Lett. 222:75-82.



Clark, W.A., Jian, X., Chen, L., and Northup, J.K. 2001. Independent and synergistic interaction of retinal G-protein subunits with bovine rhodopsin measured by surface plasmon resonance. Biochem J. 358:389-397. Day, Y.S., Baird, C.L., Rich, R.L., and Myszka, D.G. 2002. Direct comparison of binding equilibrium, thermodynamic, and rate constants determined by surface- and solution-based biophysical methods. Protein Sci. 11:1017-1025. Day, Y.S.N. and Myszka, D. 2003. Characterizing a drug’s binding site on albumin. J. Pharm. Sci. 92:333-343. de Haro, L., Ferracci, G., Opi, S., Iborra, C., Quetglas, S., Miquelis, R., Leveque, C., and Seagar, M. 2004. Ca2+ / calmodulin transfers the membrane-proximal lipid-binding domain of the v-SNARE synaptobrevin from cis to trans bilayers. Proc. Natl. Acad. Sci. U.S.A. 101:15781583. Deinum, J., Gustavsson, L., Gyzander, E., KullmanMagnusson, M., Edstrom, A., and Karlsson, R. 2002. A thermodynamic characterization of the binding of thrombin inhibitors to human thrombin, combining biosensor technology, stoppedflow spectrophotometry, and microcalorimetry. Anal. Biochem. 300:152-162. Dhalluin, C., Ross, A., Huber, W., Gerber, P., Brugger, D., Gsell, B., and Senn, H. 2005. Structural, kinetic, and thermodynamic analysis of the binding of the 40 kDa PEG-interferonalpha2a and its individual positional isomers to the extracellular domain of the receptor IFNAR2. Bioconjugate Chem. 16:518-527. Dubs, M.-C., Altschuh, D., and Van Regenmortel, M.H.V. 1991. Interaction between viruses and monoclonal antibodies studied by surface plasmon resonance. Immunol. Lett. 31:59-64. ◦

Fägerstam, L.G., Frostell, A., Karlsson, R., Kullman, M., Larsson, A., Malmqvist, M., and Butt, H. 1990. Detection of antigen-antibody interactions by surface plasmon resonance. Application to epitope mapping. J. Mol. Recognit. 3:208-214. Fisher, R.D., Wang, B., Alam, S.L., Higginson, D.S., Robinson, H., Sundquist, W.I., and Hill, C.P. 2003. Structure and ubiquitin binding of the ubiquitin-interacting motif. J. Biol. Chem. 278:28976-28984. Frostell-Karlsson, A., Remaeus, A., Roos, H., Anderson, K., Borg, P., Hämäläinen, M., and Karlsson, R. 2000. Biosensor analysis of the interaction between immobilized human serum albumin and drug compounds for prediction of human serum albumin binding levels. J. Med. Chem. 43:1986-1992.

Borch, J. and Roepstorff, P. 2004. Screening for enzyme inhibitors by surface plasmon resonance combined with mass spectrometry. Anal. Chem. 76:5243-5248.

Gilligan, J.J., Schuck, P., and Yergey, A.L. 2002. Mass spectrometry after capture and smallvolume elution of analyte from a surface plasmon resonance biosensor. Anal. Chem. 74:20412047.

Casper, D., Bukhtiyarova, M., and Springman, E.B. 2004. A Biacore biosensor method for detailed kinetic binding analysis of small molecule inhibitors of p38alpha mitogen-activated protein kinase. Anal. Biochem. 325:126-136.

Gossas, T. and Daielson, U.H. 2003. Analysis of the pH-dependencies of the association and dissociation kinetics of HIV-1 protease inhibitors. J. Mol. Recognit. 16:203-212.


Huber W. 2005 A new strategy for improved secondary screening and lead optimization using high-resolution SPR characterization of compound-target interactions. J. Mol. Recognit. 18:273-281. Jokiranta, T.S., Hellwage, J., Koistinen, V., Zipfel, P.F., and Meri, S. 2000. Each of the three binding sites on complement factor H interacts with a distinct site on C3b. J. Biol. Chem. 275:2765727662. Karlsson, R. 1999. Affinity analysis of non-steadystate data obtained under mass transport limited conditions using Biacore technology. J. Mol. Recognit. 12:285-292. Karlsson, R. and Fält, A. 1997. Experimental design for kinetic analysis of protein protein interactions with surface plasmon resonance biosensors. J. Immunol. Methods 200:121-133. Karlsson, R., Kullman-Magnusson, M., Hämäläinen, M.D., Remaeus, A., Andersson, K., Borg, P., Gyzander, E., and Deinum, J. 2000. Biosensor analysis of drug-target interactions: Direct and competitive binding assays for investigation of interactions between thrombin and thrombin inhibitors. Anal. Biochem. 278:1-13. Karlsson, R. and Larsson A. 2004. Affinity measurement using surface plasmon resonance. Methods Mol. Biol. 248:389-415. Karlsson, O.P. and Löf˚as, S. 2002. Flow-mediated on-surface reconstitution of G-protein coupled receptors for applications in surface plasmon resonance biosensors. Anal. Biochem. 300:132138. Katsamba, P.S., Park, S., and Laird-Offringa, I.A. 2002. Kinetic studies of RNA-protein interactions using surface plasmon resonance. Methods 26:95-104. Kikuchi, J., Furukawa, Y., and Hayashi, N. 2003. Identification of novel p53-binding proteins by biomolecular interaction analysis combined with tandem mass spectrometry. Mol. Biotechnol. 23:203-212. Lieberg, B., Nylander, C., and Lundström, I. 1983. Surface plasmon resonance for gas detection and biosensing. Sensors Actuators 4:299-304. Leonard, P., Hearty, S., Wyatt, G., Quinn, J., and O’Kennedy, R. 2005. Development of a surface plasmon resonance-based immunoassay for Listeria monocytogenes. J. Food Prot. 68:728735. Lopez, F., Pichereaux, C., Burlet-Schiltz, O., Pradayrol, L., Monsarrat, B., and Esteve, J.P. 2003. Improved sensitivity of biomolecular interaction analysis mass spectrometry for the identification of interacting molecules. Proteomics 3:402-412. Maesawa, C., Inaba, T., Sato, H., Iijima, S., Ishida, K., Terashima, M., Sato, R., Suzuki, M., Yashima, A., Ogasawara, S., Oikawa, H., Sato, N., Saito, K., and Masuda, T. 2003. A rapid biosensor chip assay for measuring of telomerase activity using surface plasmon resonance. Nucl. Acids Res. 31:E4-4.

McFarland, B.J. and Strong, R.K. 2003. Thermodynamic analysis of degenerate recognition by the NKG2D immunoreceptor: Not induced fit but rigid adaptation. Immunity 19:803-812. Medina, M.B. 2004. Binding interaction studies of the immobilized Salmonella typhimurium with extracellular matrix and muscle proteins, and polysaccharides. Int. J. Food Microbiol. 93:6372. Myszka, D.G. 1999. Improving biosensor analysis. J. Mol. Recognit. 12:1-6. Myszka, D.G. 2000. Kinetic, equilibrium, and thermodynamic analysis of macromolecular interactions with Biacore. Methods Enzymol. 323:325340. Myszka, D.G., He, X., Dembo, M., and Morton, T.A. 1998. Extending the range of rate constants available from Biacore: Interpreting mass transport-influenced binding data. Biophys. J. 75:583-594. Natsume, T., Taoka, M., Manki, H., Kume, S., Isobe, T., and Mikoshiba, K. 2002. Rapid analysis of protein interactions: On-chip micropurification of recombinant protein expressed in Esherichia coli. Proteomics 2:12471253. Navratilova, I., Sodroski, J., and Myszka, D.G. 2005. Solubilization, stabilization, and purification of chemokine receptors using biosensor technology. Anal. Biochem. 339:71-81. Nelson, R.W., Nedelkov, D., and Tubbs, K.A. 2000. Biomolecular interaction analysis mass spectometry. BIA/MS can detect and characterize protiens in complex biological fluids at the lowto subfemtomole level. Anal. Chem. 72:404A411A. Nordin, H., Jungnelius, M., Karlsson, R., and Karlsson, O.P. 2005. Kinetic studies of small molecule interactions with protein kinases using biosensor technology. Anal. Biochem. 340:359368. Peter, J.C., Briand, J.P., and Hoebeke, J. 2003. How biotinylation can interfere with recognition: A surface plasmon resonance study of peptideantibody interactions. J. Immunol. Methods 274:149-158. Quinn, J.G., O’Kennedy, R., Smyth, M., Moulds, J., and Frame, T. 1997. Detection of blood group antigens utilising immobilised antibodies and surface plasmon resonance. J. Immunol. Methods 206:87-96. Rich, R.L., Hoth, L.R., Geoghegan, K.F., Brown, T.A., LeMotte, P.K., Simons, S.P., Hensley, P., and Myszka, D.G. 2002. Kinetic analysis of estrogen receptor/ligand interactions. Proc. Natl. Acad. Sci. U.S.A. 99:8562-8567. Rich, R.L. and Myszka, D.G. 2001. Survey of the year 2000 commercial optical biosensor literature. J. Mol. Recognit. 14:273-294. Rini, B., Wadhwa, M., Bird, C., Small, E., GainesDas, R., and Thorpe, R. 2005. Kinetics of development and characteristics of antibodies induced in cancer patients against yeast expressed



