Handbook of Surface Plasmon Resonance

Handbook of Surface Plasmon Resonance

To Femke and Nick, Margot and Niels

Handbook of Surface Plasmon Resonance

Edited by R.B.M. Schasfoort and Anna J. Tudos University of Twente, Enschede, The Netherlands

ISBN: 978-0-85404-267-8 A catalogue record for this book is available from the British Library r The Royal Society of Chemistry 2008 All rights reserved Apart from fair dealing for the purposes of research for non-commercial purposes or for private study, criticism or review, as permitted under the Copyright, Designs and Patents Act 1988 and the Copyright and Related Rights Regulations 2003, this publication may not be reproduced, stored or transmitted, in any form or by any means, without the prior permission in writing of The Royal Society of Chemistry, or in the case of reproduction in accordance with the terms of licences issued by the Copyright Licensing Agency in the UK, or in accordance with the terms of the licences issued by the appropriate Reproduction Rights Organization outside the UK. Enquiries concerning reproduction outside the terms stated here should be sent to The Royal Society of Chemistry at the address printed on this page. Published by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 0WF, UK Registered Charity Number 207890 For further information see our web site at www.rsc.org

Foreword Make no bones about it, I love surface plasmon resonance (SPR)-based biosensor technology. After spending three years trying to measure binding constants using analytical affinity chromatography, I immediately saw the benefits of SPR the first time I sat down in front of a Biacore in 1991. Even today, no comparable technology exists to characterize molecular interactions in real time without labeling in an automated and robust fashion. But as the technology has expanded over the years, I find that there are three general attitudes towards SPR. There are the nay-sayers who hate the technology. There are long-time users who think they are experts. And there are the users who recognize they do not know everything about SPR but are eager to improve their skills. Ever since the first commercially viable instrument was unleashed in 1990 by the biosensor group at Pharmacia (which was spun out into a separate company called Biacore in 1996, only to be acquired recently by General Electric, which previously bought Amersham who at one time had merged with Pharmacia, so in fact now the biosensor group has come full circle, even though they have always shared the same cafeteria in Uppsala, Sweden), there have been critics of SPR technology. So much so that in 2003 I created a character called ‘‘Dr. Evil Pessimist’’, who represents a composite of the various detractors of SPR. Dr. Pessimist rants and rages about problems he has with the technology, including nonspecific binding, instrument drift, mass transport and avidity effects. He argues that since SPR uses a surface the rate constants we measure will never reflect solution-based binding constants. In fact, much of his resentment of the technology stems from the fact that his experiments fail or his data never fit a simple model. It has been my experience that there are two primary causes of this SPRaphobia: poor-quality reagents and/or poor experimental design. Perhaps the molecules Dr. Pessimist is studying do not in fact interact or the preparations of samples are not active to begin with. Don’t shoot the messenger. Dr. Pessimist asserts that his proteins are of high quality because they are ‘‘a single band on an SDS page gel’’. He fails to realize that this is not evidence of an active preparation or a conformationally homogeneous sample. I think biosensor experiments are akin to protein crystallography. No structural biologist I know would attempt to crystallize an impure, half-denatured preparation of protein

v

vi

Foreword

that has precipitated at the bottom of an Eppendorff tube. The sad thing is that garbage into a biosensor will often give complex responses that users misinterpret as some interesting binding event. I have found that when experiments are designed appropriately with goodquality reagents and data are processed and analyzed properly, binding responses can be routinely fit to a simple interaction model. However, unlike Dr. Pessimist, I do not expect to obtain perfect binding responses when I set up experiments on a new interaction. I realize that obtaining high-quality data is an iterative process. In my research group, we usually set up a trial experiment to verify that the binding partners actually interact. Then we will often try different coupling chemistries, surface densities, and/or buffer conditions to optimize surface activity. And when it comes to the number one complaint about SPR technology (that the surface will automatically change the thermodynamics of the system), what Dr. Pessimist fails to realize is that most biosensor experiments do not use a flat surface. Instead, the surface is coated with a dextran layer, which suspends the molecule in solution. We and others have shown with numerous systems that when experiments are performed properly, binding constants (including thermodynamic parameters) measured with SPR do in fact match those obtained from solution-based measurements. However, I agree with Dr. Pessimist in one regard. Since 1991, I have read every paper that reported using a commercial SPR biosensor and Dr. Rebecca Rich and I have composed a yearly review of the literature since 1998. This is becoming a fairly daunting task since more than 1000 research papers are published annually. More, unfortunately, is not always better. We find that the data in most biosensor articles are not worth the paper they are printed on. For example, about half the time authors even fail to present figures showing the binding responses and yet they expect us to believe the rate constants they report for their interactions. Without a visual inspection of the data, we have no idea if the experiments were run properly. And oftentimes, even when data are presented, it is clear that the investigators do not know how to utilize the technology properly. Also, while a fundamental dogma of science is to replicate and randomize samples, less than 3% of published biosensor data include replicate injections even within a single experiment. An overlay of replicate injections demonstrates the stability of the reagents and multiple independent experiments yield an average and standard deviation for the reported binding constants, yet this attention to detail in a biosensor experiment is more rare than finding a four-leaf clover in the outfield at Fenway Park. In addition, less than 5% of the authors who report kinetic constants include an overlay of the binding response with the fitted model. And finally, even from a brief glance through the literature, it is apparent that the majority of investigators do not understand that the shape of the response profile should be an exponential in both the association and dissociation phases (maybe many users do not even understand what an exponential is). It is no wonder that scientists outside the biosensor use community think SPR technology does not work. I would think the same thing if all I had to rely on was the published data.

Foreword

vii

You might ask yourself, ‘‘how did it get to this point?’’ I often wonder if all scientists are so poorly educated in basic scientific technique (which could actually explain why we haven’t found a cure for the common cold). I place the blame on the ‘‘kit mentality’’ that was introduced with molecular biology back in the early 1990s, back when we were listening to our Walkmans while typing on our IBM 286 personal computers. Nowadays you can buy a kit to clone, mutate, express and purify a protein. Well, the kit mentality continued when these same investigators got access to commercially available biosensor technology. Since these instruments are so easy to use, anyone can walk up to the machine, chuck in their proteins, collect some response, fit the data and publish the results, believing that the results must be correct because they came out of this very expensive machine. Unfortunately, it actually takes some skill and know-how to set up, execute and analyze a biosensor experiment properly. This leads me to the next group of biosensor users that give the technology a black eye. These people are the ones who have been using instruments for a long time and think they are experts. I call them ‘‘SPiRts’’. SPiRts are even more threatening than Pessimists because their complacency often leads them to perpetuate poor experimental technique. A common SPiRt mistake published in the literature is the use of multivalent analytes in solution (e.g. monoclonal antibodies or GST fusion proteins), which can produce avidity effects. All too often, SPiRts present elaborate biological justifications for the shape of their unusual binding profiles when in fact the responses are simply indicative of poor reagent quality and/or inadequate experimental optimization or data processing. Even worse, SPiRts use complex models to describe their poor-quality data. It seems that the latest fad of these model surfers is to apply a conformational change mechanism. ‘‘My data fit a conformational change model, which must mean there is a conformational change, right?’’ Wrong! To set the record straight, in 1994 my colleague and software engineer extraordinaire, Tom Morton (who I refer to as SoftEE), developed the numerical integration approach to data analysis that allows one to apply any interaction model. Before then, we were in the caveman days of linear transformation and, believe me, you don’t want to go back there. We were the first to show that a change in conformation that stabilized a bound complex would in fact produce a change in response even though there was no additional change in mass. However, in the intervening 13 years I have never needed to apply this model to describe the responses obtained from more than 1000 systems I have examined. The reason I am reluctant to use this model is that typically a data set that fits a conformational change model can be equally well described by other models such as those for ligand and/or analyte heterogeneity. Even more alarmingly, the rates for the supposed conformational changes measured on the biosensor are extremely slow, often with half-lives of 20–60 minutes if you take the time to calculate them. These rates do not make biological sense to me. A quick search of the classical conformational change literature shows that re-organizational events which occur during binding happen on a nanosecond to millisecond time-scale. The hot ‘‘new’’ trend with the SPiRts is to fit their biosensor data with a conformational change model

viii

Foreword

and then present crystal structure data of unbound and bound complexes and say ‘‘See, this change in conformation proves it’’. But an objective viewer would disagree. The fact that you see a change in conformation in the structure actually may not relate to the complex binding response you are measuring on the biosensor. Don’t be fooled by these sleight-of-hand arguments. (What would help confirm the conformational change suggested by SPR would actually be to use a time-resolved structural method such as circular dichroism or fluorescence resonance energy transfer and demonstrate that the timedependent changes are the same.) The cause of the complex binding response on the biosensor is actually more likely due to surface aggregation, nonspecific binding, molecular crowding, avidity effects or sample heterogeneity. This brings me to my favorite SPR users, who I refer to as SPiRits. SPiRits are new users or those having some familiarity of biosensor technology who have a deep desire to learn more about its features, applications and potential. They are the ones who are participating in our yearly benchmark studies, which are geared toward calibrating users’ experimental technique. They are willing to put in the effort to troubleshoot their systems and want to improve the quality of the data and not just settle for whatever the machine spits out. SPiRits will be the users who develop novel applications and implement new technologies in the future. We need SPiRits because the number and types of SPR instruments are exploding. An Internet search reveals more than 20 companies developing SPR-based biosensor systems. Lately, biosensor advances have occurred on two fronts. First, many of the recently released instruments (and others currently under development) are dedicated to specific applications ranging from small-molecule drug discovery to the characterization of complex mixtures in the clinical and food sciences. Corning’s Epic plate-based system is an example of targeting the technology for screening applications. Second, we have seen a push to increase the throughput of biosensor analyses. In the past few years, the launches of BioRad’s ProteOn XPR36 and Biacore’s A100 have dramatically impacted the biosensor field since they allow for parallel processing of multiple analytes over multiple targets simultaneously. Array-based platforms represent the next wave in biosensor development. Biacore’s Flexchip and instruments being developed by GWC Technologies, Lumera, IBIS Technologies, Genoptics and Maven open up the possibility of characterizing hundreds to thousands of interactions at one time. But not surprisingly, these array formats come with their own sets of challenges. The methods used for spotting DNA may not be optimal for producing protein arrays. Clearly, a lot of work remains to be done before protein array systems meet their full potential. As biosensor applications expand and new instruments are released, the technology’s user base also increases. I worry that higher-throughput systems may allow more users simply to generate more bad data faster. So, we clearly need to improve the skill level of both novice and seasoned users. This book is a great resource to obtain the fundamental knowledge of biosensor technology, and also discover recent developments in both

Foreword

ix

instrumentation and applications. But in order to turn professional, remember that the biosensor is just a tool. Use it wisely. Be skeptical, but keep an open mind. Know when to say when (not all systems are amenable to biosensor analysis). Go forth and become a good ShePaRd of my favorite technology. David G. Myszka University of Utah

Preface Lectori salutem, Editors, authors often claim that completion of their book was a giant, lonesome and tedious task. Often they add, their mission was ‘‘once in a lifetime’’. As they say in Germany with a sense of humor, ‘‘Einstein macht noch kein Haus1’’ meaning that the cooperation of people is in the heart of big achievements. So it has been with our book: all the authors had to find time in their busy life among other important engagements for timely writing activities, for which we cannot say often enough how grateful we are. This Handbook of Surface Plasmon Resonance is the product of an intensive interaction process and is intended for a wide audience: scientists and students intending to use the technology, the wider public interested in SPR as a phenomenon and its application, but also providers of (parts of) the technology. Although the book as a whole covers many aspects of the technology at present spanning a bridge between theory, instrumentation and applications, the chapters are written so as to be comprehensible individually as well. It is hoped that the readers of this book will share our enthusiasm for biomolecular interaction analysis based on SPR technology. We also hope that we have succeeded in revealing the potential of SPR by showing highly exciting and unique opportunities for unraveling the functional relationships of complex biological processes. Special thanks are also due to the members of the Biochip Group of the MESA+ Institute for Nanotechnology of the University of Twente who have contributed to the book: Stefan Schlautmann and Hans de Boer for technical support and some of the drawings. In addition, we thank Geert Besselink Bianca Beusink, Angelique Lokate, Dietrich Kohlheyer, Ganesh Krishnamoorthy, Dawid Zalewski, Remco Verdoold, Mayke van der Ploeg and Bjorn Harink for their input. This devoted team provided the warm and inspiring atmosphere of the Biochip Group during the two-year period from the birth of the idea to completion of the manuscript.

1

Literally: even Einstein could not build a house and the German meaning of ein stein is one stone.

x

Preface

xi

The editors would also like to thank Annie Jacob of the Royal Society of Chemistry for her clear guidance and enduring patience throughout the editorial process. Wout van Bennekom is acknowledged for final reading of several chapters. Richard Schasfoort and Anna Tudos Enschede and Amsterdam

Contents Chapter 1

Introduction to Surface Plasmon Resonance Anna J. Tudos and Richard B.M. Schasfoort 1.1

What is Surface Plasmon Resonance? 1.1.1 A Simple Experiment 1.1.2 From Dip to Real-time Measurement 1.2 How to Construct an SPR Assay? 1.2.1 The Steps of an Assay 1.2.2 Calibration Curve 1.2.3 Determination of Kinetic Parameters 1.2.4 Basics of Instrumentation 1.3 History of SPR Biosensors 1.3.1 Early History of SPR Biosensors 1.3.2 History of SPR Biosensors After 1990 1.4 How to Read This Book 1.5 Questions References

Chapter 2

1 1 2 3 4 6 7 8 9 9 11 12 13 13

Physics of Surface Plasmon Resonance Rob P.H. Kooyman 2.1 2.2 2.3

2.4 2.5

2.6

Introduction The Evanescent Wave Surface Plasmons 2.3.1 Surface Plasmon Dispersion Equations, Resonance 2.3.2 Excitation of Surface Plasmons 2.3.3 Surface Plasmon Properties 2.3.4 Choice of Experimental Parameters Analysis of Multi-layered Systems SPR Spectroscopy 2.5.1 Enhancement of Fluorescence and Absorbance 2.5.2 SPR and Metal Nanoparticles Concluding Remarks xiii

15 16 17 17 19 21 26 27 28 28 29 31

xiv

Contents

2.7 Questions 2.8 Symbols References Chapter 3

SPR Instrumentation Richard B.M. Schasfoort and Alan McWhirter 3.1

Introduction 3.1.1 From Surface Plasmon to SPR Signal 3.2 SPR Optics 3.2.1 Fan-shaped Beam 3.2.2 Fixed Angle 3.2.3 Angle Scanning 3.2.4 Grating Coupler 3.2.5 Other Optical Systems 3.2.6 SPR Imaging Instruments 3.2.7 General Optical Requirements for SPR Instruments 3.3 SPR Liquid Handling Systems 3.3.1 Flow Cell Systems 3.3.2 Cuvette Systems 3.4 SPR Instruments: State of the Art 3.4.1 Examples of Fan-shaped Beam SPR Instruments 3.4.2 Examples of Fixed-angle SPR Instruments 3.4.3 Examples of Angle Scanning SPR Instruments 3.4.4 Examples of Grating Coupler SPR Instruments 3.4.5 Examples of Other SPR Instruments 3.4.6 Examples of SPR Imaging Instruments 3.5 Protein Interaction Analysis Systems of Biacore 3.5.1 Introduction 3.5.2 Biacore T100 3.5.3 Biacore A100 3.5.4 FLEXChip 3.6 Conclusion 3.7 Questions References Chapter 4

32 32 33

35 35 38 39 40 40 41 42 43 44 45 45 48 50 50 54 59 60 60 63 69 69 70 73 75 77 78 78

Kinetic Models Describing Biomolecular Interactions at Surfaces Damien Hall 4.1

Introduction 4.1.1 Terminology of Adsorption 4.1.2 Optical Quantification of Adsorption at an Interface

81 82 86

xv

Contents

4.2

Defining Factors of the Adsorption Event 4.2.1 Mass Transfer 4.2.2 Adsorption Mechanisms 4.2.2.1 Idealized Partition Processes 4.2.2.2 Effect of Competing Reactions 4.2.2.3 Surface Functions for Different Modes of Adsorption 4.3 Summary and Conclusions 4.4 Questions 4.5 Symbols 4.6 Acknowledgements References

Chapter 5

87 90 98 99 101 102 115 116 118 119 120

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions: SPR Applications in Drug Development Nico J. de Mol and Marcel J.E. Fischer 5.1 5.2

5.3

Introduction Affinity and Kinetics of a Transport-limited Bimolecular Interaction at the Sensor Surface 5.2.1 Affinity Constants Derived from Equilibrium SPR Signals 5.2.1.1 Correction for Depletion of Free Analyte Concentration in the Cuvette 5.2.2 Affinity Constants and Rate Constants Derived from Kinetic Analysis 5.2.2.1 kobs Kinetic Analysis 5.2.2.2 Global Kinetic Analysis with a Simple Bimolecular Binding Model Detecting Mass Transport Limitation: A Practical Approach 5.3.1 Effect of Viscosity Change on the Association Phase 5.3.2 Transport Limitation in the Dissociation Phase 5.3.2.1 Rebinding Model for Transportlimited Dissociation 5.3.2.2 Competing Ligand to Prevent Rebinding During Dissociation 5.3.2.3 Experimental Procedure to Assay High Off-rates 5.3.3 Quantitative Considerations on Mass Transport Limitation 5.3.3.1 Flow or Cuvette?