Supplement 45

rDNA derived granulocyte macrophage colony stimulating factor (GM-CSF). Cytokine 29:5666. Roos, H., Karlsson, R., Nilshans, H., and Persson, A. 1998. Thermodynamic analysis of protein interactions with biosensor technology. J. Mol. Recognit. 11:204-210. Schier, R. and Marks, J.D. 1996. Efficient in vitro affinity maturation of phage antibodies using Biacore guided selections. Hum. Antibody Hybridomas 7:97-105. Schlattner, U., Dolder, M., Wallimann, T., and Tokarska-Schlattner, M. 2001. Mitochondrial creatine kinase and mitochondrial outer membrane porin show a direct interaction that is modulated by calcium. J. Biol. Chem. 276:4802748030. Schuster, S.C., Swanson, R.V., Alex, L.A., Bourret, R.B., and Simon, M.I. 1993. Assembly and function of a quaternary signal transduction complex monitored by surface plasmon resonance. Nature 365:343-346. Seet, B.T., McCaughan, C.A., Handel, T.M., Mercer, A., Brunetti, C., McFadden, G., and Fleming, S.B. 2003. Analysis of an orf virus chemokine-binding protein: Shifting ligand specificities among a family of poxvirus viroceptors. Proc. Natl. Acad. Sci. U.S.A 100:1513715142. Shuman, F.S., Hämäläinen, M.D., and Danielson, U.H. 2004. Kinetic and thermodynamic characterization of HIV-1 protease inhibitors. J. Mol. Recognit. 17:106-119. Stenlund, P., Babcock, G.J., Sodroski, J., and Myszka, D.G. 2003. Capture and reconstitution of G protein–coupled receptors on a biosensor surface. Anal. Biochem. 316:243-250. Swanson, S.J. 2003. New technologies for the detection of antibodies to therapeutic proteins. Dev Biol (Basel). 112:127-133. Thai, C.T. and Ogata, R.T. 2005. Recombinant C345C and factor I modules of complement components C5 and C7 inhibit C7 incorporation into the complement membrane attack complex. J. Immunol. 174:6227-6232. Thorpe, R. and Swanson, S.J. 2005. Assays for detecting and diagnosing antibody-mediated pure red cell aplasia (PRCA): An assessment of available procedures. Nephrol. Dial. Transplant. 20(Suppl 4):iv16-22. Tseng, M.C. and Chu, Y.H. 2005. Using surface plasmon resonance to directly identify molecules in a tripeptide library that bind tightly to a vancomycin chip. Anal. Biochem. 336:172177. Tsoi, P.Y. and Yang, M. 2002. Kinetic study of various binding modes between human DNA polymerase beta and different DNA substrates by surface-plasmon-resonance biosensor. Biochem. J. 361:317-325.


Yoshitani, N., Satou, K., Saito, K., Suzuki, S., Hatanaka, H., Seki, M., Shinozaki, K., Hirota, H., and Yokoyama, S. 2005. A structure-based strategy for discovery of small ligands binding to functionally unknown proteins: Combination of in silico screening and surface plasmon resonance measurements. Proteomics 5:1472-1480. Young, P.J., Day, P.M., Zhou, J., Androphy, E.J., Morris, G.E., and Lorson, C.L. 2002. A direct interaction between the survival motor neuron protein and p53 and its relationship to spinal muscular atrophy. J. Biol. Chem. 277:2852-2859. Zeder-Lutz, G., Zuber, E., Witz, J., and Van Regenmortel, M.H.V. 1997. Thermodynamic analysis of antigen-antibody binding using biosensor measurements at different temperatures. Anal. Biochem. 246:123-132. Zhukov, A., Schurenberg, M., Jansson, O., Areskoug, D., and Buijs, J. 2004. Integration of surface plasmon resonance with mass spectrometry: Automated ligand fishing and sample preparation for MALDI MS Using a Biacore 3000 biosensor. J. Biomol. Tech. 15:112-119.

KEY REFERENCES Dhalluin et al., 2005. See above. Excellent example of the importance of kinetic and thermodynamic information in the development of a biotherapeutic. Karlsson and Larsson. 2004. See above. A practical guide to proper experimental design when using Biacore for affinity measurements. Nordin et al., 2005. See above. Excellent example of the kinetic characterization of small molecule inhibitors of a protein kinase. Shuman et al., 2004. See above. Excellent example of kinetic and thermodynamic characterization of small molecule inhibitors of a protease.

INTERNET RESOURCES http://www.biacore.com/lifesciences/index.html Details and animations of SPR technology, Biacore instruments, application notes, a reference database, guidelines to common immobilization chemistries, an immobilization and regeneration database, self-training tools, and course listings are available. http://www.piercenet.com Maleimide coupling reagents are available from Pierce Chemical.

Contributed by Laure Jason-Moller, Michael Murphy, and JoAnne Bruno Biacore, Inc. Piscataway, New Jersey

Van Regenmortel, M.H. 2003. Improving the quality of Biacore-based affinity measurements. Dev Biol (Basel) 112:141-151.



Using Biacore to Measure the Binding Kinetics of an Antibody-Antigen Interaction

UNIT 19.14

The optical phenomenon of surface plasmon resonance (SPR) used by Biacore systems enables the detection and measurement of protein-protein interactions in real time, without the use of labels. Biacore systems are widely used for characterizing the interactions of proteins with other proteins, peptides, nucleic acids, lipids, and small molecules. Over 4000 references using Biacore to characterize protein interactions can be found in the literature. When Biacore is used to measure protein interactions, one of the interactants is immobilized onto a sensor chip surface and the other interactant is passed over that surface in solution via an integrated microfluidic flow system. The immobilized interactant is referred to as the ligand, and the injected interactant in solution is referred to as the analyte. Binding responses are measured in resonance units (RU) and are proportional to the molecular mass on the sensor chip surface and, consequently, to the number of molecules on the surface. A sensorgram is a plot of the binding response in resonance units versus time in seconds, which is displayed and recorded as a change in mass occurs on the sensor chip surface (Fig. 19.14.1). The sensorgram provides essentially two kinds of information that are relevant to different types of applications: (1) the rate of interaction (association, dissociation, or both), which provides information on kinetic rate constants and analyte concentration; and (2) the binding level, which provides information on affinity constants and can be used for qualitative or semi-quantitative applications.

Figure 19.14.1 The sensorgram is a plot of response in resonance units (RU) versus time in seconds, which is presented continuously in real time. Upon injection of an analyte, if a binding interaction occurs, then an increase in mass occurs on the sensor chip surface and association is measured. At the end of the analyte injection, as the complex decays, a decrease in mass occurs and dissociation is measured. Association and dissociation are measured as changes in response. Contributed by Michael Murphy, Laure Jason-Moller, and JoAnne Bruno Current Protocols in Protein Science (2006) 19.14.1-19.14.17 C 2006 by John Wiley & Sons, Inc. Copyright



In this unit, the application of Biacore technology to measuring a protein-protein interaction is described using an antibody and its antigen as an example. The affinity of the antibody for its antigen is determined by measuring the binding kinetics of the interaction. The protocols are divided into three major steps that are required for measuring binding kinetics using Biacore: (1) surface preparation (see Basic Protocols 1 and 2); (2) assay development (see Basic Protocol 3); and (3) kinetic analysis (see Basic Protocol 4). Note that the materials used in these protocols are available from Biacore and that the protocols are intended for use with Biacore 3000, Biacore 2000, or Biacore T100 systems. The principles described here apply to use of other Biacore systems, however, the software interface and system configuration may differ from the provided protocols. BASIC PROTOCOL 1

IDENTIFICATION OF OPTIMAL IMMOBILIZATION pH (pH SCOUTING) For most chemical coupling methods, a mechanism for attraction of the ligand to the carboxymethylated dextran surface matrix on the sensor chip is important for the efficient immobilization of proteins. The mechanism employed is an electrostatic attraction between negative charges on the carboxymethylated dextran matrix and positive charges on the protein ligand, which is referred to as preconcentration. The preconcentration strategy enables efficient immobilization from relatively dilute solutions of ligand (10 to 100 µg/ml). The protein ligand is diluted into a low-ionic-strength immobilization buffer at a pH below its isoelectric point (pI) to generate a net positive charge on the protein. The carboxymethylated dextran matrix carries a net negative charge at pH values >3.5. The diluted ligand is injected over the sensor chip surface where it will accumulate if positively charged and, in the next step (see Basic Protocol 2), undergo chemical coupling to the carboxymethylated dextran. The goal of immobilization pH scouting is to identify an appropriate pH for the immobilization buffer to establish a set of controlled conditions that provide usable and reproducible immobilization levels so that both sample and sensor chip consumption is minimized.

Materials 200 ml HBS-P+ (see recipe) 1 mg/ml anti-β2µ-globulin antibody (Biacore) 10 mM sodium acetate, pH 4.0, 4.5, 5.0, and 5.5 50 mM NaOH Biacore 3000, 2000, or T100 system Sensor Chip CM5 0.8- or 1.5-ml polypropylene tubes with rubber caps (Biacore) Set up instrument 1. Undock the maintenance chip or previously used sensor chip and remove from the instrument by clicking on the Command tab and then selecting the Undock command. When undocked, the sensor chip should easily slide out of the instrument. Note that the sensor chip indicator light on the front of the instrument is lit steadily when a chip is docked, flashing when a chip is inserted but undocked, and unlit when no chip is inserted.

2. Insert and dock the new Sensor Chip CM5. Click the Command tab and then select the Dock command. 3. Insert the appropriate instrument buffer tubing into a container containing 200 ml HBS-P+. Biacore to Measure Binding Kinetics

The buffer should be filtered (using a 0.22-µm filter) and degassed prior to use.

4. Under the Tools tab in the control software, choose Working Tools and select Prime to exchange all of the solutions in the system with fresh HBS-P+ buffer.



5. Keep the instrument in Standby (or Continue) mode when not in use. Under the Tools tab in the control software, choose Start Standby. The Standby mode maintains a slow flow rate for up to 4 days. When not performing experiments, the instrument should be kept in Standby using water with the maintenance chip docked.

Set up sample and method 6. Dilute 1 mg/ml anti-β2µ-globulin antibody to 10 µg/ml in each of the four immobilization buffers using 0.8- or 1.5-ml tubes and cap the tubes with rubber caps. Add 2 µl of 1 mg/ml anti-β2µ-globulin to 198 µl of each of the following: 10 mM sodium acetate, pH 4.0 10 mM sodium acetate, pH 4.5 10 mM sodium acetate, pH 5.0 10 mM sodium acetate, pH 5.5 7. In the instrument control software under the File tab, open the Application Wizards, select Surface Preparation, and then choose Immobilization pH Scouting. 8. Enter the names of the solutions to be used for pH scouting by selecting from the list in the wizard. Select or enter:

acetate pH 4.0 acetate pH 4.5 acetate pH 5.0 acetate pH 5.5 If the needed solution is not already a selection in the list, then additional solutions can be added.

9. Select flow path 2. 10. Provide a name for the ligand (e.g., antiB2micro), and enter values for the injection parameters: 120 sec contact time (injection) and 10 µl/min flow rate. 11. Enter 50 mM NaOH for the wash solution that is used to remove the ligand from the surface after the pH scouting experiment. Enter values for the injection parameters for the wash: 30 sec contact time and 50 µl/min flow rate. Dispense 100 µl of 50 mM NaOH into a 1.5-ml tube. 12. Next, a diagram of the reagent rack is shown and colored spots in various positions indicate samples (the samples will include the tubes of diluted ligand and one tube of NaOH). Move the samples on the diagram via clicking, dragging, and dropping or by entering the position in the table. Place the solutions in the appropriate reagent rack, taking care to match the sample content with the indicated rack positions. 13. Save the wizard template and the result file, and start the method.

Evaluate pH scouting results 14. When the method has finished, view the overlay plot of the sensorgrams for each immobilization buffer, presented automatically by the control software (or the evaluation software, if using T100). If the ligand accumulates on the sensor chip surface, then a response is generated due to the increased mass on the surface. The magnitude and slope of the preconcentration response provide information about the ligand density that can be achieved. Compare the preconcentration response for each immobilization buffer and identify the mildest pH solution that will achieve the targeted level for immobilization (1200 RU, described also in Critical Parameters).



Supplement 45

Figure 19.14.2 The effect of pH on preconcentration of anti-β2µ-globulin antibody onto the surface of Sensor Chip CM5. Electrostatic attraction of anti-β2µ-globulin antibody to the chip surface occurs for each pH tested, therefore, any pH solution that was tested can be used. The pH 5.5 sodium acetate solution was chosen because amine coupling occurs more efficiently at a higher pH.

Although the preconcentration levels were higher and achieved more quickly for pH 4.0 and pH 4.5, the preconcentration levels for pH 5.0 and pH 5.5 were acceptable. Based on the pH scouting results, the pH 5.5 solution is optimal for preconcentration of the anti-β2µ-globulin antibody (Fig. 19.14.2). BASIC PROTOCOL 2

IMMOBILIZATION OF LIGAND TO SENSOR CHIP USING AMINE COUPLING Several coupling chemistries are available for covalent immobilization of the ligand to the sensor chip surface, such as amine, aldehyde, and thiol (see UNIT 19.13). Amine coupling is the most widely applicable approach for immobilization of ligands. There are three steps in amine coupling: (1) activation of the surface, (2) contact of the ligand with the activated surface, and (3) blocking of unreacted sites. After identification of the optimal pH for preconcentration of the ligand (see Basic Protocol 1), the active and reference surfaces are prepared on Sensor Chip CM5 using, in this example, amine-coupling chemistry. The immobilization surface preparation application wizard is used to create active and reference surfaces automatically. The active surface will contain the anti-β2µ-globulin antibody ligand, and the reference surface will contain no ligand, but is activated and blocked with the same chemistry as the active surface. Generally, 75% surface activity can be expected for ligands immobilized directly by amine coupling. Note that when using the Aim for Immobilized Level option in the application wizard, the control software performs an injection of ligand to test for preconcentration prior to activating the surface to ensure that the desired immobilization level can be achieved during amine coupling. If the preconcentration response is too low to achieve the target immobilization level, the run will be stopped.

Materials

Biacore to Measure Binding Kinetics


100 mM N-hydroxysuccinimide (NHS) 400 mM N-ethyl-N-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC) 10 µg/ml anti-β2µ-globulin antibody in 10 mM sodium acetate, pH 5.5 1 M ethanolamine 50 mM NaOH Current Protocols in Protein Science

Biacore 3000, 2000, or T100 system 0.8-ml tubes Sensor chip CM5 Set up sample and method 1. Under the File tab in the instrument control software, open the Application Wizards, choose Surface Preparation and then select Immobilization. Enter CM5 for Sensor Chip and Amine Coupling for the Immobilization Method. 2. Select Aim for Immobilized Level in the immobilization procedure or set-up window. 3. Select flow cell 1 as blank. 4. Enter the ligand name (e.g., antiB2micro) for flow cell 2 and enter a target level of 1200 RU. Target levels for ligand immobilization must be calculated using the rearranged Rmax equation, which is described in Critical Parameters under Immobilization levels.

5. Enter 50 mM NaOH for the wash solution. 6. Click on Next and the wizard will recommend solution volumes and vial positions. Dispense the indicated volume of each solution into 0.8-ml tubes. Place the solutions in the appropriate reagent rack positions taking care to match the sample content with the indicated rack positions. Check the sample positions and edit if necessary. 7. Save the immobilization wizard template and result files and start the immobilization.

Evaluate immobilization results 8. When the method has finished, view the table summarizing the results of the immobilization, which is presented automatically by the control software. Note the surface

Figure 19.14.3 Immobilization of anti-β2µ-globulin antibody to the chip surface. Each step and corresponding response is labeled at the top of the sensorgram. Each X on the response represents a report point, which provides a value of the response (averaged over a 5-sec window) at a given point. Report point values are assigned a name (such as baseline or binding level) and are recorded in a table. During the preconcentration test, the ligand is injected over the unmodified sensor chip surface to determine if the desired immobilization level can be achieved. During the activation step, the carboxymethyl groups on the dextran are converted into reactive esters, and then the ligand is injected in short pulses until the desired immobilization level is achieved. During the blocking step, uncrosslinked reactive esters are blocked with ethanolamine. The immobilization level or ligand density is the final response difference from baseline, denoted by RL .



Supplement 46

density (RL ), which is the amount of immobilized ligand given by the final response in resonance units (RU). View the sensorgrams from each flow cell. The sensorgram from the reference flow cell (Fc 1) shows a bulk refractive index response for the injection of EDC/NHS and ethanolamine. The sensorgram from the active flow cell (Fc 2) shows the preconcentration injection and wash, bulk refractive index change for injection of EDC/NHS, the response from the injection of ligand, and bulk refractive index change for ethanolamine (Fig. 19.14.3). BASIC PROTOCOL 3

REGENERATION SCOUTING: IDENTIFYING A SUITABLE BUFFER FOR SURFACE REGENERATION After the active and reference surfaces have been created (see Basic Protocols 1 and 2), the binding of analyte is tested using a concentration of analyte well above the equilibrium dissociation constant (KD ) to ensure that the interaction will be observed if it occurs. If the interaction is characterized by a slow dissociation rate (108 members) to be screened against a variety of targets quite rapidly. By having everything genetically encoded, the entire method can be performed at low cost in most laboratory settings with minimal equipment, accelerating the identification of potential lead inhibitors. In contrast, synthetic chemical libraries have to be purchased or prepared at great expense and have to be screened, usually through low-throughput approaches. These features make it very difficult for many laboratories to exploit the benefits of synthetic libraries for inhibitor identification. Furthermore, genetic selection has other benefits, such as the ability to translate affinity to function. Using this protocol to identify ribonucleotide reductase protein-protein interaction inhibitors, candidate inhibitors were identified that unexpectedly bound to the smaller mR2 subunit (Horswill et al., 2004). If the target was preselected for an affinity-based method, such as phage display, these inhibitors may not have been discovered. Benefits of cyclic peptide libraries The use of cyclic peptide libraries offers additional advantages over other types of genetically encoded libraries. Cyclic peptides, like peptide aptamers, are constrained into a fixed conformation, improving the binding of a candidate inhibitor over its linear counterpart. Additionally, during biochemical characterization, it might be challenging to purify significant quantities of a peptide aptamer, and there is the constant concern over the contribution of the protein scaffold to the inhibitory effect. With cyclic peptides, the scaffold is unnecessary, and large quantities can be synthesized in a straightforward manner with chemical methods (Horswill et al., 2004; Naumann et al., 2005), facilitating biochemical studies. Moreover, the functional epitope on a cyclic peptide is presented in a structural state amenable to further synthetic optimization (Hirschmann et al., 1998), yet another benefit that can aid their development as lead compounds for pharmaceutical studies.

Critical Parameters and Troubleshooting The success of identifying small-molecule inhibitors is entirely dependent on the proteinprotein interaction in question and the ability of the fusion protein to form a complex that

behaves as a repressor to inhibit transcription of the reporter genes. If the protein target is eukaryotic in origin, post-translational modifications will not be performed in E. coli, and this could affect the level of interactions. Additionally, some proteins are deleterious to E. coli, and this toxicity can mask repression. Clearly, considerations should be made as to whether this protocol is the best suited to the target in question. One of the most common problems is poor repression of the reporter construct by the DNA-binding fusions to the target gene(s). Often this occurs with proteins that have a weak interaction or are poorly expressed in E. coli. If this is known to be an issue, it might suggest the proteins in question are not fully compatible with this system. For those not experienced with bacterial two-hybrid systems, the ones based on bacteriophage repressors tend to have smaller fold differences compared to yeast systems (Ladant and Karimova, 2000). Even if the repression level is only 3- to 4-fold, based on β-galactosidase assays, inhibitors can be identified with this protocol. If expected repression of β-galactosidase is quite poor or not observed at all, several factors should be considered. Sometimes it can be helpful to construct additional fusion constructs with a longer fragment or with the addition of a flexible peptide linker; in the case of heterodimers, the fragments can be swapped between the two DNA-binding partners (Fig. 19.15.4). Make sure a full range of IPTG concentrations is explored to produce adequate levels of fusion proteins. Although it takes time to build the constructs, integration of the gene(s) onto the chromosome can often bypass repressionrelated problems (Horswill et al., 2004). Some protein fusions perform better as singlecopy integrants rather than from multi-copy plasmids. For the genetic selection conditions, the most important step is the identification of media conditions where protein-protein interactions inhibitors will allow growth over background. If conditions are too stringent, positive candidates could be missed in the selection. Additionally, high-level expression of the fusion proteins can mask the effect of potent candidates by titrating out the inhibitor population. Using the minimum level of IPTG for fusion protein expression should help avoid this potential problem. The integration of the fusion constructs also helps maintain control over the expression level. For challenging targets, sometimes it can be helpful to perform the genetic selection at both low stringency



Supplement 46

(low 3-AT and kanamycin levels) and high stringency (high 3-AT and kanamycin levels), which can increase the pool of positive candidates available for screening.