123 125 126 128 129 130 131 135 135 137 137

139 141 142 143

xvi

Contents

5.4

Global Kinetic Analysis of Complex Binding Models 5.4.1 Global Kinetic Analysis Including Mass Transport and a Conformational Change 5.4.2 Unusual Kinetics: Intermolecular Bivalent Binding to the Sensor Surface 5.4.3 Global Kinetic Analysis: Concluding Remarks 5.5 Affinity in Solution Versus Affinity at the Surface 5.6 Thermodynamic van’t Hoff Analysis Using SPR Data 5.6.1 van’t Hoff Thermodynamic Analysis 5.6.2 Comparison of SPR Thermodynamics with Calorimetry 5.6.3 Transition State Analysis Using Eyring Plots 5.7 SPR Applications in Pharma Research: Concluding Remarks and Future Perspectives 5.8 Questions 5.9 Symbols 5.10 Acknowledgements References Chapter 6

143 144 147 152 154 158 158 161 163 165 167 168 169 169

Surface Chemistry in SPR Technology Erk T. Gedig 6.1

6.2

6.3

6.4

6.5

Introduction 6.1.1 General Aspects of Surfaces for Biomolecular Interaction Analysis 6.1.2 Selection of the Optimal Surface Adhesion Linking Layers for Gold, Glass and Plastics 6.2.1 Adhesion Linking Layers for Metal Surfaces 6.2.2 Adhesion Linking Layers for Inorganic Dielectrics 6.2.3 Adhesion Linking Layers for Plastics Bioinert Matrices 6.3.1 Non-specific Adsorption of Biomolecules 6.3.2 Bioinert Hydrogels Choosing the Optimal Nanoarchitecture 6.4.1 Two-dimensional Surfaces 6.4.2 Three-dimensional Hydrogels Coupling Procedures for Ligand Immobilization 6.5.1 Adsorptive Immobilization 6.5.2 Preconcentration Methods Prior to Covalent Immobilization 6.5.2.1 Electrostatic Preconcentration 6.5.2.2 Dry Immobilization

173 174 177 181 182 182 183 183 183 185 187 189 191 194 195 195 195 197

xvii

Contents

6.5.3

Covalent Activation Chemistries 6.5.3.1 Amine Coupling via Reactive Esters 6.5.3.2 Amine Coupling Through Reductive Amination 6.5.3.3 Thiol Coupling 6.5.3.4 Immobilization of Aldehydes Through Hydrazide Groups 6.5.3.5 Coupling Through Epoxy Groups 6.5.4 Electrostatic Methods 6.5.5 Directed Immobilization 6.5.6 Immobilization of Membrane Proteins 6.6 Conclusions and Outlook 6.7 Questions References Chapter 7

199 199 202 203 207 208 210 212 213 216 217 218

Measurement of the Analysis Cycle: Scanning SPR Microarray Imaging of Autoimmune Diseases Richard B.M. Schasfoort, Angelique M.C. Lokate, J. Bianca Beusink, Ger J.M. Pruijn and Gerard H.M. Engbers 7.1 7.2 7.3

7.4

7.5

7.6

Introduction The Analysis Cycle Buffer Solutions for Measuring the Analysis Cycle 7.3.1 Baseline Buffer 7.3.2 Regeneration Solution SPR-based Immunoassays 7.4.1 Direct Assay 7.4.2 Competition Assay 7.4.3 Inhibition Assay 7.4.4 Sandwich Assay Detection of Multiplex Analysis Cycles Using Scanning SPR Imaging 7.5.1 Dynamic Range of Scanning SPR Imaging 7.5.2 Liquid Handling Procedures 7.5.3 Determination of the Limit of Detection Using Multiplex Analysis Cycles Monitoring of Autoantibodies in Serum of Rheumatoid Arthritis Patients 7.6.1 Experimental Conditions for Serum Measurements 7.6.1.1 Serum Samples 7.6.1.2 SPR Microarray Interaction Studies 7.6.2 Results and Discussion of Monitoring Analysis Cycles for Autoantibody Screening

221 222 224 224 225 225 226 226 227 228 228 230 232 232 235 235 235 236 236

xviii

Contents

7.7

Features and Benefits of Monitoring Analysis Cycles with SPR Imaging 7.8 Conclusion 7.9 Questions 7.10 Acknowledgement References Chapter 8

Advanced Methods for SPR Imaging Biosensing Alastair W. Wark, Hye Jin Lee and Robert M. Corn 8.1 8.2

Introduction Advances in SPRI Instrumentation and Surface Chemistry 8.3 Surface Enzymatic Transformations for Enhanced SPRI Biosensing 8.3.1 Measuring Surface Enzyme Kinetics 8.3.2 RNase H–Amplified Detection of DNA 8.3.3 Fabrication of RNA Microarrays with RNA-DNA Surface Ligation Chemistry 8.4 Nanoparticle-amplified SPRI Biosensing 8.4.1 Single Nucleotide Polymorphism Genotyping 8.4.2 MicroRNA Detection 8.5 Summary and Outlook 8.6 Questions 8.7 Acknowledgements References Chapter 9

241 242 243 243 243

246 247 254 254 257 259 260 262 264 269 270 271 271

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies Wolfgang Knoll, Amal Kasry, Jing Liu, Thomas Neumann, Lifang Niu, Hyeyoung Park, Harald Paulsen, Rudolf Robelek, Danfeng Yao and Fang Yu 9.1 9.2 9.3

9.4

9.5

Introduction Surface Plasmon Fluorescence Spectroscopy (SPFS) Interface Kinetics Based on the Langmuir Adsorption Model 9.3.1 Mass Transport-controlled Kinetics 9.3.2 Interaction-controlled Kinetics 9.3.3 Equilibrium Analysis Applications of the Kinetic Model 9.4.1 Surface Hybridization Reactions of Oligonucleotides 9.4.2 Protein Binding Studies Novel Approaches to SPFS

275 278 282 284 284 286 286 286 298 300

xix

Contents

9.5.1 Grating Coupling for SPFS 9.5.2 Long-range Surface Plasmons for SPFS 9.5.3 Fluorescence Imaging and Color Multiplexing 9.6 Conclusions 9.7 Questions 9.8 Acknowledgements References

301 303 306 309 309 310 310

Chapter 10 SPR Imaging for Clinical Diagnostics Elain Fu, Timothy Chinowsky, Kjell Nelson and Paul Yager 10.1 10.2

Introduction Achieving Miniaturization and Low Cost 10.2.1 Tuning in SPR Imager Design 10.2.2 Compact Design of Additional SPR Imager Optical Elements 10.3 Optimizing Imager Performance 10.3.1 Refractive Index Resolution 10.3.2 Lateral Resolution Over the Field of View 10.4 Robust Operation 10.4.1 Effects of Temperature Fluctuations 10.4.2 Strategies to Alleviate the Effects of Temperature Fluctuations 10.4.3 Bulk RI Compensation 10.5 SPR Imaging Assays 10.5.1 Microfluidic Immunoassay Design for Small Molecule Analytes 10.5.2 Assay Compatibility with Complex Samples 10.5.3 Assay Implementation 10.6 Conclusion 10.7 Questions References

313 314 315 315 317 317 320 321 321 322 323 324 324 327 329 329 330 330

Chapter 11 The Benefits and Scope of Surface Plasmon Resonance-based Biosensors in Food Analysis Alan McWhirter and Lennart Wahlstro¨m 11.1

Introduction 11.1.1 Why Analyze Food? 11.1.2 Food Analysis Steps and SPR Assay Formats 11.2 Biacore Q and Qflex Kits – the Workhorse of Food Analysis 11.2.1 Qflex Kits for Screening Veterinary Drug Residues

333 334 334 338 339

xx

Contents

11.2.2 Qflex Kits for Quantifying Vitamin Content 11.2.3 AOAC Certification of Qflex Kits 11.3 Examples of Applications for SPR-based Biosensors in Food Analysis 11.3.1 Quantifying Antibiotics in Honey 11.3.2 Screening for Veterinary Drug Residues 11.3.3 Milk Testing 11.3.4 Detecting Antibodies to Salmonella in Meat 11.3.5 Genetically Modified Organisms 11.4 Conclusions 11.5 Questions References

343 344 345 345 346 347 350 350 351 352 352

Chapter 12 Future Trends in SPR Technology Richard B.M. Schasfoort and Peter Schuck 12.1 12.2

Introduction Trends in SPR Instrumentation 12.2.1 SPR Imaging 12.2.2 Hyphenation SPR Technology 12.2.2.1 SPR–MS 12.2.2.2 Other Hyphenated SPR Techniques 12.2.3 Nanoparticle SPR 12.3 Trends in Fluidics 12.3.1 Microarray Spotting on Gold 12.3.1.1 DNA Coding Technology 12.3.2 Prospects for SPR-based Point of Care Devices 12.3.3 Implementation of Lab-on-a-Chip Devices for SPR Systems 12.3.3.1 Pumping Liquids Using Electroosmotic Flow in Microfluidic Devices with Gold Layers 12.3.4 Lab-on-a-Chip Implementation Using Free Flow Electrophoresis and SPR Imaging for Proteomics-on-a-Chip 12.3.5 Digital Microfluidics 12.3.5.1 Cell Diagnosis and Monoclonal Antibody Screening Using SPR Imaging and Digital Microfluidics 12.4 Trends in Sensor Surfaces 12.4.1 Smart Polymer Brushes 12.4.2 Photoactivation of Surfaces for Immobilization 12.4.3 Gradient Chemistries

354 355 356 356 357 359 360 361 362 364 366 367

367

369 373

375 376 376 378 380

xxi

Contents

12.5

Trends in Measuring Reliable Kinetic Parameters 12.5.1 Introduction 12.5.2 The Model for Distribution Analysis of Rate and Equilibrium Constants 12.5.3 Examples of the Distribution Analysis Method 12.5.4 Conclusions and Perspectives of the Distribution Analysis Model 12.6 Final Comments 12.7 Questions References Subject Index

381 381 383 385 388 391 391 392 395

CHAPTER 1

Introduction to Surface Plasmon Resonance ANNA J. TUDOSa AND RICHARD B.M. SCHASFOORTb a

Shell Global Solutions International BV, P.O. Box 38000 1030 BN Amsterdam, The Netherlands; b Biochip Group, MESA+ Institute for Nanotechnology, Biomedical Technology Institute (BMTI), Faculty of Science and Engineering, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands

1.1 What is Surface Plasmon Resonance? Since its first observation by Wood in 1902 [1,2], the physical phenomenon of surface plasmon resonance (SPR) has found its way into practical applications in sensitive detectors, capable of detecting sub-monomolecular coverage. What is surface plasmon resonance? Wood observed a pattern of ‘‘anomalous’’ dark and light bands in the reflected light, when he shone polarized light on a mirror with a diffraction grating on its surface. Physical interpretation of the phenomenon was initiated by Lord Rayleigh [3], and further refined by Fano [4], but a complete explanation of the phenomenon was not possible until 1968, when Otto [5] and in the same year Kretschmann and Raether [6] reported the excitation of surface plasmons. Application of SPR-based sensors to biomolecular interaction monitoring was first demonstrated in 1983 by Liedberg et al. [7]. A historical overview of the use of the phenomenon for biosensor applications is given in Section 1.3 of this chapter. To understand the excitation of surface plasmons, let us start with a simple experiment.

1.1.1 A Simple Experiment Consider the experimental set-up depicted in Figure 1.1. When polarized light is shone through a prism on a sensor chip with a thin metal film on top, the light will be reflected by the metal film acting as a mirror. On changing the angle of incidence, and monitoring the intensity of the reflected light, the intensity of the 1

2

Chapter 1 Intensity of reflected light (%)

ϕ

A

B

Angle (ϕ)

Figure 1.1

Schematic experimental set-up of surface plasmon resonance excitation. A sensor chip with a gold coating is placed on a hemisphere (or prism). Polarized light shines from the light source (star) on the sensor chip. Reflected light intensity is measured in the detector (disk). At a certain angle of incidence (j), excitation of surface plasmons occurs, resulting in a dip in the intensity of the reflected light (A). A change in refractive index at the surface of the gold film will cause an angle shift from A to B.

reflected light passes through a minimum (Figure 1.1, line A). At this angle of incidence, the light will excite surface plasmons, inducing surface plasmon resonance, causing a dip in the intensity of the reflected light. Photons of p-polarized light can interact with the free electrons of the metal layer, inducing a wave-like oscillation of the free electrons and thereby reducing the reflected light intensity. The angle at which the maximum loss of the reflected light intensity occurs is called resonance angle or SPR angle. The SPR angle is dependent on the optical characteristics of the system, e.g. on the refractive indices of the media at both sides of the metal, usually gold. While the refractive index at the prism side is not changing, the refractive index in the immediate vicinity of the metal surface will change when accumulated mass (e.g. proteins) adsorb on it. Hence the surface plasmon resonance conditions are changing and the shift of the SPR angle is suited to provide information on the kinetics of e.g. protein adsorption on the surface.

1.1.2 From Dip to Real-time Measurement Surface plasmon resonance is an excellent method to monitor changes of the refractive index in the near vicinity of the metal surface. When the refractive index changes, the angle at which the intensity minimum is observed will shift as indicated in Figure 1.2, where (A) depicts the original plot of reflected light intensity vs. incident angle and (B) indicates the plot after the change in refractive index. Surface plasmon resonance is not only suited to measure the difference between these two states, but can also monitor the change in time, if one follows in time the shift of the resonance angle at which the dip is observed.

3

Introduction to Surface Plasmon Resonance

B Angle A

Time (s)

Figure 1.2

A sensorgram: the angle at which the dip is observed vs. time. First, no change occurs at the sensor and a baseline is measured with the dip at SPR angle (A). After injection of the sample (arrow) biomolecules will adsorb on the surface resulting in a change in refractive index and a shift of the SPR angle to position B. The adsorption–desorption process can be followed in real time and the amount of adsorbed species can be determined.

Figure 1.2 depicts the shift of the dip in time, a so-called sensorgram. If this change is due to a biomolecular interaction, the kinetics of the interaction can be studied in real time. SPR sensors investigate only a very limited vicinity or fixed volume at the metal surface. The penetration depth of the electromagnetic field (so-called evanescent field) at which a signal is observed typically does not exceed a few hundred nanometers, decaying exponentially with the distance from the metal layer at the sensor surface. The penetration depth of the evanescent field is a function of the wavelength of the incident light, as explained in Chapter 2. SPR sensors lack intrinsic selectivity: all refractive index changes in the evanescent field will be reflected in a change of the signal. These changes can be due to refractive index difference of the medium, e.g. a change in the buffer composition or concentration; also, adsorption of material on the sensor surface can cause refractive index changes. The amount of adsorbed species can be determined after injection of the original baseline buffer, as shown in Figure 1.2. To permit selective detection at an SPR sensor, its surface needs to be modified with ligands suited for selective capturing of the target compounds but which are not prone to adsorbing any other components present in the sample or buffer media.

1.2 How to Construct an SPR Assay? Now we have a basic understanding of the surface plasmon resonance signal and how to measure it in time. We know that the sensor surface needs to be modified to allow selective capturing and thus selective measurement of a target compound.

4

Chapter 1

In the following, we are going to learn more about an SPR measurement. First, the steps of an SPR assay will be discussed from immobilization through analysis to regeneration in a measurement cycle. Next, we get acquainted with a typical calibration curve, followed by examples of assay formats. Finally, a short outlook is provided on the basics of the instrumentation.

1.2.1 The Steps of an Assay In the simplest case of an SPR measurement, a target component or analyte is captured by the capturing element or so-called ligand (Figure 1.3). The ligand is permanently immobilized on the sensor surface previous to the measurement. Various sensor surfaces with immobilized ligands are commercially available, and many more can be custom-made, as explained in Chapters 6 and 7. In the simplest case, the event of capturing the analyte by the ligand gives rise to a measurable signal, this is called direct detection. Figure 1.4 shows the sensor signal step-by-step in the measurement cycle with direct detection. Each measurement starts with conditioning the sensor surface with a suitable buffer solution (1). It is of vital relevance to have a reliable baseline before the capturing event starts. At this point, the sensor surface contains the active ligands, ready to capture the target analytes. On injecting the solution containing

Flow of sample with analyte

Bound ligand

Y

Gold 50 nm

Glass

Figure 1.3

Schematic representation of direct detection: the analyte is captured by the ligands (Y) immobilized on the sensor surface. Accumulation of the analyte results in a refractive index change in the evanescent field shifting the SPR angle. Here the ligand is immobilized in a hydrogel.

5

Introduction to Surface Plasmon Resonance

SPRdip shift ∆R

Injection

Time

t1

Step:

1. baseline

Figure 1.4

2. association

3. dissociation

4. regeneration

1. baseline

Sensorgram showing the steps of an analysis cycle: 1, buffer is in contact with the sensor (baseline step); 2, continuous injection of sample solution (association step); 3, injection of buffer (dissociation step); DR indicates the measured response due to the bound target compound; 4, removal of bound species from the surface during injection of regeneration solution (regeneration step) followed by a new analysis cycle. A bulk refractive index shift can be observed at t1. See also page 222.

the analytes (2), they are captured on the surface. Also other components of the sample might adhere to the sensor surface; without a suitable selection of the ligand, this adherence will be non-specific, and thus easy to break. At this step, adsorption kinetics of the analyte molecule can be determined in a real-time measurement. Next, buffer is injected on to the sensor and the non-specifically bound components are flushed off (3). As indicated in the figure, the accumulated mass can be obtained from the SPR response (DR). Also in this step, dissociation of the analyte starts, enabling the kinetics of the dissociation process to be studied. Finally, a regeneration solution is injected, which breaks the specific binding between analyte and ligand (4). If properly anchored to the sensor surface, the ligands remain on the sensor, whereas the target analytes are quantitatively removed. It is vital in order to perform multiple tests with the same sensor chip to use a regeneration solution which leaves the activity of the ligands intact, as the analysis cycle is required to take place repeatedly for hundreds, sometimes even thousands of times. Again, buffer is injected to condition the surface for the next analysis cycle. If the regeneration is incomplete, remaining accumulated mass causes the baseline level to be increased. Often SPR measurements are carried out to determine the kinetics of a binding process. For realistic results it is vital to prevent immobilization from

6

Chapter 1

changing the ligand in a way that would influence its strength or affinity towards the target component. In addition, kinetic experiments can provide information on the thermodynamics, e.g. on the binding energy of processes. A description of the kinetic theory can be found in Chapter 4 and examples of kinetic studies in Chapter 5.