Anticipated Results In a successful genetic selection, anywhere from 50 to >300 colonies will arise on library plates for each set of 108 transformants. Of these candidates, usually 5% to 20% pass the arabinose response and retransformation tests, while the rest are probably false positives. Further filtering through growth ranking will weed out the candidate pool, often to only 1% to 5% of the original number. For example, in a case study on disrupting the ribonucleotide reductase complex, 108 total transformants of a C+5 cyclic peptide library were subjected to genetic selection as described in this unit. Of 262 candidate inhibitors processed, 24 passed the arabinose response and retransformation tests. After growth ranking, 8 candidates were selected for biochemical studies. Under these optimal conditions, a library of 108 inhibitors was pared down to 8 promising candidates through a simple series of genetic selection and processing steps (Horswill et al., 2004). Similar results have been obtained with other homodimeric and heterodimeric complexes, suggesting this example is representative.

Time Considerations The amount of time it takes to proceed from targets of interest to isolating protein-protein interaction inhibitors can vary widely when using the methods outlined in this unit. The entire procedure, going from choice of protein target to inhibitor identification, should be expected to take at least two months. Although this is a lengthy protocol, most of the individual steps are not time or labor intensive, giving the researcher flexibility to perform the project in parallel with other work. Additionally, the bulk of the time is spent on the construction of the selection strain and the determination of media conditions. Once all the conditions have been optimized, additional rounds of inhibitor selections can proceed quite quickly, allowing rapid panning of new or focused inhibitor libraries.

Literature Cited Small-Molecule Modulators of Protein Interactions

Abel-Santos, E., Scott, C.P. and Benkovic, S.J. 2003. Use of inteins for the in vivo production of stable cyclic peptide libraries in E. coli. Methods Mol. Biol. 205:281-294.

Brennan, M.B. and Struhl, K. 1980. Mechanisms of increasing expression of a yeast gene in Escherichia coli. J. Mol. Biol. 136:333-338. Clackson, T. and Wells, J.A. 1995. A hot spot of binding energy in a hormone-receptor interface. Science 267:383-386. Cochran, A.G. 2000. Antagonists of protein-protein interactions. Chem. Biol. 7:R85-94. Di Lallo, G., Castagnoli, L., Ghelardini, P., and Paolozzi, L. 2001. A two-hybrid system based on chimeric operator recognition for studying protein homo/heterodimerization in Escherichia coli. Microbiology 147:1651-1656. Geyer, C.R., Colman-Lerner, A., and Brent, R. 1999. “Mutagenesis” by peptide aptamers identifies genetic network members and pathway connections. Proc. Natl. Acad. Sci. U.S.A. 96:8567-8572. Goryshin, I.Y., Jendrisak, J., Hoffman, L.M., Meis, R., and Reznikoff, W.S. 2000. Insertional transposon mutagenesis by electroporation of released Tn5 transposition complexes. Nat. Biotechnol. 18:97-100. Grigoriev, A. 2003. On the number of proteinprotein interactions in the yeast proteome. Nucleic Acids Res. 31:4157-4161. Haldimann, A. and Wanner, B.L. 2001. Conditional-replication, integration, excision, and retrieval plasmid-host systems for gene structure-function studies of bacteria. J. Bacteriol. 183:6384-6393. Hirschmann, R., Hynes, J., Jr., Cichy-Knight, M.A., van Rijn, R.D., Sprengler, P.A., Spoors, P.G., Shakespeare, W.C., Pietranico-Cole, S., Barbosa, J., Liu, J., Yao, W., Rohrer, S., and Smith, A.B., 3rd. 1998. Modulation of receptor and receptor subtype affinities using diastereomeric and enantiomeric monosaccharide scaffolds as a means to structural and biological diversity. A new route to ether synthesis. J. Med. Chem. 41:1382-1391. Hoppe-Seyler, F., Crnkovic-Mertens, I., Denk, C., Fitscher, B.A., Klevenz, B., Tomai, E., and Butz, K. 2001. Peptide aptamers: new tools to study protein interactions. J. Steroid. Biochem. Mol. Biol. 78:105-111. Horswill, A.R., Savinov, S.N., and Benkovic, S.J. 2004. A systematic method for identifying small-molecule modulators of proteinprotein interactions. Proc. Natl. Acad. Sci. U.S.A 101:15591-15596. Huang, J. and Schreiber, S.L. 1997. A yeast genetic system for selecting small molecule inhibitors of protein-protein interactions in nanodroplets. Proc. Natl. Acad. Sci. U.S.A. 94:13396-13401. Joung, J.K., Ramm, E.I., and Pabo, C.O. 2000. A bacterial two-hybrid selection system for studying protein-DNA and protein-protein interactions. Proc. Natl. Acad. Sci. U.S.A. 97:73827387. Khlebnikov, A., Datsenko, K.A., Skaug, T., Wanner, B.L., and Keasling, J.D. 2001.



Homogeneous expression of the P(BAD) promoter in Escherichia coli by constitutive expression of the low-affinity high-capacity AraE transporter. Microbiology 147:3241-3247. Ladant, D. and Karimova, G. 2000. Genetic systems for analyzing protein-protein interactions in bacteria. Res. Microbiol. 151:711-720. Leanna, C.A. and Hannink, M. 1996. The reverse two-hybrid system: a genetic scheme for selection against specific protein/protein interactions. Nucleic Acids Res. 24:3341-3347. Marcotte, E.M., Pellegrini, M., Ng, H.L., Rice, D.W., Yeates, T.O., and Eisenberg, D. 1999. Detecting protein function and protein-protein interactions from genome sequences. Science 285:751-753. Miller, J. 1992. Procedures for working with lac. In A Short Course in Bacterial Genetics, pp. 71-80. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. Naumann, T.A., Savinov, S.N., and Benkovic, S.J. 2005. Engineering an affinity tag for genetically encoded cyclic peptides. Biotechnol. Bioeng. 92:820-830. Norman, T.C., Smith, D.L., Sorger, P.K., Drees, B.L., O’Rourke, S.M., Hughes, T.R., Roberts, C.J. Friend, S.H., Fields, S., and Murray, A.W. 1999. Genetic selection of peptide inhibitors of biological pathways. Science 285:591-595.

the biosynthesis of backbone cyclic peptide libraries. Chem. Biol. 8:801-815. Seidman, C.E., Struhl, K., and Sheen, J. 2002. Introduction of plasmid DNA into cells. In Short Protocols in Molecular Biology, 5th ed. (F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, and K. Struhl, eds.) pp. 1-29 to 1-31. John Wiley & Sons, New York. Serebriiskii, I.G., Mitina, O., Pugacheva, E.N., Benevolenskaya, E., Kotova, E., Toby, G.G., Khazak, V., Kaelin, W.G., Chernoff, J., and Golemis, E.A. 2002. Detection of peptides, proteins, and drugs that selectively interact with protein targets. Genome Res. 12:1785-1791. Tavassoli, A. and Benkovic, S.J. 2005. Genetically selected cyclic-peptide inhibitors of AICAR transformylase homodimerization. Angew. Chem. Int. Ed. Engl. 44:2760-2763. Toogood, P.L. 2002. Inhibition of protein-protein association by small molecules: approaches and progress. J. Med. Chem. 45:1543-1558. Vidal, M., Brachmann, R.K., Fattaey, A., Harlow, E., and Boeke, J.D. 1996. Reverse two-hybrid and one-hybrid systems to detect dissociation of protein-protein and DNA-protein interactions. Proc. Natl. Acad. Sci. U.S.A. 93:10315-10320. Vidal, M. and Endoh, H. 1999. Prospects for drug screening using the reverse two-hybrid system. Trends Biotechnol. 17:374-381.

Park, S.H. and Raines, R.T. 2000. Genetic selection for dissociative inhibitors of designated proteinprotein interactions. Nat. Biotechnol. 18:847851.

Walker, J.R., Roth, J.R., and Altman, E. 2001. An in vivo study of novel bioactive peptides that inhibit the growth of Escherichia coli. J. Pept. Res. 58:380-388.

Ramani, A.K., Bunescu, R.C., Mooney, R.J., and Marcotte, E.M. 2005. Consolidating the set of known human protein-protein interactions in preparation for large-scale mapping of the human interactome. Genome Biol. 6:R40.

Contributed by Alexander R. Horswill University of Iowa Iowa City, Iowa

Scott, C.P., Abel-Santos, E., Wall, M., Wahnon, D.C., and Benkovic, S.J. 1999. Production of cyclic peptides and proteins in vivo. Proc. Natl. Acad. Sci. U.S.A. 96:13638-13643.

Stephen J. Benkovic The Pennsylvania State University University Park, Pennsylvania

Scott, C.P., Abel-Santos, E., Jones, A.D., and Benkovic, S.J. 2001. Structural requirements for



Supplement 46

Determination of Protein Contacts by Chemical Cross-Linking With EDC and Mass Spectrometry Chemical cross-linking of proteins is an established method for obtaining rough structural information about protein complexes. By identifying the cross-linked residues of neighboring proteins, spatial alignments can be deduced and a rough three-dimensional model can be built. Similarly, in monomeric proteins, cross-links can reveal information about protein folding and domain orientation. In addition, analysis of the cross-linked products on SDS-PAGE gels can help partially distinguish intramolecular and intermolecular cross-links (see Anticipated Results).