1.2.2 Calibration Curve

SPR-dip shift

Apart from kinetic and thermodynamic studies, SPR measurements can also be used for the determination of the concentration of the analyte in a sample (quantitative analysis). In this case, first different concentrations of the analyte are applied in separate analysis cycles. The sensorgrams measured at different concentrations give an overlay plot similar to that depicted in Figure 1.5, with the plateaus of the association step increasing at increasing analyte concentration [8]. A calibration curve can be constructed by plotting the response (DR) after a certain time interval (t1) versus concentration. When analyzing samples with an unknown concentration of the analyte, usually multiple dilutions are made, for example 10, 100 and 1000 times, or for more accurate determinations serial dilutions by a factor of 2. If the concentration

dR/dt

R

Injection t0

Step:

1. baseline

Figure 1.5

2. association

t1

3. dissociation

time

4. regeneration 1. baseline

Typical overlay plot of sensorgrams from serial diluted analyte concentrations. Just after injection at t0 a sample specific binding of the analyte occurs and mass transport to the surface is rate limiting and linearly dependent on the concentration. From the slopes of a positive control (dR/dt), the concentration of an unknown sample can be determined. During the association phase the number of unbound ligand molecules decreases and dissociation takes place. The off-rate constant or dissociation constant (kd) can be determined after injecting dissociation buffer at t1. See for more details chapter 4 and 5 of this book.

7

Introduction to Surface Plasmon Resonance

of the analyte in the sample is very high, the undiluted sample will yield results on the upper plateau range of the calibration curve. The diluted solutions, however, might yield points along the lower, concentration-dependent sections of the calibration curve and the concentration of the analyte can be determined. As mentioned above, SPR sensing means detection of refractive index changes at the sensor surface, which in practice translates to the amount of mass deposited at the sensor surface. Direct detection is only possible if the capturing event of the analyte brings about measurable refractive index changes. This is easier to achieve if the molecular weight of the analyte is high (i.e. around 1000 Da or higher). However, for small molecules to produce a measurable refractive index change, large numbers would be required, making the analysis intrinsically less sensitive. If the analyte is a small molecule (MWo1000 Da), often direct detection is not viable. Detection of small molecules can be carried out using a different strategy. Most often, small molecules are detected in a sandwich, competition or inhibition assay format. In all assay formats, not only the lower detectable concentration is limited, but also the physical number of immobilized elements on the sensor surface, which provides a maximum limiting value. Discussion of the different assay formats can be found in Chapter 7 and other methods for concentration determination are described in Chapters 4 and 5.

1.2.3 Determination of Kinetic Parameters The most prominent benefit of direct detection using SPR biosensor technology is the determination of kinetics of (bio)molecular interactions. Reaction rate and equilibrium constants of interactions can be determined, e.g. the interaction A + B - AB can be followed in real time with SPR technology, where A is the analyte and B is the ligand immobilized on the sensor surface. Table 1.1 contains the most relevant kinetic parameters, the association and dissociation constants, for the simplest case A + B - AB. The association constant is the reaction rate of complex (AB) formation, giving the number of complexes formed per time at unit concentration of A and B. As soon as the complex AB is formed, its dissociation can commence. The dissociation rate constant describing this process expresses the number of AB complexes Table 1.1

Definitions of the most relevant kinetic parameters: the association and dissociation constants.

Definition Description

Units Typical range

Association rate constant, ka

Dissociation rate constant, kd

A + B - AB Reaction rate of AB formation: number of AB complexes formed per unit time at unit concentration of A and B l mol
1 s
1 103–107

AB - A + B Dissociation rate of AB: number of AB complexes dissociating per unit time s
1 10
1
5  10
6

8

Table 1.2

Chapter 1

Definition of the equilibrium association and dissociation constants.

Definition Description Unit Typical range

Equilibrium association constant, KA

Equilibrium dissociation constant, KD

[AB]/[A][B] ¼ ka/kd Affinity to association: high KA, high affinity to associate l mol
1 105–1012

[A][B]/[AB] ¼ kd/ka Stability of AB: high KD, low stability of AB mol l
1 10
5–10
12

dissociating per unit time. Note that the unit dimensions for the association and dissociation rates are different and can vary with the stoichiometry of the complex. The typical range of the association and dissociation constant shows large variations and is dependent on, among other things, the temperature. When association of A and B starts, no product is yet present at the sensing surface. At this point, the rate of the association reaction is highest and that of the dissociation reaction is lowest. As the process progresses, more and more of the AB complex is produced, enhancing the rate of dissociation. Due to decreasing A and B concentration, the rate of association might decrease. Equilibrium is reached when the rates of the association and dissociation reactions are equal; the definitions and unit dimensions are given in Table 1.2. As can be seen, the equilibrium association and dissociation constants, which represent the affinity of an interaction, have a reciprocal relationship with each other. The effect of parameters such as temperature is described in later chapters. The rate constants (Table 1.1) and equilibrium constants (Table 1.2) of (bio)molecular interactions provide information on the strength of association and the tendency of dissociation. Various aspects of kinetics, models and calculation of affinity constants are described in Chapters 4, 5 and 9.

1.2.4 Basics of Instrumentation Studying biomolecular interactions using SPR does not require a detailed understanding of the physical phenomena. It is sufficient to know that SPRbased instruments use an optical method to measure the refractive index near a sensor surface (within B200 nm to the surface). SPR instruments comprise three essential units integrated in one system: optical unit, liquid handling unit and the sensor surface. The features of the sensor chip have a vital influence on the quality of the interaction measurement. The sensor chip forms a physical barrier between the optical unit (dry section) and the flow cell (wet section). SPR instrumentation can be configured in various ways to measure the shift of the SPR-dip. In general, three different optical systems (Chapter 2) are used to excite surface plasmons: systems with prisms, gratings and optical waveguides. Most widespread are instruments with a prism coupler, also called ‘‘Kretschmann configuration’’ [9]. In this configuration, which is shown in Figure 1.1, a prism couples p-polarized light into the sensor coated with a thin metal film. The light is

Introduction to Surface Plasmon Resonance

9

reflected on to a detector, measuring its intensity, using a photodiode or a camera. In instruments with a grating coupler [10], light is reflected at the lower refractive index substrate. In practice, this means that light travels through the liquid before photons generate surface plasmon waves as in ellipsometric instruments [11]. Besides the grating couplers, some instruments apply optical waveguide couplers [12] or measure the SPR wavelength shift as a result of the biomolecular interaction process (see Chapter 2 and ref. [13]). All configurations share the same intrinsic phenomenon: the direct, label-free and real-time measurement of refractive index changes at the sensor surface. SPR sensors offer the capability of measuring low levels of chemical and biological compounds near the sensor surface. Sensing of a biomolecular binding event occurs when biomolecules accumulate at the sensor surface and change the refractive index by replacing the background electrolyte. Protein molecules have a higher refractive index than water molecules (Dn E 10
1). The sensitivity of most SPR instruments is in the range Dn E 10
5 or 1 pg mm
2 of proteinous material. Often in real-time biosensing absolute values are not a prerequisite, only the change is monitored as a result of biospecific interaction at the sensor surface. A detailed description of commercial instruments is given in Chapter 3.

1.3 History of SPR Biosensors The term biosensor was introduced around 1975, relating to exploiting transducer principles for the direct detection of biomolecules at surfaces. Currently the most prominent example of a biosensor is the glucose sensor, reporting glucose concentration as an electronic signal, e.g. based on a selective, enzymatic process. Some argued that all small devices capable of reporting parameters of the human body were biosensors (e.g. ion-sensitive field-effect transistors (ISFETs) measuring pH). But then, a thermometer recording fever should also be called biosensor. According to the present definition, in biosensors the recognition element (ligand) of the sensor or the analyte should originate from a biological source.

Biosensors are analytical devices comprised of a biological element (tissue, microorganism, organelle, cell receptor, enzyme, antibody) and a physicochemical transducer. Specific interaction between the target analyte and the biological material produces a physico-chemical change detected by the transducer. The transducer then yields an analog electronic signal proportional to the amount (concentration) of a specific analyte or group of analytes.

1.3.1 Early History of SPR Biosensors Application of SPR-based sensors to biomolecular interaction monitoring was first demonstrated in 1983 by Lundstrom’s pursuit towards physical

10

Chapter 1

methods for label-free, real-time detection of biomolecules [7]. The intrinsic properties of the molecules, e.g. mass, refractive index and/or charge distribution [14], were probed using ellipsometry, refractometry, surface plasmon resonance, photothermic detection methods and others. At the National Defense Research Laboratory of Sweden, protein–protein interactions were monitored in real time, label-free, using ellipsometry. Most importantly, the refractive index change at a light-reflecting surface was the operating transducer mechanism. Although successful in the detection of refractive index change due to the binding of biomolecules on optical transducer surfaces, a disadvantage of the ellipsometer is that light passes through the bulk of the sample solution, hence light-absorbing or particle-containing samples cannot easily be measured. Among other research laboratories in the same period, the University of Twente (The Netherlands) was active in the search for finding new transduction principles for measuring immunochemical reactions at field effect transistor devices (ImmunoFET) [15] and at surfaces with an optical read-out (immunochemical optical biosensor, IMOB). Optical transducer principles [16] including ellipsometry, surface plasmon resonance and interferometric principles (Mach Zehnder) showed promise for direct transduction of the biomolecular binding event. Successful measurements of immunochemical reactions using SPR were carried out as early as in the mid-1980s [17]. Pharmacia Biosensor AB chose SPR as their platform technology for direct sensing of biomolecular interactions. The Kretschmann configuration offered advantages in freedom of design of the liquid handling system. Coming from the higher refractive index medium (the prism), light does not pass through the liquid, but is reflected at the sensor surface covered with a thin metal layer. Gold was chosen as the best inert metal film required for surface plasmon resonance, although from a physical point of view silver provides a better SPR effect (see Chapter 2). Studies on the surface chemistry led to modification of the gold with a selfassembling layer of long-chain thiols to which a hydrogel could be attached. Carboxylated dextran was immobilized at the surface, which provides a substrate for efficient covalent immobilization of biomolecules, in addition to a favorable environment for most biomolecular interactions. The thickness of the dextran hydrogel of 100 nm is perfectly compatible with the ca. 200 nm evanescent field (see Figure 1.3). The reliable production of these high-quality sensor chips was unequivocally the basis for the successful launch of SPR instruments. Techniques were developed to etch silica to form a casting mold for the manufacture of microfluidic flow channels. Also, development proceeded on optogels for use between the prism in the optical unit of the instrument and the sensor chip. The optogel ensures optical contact with the prism, allowing simple replacement of the sensor chip. These efforts in research and development relied on the combination of three unrelated fields: optics, microfluidics and surface chemistry, and resulted in the successful development of the instrumental concept of biomolecular interaction analysis (BIA).

Introduction to Surface Plasmon Resonance

11

1.3.2 History of SPR Biosensors After 1990 In 1990, Pharmacia Biosensor AB launched the first commercial SPR product, the Biacore instrument [18]. The instrument was the most advanced, sensitive, accurate, reliable, reproducible direct biosensor technique and SPR became (and still is) the ‘‘golden standard’’ of transducer principles for measuring realtime biomolecular interactions. Since the early 1990s, producers have been struggling to meet the standards set by Biacore. Fisons Instruments1 [19] made serious attempts to compete with Biacore’s technology; their cuvette-based IAsys instrument uses evanescent field-based technology, essentially not SPR, for the study of biomolecular interactions. The Biacore 2000 instrument was introduced in 1994 with improved detection and a different flow system so that the sample could interact at four spots on the sensor. Data of the reference spot could be used for signal correction. With the introduction of Biacore 2000 it also became possible to monitor directly interactions of small molecule analytes reacting with immobilized protein ligands [20]. In 1995, the cuvette based SPR system of IBIS Technologies was launched. The instrument was compatible with the Biacore sensor chip. In 1997, the IBIS II, a two-channel cuvette-based SPR instrument with autosampler operation, was introduced [21]. Following the merger with the sensor chip coating company Ssens BV in 1999, the development of an SPR imaging instrument was initiated at IBIS Technologies. In 2007, the development of the IBIS-iSPR instrument, with the scanning angle principle, resulted in the required reliability and accuracy for microarray imaging of multiple biomolecular interactions (4500). The potency of the instrument is demonstrated in Chapter 7. Biacore X, a two-spot instrument introduced in 1996, was followed by the Biacore 3000 in 1998. The latter was later extended with recovery tools to improve interfacing with mass spectrometry [22]. Biacore Q was introduced for the food analysis market in 2000 (Chapter 11). Positioned for small molecule analysis and drug discovery, the introduction of the Biacore S51 marked a technology shift in terms of detection, flow cell design and sample capacity: the area of the detected spot was reduced from 1 to 0.01 mm2 and the number of spots was increased from four to six. In 2004, a high-end instrument was introduced with four channels each with five sensor spots (Biacore A100). Combining the flow cell of the Biacore S51 instrument and the performance of the four-channel Biacore 3000, this instrument has 20 in-line sensors to monitor biomolecular interactions in the flow cells. The technology is not suitable, however, to image the surface. In Chapter 3, other Biacore instruments (T100 and X100) are described. In order to measure up to 400 interactions simultaneously, in 2005 Biacore acquired the grating coupler SPR system of HTS Biosystems, co-developed with Applied Biosystems (8500 Affinity Analyzer), which was capable of imaging the sensor surface. After restyling, this product (named Flexchip) was launched in 2006 [10]. 1

Later Affinity Sensors.

12

Chapter 1

Although it is impossible to describe accurately the history of the developments of the 25 companies producing SPR (related) instruments (see Chapter 3), it is justified to treat the history along the Biacore product line. During the years following the introduction of the first SPR instrument, detection sensitivity has improved by roughly 20-fold. The range of affinity and kinetic data that can be determined has been extended at least 100-fold as a consequence of the increased sensitivity and due to improvements in data analysis. The amount of independent sensor surfaces grew from four channels in 1990 (Biacore) to at least 500 in the new IBIS SPR imaging instrument. The carboxymethylated dextran surface introduced in 1990 [23], still the first choice for many applications, has been complemented with a range of other surfaces. Systems for dedicated applications have been introduced by various manufacturers as complements to all-purpose research instrumentation [24]. A good gauge of the success of biosensor technology is that more than 1000 publications each year include data collected from commercial biosensors. In the paper entitled ‘‘Survey of the 2005 commercial optical biosensor literature’’, Rich and Myszka [25] gave an outstanding overview of the SPR literature, including practical lessons in performing and interpreting biomolecular interaction analysis experiments. The majority of the publications (985) in 2005 employed Biacore technology (87%), indicating the relevance of Biacore’s technology in the market. Affinity Sensors was the runner-up company with 40 publications (B4%), Eco Chemie/Windsor Scientific (distributor) totaled 18 publications. which was essentially from the same technology provider (originally IBIS Technologies), Texas Instruments scored 17 publications in 2005 and 60 publications (B6%) were attributed to 13 other companies. With the introduction of a number of new SPR instruments (Chapter 3) and a series of novel sensor surfaces and chemistries, the impact of SPR biosensors on molecular interaction studies will continue to grow. With improved experimental design, including SPR imaging instruments and advanced data analysis methods, high-quality data for the determination of kinetic parameters of biomolecular interaction phenomena can be obtained. These data promise additional insights into the mechanisms of molecular binding events, which will be important for function–regulatory protein interaction studies in order to unravel the exciting processes in living species.

1.4 How to Read This Book Although most chapters can be read as stand-alone literature on different aspects of SPR technology, this handbook aims to provide the reader with a total coverage of the basics of the technique and applications and the most relevant developments at the time of reviewing. The book starts with a description of the physics of surface plasmons and SPR in its original form and some novel applications, for example, nanoparticle SPR. The description of SPR instrumentation and a survey of currently available commercial products from 25 companies follows in Chapter 3. An introduction on how to obtain kinetic information from SPR measurements can be found in

Introduction to Surface Plasmon Resonance

13

Chapter 4, followed by Chapter 5 illustrating kinetic and thermodynamic analysis of ligand–receptor interactions, probing the validity of this approach in pharmaceutical applications. Chapter 6 brings the reader closer to the surface architecture and chemical design strategies of SPR sensors. An in-depth treatise on the analysis cycle and modern assay architecture, including SPR microarray imaging, is provided in Chapter 7, followed by advanced methods for SPR imaging biosensing in Chapter 8. Specific application areas are highlighted in the last few chapters of the book, revealing Surface Plasmon Fluorescence Techniques (Chapter 9) and the future of medical applications at the point of patient care (Chapter 10) and for food safety (Chapter 11). Finally, Chapter 12 gives an outlook on future trends in SPR technology, including ‘‘lab-on-a-chip’’ microfluidics and trends for measuring reliable kinetic parameters.

1.5 Questions 1. SPR technology for direct and label-free detection of biomolecular interactions dominates affinity biosensor technologies to a great extent and it is expected that in 2007 more than 1000 papers regarding SPR results will be published. What are the technical reasons for the success of SPR? 2. In SPR, the intrinsic refractive index of a protein which accumulates on the sensor surface is measured. Explain how we can distinguish between the refractive index of the buffer and that of the adsorbed protein. 3. Why should we express the sensitivity of an SPR instrument in accumulated mass per square surface and not in moles per liter? 4. Consider the monophasic reversible interaction A + B " AB, where A is the analyte and B is the immobilized ligand. The sample is injected and shows a higher background electrolyte refractive index. Draw the sensorgram of two analysis cycles of injection of a sample with the second analysis cycle a two times diluted sample. Consider 100% regeneration after each analysis cycle. 5. The response DR gives us an indication of the amount of accumulated mass per unit surface area. How can we determine the concentration of an analyte in solution from these responses? 6. The study of the rate constants of biomolecular interactions is an important feature of surface plasmon resonance biosensors. Why?

References 1. 2. 3. 4. 5.

R.W. Wood, Philos. Mag., 1902, 4, 396–402. R.W. Wood, Philos. Mag., 1912, 23, 310–317. Lord Rayleigh, Proc. R. Soc. London, Ser. A, 1907, 79, 399. U. Fano, J. Opt. Soc. Am., 1941, 31, 213–222. A. Otto, Z. Phys., 1968, 216, 398–410.