UNIT 19.16

BASIC PROTOCOL

Mass spectral analysis of cross-linked proteins has emerged as a simple and efficient technique for the analysis of protein folding and protein-protein interactions. When a folded monomeric protein or a protein complex (multimeric protein) in a native conformation is chemically cross-linked, a covalent link between two amino acids is formed. The distance between these linked amino acids is defined by the three-dimensional folding of the protein and is “frozen” by the spacer length of the cross-linker. This distance is maintained during all subsequent purification and analysis stages. The more cross-links made and identified, the greater the number of conformational restraints that can be applied for low-resolution structural modeling. Currently many protein cross-linking reagents are commercially available and a large number of custom-synthesized reagents have been reported (Sinz, 2003; Trakselis et al., 2005). These include cross-linkers reactive to different amino acids and of various spacer lengths. The use of the cross-linking reagent 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride (EDC) is described in this unit as an example. The general rules discussed below for the use of EDC (e.g., the use of buffers nonreactive to the cross-linker; cross-linker to protein ratio) apply for all cross-linkers. These and other guidelines are also found in the supplier’s instructions. EDC is known as a “zero-length” cross-linker and introduces a chemical bond between a negatively charged side chain carboxyl group (Glu or Asn) and a positively charged side chain amino group (Lys). The close proximity of the reacting groups allows a very precise pinpointing of the sites of protein interactions. The yield of cross-linked protein from this type of experiment varies significantly, depending on the nature and folding state of the protein, the accessibility of the reactive amino acids and the concentration of the cross-linker. It is recommended to start with a small amount of material and perform trial reactions with varied amounts of cross-linker. The conditions below represent a good starting point. NOTE: Throughout the protocol acetonitrile is used at different concentrations. It may extract additives and plasticizers from the plastic tubes which will result in contaminating peaks in the mass spectra. Before the experiment, half-fill all tubes with 70% acetonitrile in 0.1% trifluoroacetic acid. Close the tubes. Shake or vortex vigorously for about 1 min. Discard wash. Rinse with water and allow to air dry.

Materials Purified protein(s) to be cross-linked, of a known concentration (6 to 8 nmol minimum) 0.5 M sodium N-morpholinoethanesulfonate (MES buffer), pH 6.5 (see recipe) Identification of Protein Interactions Contributed by Lyuben N. Marekov Current Protocols in Protein Science (2007) 19.16.1-19.16.8 C 2007 by John Wiley & Sons, Inc. Copyright


0.2 M [1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride] (EDC; Pierce) in 50 mM MES; prepare immediately before use 0.02 M sulfo-N-hydroxysuccinimide (Sulfo-NHS;Pierce) in 50 mM MES; prepare immediately before use. 0.5 M hydroxylamine hydrochloride 25 mM Tris·Cl, pH 8.0 (APPENDIX 2E) in acetonitrile:water (1:1) 100% acetonitrile Trypsin working solution (see recipe) 25 mM Tris·Cl, pH 8.0 in water 25 mM Tris·Cl, pH 8.0 in 20% (v/v) acetonitrile 80% H2 16 O/20% (v/v) acetonitrile 80% H2 18 O/20% (v/v) acetonitrile Modified trypsin, sequencing grade (Promega) 95% isotopically enriched H2 18 O (Sigma-Aldrich) 2.5% (v/v) trifluoroacetic acid in water 0.1% (v/v) trifluoroacetic acid in 50% (v/v) acetonitrile 0.1% (v/v) trifluoroacetic acid in water 0.1% formic acid in 50% (v/v) acetonitrile or methanol Dialysis membrane or micro dialysis cups of appropriate MWCO (Pierce) or gel filtration spin column (MacroSpin columns; http://nestgrp.com, no. SMM 025.25; also see UNIT 8.3) Scalpel Speedvac evaporator Clear 0.5- or 1.5-ml microcentrifuge tubes 30◦ C incubator Reversed-phase C18 ZipTip pipet tips (Millipore) or OMIX Tip C18 (Varian) Additional reagents and equipment for dialysis (APPENDIX 3B) or buffer exchange using a gel filtration column (UNIT 8.3), mass spectrometry (Chapter 16), SDS-PAGE gel electrophoresis (UNIT 10.1), and gel staining and destaining (UNIT 10.5) 1. Prepare the protein or proteins of interest in a solution of 50 mM MES buffer, pH 6.5, so that the concentration is ∼1 mg/ml. 2. Dialyze (APPENDIX 3B) extensively against 50 mM MES buffer, pH 6.5. Alternatively, perform a buffer exchange with a spin column. This step is crucial as trace amounts of carbonate, acetate, phosphate, or aminocontaining buffers from the protein preparation procedures will interfere with the crosslinking.

3. Add 10-fold molar excess of EDC over the carboxylic groups. To calculate the amount of carboxyl groups use the equation: (µg protein/molecular weight [kDa])×(number of glutamic and aspartic acid residues) = nmol protein carboxyls Example: 0.2 mg (200 µg) of a 40-kDa protein with 18 carboxylic groups contains: (200 [µg]/40 [kDa]) × (18) = 90 nmol carboxylic groups To calculate the amount of EDC required 90 (nmol carboxylic groups) × 10-fold molar excess = 900 nmol EDC 0.2 M EDC = 200 nmol/µl. Add 900/200 = 4.5 µl Protein Contacts by Chemical Cross-Linking & MS

4. Add 0.9 µl of 0.02 M sulfo-NHS to achieve a 50:1 EDC/sulfo-NHS ratio. 5. Incubate up to 4 hr at room temperature.



6. At various time intervals (e.g., 15, 30, 60, 120 min) withdraw aliquots containing sufficient protein to be visualized on an electrophoresis gel. Immediately quench the reaction by adding hydroxylamine hydrochloride to a final concentration of 10 mM. Add the appropriate amount of (2× or 4×) reducing, denaturing SDS sample buffer. 7. Analyze the aliquots by denaturing SDS gel electrophoresis (UNIT 10.1), stain and destain the gels (UNIT 10.5). Depending on the results, repeat the experiment with modified conditions to optimize the reaction. 8. Perform a full-scale reaction. 9. Separate the products by 1-D SDS-PAGE (UNIT (UNIT 10.5).

10.1),

stain and destain the gels

Perform in-gel protein digestion Also see UNITS 11.3 and 16.4 for more information on enzymatic digestion of proteins for mass spectrometric analysis. 10. Using a clean scalpel cut out the stained band(s) of interest. 11. Estimate the volume of the excised bands using the equation below: Volume of band [mm3 ] = width × height × gel thickness. Sum the volume of all the bands. Multiply by 1.2 to determine the volume [µl] of trypsin solution needed for step 19.

12. Cut each gel slice into small pieces (> 1 µM

Signal (RU)

400

300

200

100

0 0

B

20

40

60

80

100

1400 KD = 0.3 µM

Signal (RU)

1200 1000 800 600 400 200 0 0

100

200

300

400

Time (sec)

Figure 20.2.7 Effects of the ligand density on the binding kinetics for interactions with relative low affinity (A) and medium affinity (B). Each panel shows the interaction of immobilized superantigen with single-chain TCR, observed at high ligand surface density (1700 RU, solid lines) and low ligand surface density (700 RU, dashed lines). For easier comparison, the signal obtained at the lower density surface was scaled proportionally. In panel A, a slow phase of binding in the association phase and a residual binding (possible slow dissociation of multivalently bound aggregates) is introduced at high ligand density, under otherwise identical conditions. For the interaction in panel B, the chemical off-rate constant is smaller than for the low-affinity interaction shown above. Nevertheless, from comparison of the binding curves at different ligand surface densities (under otherwise identical conditions) it is obvious that an increased ligand density has significant effects on the surface binding kinetics. This observation could be due to aggregation or to mass transport limitations, both of which are more likely at higher ligand densities.



Supplement 17

2. Increased dissociation rate when a soluble form of the ligand is injected (Fig. 20.2.5D). This always strongly indicates rebinding (mass transport limitation) if detected. Since the ligand by itself should not interact with the surface, its effect is the binding to analyte near the sensor surface, preventing it from rebinding and allowing diffusion and washout from the surface (Fig. 20.2.5C). Unfortunately, this effect may not be present for large ligands or ligands/analytes with high nonspecific binding if they exhibit limited diffusivity. 3. Double exponential dissociation phase. This will be seen only after dissociation from more than 50% occupation of all available surface sites. 4. Weak dependence on the flow rate (stirring speed in cuvette-based systems). Since the transport parameters only change with the cube root of the flow rate, i.e., generating only a factor of 2 when changing the flow rate by a factor of 10, this flow rate dependency can be difficult to detect. This is particularly true if the flow rate is only varied by a factor of 2, and the reaction is only partially transport influenced. Then, only a ∼10% change in the apparent koff would be expected in cases where the true koff is 100% larger than the apparent koff. 5. Linear association phase. This effect may not always be present, in particular at substantial transport limitation (Schuck, 1996; Schuck, 1997b; Schuck and Minton, 1996b). Generally, tests 1 and 2 are the most reliable. They also lead to experimental techniques that can be utilized to reduce or eliminate mass transport artifacts. The most effective way for reducing mass transport influence is lowering the surface density of the immobilized ligand. Higher flow rates give only comparatively very small improvements, but are connected with strongly increased sample volume requirements. If the surface density of the ligand cannot be reduced further without leading to an insufficient signal-to-noise ratio, then switching from kinetic experiments to steady-state or competition steady-state experiments is the best solution. This will give information on the equilibrium constant. The kinetic rate constants can then be estimated best from a saturation experiment (approaching complete saturation of all surface sites), followed by a dissociation phase during which soluble ligand is coinjected. The soluble ligand will minimize rebinding and allow the estimation of the chemical off-rate constant, from which the chemical on-rate constant can be determined via Equation 20.2.3. Analyte Aggregates Oligomeric aggregates of analyte can be troublesome in biosensor experiments in two different ways. First, if trace amounts of higher oligomers are present in the analyte sample, this will lead, in the association phase, to a slow accumulation at the sensor surface, which can be visible as a slower second phase of binding. As depicted in Figure 20.2.2, these multimeric analyte aggregates can have multiple interactions with immobilized ligand molecules, and therefore they will dissociate much more slowly than the monomeric analyte. Consequently, they will appear in the dissociation phase as a submoiety with a very low off-rate constant (Davis et al., 1998). The troublesome trace amounts of oligomers can be eliminated by careful chromatographic purification, or their influence can be minimized by exchanging the role of analyte and ligand (Davis et al., 1998; Andersen et al., 1999). This is illustrated in the example of Figure 20.2.6, which demonstrates the importance of both the sample preparation and the choice of ligand and analyte. Measuring Protein Interactions by Optical Biosensors