14

Chapter 1

6. E. Kretschmann and H. Reather, Z. Naturforsch., Teil A, 1968, 23, 2135–2136. 7. B. Liedberg, C. Nylander and I. Lundstrom, Sens. Actuators, 1983, 4, 299–304. 8. S. Geib, G. Sandoz, K. Mabrouk, A. Matavel, P. Marchot, T. Hoshi, M. Villaz, M. Ronjat, R. Miquelis, C. Leveque and M. de Waard, Biochem. J., 2002, 364, 285–292. 9. E. Kretschmann, Z. Phys., 1971, 241, 313–324. 10. D. Wassaf, G. Kuang, K. Kopacz, Q.L. Wu, Q. Nguyen, M. Toews, J. Cosic, J. Jacques, S. Wiltshire, J. Lambert, C.C. Pazmany, S. Hogan, R.C. Ladner, A.E. Nixon and D.J. Sexton, Anal. Biochem., 2006, 351, 241–253. 11. M. Pe´rez-Moralesa, J.M. Pedrosab, E. Mun˜oza, M.T. Martı´ n-Romeroa, D. Mo¨biusc and L. Camachoa, Thin Solid Films, 2005, 488, 247–253. 12. J. Ctyrokya, J. Homola and M. Skalskya, Opt. Quantum Electron., 1997, 29, 301–311. 13. C.R. Yonzon, E. Jeoung, S. Zou, G.M. Mrksich and R.P. van Duyne, J. Am. Chem. Soc., 2004, 126, 12669–12676. 14. Z. Salamon, H.A. Macleod and G. Tollin, Biochim. Biophys. Acta, 1997, 1331, 117–129a. 15. R.B.M. Schasfoort, R.P.H. Kooyman, P. Bergveld and J. Greve, Biosens. Bioelectron., 1990, 5, 103–125. 16. J. Homola, Surface Plasmon Resonance Based Sensors. Springer Series on Chemical Sensors and Biosensors, Series Ed. O.S. Wolfbeis, Vol. 4, Springer, Berlin, 2006. 17. R.P.H. Kooyman, H. Kolkman, J. Van Gent and J. Greve, Anal. Chim. Acta, 1988, 213, 35–45. 18. U. Jonsson, L. Fagerstam, B. Ivarsson, B. Johnsson, R. Karlsson, K. Lundh, S. Lofas, M. Malmqvist, BioTechniques, 1991, 11, 620–622, 624. 19. L.A. Chtcheglova, M. Vogel, H.J. Gruber, G. Dietler and A. Haeberli, Biopolymers, 2006, 83, 69–82. 20. R. Karlsson and R. Stahlberg, Anal. Biochem., 1995, 228, 274–280. 21. T. Wink, J. de Beer, W.E. Hennink, A. Bult and W.P. van Bennekom, Anal Chem., 1999, 71, 801–805. 22. D. Nedelkov, A. Rasooly and R.W. Nelson, Int. J. Food Microbiol., 2000, 60, 1–13. 23. B. Johnsson, S. Lofas and G. Lindquist, Anal. Biochem., 1991, 198, 268–277. 24. R.L. Rich and D.G. Myszka, J. Mol. Recognit., 2006, 19, 478–534. 25. R.L. Rich and D.G Myszka, Anal. Biochem., 2007, 361, 1–6.

CHAPTER 2

Physics of Surface Plasmon Resonance ROB P.H. KOOYMAN Biophysical Engineering Group, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands

2.1 Introduction In the last two decades, surface plasmon resonance (SPR) has evolved from a fairly esoteric physical phenomenon to an optical tool that is widely used in physical, chemical and biological investigations where the characterization of an interface is of interest. Recently, the field of SPR nano-optics has been added where metallic structures on a nanoscale can be designed such that they can perform certain optical functions. This chapter will be mainly concerned with the more conventional, well-understood SPR theory used in sensor applications and it will touch upon some of the newer developments relevant for this area. Essential for the generation of surface plasmons (SPs) is the presence of free electrons at the interface of two materials – in practice this almost always implies that one of these materials is a metal where free conduction electrons are abundant. This condition follows naturally from the analysis of a metaldielectric interface by Maxwell’s equations. From this analysis, the picture emerges that surface plasmons can be considered as propagating electron density waves occurring at the interface between metal and dielectric. Alternatively, surface plasmons can be viewed as electromagnetic waves strongly bound to this interface; it is found that the surface plasmon field intensity at the interface can be made very high, which is the main reason why SPR is such a powerful tool for many types of interface studies. Experimental research on SPs started with electron beam excitation; in 1968, optical excitation was demonstrated by Otto [1] and Kretschmann and Raether [2]. This last approach turned out to be much more versatile, so in this chapter the focus will be on the optics of SPR. The following is by no means intended as an 15

16

Chapter 2

in-depth treatment of surface plasmons, rather it is an attempt to provide a lowthreshold introduction to the physics of SPR for those who are actually involved in SPR work and want to understand a bit more than ‘‘measuring the shift of the SPR dip’’.

2.2 The Evanescent Wave Before we discuss SPs in more detail, it may be appropriate to provide a mathematical description of the evanescent wave, which is so central in the concept of SPR sensing. This is conveniently done by considering the phenomenon of total internal reflection. An electromagnetic plane wave that propagates in a medium with refractive index n can mathematically be described by an electric field E:
 E ¼ E0 expðjot
jk  rÞ ¼ E0 exp jot
jkx x
jky y
jkz z ð2:1Þ where E0 is the amplitude of the electric field, o is the angular frequency, k is the wavevector, r ¼ (x,y,z) is the position vector and j ¼ O
1. Note that eq. (2.1) only represents a traveling wave if the exponent is complex. In the present context, we will mainly be concerned with the wavevector k: its direction is parallel to that of the wave propagation; its magnitude is given by qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2p o k ¼ k2x þ k2y þ k2z ¼ n ¼ n ð2:2Þ l c where l and c are the wavelength and propagation velocity in vacuum, respectively. Next we consider the refraction of such a wave at an interface between two media 1 and 2 with refractive indices n1 and n2, respectively (see Figure 2.1). Without loss of generality, we can choose the direction of the light beam such that kz ¼ 0 and our problem becomes essentially two-dimensional. From elementary physics we know that for this situation Snell’s law holds: n1 sin a ¼ n2 sin b

ð2:3aÞ

kx1 ¼ kx2  kx

ð2:3bÞ

or, equivalently,

By using eqs. (2.2) and (2.3b), we can find an expression for the component of the wavevector ky perpendicular to the interface1:   2  2 n2 2 2 2 2p
sin a ð2:4Þ k y 2 ¼ n1 l n21 Now, let us assume that n1 4 n2. From eq. (2.4), it is seen that for sin a 4 n2/n1 the right part is negative, and, consequently, ky is purely imaginary. Returning to 1

Note that the direction y is in this chapter always perpendicular to the surface.

Physics of Surface Plasmon Resonance

Figure 2.1

17

Refraction of light at an incident angle a, at an interface of two materials with refractive indices n1 and n2. Definition of axis system and quantities.

eq. (2.1), we conclude that for this case in medium 2 there is only a traveling wave parallel to the interface: E2 ¼ E0 e
ky2 y expðjot
jkx xÞ

ð2:5Þ

with the amplitude of the electric field exponentially decaying along the y-direction with a characteristic distance 1/ky2  1/jky2. For obvious reasons, this field in medium 2 is denoted as the evanescent field. Eq. (2.4) can be used to calculate its penetration depth, which is of the order of half a wavelength. This explains the interface sensitivity of the evanescent field: only close to the interface is an electromagnetic field present; therefore, only a changing dielectric property (e.g. a changing refractive index) in the vicinity of the interface will influence this field. We will see that also in SPR an evanescent field is generated.

2.3 Surface Plasmons 2.3.1 Surface Plasmon Dispersion Equations, Resonance There are several approaches that all result in the dispersion relation for an SP, that is, a relation between the angular frequency o and the wavevector k. In his last standard treatise on SPs, Raether [3] calculated the SP dispersion relation from first principles, viz. Maxwell’s equations. A particularly elegant approach was suggested by Cardona [4] and we will adopt it here. For reasons that will become clear in the course of Section 2.3.2, we will only discuss p-polarized2 2

p-Polarized light has its electric field vector in the plane of incidence.

18

Chapter 2

light interacting with an interface. For any interface between two media, the complex reflection coefficient rp for p-polarized incident light electric field is described by Fresnel’s equations (see, e.g., ref. [5] for a derivation on the basis of Maxwell’s equations):   Ei   jj tanða
bÞ jj e rp ¼ ¼ rp e ¼  ð2:6aÞ tanða þ bÞ Er where Ei and Er are the incident and reflected electric fields, respectively, and the angles a and b are defined as shown in Figure 2.1.3 Of course, the angles a and b are again related by Snell’s law [eq. (2.3)]; in addition, a phase change j of the reflected field relative to the incident field occurs, depending on the refractive indices of the materials involved. For the reflectance, defined as the ratio of the reflected intensities, the following relation holds:  2 R p ¼ rp  ð2:6bÞ Now, following Cardona [4], two special cases exist: if a+b ¼ p/2, then the denominator of eq. (2.6a) becomes very large and thus Rp becomes zero. This situation describes the Brewster angle, where there is no reflection for p-polarized light. The other special case occurs when a
b ¼ p/2: we see from eqs. (2.6a) and (2.6b) that Rp becomes infinite: there is a finite Er for a very small Ei. This circumstance corresponds to resonance. From this relation between a and b we can deduce the dispersion relation if a
b ¼ p/2, then cosa ¼
sinb and tana ¼ k1x/k1y ¼
n2/n1. For the components of the wavevector k ¼ (kx, ky), we can write e1 e2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffi o e2i kyi ¼ c e1 þ e2

k2x ¼ k21
k2y1 ¼ k21
k2x o kx ¼ c

rffiffiffiffiffiffiffiffiffiffiffiffiffiffi e1 e2 and e1 þ e2

ð2:7Þ ð2:8Þ

where e1 and e2 are the dielectric constants4 of materials 1 and 2, respectively, and i ¼ 1 or 2. Equation (2.8) is the sought SPR dispersion equation for an interface between two half-infinite media. Next, we investigate the case where medium 2 is a metal. This medium then contains a large number of free electrons and the consequence is that at an angular frequency ooop its dielectric constant e2 will be negative (see, e.g., ref. [5]): o2p o2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi op ¼ 4pne e2 =me e2 ðoÞ ¼ 1


3 4

Note that in Figure 2.1 the direction y (instead of z) is perpendicular to the surface. Dielectric constant and refractive index are related by e  n2.

ð2:9aÞ ð2:9bÞ

Physics of Surface Plasmon Resonance

19

where op is the so-called plasma frequency, ne is the free electron density and e and me are the electron charge and mass, respectively. Generally, this implies that for ooop no electromagnetic field can propagate in a metal [cf. eqs. (2.1) and (2.2)]. More specifically, provided that e2 4
e1, we find for the interface that kyi is imaginary, whereas kx remains real. Thus an electromagnetic wave exists, propagating strictly along the interface, with evanescent tails extending into both sides of the interface [cf. eq. (2.5)]. To get a feeling for the quantities involved, it is instructive to calculate penetration depths for a real case, on the basis of eq. (2.8). We take l ¼ 700 nm, thus o ¼ 2.69  1015 s
1 and a gold/water interface. At this wavelength egoldE
16 and ewaterE1.77. We calculate for the penetration depths 1/ky,water ¼ 238 nm and 1/ky,gold ¼ 26 nm. Now all ingredients are available to appreciate the use of SPR in sensor applications. Let us assume that we have a situation where molecules X are allowed to adsorb to the water/metal interface. We can view this as a process where water molecules are replaced by molecules X. Because, generally, eXaewater, the average dielectric constant close to the interface will change. Equation (2.8) then describes the concomitant change of the wavevector kx. Because the SP field is evanescent in the direction perpendicular to the interface, a change of the dielectric constant e2 is only detectable in SP characteristics if this change occurs within the penetration depth of the SP field: an SPR sensor will only be sensitive to molecular processes (binding, adsorption, etc.) that occur at a distance to the metal surface that is roughly half the wavelength of the used light.

2.3.2 Excitation of Surface Plasmons By substitution of eqs. (2.9a) and (2.9b) into eq. (2.8), we obtain a graphical representation of the SPR dispersion relation as shown in Figure 2.2 (line I). In the same figure, the dispersion relation for ‘‘normal’’ light is depicted (line a). We immediately see that, apart from the origin, there is no point where the SPR curve and the light curve intersect, implying that in this geometry ‘‘normal’’ light cannot simultaneously provide the correct wavevector and angular frequency to excite a surface plasmon. One way to circumvent this problem is to introduce a second interface, as depicted in the inset of Figure 2.2. Here a thin metal layer (dielectric constant em) is sandwiched between two dielectric materials 1 and 3 with different dielectric constants e1 and e3, with e1 4 e3. By applying Fresnel’s equations to the two interfaces, more complicated dispersion equations are found than eq. (2.8); however the essential physics remains unchanged. We now find two dispersion equations for kx, one for each interface, and we see that the line representing the dispersion relation for ‘‘normal’’ light in medium 1 (line b) intersects the SP dispersion line for the metal/medium 3 interface. This indicates that light incident from medium 1 can excite SPs: by proper adjustment of the incoming angle a (Figure 2.2, inset), we can tune the incoming wavevector kx ¼ kn1sina to

20

Figure 2.2

Chapter 2

Dispersion relation for surface plasmons. Curves I and II represent the SP dispersion for the interfaces e3/em and e1/em, respectively. The lines a and b are the dispersion relations for ‘‘normal’’ light in medium e3 and e3, respectively, which are dependent on the angle of incidence a in the experimental setup as indicated in the inset. By varying a, any line c between the lines a and b can be realized.

match the wavevector necessary for SP excitation. In this way, any kx between the two lines, labeled a and b in Figure 2.2, can be set. As an example, one such line, labeled c, is indicated. This so-called attenuated total reflection (ATR) technique was first demonstrated5 by Kretschmann and Raether [2] and has since then almost become the standard technique for SP excitation. Another way of providing a wavevector appropriate for SP excitation is the use of a metal layer on which a periodic structure is prepared [6] as illustrated in Figure 2.3. When light with wavevector kx ¼ 2p/l.nisiny falls on such a structure, this acts as a diffraction grating and diffraction orders m ¼ 0, 1, 2, . . . are generated in the reflected light (see, e.g., ref. [7]). The generated wavevector kx,net parallel to the interface can be written as kx;net ¼ kx þ m

5

2p L

In fact, Otto was the very first to demonstrate this in a somewhat less versatile form.

ð2:10Þ

Physics of Surface Plasmon Resonance

Figure 2.3

21

Schematic view of a grating coupler. By diffraction of an incident light beam, the grating produces kx values larger than that of the incident light. By adjusting the incident angle a the wavevector can be tuned to kx required to produce a surface plasmon.

where L is the periodicity of the grating. Again, the wavevector kx,net can be tuned to the SPR wavevector, given by eq. (2.8),6 by changing the incident angle. Up to now the required polarization direction of the incoming light remained unmentioned. As already pointed out, SPs are conductivity fluctuations brought about by collective surface charge density oscillations. These charge density waves have to be excited by an external electric field. Only an electric field with a component perpendicular to the interface can induce a surface charge density; only p-polarized light has a perpendicular electric field component.

2.3.3 Surface Plasmon Properties With SPs, a number of specific properties are associated that are particularly relevant to sensor applications: (1) the field enhancement, (2) the phase jump of the reflected field upon SP excitation and (3) the SP coherence length. Field enhancement. A calculation of the electric field transmission coefficient on the basis of Fresnel’s equations for the interface reveals that the electric field at the low index side of the metal can be much larger than that at the other side of the metal layer. In Figure 2.4, the intensity enhancement is depicted as a function of the angle of incidence of incoming light for a number of different thicknesses of a gold layer. It is found that very close to the SPR angle the intensity can be enhanced by a factor of more than 30. This circumstance accounts for much of the 6

The dispersion equation eq. (2.8) is hardly affected if the metal surface has a shallow corrugation.

22

Chapter 2

Figure 2.4 Field enhancement for various values of the thickness of the gold layer. Wavelength of the excitation light is 700 nm; the low-index side of the metal layer consists of water (e3 ¼ 1.77).

remarkable sensitivity that the SPR condition has for a changing dielectric environment. Phase jump. As already mentioned in Section 2.3.1 and expressed in eq. (2.6), a reflection event at an interface is generally accompanied by a phase jump of the reflected field. This is illustrated in Figure 2.5a for a prism–gold–water system. For comparison, the ‘‘conventional’’ SPR dip is shown in Figure 2.5b for the same layer system. We see that around the SPR dip the phase of the reflected electric field undergoes a relatively large change. The significance of this phenomenon for sensing purposes is more clear when we plot the reflectance and phase changes as a function of incident angle for a certain change in dielectric constant of the water. This is depicted in Figure 2.5c and d. In the following rough calculation, we assume that both the change in reflection coefficient DR and the phase change Dj are proportional to De. From Figure 2.5c we estimate that DR/De E30, whereas from Figure 2.5d we find that Dj/De E250. Experimentally, a minimum DRE10
3 can be measured, whereas a minimum Dj E 10
3 is feasible, using interferometric techniques. The conclusion is that on the basis of reflectance measurements a minimum DeE4  10
5 can be detected, whereas a phase measurement provides a sensitivity of De E4  10–6. In view of the very approximate character of this calculation,

Physics of Surface Plasmon Resonance

Figure 2.5

23

Comparison of the angle-dependent phase changes (a, c) and reflectance changes (b, d) for variation of the dielectric constant at the low-index side of the metal layer. A gold layer is used, SPs are excited at l ¼ 700 nm. (c) and (d) depict the differential phase and reflectance, respectively, for a change in the medium’s dielectric constant of 0.01.

the absolute values found are of limited validity; however, the finding that a phase measurement provides an order of magnitude better sensitivity is a hard conclusion and, indeed, this was demonstrated by Nikitin and co-workers [8,9]. The only drawback of this approach seems to be the much more complicated experimental setup. SP coherence length. Generally, the metal’s dielectric constant e2 is complex and this circumstance results in a complex propagation constant kx ¼ kx 0 0 + jkx 0 0 [cf. eq. (2.8)], where kx 0 0 and kx 0 0 are real and imaginary parts, respectively. For a surface plasmon, traveling along the interface with wavevector kx, this implies that the field intensity decays with a characteristic distance 1/2kx 0 0 . For gold and silver, the standard metals in sensor applications,

24

Figure 2.6

Chapter 2

SPR response to a dielectric step at several wavelengths. For each wavelength the light angle of incidence is set such that outside the strip (extending from 0 to 125 mm) the interrogating kx is resonant with the surface plasmon wavevector. The surroundings of the strip has dielectric constant e3 ¼ 1.

the imaginary part of the dielectric constant increases with decreasing wavelength and the SP propagation length decreases accordingly. This is illustrated in Figure 2.6: here a layer system was prepared where a 30 nm SiO2 strip was deposited on a 50 nm silver layer. For a series of wavelengths the angle of incidence was chosen such that SPs were excited in the area

Physics of Surface Plasmon Resonance

25

outside the strip and for each wavelength the whole area was illuminated with a collimated light beam under a constant angle of incidence. Because of the contrast in dielectric constant between the strip and its surroundings (air). The SP resonance condition is not fulfilled in the area below the strip and we see the decaying SP (increasing reflectance) at the left edge of the strip. Beyond the right edge of the strip the SPR condition is again fulfilled and the SP resonance builds up. The figure nicely demonstrates that with decreasing wavelength the SP propagation length becomes shorter: the blurring on the left side of the strip becomes less prominent for shorter wavelengths. It turns out that in the wavelength range 500–00 nm the propagation length varies between o10 and 40 mm. For a quantitative description of the findings depicted in Figure 2.6, we have to analyze the interference between the several fields that are present in the layer. In Figure 2.7, the layer system and the fields involved are indicated: the resonant SP field, the non-resonant SP field and the external exciting field with amplitudes E1, E2, E3, respectively. For a resonant SP that enters the SiO2 stripcovered layer, the total field reaching the detector can be written as [10,11]
0 00  0 0 j k1 þjk1 x þ E2 ejk0 x
E3 ejk0 x ð2:11Þ Etot ðxÞ ¼ ðE1
E2 Þe where k 0 1 is the wavevector corresponding to resonance in the covered area and k 0 0 is the wavevector that excites SPs in the uncovered area. Defining A ¼ E1
E2 and B ¼ E2
E3, the resulting intensity at the detector becomes  0  00 00 0 Itot ðxÞ ¼ B2 þ A2 e
2k1 x þ 2ABe
k1 x cos k1
k0 x ð2:12Þ When a non-resonant SP leaves the covered area, the resonant SP builds up and the intensity at the detector decreases accordingly [11]: h  00 i2 Itot ðxÞ ¼ B þ A 1
e
k0 x

ð2:13Þ

From Figure 2.6, we see that this model gives a very accurate description of the experiments.

Figure 2.7

Definition of wave vectors and fields for the system consisting of a dielectric strip on top of a metal layer.

26

Chapter 2

Both this model and the experiments indicate that a plasmon needs roughly four times the propagation length Lx for a full decay or for a full build-up; this propagation length can be loosely defined as Lx ¼

1 2k00x

ð2:14Þ

This implies that SPs with mutual distances significantly larger than Lx are independent. This is a very important conclusion because it is the fundament of surface plasmon microscopy [12,13], with its many applications in SPR imaging and SPR multisensing: on a substrate we can define areas that in an SPR experiment will behave mutually independently, provided that these areas are significantly larger than Lx2. For SPs on gold, excited at l ¼ 632 nm, LxE7 mm and on a total sensor area of 1 cm2 more than 104 independent sensor ‘‘patches’’ that each have an area of somewhat smaller than 100  100 mm2 can in principle be defined, of which the optical responses can be simultaneously read out by using an imaging system. As practical aspects are outside the scope of this chapter, the interested reader should consult Chapter 7 for more details.

2.3.4 Choice of Experimental Parameters It is impossible to define a general set of optimum SPR parameters, for instance, optimal spatial resolution in an SPR microscopy/imaging setup requires values of the experimental parameters other than those to obtain maximum sensitivity for dielectric changes. Therefore, this section provides only some general guidelines, based on consideration of the properties of the metal layer. To obtain maximum sensitivity, it is advantageous to maximize the steepness of the reflectance as a function of the angle of incidence, because this allows for a more accurate determination of the angle of minimum reflectance (cf. Figure 2.8). This implies optimization of the reflectance minimum Rmin and minimizing the width of the resonance curve. Rmin can be made very close to zero by selecting the appropriate thickness of the metal layer; as can be seen in Figure 2.8, optimum thicknesses are somewhat dependent on the applied wavelength and are between 40 and 50 nm. The width of the resonance curve is mainly determined by the complex value of the metal’s dielectric constant. Generally, a large (negative) real part, together with a small imaginary part, results in narrow resonance curves. In practice only two options are available for the choice of the metal layer: gold or silver. As seen in Figure 2.8, silver has the better SPR characteristics in view of the larger real part of its dielectric constant; however, it is chemically less inert. In Figure 2.8, it is also seen that the use of higher excitation wavelengths has an appreciable effect on the width of the resonance curve. This is one of the reasons why (near-) infrared SPR experiments are attracting attention [14,15]. However, it should be realized that narrowing the reflectance curve necessarily implies increasing the SPR propagation length [eq. (2.13)], which can be a disadvantage in certain SPR imaging applications. For a

Physics of Surface Plasmon Resonance

Figure 2.8

27

The SPR ‘‘dip’’ for 46 nm of silver (dashed) and gold (solid) with water on the low-index side, for several excitation wavelengths. Dielectric data for the metal layer obtained from refs. [3] and [33].

gold layer, it can be calculated that an increase in wavelength from 450 to 1500 nm results in a change in the propagation length from 100 to almost 1 mm. Finally, it should be mentioned that an increase in wavelength results in an increase in the penetration depth 1/ky [cf. eqs. (2.4) and (2.5)], with the consequence that the reflectance minimum will become more sensitive to dielectric changes relatively far from the metal/dielectric interface; hence the surfacesensitive character of SPR becomes less prominent. This implies that for detection of the growth of thin layers the optimum choice of wavelength will be different from that in a situation where a more bulk-like change in refractive index has to be detected [33].

2.4 Analysis of Multi-layered Systems In most SPR-based sensor applications, the system of interest consists of a gold or silver layer on which one or more thin layers are deposited in an aqueous environment. Often the desired parameters are the thicknesses of the several layers, which can be converted into surface concentrations of the layer-composing molecules (cf. Chapters 4 and 5). One way to obtain these parameters is a repeated application of the Fresnel equation [eq. (2.6a)]. The following relation holds for a system consisting of N

28

Chapter 2

layers with dielectric constants and thicknesses ei and di, respectively, placed between a prism with dielectric constant ep and a medium (e.g. water) with dielectric constant ew (see. e.g., ref. [5]):     k ky;p ky;w M11 þ M12 ey;w
M þ M 21 22 ew ep w    rp ðaÞ ¼  ð2:15aÞ ky;w ky;p k M11 þ M12 ew ep þ M21 þ M22 ey;w w where M is the so-called transfer matrix: M ¼ M1  M 2  . . .  MN

ð2:15bÞ

with " Mi ¼



jei
 # cos ky;i di ky;i sin ky;i di


jky;i cos ky;i di ei sin ky;i di

ð2:15cÞ

The angular dependence of rp is contained in the wavevectors ky,i, perpendicular to the layer system; these can be calculated using eq. (2.4). The reflectance can be obtained by application of eq. (2.6b). Provided all thicknesses di and dielectric constants ei are known, eq. (2.15) gives an accurate description of the SPR experiment. Of course, in practice one is concerned with the inverse problem and a priori knowledge, such as the dielectric constant and dimensions of the molecules composing a certain layer, is required for a satisfactory analysis of experimental SPR results. This precludes an unambiguous analysis of more than two or three layers. Another, more intuitive, approach is the introduction of an effective dielectric constant, eeff. Here, the actual multi-layered system is replaced by a twolayer system, where e1 in eq. (2.8) is replaced by the effective dielectric constant eeff, given by the average of all dielectric constants in the layer system, weighted by the penetration depth y0 of the SPR evanescent field [16,17]: eeff

2 ¼ y0

ZN

eðyÞe
2y=y0 dy

ð2:16Þ

0

Of course, with the use of this equation. we face the same problems as those when we use eq. (2.15).

2.5 SPR Spectroscopy 2.5.1 Enhancement of Fluorescence and Absorbance Up to now we have only considered, apart from the metal layer, transparent layers, i.e. layers that are characterized by a positive, real dielectric constant. When one or more layers contain a light-absorbing compound, SPs can boost the fluorescence intensity from a thin layer more than 40-fold [18]. This effect is due solely to the large field enhancement that occurs on the low index side of

Physics of Surface Plasmon Resonance

29

the metal layer when an SP resonance condition is established (cf. Figure 2.4). More information on this phenomenon can be found in Chapter 9. Another, more subtle, feature of the interaction between SPs and light-absorbing molecules is the increased sensitivity for the detection of absorbances in thin layers. The addition of a light-absorbing layer results in two effects on the SPR angular-dependent curve, which are quantitatively described by eq. (2.15): (1) the SPR dip shifts to a larger angle of incidence and (2) the value of the reflectance minimum increases. This last effect can readily be understood as light absorption necessarily results in a decreased reflectance. In addition, the SPR field enhancement on the low-index side of the metal layer will result in increased sensitivity for dielectric changes [19] and therefore also for changes in the absorbance. The first effect is a consequence of the Kramers–Kronig relation (see, e.g., ref. [20]), which in the present context can be expressed as the statement that any change in the imaginary part of the dielectric constant will be accompanied by a change in the real part; in SPR it is mainly the real part of the dielectric profile on top of the metal layer that determines the angular position of the SPR dip. It has been demonstrated [21] that an SPRassisted monolayer absorbance measurement can result in a 40-fold reflectance increase as compared with a metal-lacking ATR system. In addition, it is possible to extract unambiguously the thickness and the dielectric constant of an absorbing layer from a single SPR experiment [22].

2.5.2 SPR and Metal Nanoparticles SPR phenomena are not restricted to planar multilayers as discussed so far; it turns out that for metal particles with dimensions much smaller than the wavelength of the interacting light, SP effects can be much more prominent. Generally, the net electric field Etot around a dielectric particle is composed of the superposition of an external applied field E0 and the induced (dipole) field in the particle. For a polarizable spherical particle with radius rm and dielectric constant e, placed in a medium with dielectric constant e1, the following expression is found (see, e.g., ref. [5]) for the field gain G:  3 Etot ðoÞ eðoÞ
e1 rm E1 þ ð2:17Þ GðoÞ ¼ E0 ðoÞ eðoÞ þ 2e1 r þ rm It is seen that G can reach enormous values for e close to
2e1; in a ‘‘normal’’ dielectric medium where e1 4 0, this condition points to the use of a metal, where e can be negative; additionally, the imaginary part of e should be as small as possible. It turns out that this condition corresponds to the excitation of a surface plasmon in the metallic nanoparticle [23]. Particularly in the field of Raman spectroscopy this can result in enormous sensitivity enhancements7 (for a review, see ref. [24]). 7

Conventional Raman spectroscopy suffers from a very low scattering efficiency which can be 12 orders of magnitude lower than that of fluorescence.

30

Chapter 2

Now consider a Raman-active molecule near a metal nanoparticle. The detected Raman intensity I(o,osc) can be expressed as 2 2 IRaman ðo; osc Þ ¼ gEexc  Esc ¼ gG2 ðoÞG2 ðosc ÞI0 ðoÞI0;sc ðosc Þ

ð2:18Þ

where Eexc is the total excitation electric field to which the molecule is exposed and Esc is the total Raman-scattered field. The constant g is an experimental constant that is unimportant in the present discussion. By choosing an angular frequency o that excites surface plasmons in the metal (usually gold or silver) and detecting scattering frequencies osc not too far from the excitation frequency, both the excitation and the scattered field are enhanced by the presence of the metal particle. By substituting eq. (2.17) into eq. (2.18), we see that the distance dependence of the net Raman scattering intensity changes with the power –12 of the molecule–nanoparticle distance! Indeed, it has been found experimentally that a surface-enhanced Raman spectroscopy (SERS) experiment can result in experimental Raman signals that are enhanced 1012–1014 times compared with those obtained from non-surfaceenhanced experiments. It should be added that apart from this SPR enhancement mechanism, another chemical enhancement effect is operational, which accounts for a 10–100-fold amplification of the bare Raman signal [24]. It has been demonstrated [25,26] that SERS is able to detect single molecules. Together with its very high molecular specificity, this offers great promise as a detection tool for very low concentrations of biomolecules, such as DNA strands or proteins. In principle, these field enhancements should also be important in the detection of fluorescent molecules near a metal nanoparticle. However, the nearby presence of a metal layer leads to additional non-radiative decay paths of the electronic excited states of a nearby molecule, with the net result that in many cases the fluorescence will be largely quenched. So far, metal nanoparticles were considered as surface plasmon-assisted field amplifiers. However, these particles can also be exploited as intrinsic refractive index sensors, analogous to the more familiar planar SPR experiments (for reviews, see refs. [27] and [28]). The physical basis of this application is the light extinction (absorption and scattering), which is heavily dependent on the nanoparticle’s dielectric constant, size and geometry and also on the dielectric constant e1 of the surrounding medium. Mie theory gives a reasonably adequate description of the extinction coefficient Aext and for spherical particles with diameter less than about 20 nm the following expression is found [23]: 3=2

Aext ¼

18pNp Ve1 ImðeÞ l ½ReðeÞ þ 2e1 2 þ½ImðeÞ2

ð2:19Þ

where Np is the number of nanoparticles, each of which has a volume V, and l is the wavelength of the applied light.

Physics of Surface Plasmon Resonance

31

Again we see the pronounced influence of the occurrence of SPs: at Re(e) ¼ –2e1 we find a maximum in the extinction coefficient, which can reach large values for low values of Im(e). Hence also in this situation we are led to the use of gold or silver as a metal nanoparticle. More sophisticated models (see, e.g., ref. [29]) also account for the size and shape of the nanoparticles and computer programs are available in the public domain that can predict the extinction spectrum of nanoparticles of any shape [30], by modeling the particle as a series of dipoles placed in an oscillating electric field. However, the main features of nanoparticle extinction remain contained in eq. (2.19). For one nanoparticle with a diameter around 25 nm, excited close to its SP resonance, eq. (2.19) results in Aext of the order of 10–16 m2, which corresponds to the more familiar molar extinction coefficient in the order of 109 l mol
1 cm
1. This value, which indeed was observed experimentally [29], is more than three orders of magnitude larger than that of strong light-absorbing organic dye molecules, allowing for relatively simple optical detection and characterization of individual nanoparticles [31,32]. In another series of experiments, the shift of the extinction maximum as a function of the refractive index of the surrounding medium was investigated [28]. It was found experimentally that for silver nanoparticles the spectrum could shift as much as 20 nm for a change of 0.1 in the refractive index. Because molecules that adsorb to a nanoparticle change the refractive index around the particle, it is obvious that this, analogous to conventional SP resonance, can be used as a sensor principle. Indeed, it has been demonstrated experimentally that the full coverage of a silver nanoparticle with low molecular mass molecules resulted in a spectrum shift of approximately 40 nm. The full coverage corresponded to only 4  104 molecules. Together with the single particle detection capability, this promises enormous sensitivity, allowing for near single molecule detection [32].

2.6 Concluding Remarks The phenomenon of SPR is one of the many examples where an interesting physical phenomenon leads to applications that are highly important to both applied science and society. In a planar SPR system, it is particularly the combination of field enhancement and relatively short coherence length that allows for a unique sensor concept that provides both multiplexing capabilities and very high sensitivity. The general physical picture is well understood; however, some areas are still in vivid scientific debate (SERS, optics of nanoparticles). From a technological point of view, the emerging field of nanotechnology will enable us to exploit to its full potential the SP phenomena of tailored nanoparticles. It is the author’s firm conviction that merging of (bio-)nanotechnology and SP phenomena of nanoparticles will ultimately lead to sensor concepts and sensor realizations that will really be important in numerous

32

Chapter 2

aspects of society, varying from food safety monitoring and high-throughput screening to early in vivo detection of tumor growth.

2.7 Questions 1. Derive eq. (2.4) from eqs. (2.2) and (2.3). 2. Calculate for a gold/water interface at l ¼ 700 nm (
gold ¼
16;
water ¼ 1:770) the angular shift of the SPR dip when
water increases to 1.775. For the light-incoupling we use a semi-circular glass piece with refractive index nglass ¼ 1.5. 3. Estimate the effective dielectric constant for the following system, when this interface is probed with a wavelength l ¼ 700 nm (for dielectric constants of gold and water, see previous question) water gold

The squares in the figure represent cubes of protein molecules in an aqueous environment, adsorbed to the gold surface. Each protein molecule has a volume of 5*5*5 nm3 and a dielectric constant
protein ¼ 2.30. The average distance between the edges of the cubes is 7 nm. 4. A particular SPR application could be the detection of micro-organisms in, e.g., waste water, by detecting changes in bulk refractive index. Which SPR excitation wavelength region would be more favourable, the blue/UV or the infrared region? 5. Express the extinction coefficient Aext (eq. 2.16) in units M
1cm
1.

2.8 Symbols A Aext B const c di E e E0 E0 Eexc Ei Er

A ¼ E1
E2 extinction coefficient B ¼ E2
E3 experimental constant in surface-enhanced Raman spectroscopy propagation velocity in vacuum film or layer thickness electric field strength charge of electron amplitude of electric field external applied field E0 (Section 2.5.2) total excited electric field (Raman) incident electric field reflected electric field

Physics of Surface Plasmon Resonance

Esc Etot G I IRaman k ky kx, net Lx me m N ne n Np ni r rm rp Rp V y0 a b e e0 eeff ep ew k0 k1 l L j o op osc

33

total scattered field (Raman) net electric field around nanoparticles field gain of nanoparticle SPR intensity Raman intensity near an SPR nanoparticle wavevector y component of wavevector x component of wave vector propagation length of a full decay or build-up mass of electron diffraction order number of layers in system free electron density refractive index number of nanoparticles refractive index of material i position vector radius of polarizable spherical (nano)particle reflection coefficient (complex) reflectance volume of nanoparticles penetration depth of SPR evanescent field incident angle of light refraction angle of light dielectric constant dielectric constant of medium effective dielectric constant dielectric constant of prism dielectric constant of water wavevector excited in uncovered area wavevector corresponding to resonance of strip-covered area wavelength in vacuum periodicity of grating phase change of the reflected field relative to the incident field angular frequency plasma frequency scattering frequency

References 1. A. Otto, Z. Phys., 1968, 216, 398. 2. E. Kretschmann and H. Raether, Z. Naturforsch., 1968, 230, 2135. 3. H. Raether, Surface Plasmons on Smooth and Rough Surfaces and on Gratings, Springer, Berlin, 1988. 4. M. Cardona, Am. J. Phys., 1971, 39, 1277.