The second potentially problematic form of aggregation is a surface-induced multimerization of the analyte. Because the local macromolecular concentrations at the sensor surface are very high (e.g., in the order of 10mg/ml at a signal of 1000 RU in Biacore



instruments), local crowding effects combined with non-specific interactions of the analyte can promote oligomer formation at the sensor surface (Fig. 20.2.7A; Minton, 1995, 1998). As with the influence of preformed aggregates, this process will lead to biphasic association and dissociation profiles, with the slower phase resulting from oligomer accumulation and dissociation, respectively. This process will be favored by higher surface concentrations—i.e., higher density of immobilized ligand, and by higher ligand affinity (Fig. 20.2.7). As with mass transport limitations, they can be detected by Table 20.2.1

Troubleshooting Guide for Measuring Protein Interactions by Optical Biosensors

Problem

Solution

No electrostatic preconcentration achieved; poor immobilization

Desalt protein (e.g., using spin column or microdialysis); decrease pH of buffer used for immobilization (should be below pI of protein) Analyte may be too hydrophobic, or there may be electrostatic interaction with surface. Increase salt or detergent concentration in running buffer.a If this does not result from high analyte concentrations, dialyze the analyte against running buffer, or use a spin column for buffer exchange Use increasingly harsher conditions for regeneration; test procedures used in affinity chromatography; check for strong nonspecific binding of the analyte; check for possible incomplete blocking of activated surface sites after immobilization Check for presence of free biotin in sample. Biotin on the analyte may not be accessible by surface-immobilized streptavidin; in this case try biotinylated linker Check for mass-transport artifacts; check for possible traces of aggregates and for formation of aggregates at the surface; change immobilization method (avoid random coupling, avoid large surface densities). If this is not successful, go to steady-state analysis methods. Use longer injections by increasing injection volume and/or decreasing flow rate. If sample volume is limiting, try an equilibrium titration Decrease immobilization density.b If immobilization density cannot be lowered further, go to steady-state analysis, combined with establishment of lower limit of kdis from dissociation after saturation. Potential signature of mass transport; lower the immobilization density Signature of mass transport limitation; lower the immobilization density

Nonspecific binding is high

High signal from buffer; refractive index changes

Analyte does not come off after regeneration

No binding of biotinylated sample to streptavidin surface

Kinetics does not follow 1:1 binding

No steady-state binding is reached in a flow system Mass transport limitation

Increasing slope in the association phase Increasing signal in the dissociation phase

aBe aware of effects of nonspecific binding on the binding kinetics, which substantially decreases the diffusivity of the

analyte across the sensor surface, potentially leading to mass transport artifacts that cannot be detected through change of the ligand density. bIncreasing the flow rate affects transport much less, but consumes much more sample.



Supplement 17

variation of the surface density of the ligand. They also can be reduced by lowering the surface density of the immobilized ligand, by size-exclusion chromatography immediately prior to the experiment, or by exchanging the role of ligand and analyte. Troubleshooting As with other complex biophysical techniques and all investigations involving biological samples, it is impossible to give guidelines which are even nearly complete or generally helpful for troubleshooting. Nevertheless, Table 20.2.1 presents a small list of possible solutions to potential problems, given in the hope that some readers may find them helpful. SUMMARY Binding studies with optical biosensors can be very powerful and versatile. Among the most important virtues are their high sensitivity and utility for a broad range of affinities, real-time detection allowing studies of binding kinetics, and relatively low requirements of sample volume. We have outlined some general strategies and described some of the most commonly used techniques. The work with protein interactions at a surface can introduce additional experimental difficulties as compared to solution methods. As a general rule in experiments with optical biosensors, it is highly recommended that analyses be performed in different ways so that the consistency of the results can be tested. Biosensors can be an excellent tool in the study of protein interactions, and become particularly powerful if combined with other methods. ACKNOWLEDGMENT The authors are grateful for helpful comments by D. Margulies in the preparation of the manuscript. They acknowledge D. Kranz and K. Karjalainen for providing the samples used for some of the experiments shown. Peter Andersen acknowledges grant support by the Danish Natural Sciences Research Council. LITERATURE CITED Andersen, P.S., Lavoie, P.M., Sekaly, R.P., Churchill, H., Kranz, D.M., Schlievert, P.M., Karjalainen, K., and Mariuzza, R.A. 1999. Role of the T cell receptor alpha chain in stabilizing TCR-superantigen-MHC class II complexes. Immunity 10:473-483. Buckle, P.E., Davies, R.J., Kinning, T., Yeung, D., Edwards, P.R., Pollard-Knight, D., and Lowe, C.R. 1993. The resonant mirror: A novel optical sensor for direct sensing of biomolecular interactions. Part II: Applications. Biosens. Bioelectron. 8:355-363. Davis, S.J., Ikemizu, S., Wild, M.K., and van der Merwe, P.A. 1998. CD2 and the nature of protein interactions mediating cell-cell recognition. Immunol. Rev. 163:217-236. Edwards, P.R., Gill, A., Pollard-Knight, D.V., Hoare, M., Buckle, P.E., Lowe, P.A., and Leatherbarrow, R.J. 1995. Kinetics of protein-protein interactions at the surface of an optical biosensor. Anal. Biochem. 231:210-217. Measuring Protein Interactions by Optical Biosensors

Edwards, P.R., Maule, C.H., Leatherbarrow, R.J., and Winzor, D.J. 1998. Second-order kinetic analysis of IAsys biosensor data: Its use and applicability. Anal. Biochem. 263:1-12.

Garland, P.B. 1996. Optical evanescent wave methods for the study of biomolecular interactions. Q. Rev. Biophys. 29:91-117. Gershon, P.D. and Khilko, S. 1995. Stable chelating linkage for reversible immobilization of oligohistidine tagged proteins in the Biacore surface plasmon resonance detector. J. Immunol. Methods. 183:65-76. Glaser, R.W. and Hausdorf, G. 1996. Binding kinetics of an antibody against HIV p24 core protein measured with real-time biomolecular interaction analysis suggest a slow conformational change in antigen p24. J. Immunol. Methods. 189:1-14. Hall, D.R. and Winzor, D.J. 1997. Use of a resonant mirror biosensor to characterize the interaction of carboxypeptidase A with an elicited monoclonal antibody. Anal. Biochem. 244:152-160. Hermanson, G.T. 1996. Bioconjugate techniques. Academic Press, San Diego. Khilko, S.N., Jelonek, M.T., Corr, M., Boyd, L.F., Bothwell, A.L.M., and Margulies, D.H. 1995. Measuring interactions of MHC class I molecules using surface plasmon resonance. J. Immunol. Methods. 183:77-94.



Knoll, W. 1998. Interfaces and thin films as seen by bound electromagnetic waves. Annu. Rev. Phys. Chem. 49:569-638. Leckband, D.E. 1997. The influence of protein and interfacial structure on the self-assembly of oriented protein arrays. Adv. Biophys. 34:173-190. Leckband, D.E., Kuhl, T., Wang, H.K., Herron, J., Muller, W., and Ringsdorf, H. 1995. 4-4-20 antifluorescyl IgG Fab′ recognition of membrane bound hapten: Direct evidence for the role of protein and interfacial structure. Biochemistry 36:11467-11478. Lukosz, W. 1991. Principles and sensitivities of integrated optical and surface plasmon sensors for direct affinity sensing and immunosensing. Biosens. Bioelectronics. 6:215-225. Malmborg, A.C. and Borrebaeck, C.A. 1995. Biacore as a tool in antibody engineering. J. Immunol. Methods. 183:7-13. Margulies, D.H., Corr, M., Boyd, L.F., and Khilko, S.N. 1993. MHC Class I/peptide interactions: Binding specificity and kinetics. J. Mol. Recognit. 6:59-69. Margulies, D.H., Plaksin, D., Khilko, S.N., and Jelonek, M.T. 1996. Studying interactions involving the T-cell antigen receptor by surface plasmon resonance. Curr. Opin. Immunol. 8:262-270. Minton, A.P. 1995. Confinement as a determinant of macromolecular structure and reactivity. 2. Effects of weakly attractive interactions between confined macrosolutes and confining structures. Biophys. J. 68:1311-1322. Minton, A.P. 1998. Molecular crowding: Analysis of effects of high concentrations of inert cosolutes on biochemical equilibria and rates in terms of volume exclusion. Methods Enzymol. 295:127149. Muller, K.M., Arndt, K.M., and Plückthun, A. 1998. Model and simulation of multivalent binding to fixed ligands. Anal. Biochem. 261:149-158. Myszka, D.G., Jonsen, M.D., and Graves, B.J. 1998. Equilibrium analysis of high affinity interactions using Biacore. Anal. Biochem. 265:326-30. Nieba, L., Krebber, A., and Plückthun, A. 1996. Competition Biacore for measuring true affinities: Large differences from values determined from binding kinetics. Anal. Biochem. 234:155165. Ober, R.J., and Ward, E.S. 1999a. The influence of signal noise on the accuracy of kinetic constants measured by surface plasmon resonance experiments. Anal. Biochem. In press. Ober, R.J. and Ward, E.S. 1999b. The choice of reference cell in the analysis of kinetic data using Biacore. Anal. Biochem. 271:70-80. O’Shannessy, D.J. and Winzor, D.J. 1996. Interpretation of deviations from pseudo-first-order kinetic behavior in the characterization of ligand binding by biosensor technology. Anal. Biochem. 236:275-283.