34

Chapter 2

5. J.R. Reitz, F.J. Milford and R.W. Christy, Foundations of Electromagnetic Theory, Addison-Wesley, New York, 1993. 6. D.C. Cullen, R.G. Brown, C.R. Lowe, Biosensors, 1987/88, 3, 211. 7. E. Hecht, Optics, Addison-Wesley, New York, 2002. 8. A.N. Grigorenko, P.I. Nikitin and A.V. Kabashin, Appl. Phys. Lett., 1999, 75, 3917. 9. A.V. Kabashin and P.I. Nikitin, Opt. Commun., 1998, 150, 5. 10. B. Rothenha¨usler, W. Knoll, J. Opt. Soc. Am. B, 1988, 5, 1401. 11. C.E.H. Berger, R.P.H. Kooyman and J. Greve, Opt. Commun., 1999, 167, 183. 12. E. Yeatman and E. Ash, Electron. Lett., 1987, 23, 1091. 13. B. Rothenha¨usler and W. Knoll, Nature, 1988, 332, 615. 14. B.P. Nelson, A.G. Frutos, J.M. Brockman and R.M. Corn, Anal. Chem., 1999, 71, 3928. 15. S. Patskovsky, A.V. Kabashin, M. Meunier and J.H.T. Luong, J. Opt. Soc. Am. A., 2003, 20, 1644. 16. K. Tiefentahler, W. Lukosz, J. Opt. Soc. Am. B, 1989, 6, 209. 17. R.G. Heideman, ‘‘Optical waveguide based evanescent field immunosensors’’, PhD Thesis, University of Twente, Enschede, 1993. 18. T. Liebermann and W. Knoll, Colloids Surf. A, 2000, 171, 115. 19. H. Kogelnik, in ‘‘Topics in Applied Physics’’, vol 7, ed. T. Tamir, Springer, Berlin, 1975. 20. R.W. Ditchburn, Light, Blackie, London, 1963. 21. S. Wang, S. Boussaad and N.J. Tao, Rev. Sci. Instrum., 2001, 72, 3055. 22. P.S. Vukusic, J.R. Sambles and J.D. Wright, J. Mater. Chem., 1992, 2, 1105. 23. U. Kreibig and M. Vollmer, Optical Properties of Metal Clusters, Springer, Berlin, 1995. 24. K. Kneipp, H. Kneipp, I. Itzkan, R.R. Dasari and M.S. Feld, J. Phys. Condensed Matter, 2002, 14, R597. 25. S. Nie and S.R. Emory, Science, 1997, 275, 1102. 26. K. Kneipp, Y. Wang, H. Kneipp, L.T. Perelman, I. Itzkan, R. Dasari and M.S. Feld, Phys. Rev. Lett., 1997, 78, 1667. 27. C.R. Yonzon, D.A. Stuart, X. Zhang, A.D. McFarland, C.L. Haynes and R.P. Van Duyne, Talanta, 2005, 67, 438. 28. C.L. Haynes, R.P. Van Duyne, J. Phys. Chem. B, 2001, 105, 5599. 29. S. Link, M.A. El-Sayed, J. Phys. Chem. B, 1999, 103, 8410. 30. http://www.astro.princeton.edu/Bdraine/DDSCAT.6.1.html 31. S. Schultz, D.R. Smith, J.J. Mock and D.A. Schultz, Proc. Natl. Acad. Sci. USA, 2000, 97, 996. 32. A.J. Haes, D.A. Stuart, S. Nie and R.P. Van Duyne, J. Fluor., 2004, 14, 355. 33. K. Johansen, H. Arwin, I. Lundstro¨m and B. Liedberg, Rev. Sci. Instrum., 2000, 71, 3530.

CHAPTER 3

SPR Instrumentation RICHARD B.M. SCHASFOORTa AND ALAN MCWHIRTERb a

Biochip Group, MESA+ Institute for Nanotechnology, Biomedical Technology Institute (BMTI), Faculty of Science and Engineering, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands; b General Electric Healthcare, Biacore AB, Rapsgatan 7, SE-754 50 Uppsala, Sweden1

3.1 Introduction General aspects of surface plasmon resonance (SPR) instrumentation are described in this chapter, providing the reader with an insight into the development of SPR technology to its current state. SPR instruments comprise three essential units integrated in one system: optics, liquid handling unit and the sensor chip. Instruments may differ optically, in the liquid handling system or in the degree of automation, all influencing their performance. Additionally, the quality and features of sensor chips determine the quality of the measurement of biomolecular interactions; however sensor surfaces are treated in detail in Chapter 6. In this chapter, first general aspects of the different SPR optical systems are discussed, followed by aspects of liquid handling and operational automation, concluded by a discussion of various SPR instruments. The latest instruments of Biacore, the largest company in the SPR market with a complete SPR product line, are discussed separately in Section 3.5; however, the Biacore Q, dedicated to food analysis, is treated in Chapter 11 on food analysis.

3.1.1 From Surface Plasmon to SPR Signal Real-time and label-free biomolecular interaction sensing is unthinkable at present without the surface plasmon resonance principle [1,2]. Currently, instruments on the market differ in performance based on differences in their 1

Author of the section on Biacore instruments (Secton 3.5).

35

36

Chapter 3

2

3

liquid handling system

sensor

chip

1

Figure 3.1

SPRoptics

Schematic view of the three main units of an SPR system: 1, SPR optics; 2, liquid handling unit; 3, sensor chip. The biomolecular interaction takes place on the wet side of the sensor.

optical systems and their degree of development and automation. SPR optics, liquid handlings and the sensor chip are integrated in the SPR instrument as depicted schematically in Figure 3.1. As indicated, the sensor chip forms a physical barrier between the optical system (dry section) and the liquid handling unit e.g. flow cell (wet section). How is the refractive index measured at the surface? SPR causes an intensity decrease or dip in the reflected light at the sensor surface, as discussed in Chapter 2. In the left section of Figure 3.2, the reflectivity, i.e. the ratio of incoming and reflected light vs. angle of incidence passes through a minimum depending on the angle of incidence of the p-polarized light. The angle at which the minimum is found depends on the refractive index of the medium in the immediate vicinity of the sensor surface. On adsorption or binding of molecules, the refractive index at the sensor surface will change, causing a shift of the reflectivity curve. This shift can be observed in real time, as shown in the sensorgram in the right section of Figure 3.2. A sensorgram can be obtained in at least three ways: (1) reflectivity shift versus time, (2) angle shift versus time or (3) wavelength shift versus time. In Figure 3.3, a sensorgram of analyte binding versus time is presented, together with a rotated view of an SPR dip. While the SPR angle is the most

37

SPR Instrumentation Reflectivity in %

X11

X1

A X

Figure 3.2

B Time (s)

Angle in m°

Shift of the SPR dip from A to B. A derivative parameter of the shift of the SPR curve is the reflectivity (y-axis) measured at a fixed angle at the left flank of the SPR dip (see arrow showing reflectivity shift of X1 to X11). In the sensorgram on the right the reflectivity shift as a function of time is shown. Angle in m°

B

A X11

X1

Reflectivity

Figure 3.3

X

Time (s)

Rotated presentation of the SPR dip (left section) that forms directly the pointer of the sensorgram (right section). The angle shift of the SPR dip is determined (left, A to B), followed by plotting the angle of the SPR minimum in the sensorgram vs. time (right). Here the SPR dip minimum of the initial curve (A) shifts with time towards a larger angle (B).

representative parameter of the biomolecular interaction process in time, one should follow the angle of the SPR minimum (dip). As the dip shifts in the left section from A to B, the SPR dip is followed in the right section of Figure 3.3. The angle shift as a function of time is shown called the sensorgram.

38

Chapter 3

Usually, the SPR curve is measured or fitted and the angle of the SPR dip is determined, followed by plotting the angle of the SPR minimum in the sensorgram (right part of Figure 3.3). A sensorgram can be obtained from a monochromatic light source using either the reflectivity change at fixed angle (y-axis in %R) or the SPR dip shift (y-axis in RU or millidegrees), as illustrated in Figures 3.2 and 3.3, respectively. However, if a sensorgram of an equal experiment is either plotted as reflectivity (Figure 3.2) or as SPR angle (Figure 3.3) on the y-axis versus time, the two sensorgrams are not equal and large deviations of the presented binding curve may occur! A reflectivity change cannot be exchanged with a shift of the SPR dip. From theory as described in Chapter 2, the amount of biomolecules accumulated in the evanescent field results an (almost) linear shift of the SPR dip. Therefore, only real-time measurement of the SPR dip as indicated in Figure 3.3 is the preferred mode of operation [3]. Reflectivity change is only a derivative parameter of this shift. Therefore, optimally the SPR dip shift should be detected continuously to follow changes at the surface, e.g. a biomolecular interaction event [4]. The angle shift can be expressed in different units in the various instruments, for example response units (RU, Biacore [5]) or in millidegrees (IBIS and Eco Chemie systems [6]), which directly depict the SPR angle shift, and also in percentage reflectivity (%R) (GWC Technologies system, Genoptics, K-MAC [7,8]) or wavelength shifts (Thermo system, Graffinity Pharmaceuticals [9]). Because of a near-linear relationship between the amount of surface-bound protein and the SPR signal, the sensorgram of the SPR dip shift provides quantitative, real-time data. Errors in the position of the starting angle on the initial SPR curve can result in a highly unreliable sensorgram. Importantly, a bulk refractive index shift or background response will be generated if the refractive index of the medium changes, e.g. due to differences in solvent. Reflectivity measurements should be checked with the measurement of the angle- (or wavelength-) resolved SPR curve. Moreover, the shape of SPR dip can reveal the quality of the sensor surface. For example, if inhomogeneities or agglutinates (particles) are present on the surface, the dip becomes shallower and broader as a result of spatial differences in resonance conditions within a sensing area. Therefore, checking the SPR dip is a prerequisite for good SPR measurement practice. However, most of the (cost-effective) instruments described in this chapter only support reflectivity measurements.

3.2 SPR Optics SPR instrumentation can be configured in various ways to measure the shift of the SPR dip. In general, three different optical systems are used to excite surface plasmons: systems with prisms, gratings and optical waveguides, as explained in Chapter 2 and more extensively by Homola [10]. Most widespread are instruments with a prism coupler, also called the ‘‘Kretschmann configuration’’ [11]. In this configuration, a prism couples p-polarized light into the SP film and

SPR Instrumentation

39

reflects the light on to a light intensity detecting device, e.g. a photodiode or even a camera. This configuration can be further divided into three subgroups: fanshaped beam, fixed-angle and angle scanning SPR instruments, as explained in the following sections. In instruments with a grating-coupler [12], light is reflected at the lower refractive index substrate. In practice, this means that light travels through the liquid before photons generate surface plasmon waves as similar setup to ellipsometer instruments [13]. Besides the grating couplers, some instruments apply optical waveguide couplers [14] or apply the SPR wavelength shift as a result of the biomolecular interaction process [15]. In the following sections, basic features and characteristics of the different optical SPR systems are treated. In the final section, SPR imaging instruments will be discussed [16].

3.2.1 Fan-shaped Beam In a fan-shaped beam instrument, a converging or diverging beam of p-polarized light is coupled in the higher refractive index medium using a cylindrical or triangular prism. In a converging beam fan-shaped instrument, the beam is focused on to an infinitely narrow line on the sensor chip. A photodiode array is used to detect the reflected diverging beam with the SPR dip. A line on the sensor can be imaged on a camera by measuring, e.g., 4–20 sensor spots. This principle has been successfully applied in Biacore instruments since 1990 Figure 3.4. Other instruments in this category often use a diverging fan-shaped beam, resulting in a not spatially defined location of the SPR dip on the sensor chip: the position of the minimum ‘‘walks’’ over the sensor chip while the biomolecular interaction process proceeds.

Figure 3.4

In a fan-shaped beam instrument the SPR dip can be followed in real time without moving parts. Here the reflected beam is drawn reciprocal meaning that black corresponds with high intensity while white has low intensity.

40

Chapter 3

3.2.2 Fixed Angle Fixed-angle SPR instruments (Figure 3.5) measure the reflectivity on the left flank of an SPR curve resulting from the shift of the SPR dip, as shown by the vertical line between X1 and X11 in Figure 3.2. In many instruments the angle can be adjusted to find the inflection point of the SPR curve in order to obtain the maximum reflectivity change. Sometimes a stepper motor-operated angle positioner (or mirror) is applied to automate the setting of the (optimal) starting angle. An inflection point can be determined from the second derivative of the SPR curve, which gives the user the most sensitive starting angle or highest slope of the reflectivity shift versus angle shift. However, two differently treated spots usually require different optimal starting angles and therefore the reflectivities cannot be quantitatively compared without correction for the shape of the SPR curve.

3.2.3 Angle Scanning The SPR curve can be scanned fast (see the double arrow in Figure 3.6) in a so-called angle scanning SPR instrument. An angle-controlled mirror is applied to follow the position of the SPR dip in real time. The surface of the sensor chip is at a certain angle in full resonance with the advantage that the area of the light beam is averaging the reflectivity and therefore potential defects of homogenous coated sensor chips. To acquire the data of a full scan very fast, an angle-controlled mirror (or angle scanner) is preferred. Instruments of Eco Chemie (SPRINGLE and ESPRIT), for example, scan the full SPR curve 76 times per second. In this instrument, averaging of the acquired dip positions during the time interval is applied, resulting in an improvement of the signal-to-noise ratio by the square root of the number of averages. For example, an interval time of 1 s with 76 scans

Figure 3.5

In a fixed-angle SPR instrument the reflectivity is detected at a set angle. The angle position can be set manually or using an angle stepper motor.

SPR Instrumentation

41

Figure 3.6

In angle scanning SPR instruments an angle scanner scans the sensor surface in a small angle window (e.g. 51) very fast. At a certain moment the total surface, not just part of it, is in full resonance, resulting improved sensitivity.

Figure 3.7

In a grating coupler, the SPR phenomenon is exploited differently, as the light beam travels through the flow cell.

averaged results in an improvement of the signal-to-noise ratio of O76 or a factor better than 8. The recently introduced SPR imaging instrument of IBIS Technologies (see Section 3.2.6 and Chapter 7) also applies such a scanning principle, with additional, angle-controlled scanning features and curve-fitting routines, while measuring the angle shift of multiple regions of interests [17].

3.2.4 Grating Coupler Although grating coupler instruments (Figure 3.7) are regarded as SPR instruments, they are quite different. Whereas in prism-coupled SPR optical cells the light never passes through the sample, in grating coupler instruments the light passes through the sample solution in the flow cell, resulting in decreased stability of the sensor signal in comparison with the Kretschmann

42

Chapter 3

configuration. To avoid unwanted internal reflection effects, flow cells of grating coupler SPR devices have an increased height. Therefore, the sample volume in grating coupler instruments is normally larger than that in instruments using the Kretschmann configuration, where the height of the flow cell always exceeds the evanescent field of the reflected light. The attractive feature of grating coupler SPR devices is the option of disposable gratings, which can be mass produced by injection molding replication techniques. The optical configuration of grating coupler SPR instruments is the same as for ellipsometry platform instruments (e.g. Nanofilm; see section 3.4.2).

3.2.5 Other Optical Systems Some other instruments measure the refractive index change in the evanescent field of the optical device by modulation of the coupling wavelength as shown in Figure 3.8. A spectrophotometer is applied as detector to follow the shift of the ‘‘plasmon’’ wavelength. A class of instruments uses modulation of the phase or polarization [14]. This phenomenon is exploited in ‘‘SPR-like’’ resonant mirror instruments [15]. The principle of resonant mirror measurements, as depicted later in Figure 3.26, is that two phases of the reflected light are measured and, due to relative phase differences of the two modes (TE and TM components), a signal can be obtained as a result of refractive index changes in the evanescent field. Although plasmons are not generated and the enhancement of the evanescent field does not occur, multiple reflections in the resonant mirror contribute to a high-sensitivity measurement of refractive index changes in the evanescent field. Because the system utilizes a less inert oxide layer than gold, more uncontrolled drift behavior can be observed than in conventional

Figure 3.8

A wavelength interrogation-based SPR device applies a polychromatic light source to follow the wavelength-dependent SPR shift, i.e. the color of the light changes as a result of the biomolecular interaction.

SPR Instrumentation

43

gold-based SPR systems. Because of this unreliable drift, the sensitivity of this instrument is lower than in other ‘‘real’’ SPR systems. Interferometric read-outs of optical waveguides also use the refractive index change as a result of the biomolecular interaction event at the surface of the device. An evanescent field is formed when reflection occurs at the medium of higher refractive index. Although physically ‘‘SPR-like’’ instruments including interferometers might be more sensitive than SPR-based instruments, they lack the inertness of the gold surface. The oxides with the required refractive indices often show uncontrolled and unreliable baseline drifts including memory effects from previous sample/salt additions. In the literature, various examples are described [10]; however, only a few have been commercialized [18].

3.2.6 SPR Imaging Instruments In SPR imaging instruments, the sensor surface is optically imaged by a camera. Although some SPR instrument manufacturers use cameras for measuring the SPR shift, a prerequisite for classification as an SPR imaging instrument is that a microscopic view of the SPR sensor surface is generated. Any inhomogeneities, surface coating defects, errors of spotting, including e.g. missing spots, and effects of drying of the surface can be directly observed in the microscopic reflectivity image of the SPR sensor surface. If, for example, air bubbles adsorb on the surface or appear in the flow system, it can be seen directly on the SPR image monitor. Prior to performing the analysis cycle, immobilization of capturing entities or ligands is usually carried out off-line by spotting selective ligands in a microarray format.2 In 1988, Knoll3 et al. described SPR microscopy [19], and later Kooyman4 et al. established the principles of SPR imaging in reflectivity mode as well. Berger et al. showed that in total 16 different sensor spots could be imaged [20] in a two-dimensional array of antibody–antigen sensor surfaces in real time. A serious drawback of many SPR imaging instruments is that they measure reflectivity as a parameter derived from the refractive index in the evanescent field. For accurate kinetic measurements of biomolecular interactions, only the shift of the SPR dip of the microarray spots is reliable, because an optimal fixed angle position based on the inflection point of the SPR dip can only be found for one spot and not for hundreds of spots. Using the scanning angle principle, the recently introduced IBIS microarray imaging instrument can follow the real shift of the SPR dip of all the regions of interests simultaneously even when spots deviate 42000 millidegrees in resonance angle, as shown in Chapter 7. The in-house instrument of Graffinity Pharmaceuticals images the wavelength shifts of the biomolecular interactions [21]. An example of a grating coupler imaging SPR instrument is the FLEXChip of Biacore (Section 3.5). 2

A microarray contains a number of predefined selective regions or spots regularly ordered in rows and colums. 3 Author of Chapter 9 in this volume. 4 Author of Chapter 2 in this volume.