O’Shannessy, D.J., Brigham-Burke, M., and Peck, K. 1992. Immobilization chemistries suitable for use in the Biacore surface plasmon resonance detector. Anal. Biochem. 205:132-136. O’Shannessy, D.J., Brigham-Burke, M., Soneson, K.K., Hensley, P., and Brooks, I. 1993. Determination of rate and equilibrium binding constants for macromolecular interactions using surface plasmon resonance: Use of nonlinear least squares analysis methods. Anal. Biochem. 212:457-468. O’Shannessy, D.J., O’Donnell, K.C., Martin, J., and Brigham-Burke, M. 1995. Detection and quantitation of hexa-histidine-tagged recombinant proteins on western blots and by a surface plasmon resonance biosensor technique. Anal Biochem. 229:119-24. Plant, A.L., Brigham-Burke, M., Petrella, E.C., and O’Shannessy, D.J. 1995. Phospholipid/alkanethiol bilayers for cell-surface receptor studies by surface plasmon resonance. Anal. Biochem. 226:342-348. Ramsden, J.J., Bachmanova, G.I., and Archakov, A.I. 1996. Immobilization of proteins to lipid bilayers. Biosens. Bioelectronics 11:523-528. Schuck, P. 1996. Kinetics of ligand binding to receptor immobilized in a polymer matrix, as detected with an evanescent wave biosensor. I. A computer simulation of the influence of mass transport. Biophys. J. 70:1230-1249. Schuck, P. 1997a. Use of surface plasmon resonance to probe the equilibrium and dynamic aspects of interactions between biological macromolecules. Annu. Rev. Biophys. Biomol. Struct. 26:541-566. Schuck, P. 1997b. Reliable determination of binding affinity and kinetics using surface plasmon resonance biosensors. Curr. Opin. Biotechnol. 8:498502. Schuck, P., and Minton, A.P. 1996a. Minimal requirements for internal consistency of the analysis of surface plasmon resonance biosensor data. Trends Biochem. Sci. 252:458-460. Schuck, P. and Minton, A.P. 1996b. Analysis of mass transport limited binding kinetics in evanescent wave biosensors. Anal. Biochem. 240:262-272. Schuck, P., Millar, D.B., and Kortt, A.A. 1998. Determination of binding constants by equilibrium titration with circulating sample in a surface plasmon resonance biosensor. Anal. Biochem. 265:79-91. Schuster, S.C., Swanson, R.V., Alex, L.A., Bourret, R.B., and Simon, M.I. 1993. Assembly and function of a quaternary signal transduction complex monitored by surface plasmon resonance. Nature 365:343-347. Sigal, G.B., Bamdad, C., Barberis, A., Strominger, J., and Whitesides, G.M. 1996. A self-assembled monolayer for the binding and study of histidine-tagged proteins by surface plasmon resonance. Anal. Chem. 68:490-497.



Supplement 17

Silhavy, T.J., Szmelcman, S., Boos, W., and Schwartz, M. 1975. On the significance of the retention of ligand by protein. Proc. Natl. Acad. Sci. U.S.A. 72:2120-2124.

O’Shannessy et al., 1992. See above.

Stein, T., and Gerisch, G. 1996. Oriented binding of a lipid-anchored cell adhesion protein onto a biosensor surface using hydrophobic immobilization and photoactive crosslinking. Anal. Biochem. 237:252-259.

General review of the method and its application.

van der Merwe, P.A. and Barclay, A.N. 1996. Analysis of cell-adhesion molecule interactions using surface plasmon resonance. Curr. Opin. Immunol. 8:257-261.

Web site for Biacore; extensive list of published biosensor applications

Yarmush, M.L., Patankar, D.B., and Yarmush, D.M. 1996. An analysis of transport resistances in the operation of Biacore; Implications for kinetic studies of biospecific interactions. Mol. Immunol. 33:1203-1214.

Key References Davis et al, 1998. See above. Contains a detailed description of analyte aggregation effects on the measured surface binding. Nieba et al., 1996. See above.

Collection of immobilization techniques. Schuck, 1997b. See above.

Internet Resources http://www.biacore.com

http://www.affinity-sensors.com Web site for Affinity Sensorsl; extensive list of published biosensor applications.

Contributed by Peter Schuck and Lisa F. Boyd National Institutes of Health Bethesda, Maryland Peter S. Andersen University of Maryland Rockville, Maryland

Demonstration how competition approaches can be used to circumvent kinetic artifacts.

Measuring Protein Interactions by Optical Biosensors



Analytical Centrifugation: Equilibrium Approach Equilibrium centrifugation, also called equilibrium sedimentation and equilibrium analytical ultracentrifugation, is a classical method of biochemistry for providing firstprinciple thermodynamic information about the molar mass, association energy, association stoichiometry, and thermodynamic nonideality of molecules in solution. Because it relies on the principal property of mass and the fundamental laws of gravitation, equilibrium sedimentation can be used to analyze the solution behavior of proteins of any type in a wide range of solvents and over a wide range of protein concentrations. For many questions about protein solution behavior, there is no satisfactory substitute method of analysis. This unit presents (1) background information about equilibrium sedimentation, (2) a review of the basic theory of equilibrium sedimentation, (3) a strategy to use when planning equilibrium sedimentation experiments, and (4) an overview of the analysis of sedimentation equilibrium data. The early literature covering the invariant foundations of sedimentation is still relevant (see references in Laue and Stafford, 1999), and beginners should review these early papers. This unit will stress what questions may be answered when performing sedimentation equilibrium analysis using the absorbance and interference optical detectors on the BeckmanCoulter XLI ultracentrifuge. In so doing, it will neglect techniques that use preparative centrifuges and postsedimentation analysis (Rivas and Minton, 1993), as well as density gradient sedimentation (Rickwood, 1984). Furthermore, space limitations prevent complete descriptions of the methods available for data analysis; therefore, references are made to the pertinent literature. The analytical ultracentrifuge is similar to a high-speed preparative centrifuge in that a spinning rotor provides a gravitational field large enough to sediment molecules (Ralston, 1993). The analytical ultracentrifuge is distinguished from the preparative centrifuge by specialized rotors, sample holders, and optical systems that permit the observation of samples during sedimentation (Fig. 20.3.1). Using present technology, a maximum of 32 samples may be analyzed in a single experiment (Ralston, 1993; Laue, 1995).

Contributed by Tom Laue Current Protocols in Protein Science (1999) 20.3.1-20.3.13 Copyright © 1999 by John Wiley & Sons, Inc.

UNIT 20.3

The defining characteristic of sedimentation equilibrium is the time-invariant concentration gradient that develops as the flux of sedimenting molecules is exactly balanced by the flux of diffusing molecules at each point in the sample holder, referred to as a cell. No distinct boundaries are observed; instead, a smooth concentration gradient is observed (Fig. 20.3.2). The accurate determination of this concentration gradient is the fundamental measurement needed for the analysis of a sedimentation equilibrium experiment. To make this measurement, the cells are passed through the optical paths of detectors by the spinning rotor (Fig. 20.3.1). Presently, there are two optical detectors for the XLI, one that uses absorbance and one that uses refractive index to measure the concentration at closely spaced radial intervals. A review of these two optical systems is available (Laue, 1996). Analysis of the concentration profiles provided by the optical systems provides rigorous insight into the solution behavior of sedimenting molecules. Interpretation of these data requires knowledge of the rotor speed, temperature, elapsed time, and buoyancy (Laue and Stafford, 1999).

THEORY Sedimentation equilibrium is a powerful method for describing the solution thermodynamic behavior of proteins. The most common thermodynamic behaviors exhibited by proteins include (1) single ideal component, (2) ideal, reversible self-association, (3) heterogeneous mixture of components, (4) ideal, reversible association between dissimilar proteins, and (5) nonideality. The theory behind these behaviors may be found in Laue and Stafford (1999) and references therein. These behaviors are not necessarily mutually exclusive, and it is often found that a protein exhibits both self-association and nonideality, or self-association and heterogeneity. Experimentally, then, it is important to be able to diagnose which of these five behaviors are important to a protein’s solution behavior. The goal of this section is to provide a brief mathematical description of the equilibrium concentration profiles for the five common behaviors. Based on these descriptions, a set of diagnostic tests will be presented.



F

H A

G

B F

D

C

E

D

E

Figure 20.3.1 A schematic representation of an analytical ultracentrifuge. The rotor (C) has holes through it to hold sample containers commonly called cells (D). Each cell (see inset) consists of a centerpiece (G) with open-sided chambers called channels (H) to hold the liquid samples. There are two channels per sample, one containing the protein in its solvent and an adjacent one (not shown) containing only solvent to serve as an optical reference. The centerpiece, in turn, is sealed between windows (F) to permit the passage of light through the channels, thus allowing the cell contents to be viewed. Centerpieces are made out of a variety of tough inert materials. Depending on the type of experiment that will be performed, centerpieces can hold one, three, or four samples each. Rotors can hold either four or eight cells. As the rotor spins, each cell passes through the optical paths of two different detectors. The absorbance detector uses a pulsed xenon lamp (A) to provide a burst of light when a cell is aligned in the beam. The absorbance detector (E) uses a narrow slit and photomultiplier tube to determine the light intensity after the beam has passed through the sample. The slit is moved radially by a motor so that the absorbance profile, called an absorbance scan, can be determined. The second detector determines the refractive index difference between the sample and reference channels at each radial position. The light source (B) uses a laser diode to produce two radially directed narrow stripes of light, one that passes through the sample channel and one that passes through the reference channel. These two stripes of light are brought together to produce an interference image (Fig. 20.3.2) in which the difference in the refractive index at each radial position is displayed as the vertical displacement of a set of fringes.