44

Chapter 3

3.2.7 General Optical Requirements for SPR Instruments The dynamic range of an instrument in the Kretschmann configuration for studying biomolecular interactions is determined by its range of angles. The larger the dynamic range, the smaller the angle resolution of an instrument should be in order to maintain the required sensitivity. Therefore, an instrument is usually built to zoom in to a small angle window (e.g. 2000 millidegrees) to measure tiny angle shifts, e.g. of the order of millidegrees. However, if the angle range is very small, the SPR dip might be outside the available range of angles. A way to overcome this shortcoming is to apply the wavelength mode (see Figure 3.8), with the disadvantage that the penetration depth of the evanescent field and the propagation length of the plasmon are not constant for different frequencies of light. The penetration depth of the evanescent field, and also the lateral resolution of SPR, are dependent on the wavelength of the incident light (see also Chapter 2). The lateral resolution at 680 nm wavelength incident light is about 10 mm [17]. If different wavelengths are used to excite SPR, the instrument needs a complex algorithm to compensate for the variable volume of the evanescent field. Hence most instrument manufacturers choose a single wavelength window for SPR excitation. In every instrument the polarity of the excited light needs to be kept perpendicular to the surface (p-polarized light). This requires trimming of the polarization filter at full resonance condition to obtain an optimal reflectance minimum. In some instruments, e.g. GWC [22], not only p-polarized light (perpendicular to the surface, but also s-polarized light (in plane with the surface) is needed to calibrate the intensity of the light source. An additional experimental requirement is to provide refractive index matching between the coated gold surface of the sensor chip and the optical element (prism, hemisphere, etc.). Biacore solved this problem by applying an optogel5 to coat the prism and ensure refractive index matching. In other instruments, an optical matching oil is used; however, it is a tedious task to replace a sensor chip and clean the prism between measurements. Other manufacturers have solved the optical matching problem by applying costly, disposable, gold-covered prisms. At present, gold on sensor chip surfaces is the ‘‘gold standard’’ in SPR. Although from a physical point of view silver is better, gold provides more chemical inertness. Attempts to protect the thin, non-inert silver layer have so far failed and the performance of these sensors usually decreases rapidly to an unacceptable level. Protection of the silver with, e.g., deposited oxides leads to unwanted memory effects and drifts of the sensor signal. For this reason, currently all manufacturers apply the physically second-best metal for SPR: gold. The refractive index of aqueous liquids is highly temperature dependent: at least –14 millidegree shift (or –115 Biacore resonance units) per degree Celsius is observed. Therefore, temperature control of the instrument is essential. Temperature stabilization (better than 0.01 1C) is a prerequisite for reliable measurements in the millidegree range or in refractive index ranges below 10
4. 5

Proprietary product of Biacore.

SPR Instrumentation

45

Moreover, reference measurements are essential to reject common mode effects, e.g. temperature drift or fast temperature jumps and also a bulk refractive index shift, but cannot compensate for the temperature effect on the affinity constant of biomolecular interactions. In order to determine accurately the kinetics of biomolecular interactions [23], the shift of the SPR dip is the parameter that reflects best the amount of accumulated mass at the sensor surface, usually expressed as pg mm
2. In Section 3.1.1 we pointed out that the measurement of reflectivity change at a fixed angle position of the SPR curve is an inaccurate parameter and is not appropriate for quantitative determination of the refractive index change in the evanescent field of the sensor surface. Many cost-effective instruments on the market apply the sensorgram reflectivity (in %R), including most of the SPR imaging instruments. The quality of the optical components and optical alignment of the beam including lenses, noise of both light source and photodetector or camera all contribute to the quality of the measurements. Finally, the software including the algorithms for averaging, subtracting or eliminating raw data points and the performance of hardware contribute to the quality of the SPR instrument in terms of sensitivity, repeatability, accuracy and robustness.

3.3 SPR Liquid Handling Systems In combination with the optical unit and sensor chips, the liquid handling system forms a vital part of SPR measurement systems. As shown in Figure 3.1, the bottleneck remains the integration of these parts in one instrument provided that the quality of the weakest part will be the limiting factor for the overall quality of the system. For instance, high-quality optics can never compensate for a bad-quality sensor chip. Additionally, the way in which the sample is exposed to the sensor surface determines the kinetic profiles in terms of rate constants, mass transport limitation, stagnant layer and diffusion gradient and depletion at the surface. Three main liquid handling systems can be identified: flow cells, cuvettes and microfluidic (bio)chips. The first two systems are described in this section. An outlook on SPR microfluidic biochips or lab-on-a-chip devices is given in Chapter 12. In most instruments, flow systems are applied with various degrees of automation ranging from simple to highly automated cartridges. Samples can be transported using syringes or peristaltic pumps, either with or without pneumatic valves and sample loops. Also, cuvettes can be operated automatically using a liquid handler. In cuvette systems, different mixing methods are applied to stir the sample solution. The cuvette can be described as a multiparameter, controllable batch reactor, in which binding events take place at the bottom or sensor surface of the cuvette.

3.3.1 Flow Cell Systems In SPR instruments, liquids are transported into the flow cell to study (bio) molecular interactions [24] at the sensor surface. Peristaltic or syringe pumps

46

Chapter 3

are applied to pump the liquid along the sensor chip surface. Some instruments apply automatic injection systems, definitely improving performance. In flow cells with laminar flow the sample is sometimes separated from the buffer by an air bubble; however, air bubbles may adhere and affect the surface properties of the sensor chips if hydrophobic coatings are used. Flow cells are formed by pressing a micromachined device with preformed microfluidic channels against the sensor surface. In 1990 Biacore introduced a flexible microfluidics system for its SPR technology based around integrated micro fluidics cartridges (IFC) technology (Figure 3.9). The IFC allows analyte to pass over the sensor surface in a continuous, pulse-free and controlled flow, maintaining constant analyte concentrations at the sensor surface. A liquid handling system is used to automate the biospecific interaction analysis procedures. Three major flow cell configurations are generally applied in SPR systems (Figure 3.10). Most commonly used are the planar flow cells. Confined wall-jet IFC channels

IFC Flow cells, formed by contact of the IFC on the sensor surface

Sensor surface

Glass slide Prism

Liquid Handling Minaturized system Low volume of reagents Intergrated and automated liquid handling

Buffer Sample Valve

Figure 3.9

The integrated microfluidic cartridge (IFC) of the Biacore 3000 instrument. Flow cells are formed when a microfluidic cartridge is pressed against a sensor surface (top). Pneumatic valves are used to guide the sample in the injection loop followed by flowing the sample over the sensor surface.

SPR Instrumentation

Figure 3.10

47

Top and side views of three flow cells used in SPR instruments. The hydrodynamic addressing flow cell is discussed in Section 3.5.

flow cells and hydrodynamic addressing flow cells are less common. A planar flow cell contains a simple inlet and outlet and a single broad channel through which the sample flows and interacts with the sensor surface. Owing to the small dimensions of the chamber6 and the resulting low Reynolds numbers, the flow is laminar [25]. Usually, the flow cell is inserted in the SPR instrument after the sensor surface has been prepared by the user; in some configurations also the immobilization cycle can be performed in a closed loop. The pneumatic system of a microfluidic cartridge in the Biacore instruments (Figure 3.9) opens and closes valves in user-defined protocols to control the flow in the flow cell. Sample loops can be manufactured at a very close distance to the sensor surface, minimizing dead volumes and dispersion, creating clearly defined injection plugs. Small deviations are related to the dead volumes (o1 ml) and to dispersion between flow cells, if e.g. four parallel flow cells are used. Wall-jet flow cells are mainly applied when high mass transport to the sensor surface is required at very low flows. In a confined wall-jet flow cell, the direction of the jet is radial along the sensor surface. Wall jet cells are often used to monitor fast surface processes [26]. Polar coordinates (circular approach) are applied to describe the radial gradient of flow velocity. As the flow decreases with increase in the radius of the wall-jet flow cell, the effect of flow rate on the biomolecular interaction process can be studied in one experiment on uniformly coated sensor discs using SPR imaging instruments. Hydrodynamic addressing can be used for the simultaneous measurement of multiple interactions in a single flow cell and is shown later in Figures 3.35 and 3.36 [27]. By adjusting the relative flow at the two inlets (one for the immobilized partner and one for buffer), liquids can be directed to different addressable detection spots. The flow cell design allows rapid and efficient switching of flow between buffer and sample solutions and the transverse arrangement of the detection spots ensures simultaneous access of sample to all spots. As there is no lag time between interactions, highly accurate reference subtraction allows the measurement of very rapid kinetics. By immobilizing several ligands in one 6

The smaller the sample volume, the better.

48

Chapter 3

flow cell, comparative binding properties can be directly examined under identical experimental conditions.

3.3.2 Cuvette Systems Cuvette liquid handling systems consist of an open container filled manually or automatically by a liquid handling robot or liquid handler (Figure 3.11). The cuvette must be mounted leak-free on the sensor chip and a correct configuration of the optics (from the bottom) is required to apply the open container or cuvette. The biomolecular interaction takes place on the replaceable sensor chip at the bottom of the cuvette. If the liquid is not mixed, sensorgrams are unreliable due to inhomogeneous, uncontrolled molecular transport to the surface. Therefore, cuvette systems are usually equipped with a mixing system. In contrast with flow cells, cuvette systems are not prone to clogging, hence liquid samples with solid particles can be measured in cuvette systems, provided that the sensor surface is not damaged by impinging particles. Examples of such samples are fermentation media, blood plasma, cell cultures and food products. Because the sample stays in the cuvette during the entire biomolecular interaction process or analysis cycle, the undiluted sample can be recovered almost completely. In principle, a sample volume in the range of 25 ml is sufficient to study biomolecular interactions for up to an hour. A disadvantage of the cuvette system is the open architecture, allowing uncontrolled evaporation of the

Figure 3.11

Cuvette of the ESPRIT, Eco Chemie Netherlands, system with two containers, two drain connections for each container and two inlet connections. The cuvette can be automatically filled and drained while mixing takes place with two needles inserted in the containers connected to two syringe pumps.

SPR Instrumentation

49

sample solution. When low-concentration samples are used, depletion of the analyte occurs and kinetic models should compensate for this effect, as described in Chapter 5. Ideally, an SPR instrument would be compatible with both cuvette or flow cell, allowing the user to choose for the best performance and quality of the biomolecular interaction process of interest. Various mixing systems can be used to agitate the sample in the cuvette, including stirrers [28], vibrating acoustic plates [30] and aspirate/dispense mixing [6]. In principle, every device can be applied that controllably and effectively induces turbulence to homogenize the sample solution and to minimize the stagnant layer at the sensor surface. For example, in early resonant mirror instruments the sample solution was stirred by a high-speed rotating stirrer [29]. A disadvantage of this system is that mass transport to the surface is not effectively controlled even at fast rotation speeds. Later, a vibrating plunger/ piston was developed, showing better mass transport characteristics [30]. Mixing can also be achieved by a syringe pump with a controllable automatic aspirate/dispense mixing needle.7 In these instruments, a syringe pump is constantly aspirating and dispensing the sample or buffer in the cuvette during real-time measurements to obtain reproducible hydrodynamic conditions at the sensor surface. During a mixing cycle of set volume and speed, first part of the cuvette volume is aspirated. During the dispense action, a free wall-jet of the sample solution can be forced to flow in the diffusion layer of the surface. As a result, the mass transport can be dramatically increased. Relevant hydrodynamic parameters of an aspirate/dispense mixing system are sample volume, mixing volume, which should be part of the sample volume, speed or frequency of mixing, diameter of the nozzle (inner diameter of the needle), distance and position of the nozzle to the sensor surface, diameter of the cuvette, viscosity and temperature of the sample solution. The cuvette needs to be drained after the interaction process, but the surface should not become dry, therefore hydrophilic coatings are used to prevent the sensor surface from being exposed to air. If the hydrogel dries out, however, irreproducible effects will occur. Physical transport determines how fast the sample molecules are transported to the surface. Mass transport limitations arise when the concentration of the analyte at the sensor surface is lower than the sample bulk concentration (for a detailed description, see Chapter 4). In aspirate/dispense mixing systems, mass transport to the surface is greatly increased and the diffusion-controlled stagnant layer at the surface is strongly reduced. The process can be described by the dynamic free wall-jet principle first published by Glaubert [31]. As various parameters affect the biomolecular interaction process, the reproducibility of the measurement can be maintained best in automated operation at constant hydrodynamic parameters. The reader is referred to Chapter 5 for details on kinetic measurements and kinetic evaluation software for biomolecular interaction parameters, and a practical example is given in Chapter 9. 7

The systems of IBIS, SPRINGLE and ESPRIT of Eco Chemie and the instruments derived from the PLASMOON system of Biotul, e.g. by Plasmonic Biosensor.

50

Chapter 3

3.4 SPR Instruments: State of the Art Currently, SPR instruments are available from various manufacturers. In Table 3.1 and the following sections, a short description is given of commercially available SPR instruments, sorted by their optical configuration, describing the main features and benefits of the instruments. In addition, other instruments can be identified that detect biomolecular interactions in real time and label free which are not directly based on SPR. Information on these products and manufacturers is compiled in Table 3.2. It should be noted that this list might not be complete but it provides an illustration on the range of products and manufacturers in the label-free biosensor area.

3.4.1 Examples of Fan-shaped Beam SPR Instruments Biacore8 (Uppsala, Sweden) dominates the SPR market with more than 90% of installed products and 87% of the publications in 2005 [39] and dictates the standards serving as reference for all other instrument manufacturers (Figure 3.12). The principle of fan-shaped SPR instruments has been used by Biacore since the launch of the first product in 1990. In Biacore instruments a light-emitting diode (l ¼760 nm) is used and a convergent light beam reflects at an exact position at the sensor surface. A photodiode array is used to determine accurately the SPR dip position. Extending the fan of light into a wedge and using a two-dimensional detector allows detection along a line on the sensor surface. The benefit of using a fan of light and a linear array detector is that no moving parts are required to carry out SPR assays. A computer algorithm calculates the exact position of the minimum to a fraction of the size of a single diode. The accuracy of the optical system corresponds to about 0.1 millidegree shift of the SPR angle. Biacore applies an optogel to optically match the sensor chip with the cylindrical prism. The latest instruments have 20 sensor spots in four flow cells. With different probes immobilized at defined spots along a line, simultaneous measurement of different biomolecular interactions is realized. Biacore applies several types of flow cells connected to a microfluidic cartridge. The BI-SPR of Biosensing Instruments (BI) (Tempe, AZ, USA) uses an innovative method to detect the SPR angle, key to the high performance of the instrument (Figure 3.13). A shift of the surface plasmon resonance angle results in a shift in location of the SPR dip on the detector surface. The two flow channels can be used in conjunction with two valves for simultaneous measurement of two samples. Alternatively, one channel can be used for background subtraction. The system provides a quick and easy setup with various cell modules for DNA sequencing, protein–protein interaction, ligand/receptor recognition, drug development applications and the optional EC cell module (not included with the system) for electrochemical SPR measurements. 8

Acquired by GE Healthcare in 2006.

a

Biacore Biosensing Instrument Nomadics DKK-TOA Reichert Plasmonic Biosensor Resonant Probes Moritex Optrel GBR Analytical m-Systems Sensia K-MAC Nanofilm Surface Analysis EcoChemie Biacore Thermo Electron Corp. NeoSensors GWC Technologies GenOptics/Horiba Jobin Yvon Bio-Rad Laboratories Toyobo Lumera Graffinity Technologies IBIS Technologies

Manufacturer

Biacore product line, including: Bialite, Biacore J, -X, -1000, -2000, -3000, -C, -S51, -Q, -A100, -T100, -X100.

Biacore product line BI-SPR1000 SensiQ SPR-20 SR7000 Plasmonic SPTM SPR 670 M Multiskop BIOSUPLAR-321 Sensia b-SPR SPRLAB and SPRi Nanofilm EP3 SPRINGLE/ESPR IT FLEXChip SPR 100 IAsys SPRimager II SPRi-Plex ProteOn XPR36 MultiSPRinter Proteomic Processor Plasmon Imager IBIS iSPR

Fan shaped, converging Fan shaped, converging Fan shaped, diverging Fan shaped, converging Fan shaped, diverging Fan shaped, diverging Fixed angle Fixed angle Fixed angle Fixed angle Fixed angle Fixed angle+imaging Fixed angle+imaging Angle scanning Grating coupler Wavelength Resonant mirror Imaging Imaging Imaging Imaging Imaging Imaging/wavelength Imaging/scanning

a

SPR system

Overview of SPR instruments and their manufacturers [32,33].

Type

Table 3.1

www.biacore.com www.biosensingusa.com www.nomadics.com/ www.dkktoa.net/ www.reichertai.com/ www.plasmonic.de/ www.resonant-probes.de/ www.moritex.com www.optrel.de www.biosuplar.de www.sensia.es www.k-mac.co.kr/ www.nanofilm.de/ www.ecochemie.nl www.biacore.com www.thermo.com www.neosensors.com/ www.gwctechnologies.com www.genoptics-spr.com/ www.bio-rad.com www.toyobo.co.jp www.lumera.com/ www.graffinity.com www.ibis-spr.nl

URL address

SPR Instrumentation 51

52

Table 3.2

Chapter 3

Non-SPR instruments for real time and label-free monitoring of biomolecular interactions.