Single Ideal Component Dilute solutions of a highly purified protein at physiological salt concentrations often behave like an ideal system with the protein as the only sedimenting component. For a single thermodynamically ideal component, the basic equation (Eqn. 20.3.1) describing the concentration as a function of radial position, c(r), is a simple exponential: 2

c( r ) = c0 e

σ

r − r0 2 2

= c0 e σξ

Equation 20.3.1 Analytical Centrifugation: Equilibrium Approach

where c0 is the concentration at a reference radius (r0, in cm), and σ is the reduced molecular weight:

σ=

Mbω 2 RT

Equation 20.3.2

In Equation 20.3.2, Mb is the buoyant mass (discussed below), ω is the centimeter-gramsecond (cgs) rotor speed (rpm π/30, in radians/sec), R is the cgs gas constant (3.14 × 107 erg/mole/°K), and T is the absolute temperature (°K). Notice that σ is directly proportional to the buoyant mass and the square of the rotor speed. Because of these dependencies, equilibrium sedimentation can be used to measure molecular weights ranging from ∼1000 to 20 × 106 Da. Experimentally, a single ideal component yields the same value of Mb regardless of the sample concentration or rotor speed.



meniscus

solution

base

increasing c sedimentation diffusion increasing r increasing g

Figure 20.3.2 A Rayleigh interference refractive optical system image of one channel of a six-channel centerpiece (Fig. 20.3.4). This optical system provides an image in which the concentration at each radial position is represented by the vertical displacement of a set of equally spaced horizontal fringes. The center of rotation is to the left and the edge of the rotor is to the right. An air/liquid meniscus forms at the top of the channel, one in the reference sector and one in the sample sector. As slightly different volumes of liquid are used to fill these two sectors, the menisci do not appear at exactly the same radius. If the two menisci do become superimposed over the course of an experiment, there is a leak between the sample and reference sectors, indicating that the centerpiece is scratched and needs polishing. The sample is a protein at sedimentation equilibrium, where the flux of molecules towards the cell bottom due to sedimentation is exactly balanced by the flux of molecules towards the meniscus due to diffusion. The balance of fluxes yields an exponential concentration distribution. The base of the cell is formed by FC-43, a clear fluorocarbon liquid with a refractive index very close to that of water.

Ideal Reversible Self-Association The equation describing reversible self-association in an otherwise ideal solution is obtained by expanding Equation 20.3.1 in terms of the species:

c( r ) = ∑ c0 i e c01e

σi ξ

σ1ξ

=

+ c02 e

σ 2ξ

+ c03 e

σ 3ξ

+...

c0 n

Kal ⇔n =

c01

n

Equation 20.3.4

where [c0n] is the oligomer concentration and [c01] is the monomer concentration. Substituting Equation 20.3.4 into Equation 20.3.3: i

Equation 20.3.3

where i is the index for the components, c01 and σ1 refer to reference concentration and reduced buoyant molecular weight of the monomer, c02 and σ2 refer to these quantities for the dimer, and so forth. For a reversible mass-action association, all σ’s are related so that σ2 = 2σ1, σ3 = 3σ1, and so on. The monomer reference concentration, c01, is linked to the other reference concentrations, c0n, through the equilibrium constant:

c( r ) = ∑ ci ( r ) = ∑ c0 i K al ⇔ i e = c01e 3

σ1ξ

2

+ c01 K al ⇔ 2 e

c01 K al ⇔ 3 e

3σ1ξ

iσ i ξ

2 σ1ξ

+

+...

Equation 20.3.5

It is important to note in Equation 20.3.5 that the abundance of all of the associated species depends only on the reference concentration of the monomeric protein, c01. As c01 increases, the abundance of dimers, trimers, and so on



Supplement 18

increases relative to the concentration of the monomer, whereas dilution decreases their relative abundance. One way to describe the changes in the relative abundance of species is to use an average molecular weight (Van Holde, 1985). The weight-averaged molecular weight, Mw, is defined as:

Mw ≡

∑ ci Mi ∑ ci

Equation 20.3.6

There are procedures that determine Mw at each point in an equilibrium sedimentation distribution, and graphs of Mw as a function of either concentration or rotor speed are useful diagnostics (Yphantis, 1964; Roark and Yphantis, 1969). For a reversible ideal self-association, Mw increases with increasing protein concentration and decreases as a solution is made more dilute. Most importantly, Mw depends only on the concentration of the monomer and is independent of rotor speed.

Heterogeneous Mixtures Formation of irreversible aggregates is a common behavior of proteins. These aggregates can be of any size, resulting in a mixture of dimers, trimers, and so on whose concentration is not a reflection of the monomer concentration. The equation describing the sedimentation equilibrium of such a heterogeneous mixture is obtained by expanding Equation 20.3.1 in terms of distinct components:

Ideal Reversible Heterogeneous Association There are two common situations that lead to heterogeneous associations. The first is when monomers of an otherwise seemingly homogeneous solution self-associate, but with a range of equilibrium constants (Yphantis et al., 1978; Senear and Teller, 1981). In fact, careful scrutiny of nearly all self-associations seems to reveal some heterogeneity. Experimental treatment of these systems is outside the scope of this chapter, but the reader should be aware that this can be observed. The second situation occurs when unlike molecules within a mixture associate. Some of the most interesting interacting biochemical systems involve reversible associations between nonidentical components. More complicated expressions must be derived that explicitly take into account the concentrations of each component and the various species:

c( r ) = ∑ ci ( r ) = c01e

σ1ξ

m ′ ⇔m e Kal + ∑ c01

mσ1ξ

m

(self − association of component 1) + c02 e

σ2ξ

n ′′ ⇔n e Kal + ∑ c02

nσ 2 ξ

n

(self − association of component 2) j k ′′′⇔ j + k e Kal + ∑ c01c02

( jσ1 + kσ 2 ) ξ

j ,k

( heteroassociation ) Equation 20.3.8

c( r ) = ∑ c 0 i e

σi ξ

σ ξ c01e 1

=

+ c02 e

σ 2ξ

+ c03e

σ 3ξ

+...

Equation 20.3.7

Analytical Centrifugation: Equilibrium Approach

Each term on the right-hand side of this equation describes the separate sedimentation of one size of aggregate. However, unlike a reversible association (Equation 20.3.5), there is no relationship linking the different reference concentrations. Experimentally, a heterogeneous mixture leads to values of Mw that are either independent of protein concentration or, for reasons outside this discussion, decrease with increasing protein concentration. Furthermore, as higher molecular weight components are pelleted from the solution, Mw decreases, often dramatically, with increasing rotor speed.

where m, n, j, and k are stoichiometric coefficients describing the formation of species from a hetergeneous mixture of components. Although extracting thermodynamic quantities for extremely complicated systems is still beyond the capabilities of sedimentation equilibrium, two-component associations are amenable to study. The first step in studying such a system is to characterize the behavior of the individual components. Should one of the components reversibly self-associate, then the analysis becomes more difficult, though not intractable. Often, the analysis of a heterogeneous association is made easier by discriminating between components, either by spectral enhancement (Laue et al., 1993) or by spectral deconvolution (Schuck, 1994). Characterizing heterogeneous associations is an experimental challenge that requires the analysis of a large amount of data acquired over



a wide range of concentrations and at different concentration ratios of the components. The analysis of such data is outside the scope of this unit, but has been described elsewhere (Luckow et al., 1989). Diagnostically, a heterogeneous association exhibits increasing Mw with increasing protein concentration, though the extent of the increase in Mw will depend on the ratio of the concentrations of the components initially in the cell, as well as on the stoichiometry of assembly of the complex between the components. Also, there is likely to be some decrease in Mw with increasing rotor speed, especially if the two components differ significantly in Mb.

Nonideality Although a reversible association is a form of nonideality, in the context of sedimentation equilibrium the term nonideality is usually reserved for the repulsive interactions between protein molecules. These interactions tend to increase the apparent concentration, or chemical activity, ai, of a component, i, so that ai = γici, with the nonideality coefficient γi > 1. It is ai that is balanced by the gravitational potential, but ci that is measured by the optical systems. The effect of nonideality, therefore, is to suppress the observed concentration gradient (i.e., ci = ai/γi, with γi > 1). The two contributions to nonideality are the volume occupied by the protein molecules (called the excluded volume effect) and the charge-charge repulsion between like-charged

Table 20.3.1

proteins. Excluded volume contributions depend on the shape and flexibility of a molecule and are entropic. Charge-repulsion contributions are enthalpic and increase as the pair-wise product of the charges on the proteins, but decrease inversely with supporting electrolyte concentration. The two contributions to nonideality are additive, with charge-charge repulsion usually being the major contributor. Should the effects not be too severe, the effects of nonideality can be included in Equation 20.3.1 or Equation 20.3.5 by approximating the activity as a virial expansion of the concentration (Yphantis, 1964; Laue, 1995). Experimentally, nonideality is observed as a decrease in Mw with increasing concentration. For globular proteins having a moderate net charge (less than ∼10), nonideality will be minimal at concentrations

Current Protocols in Protein Science: 001

Current Protocols in Protein Science

Current Protocols in Cytometry

Current Protocols in Toxicology

Current Protocols in Molecular Biology

RNA-Protein Interaction Protocols

Membrane Protein Protocols

Protein Kinase Protocols

RNA-Protein Interaction Protocols

Protein Purification Protocols

Protein Folding Protocols

Adhesion Protein Protocols

Protein Lipidation Protocols

Protein Sequencing Protocols

Prion Protein Protocols

Adhesion Protein Protocols

Protein Targeting Protocols

Protein Phosphatase Protocols

Membrane Protein Protocols

Membrane Protein Protocols

Protein Targeting Protocols

Protein Sequencing Protocols

Protein Protocols Handbook

RNA-Protein Interaction Protocols

Protein Phosphate Protocols

Current Protocols in Molecular Biology

Current Protocols in Molecular Biology

Protein Kinase Protocols

Protein Purification Protocols

Adhesion Protein Protocols

Protein Targeting Protocols

Current Protocols in Protein Science: 001