Manufacturer and product

Working principle

URL address

Axela Biosensor: dotLab System Farfield Scientific (UK): AnaLight Silicon Kinetics: nanoporous silicon (npSi) SRU Biosystems (Woburn, MA, USA): BIND system Corning: Epic system, highthroughput optical biosensor ForteBio: Octet system

Grating-based light diffraction Dual-polarization interferometry Optical biosensor substrate Nanostructured optical grating Dynamic mass redistribution (DMR) effect Bio-layer interferometry (BLI) optical fiber biosensor Ellipsometry in Kretschmann configuration Acoustic effect

www.axelabiosensors.com

Maven Biotechnologies: LFIRE Akubio: acoustic biosensor

www.farfield-scientific.com www.siliconkinetics.com www.srubiosystems.com www.corning.com/ lifesciences www.fortebio.com

www.mavenbiotech.com www.akubio.com

Nomadics (Dallas, TX, USA) markets SPR technology in a miniaturized SPR sensing device named SensiQ, in collaboration with Texas Instruments (TI) (Figure 3.14). TI’s SPREETA SPR sensor, coupled with a high-resolution electronic interface and flow equipment in the SensiQ device, allows analysis using Windows-based computers. SensiQ is a dual-channel, semi-automated SPR system which utilizes advanced microfluidics and proven surface attachment methodologies. Concepts can be evaluated quickly with minimal or no hardware and software development. The Model SPR-20 instrument of DKK-TOA (Takadanobaba, Tokyo, Japan) performs measurements using a 101 convergent beam directed into the prism and a CCD camera as the detector (Figure 3.15). The SPR angle can be adjusted manually between 35 and 851. Standard equipment includes both flow-type and cuvette-type analysis cells. Measurements at two sensor spots in a flow cell can be performed without the need to move the mobile stage because the convergent beam has sufficient angle resolution. The instrument is suited for both gas- and liquid-phase samples and has a resolution of 3 millidegrees. The software associated with this instrument is run on a standard Apple Macintosh personal computer. The Reichert (New York, USA) SR7000 DC surface plasmon resonance dualchannel spectrometer uses a fan-shaped divergent beam, unlike in Biacore instruments, where a convergent fan shaped beam is used (Figure 3.16). The advantage of this setup is that no moving parts are required. However, SPR does not occur at a fixed position on the sensor but ‘‘walks’’ along the surface while the biomolecular interaction is proceeding and therefore the SPR path is moving,

SPR Instrumentation

Figure 3.12

53

Part of the SPR product line of Biacore: Biacore J, X, 3000, C, S51 and X100. The other Biacore systems [A100, T100, Q (see Chapter 11) and FLEXChip instrument, Section 3.5] are not shown here but separately highlighted in Section 3.5.

limiting the use of small flow cells. The Peltier temperature-controlled instrument is connected to a liquid handler (Endurance, Spark, The Netherlands). Plasmonic Biosensor (Wallenfels, Germany)9 acquired autosampler technology from Biotul (Munich, Germany) and developed a new, patented optical detection system with disposable prisms (Figure 3.17). The optical system generates surface plasmon waves by a defocused laser beam which reflects at the surface of a gold-covered prism. With this optical cell, adhesion phenomena can be detected by using the SPR effect. The Plasmonic instrument allows the real-time analysis of unpurified samples contained in an array of eight cuvettes in parallel. The sample handling is carried out with a patented, removable, accurately positioned pipette tip system with no cross-contamination. 9

Formerly Jandratek GmbH.

54

Chapter 3

Figure 3.13

The BI-SPR of Biosensing Instruments.

Figure 3.14

Nomadics SensiQ with SPREETA chip. A divergent fan-shaped beam reflects at the sensing surface.

3.4.2 Examples of Fixed-angle SPR Instruments The SPTM instrument of Resonant Probes (Goslar, Germany) is a simple SPR spectrometer. The setup consists of a high-precision goniometer10, a versatile and accessible sample holder, a laser and a detection unit based on lock-in amplification. The lock-in detection makes the system insensitive to stray light, even though the setup is completely open. SPTM is ideally suited for research environments. The coupling angle, yc, is determined with an accuracy of 0.011. Moritex (Myutron, Tokyo, Japan) has a complete product line of biosciencerelated laboratory instruments including an SPR platform11. The Moritex SPR 10 11

To determine a precise angular position. Moritex Corporation acquired SPR technology from Nippon Laser and Electronics Lab in 2004.

SPR Instrumentation

55

Figure 3.15

Photograph and scheme of the DKK-TOA SPR-20 instrument. The angle position can be set in a large range.

Figure 3.16

The SR7000 DC instrument of Reichert (right) connected to an autosampler (middle) controlled by syringe pumps (left).

Figure 3.17

The SPR instrument of Plasmonic Biosensor.

56

Figure 3.18

Chapter 3

The Multiskop of Optrel GBR in a horizontal (left) and vertical (right) optical setup. The instrument can be applied either in an SPR Kretschmann configuration or in an ellipsometric setup.

670M system operates with two sensing spots and employs a flow system using a peristaltic pump. The system is currently under redesign before market introduction. Optrel GBR (Kleinmachnow, Germany) was founded in 1996 and focuses on the design of instrumentation for the investigation of interfaces and thin films. The Multiskop of Optrel GBR is a unique, single, modular system which incorporates surface plasmon spectroscopy (SPS) and ellipsometry in a single instrument (Figure 3.18). The modular design permits many different configurations. The BIOSUPLAR series of Analytical m-Systems/Mivitec (Sinzing, Germany) were introduced to the market at the end of the 1990s. At present, the third generation is in production (BIOSUPLAR-321). The device has two optical channels which can be used independently or as reference and sensing channels for differential measurements. The flexible setup allows measurements of angle dependence of the reflected light, monitoring of the time dependence of the resonance angle and measurements of light intensity at a fixed angle. The open system can be used for setting up an SPR experiment. Moreover, the basic configuration allows practical training and tutorials for new SPR users (Figure 3.19). The b-SPR research platform of Sensia (Madrid, Spain) uses the Kretschmann configuration to achieve total internal reflection resonance conditions (Figure 3.20). Polarized light from a laser diode reflects at the gold surface of two flow cells, with a volume of 300 nl each, suited for the simultaneous measurement of two biomolecular interactions. The platform operates with a sample and reference cell. An advantage is that the technology is simple from the user’s point of view; however, alignment of the beamsplitter and guiding reflected light from the small flow cell make the system complicated from a technology point of view. A prism and a multi-photodiode are located on

SPR Instrumentation

Figure 3.19

The Bio-Suplar is a compact instrument.

Figure 3.20

The b-SPR instrument of Sensia.

57

a concentric rotary stage with an angular resolution of 0.011. The b-test SPR system has a refractive index detection limit of 10
5. K-MAC (Korea Materials and Analysis Corp., Daejeon, Korea) is a venture company for analytical instruments. In 2003, K-MAC started the development of SPR instruments with the SpectraBio2000, recently renamed SPRLAB , a motor-operated, fixed-angle instrument (Figure 3.21). A second instrument (SPRi ) is now available based on SPR imaging technology for protein analysis in drug discovery and disease diagnosis applications.

58

Figure 3.21

Chapter 3

The SPRLAB (A) and the SPR imaging SPRi system (B) of K-MAC.

SPRLAB is an angle scanning instrument with gold-covered prisms; its liquid handling system combines a microfluidic flow cell with a precision syringe pump. The instrument is characterized by a wide dynamic range combined with high sensitivity to detect mass changes on the gold surface. It is suited to work with strong acids and bases, organic solvents including DMSO and CCl4. The stepper motor-operated angle position instrument can be tuned to follow the angle shift as a function of time. The recently launched SPRi imaging system is a reflectivity-based SPR imaging instrument with manual setting of the angle of incidence, designed for rapid monitoring of the biochip sensor, an array of various biomolecules, such as proteins, recombinant proteins, cells or other microorganisms. SPR imaging is used for high-throughput analysis of bioaffinity interactions by fabricating DNA/protein arrays on biochip gold surfaces. The company applies a range of sensor chips, including prism coated gold slides. Nanofilm Surface Analysis (Go¨ttingen, Germany) developed the Ellipsometry Platform EP3 with an additional SPR experimental option in the classical Kretschmann configuration (Figure 3.22). The SPR cell is specially designed for kinetic measurements of biomolecular interactions as an alternative for ellipsometric measurements on transparent substrates, e.g. glass slides (OptiSlides). The sensitivity of the OptiSlide ellipsometric measurements, attained without

SPR Instrumentation

Figure 3.22

59

The Ellipsometry Platform EP3 including SPR option of Nanofilm.

the gold layer, is one order of magnitude less than that of conventional SPR. However, the lateral resolution is at least 10 times better, permitting the labelfree study of the quality of microarray spots. The Nanofilm EP3 measures the ellipsometric parameters Psi and Delta in addition to the reflected light intensity in the SPR mode of operation. Psi is analogous to the reflected light intensity in classical SPR, whereas the phase signal Delta gives additional information on, e.g., surface roughness, exceeding the limitations of standard SPR. The combination of SPR and imaging ellipsometry allows for quality control of microarrays, as it measures surface morphology through 2D thickness maps and 3D profiles of the inner layers of stacked multilayers, in addition to microarray spots with improved lateral resolution. The most frequent application of SPR imaging ellipsometry is marker-free parallel kinetics recording of up to 100 spots.

3.4.3 Examples of Angle Scanning SPR Instruments The SPR system of EcoChemie (Utrecht, the Netherlands) applies angle scanning optics. The minimum of the SPR dip is detected as fast as 76 Hz, hence it is fair to say that this instrument detects the true SPR dip continuously. The initial design of the instrument12 has proven its reliability and flexibility for a decade. Redesigned by EcoChemie, the SPRINGLE is sold as a single-beam instrument whereas the ESPRIT system operates with two channels and has automated sample transfer with an autosampler (Figure 3.23). The SPR angle resolution and noise levels are better than 0.1 millidegrees, similar to those of 12

Previously sold by IBIS Technologies as IBIS I and IBIS II instruments.

60

Figure 3.23

Chapter 3

The double-beam ESPRIT instrument (left) and the single-beam SPRINGLE SPR scanning instrument of EcoChemie.

the best high-end SPR instruments on the market. The cuvette-based instrument applies aspirating-dispensing mixing through a syringe, whereas the interaction process is followed in real time. EcoChemie sells an adaptor for using a Biacore sensor chip with the instruments. A unique feature is the integrated AUTOLAB electrochemical measurement workstation to perform electrochemical SPR (E-SPR) measurements. In this application, the gold surface is used as an electrode connected to the potentiostat output of the AUTOLAB workstation. Real-time SPR can be combined with electrochemical measurements in one experiment. The prices of both EcoChemie systems make them competitive single- and double-beam type instruments. In the literature [38], Windsor Scientific, which is a former distributor of Eco Chemie and IBIS Technologies, is incorrectly denoted as the instrument manufacturer of IBIS instruments.

3.4.4 Examples of Grating Coupler SPR Instruments The FLEXChip system13 of Biacore (GE Healthcare) provides an open, arraybased platform for kinetic screening, allowing the simultaneous study of the interactions of hundreds of proteins and peptides against a single sample (Figure 3.24). The FLEXChip system utilizes a grating coupler for SPR analysis and is also an imaging apparatus. For more details, see Section 3.5.4.

3.4.5 Examples of Other SPR Instruments The SPR 100 module of Thermo Electron Corp. (Waltham, MA, USA) is based on wavelength-specific SPR detection (Figure 3.25). An added Fourier transform infrared (FT-IR) spectrometer module provides fully fledged infrared spectroscopic capabilities providing chemical information on the bound 13

Originally from HTS Biosystems (Hopkinton, MA, USA).

SPR Instrumentation

61

Figure 3.24

Photograph including optical setup of Biacore’s FLEXChip instrument.

Figure 3.25

The SPR 100 module of Thermo Electron Corp. This instrument uses the wavelength interrogation SPR principle (see Figure 3.8).

species. FT-IR-SPR measures the wavenumber corresponding to the minimum reflectivity, resulting in significant sensitivity advantages over the traditional angle shift-based SPR technology. The SPR 100 captures quantitative data for an extraordinarily diverse range of applications. The unexceeded broad dynamic range is due to the combination of broad spectral range with broad incident light angles from B40 to 701. Measurements can be made in the liquid or gas phase in a single instrument. The module was specifically designed for use with Thermo Electron’s FT-IR spectrometers. At present, GWC Technologies (Madison, WI, USA) markets the SPR 100 and also the SPRimager II instrument as described in the SPR imaging section (Section 3.4.6). In the mid-1990s, the IAsys system of NeoSensors. (Sedgefield, UK) was the cuvette-based alternative to Biacore instruments and deserves special attention here (Figure 3.26). It is still the second label free biosensor, with 40 publications in 2005. The cuvette was mixed with a stirrer, later a vibrating acoustic plate,

62

Chapter 3

Sample well Fibrin monomers

Vibro-stirrer

Fibrinogen layer

Ligand attachment surface Resonant mirror Low index resonant layer

Incident laser light

Figure 3.26

Prism

Resonated reflected light

The IAsys system of NeoSensors (from Ref. [30]). The resonant mirror principle of the IAsys applies detection in the evanescent field but is not SPR.

and operated with a liquid handler to inject the sample automatically into the cuvette. Although the resonant mirror technology of the IAsys system14 is not aimed at SPR, it provides an alternative technology to measure refractive index changes in the evanescent field at the sensor surface (see Section 3.2.5). The IAsys system is a user-friendly instrument with elaborate sensor surface chemistry; fluidics (cuvettes) and applications are similar to those of SPR biosensors. 14

Originally developed by Fisons (UK) and acquired by Thermo Instruments and Affinity Sensors (Cambridge, UK), now NeoSensors (Sedgefield, UK), which is a division of the Farfield group.

SPR Instrumentation

Figure 3.27

63

The SPRimager II of GWC Technologies. The SPR angle should be set manually using a spindle.

3.4.6 Examples of SPR Imaging Instruments The SPRimager II of GWC Technologies (Madison, WI, USA), originally developed in the group of Robert Corn15, contributor of Chapter 8 of this book [34], captures data on the entire sensor surface simultaneously with a CCD camera (Figure 3.27). A fixed angle is set manually at the left angle flank of the SPR dip. As described in Section 3.1.1, the optimal angle in terms of maximal reflectivity is in the inflection point of the SPR dip. However, this angle can only be set for a reference spot and not for all different spots simultaneously. Under appropriate conditions, the SPR response is a linear function of the surface coverage provided D%R r 10%. The instrument can be used to study biomolecular interactions at different spots, for example in a 5  5 microarray. The instrument provides various manually operated control features to check the quality of the interactions. The degree of automation is limited and a single peristaltic pump is used to pump the sample through a vertically positioned, gold-covered sensor mounted in a flow cell. The instrument uses disposable sensors of SF10 glass. These high refractive index substrates are not compatible with standard optical glass substrates (e.g. K5 or BK7), hence the prism is also made of SF10. Refractive index matching oil is used between the disposable sensor and the prism. GWC Technologies also markets the Thermo Electron product the SPR100 (see Section 3.4.5). 15

Present affiliation: University of California, Irvine, CA, USA.

64

Chapter 3

GenOptics (Orsay, France) is a laboratory instrumentation provider of highperformance systems with advanced optical imaging (SPRi) to quantify and monitor biomolecular interactions. GenOptics instruments use the Kretschmann configuration to excite surface plasmons and are equipped with a rotating mirror for scanning precisely the angle of incidence in the SPR reflectivity dip. This scan allows the selection of the best reflectivity performance of the sensor chip to monitor protein–protein interactions in real time. A broad monochromatic polarized light illuminates the whole functionalized area of the SPRi-Biochip, which is combined with a detection chamber. Information can be quickly obtained from the interaction process while monitoring reflectivity variations against time. GenOptics currently markets two protein-array platforms: the SPRi-Plex for large-scale, automated screening and SPRi-Lab+ with an open configuration for biomolecular interaction experiments and application development (Figure 3.28). Horiba Jobin Yvon (Longjumeau, France) obtained exclusive, world-wide distribution rights for both the SPRi-Lab+ detector and SPRi-Plex. The ProteOn XPR36 protein interaction array system of BioRad Laboratories (Hercules, CA, USA) is a SPR imaging biosensor with a multi-channel module and interaction array sensor chip for analysis of up to 36 protein interactions in a single injection step (Figure 3.29). The capability of the ProteOn XPR36 system to generate rapidly a 6  6 interaction array between six ligands and six analytes greatly increases the throughput, flexibility and versatility of experimental design for a wide range of biomolecular interaction studies. A 901 mechanical switch allows the placement of six flow lines perpendicular to each other on the sensor surface. Prior to the analysis cycle, six different ligands can be immobilized on the sensor surface. Then perpendicular to this immobilized ligands in lines, the user can inject the analyte and acquire kinetic data in so-called one-shot kinetics of six biomolecular interactions in six analyte dilutions in a single run. Toyobo (Osaka, Japan) was founded in 1882 to manufacture high-tech materials and currently offers products for the life sciences, including biosensor equipment. Toyobo’s MultiSPRinter is an array-based SPR sensor platform distributed in Japan. This system provides a complete detection system including a spotter. The platform has been used to monitor DNA hybridization and kinetics for protein binding on an array of related double-stranded DNAs. In collaboration with RIKEN (www.rikenresearch.riken.jp), Toyobe launched the so called Photo-linker Chip for the MultiSPRinter for fast microarray production. The photo-linker chip is able to provide up to 96 low molecular weight spots on a microarray. The LFIRE from Maven Biotechnologies (Pasadena, CA, USA) allows precise, real-time measurements of specific interactions between molecular entities in a microarray or well-plate format (Figure 3.30). These molecular entities can be proteins, nucleic acids, lipids, small molecules such as drugs or steroids or even whole cells. LFIREis based on ellipsometry, a technique that measures changes in the polarization of light upon reflection from the interface between materials. However, the optical setup is different from the Nanofilm configuration shown in Section 3.4.2, but similar to the Kretschmann configuration commonly used in

65

SPR Instrumentation

PEEK tubing Waste

Light source

Flow cell

Injection valve Buffer Syringe pump SPRi Biochip

Polarizer Optical system 1

Optical system 2 CCD Camera

Mirror

Figure 3.28

The optical setup of the GenOptics SPRi-Plex and SPRi-Lab+ system. By changing the mirror angle an optimal reflectivity contrast can be obtained as shown in the image with 460 spots.

SPR instruments. As a ‘‘real-time’’ system, it monitors reactions as they happen, providing kinetic information on biomolecular interactions. The Proteomic Processor of Lumera (Bothell, WA, USA) uses SPR microscopy in which a beam of light is directed on to a spot of a microarray (Figure 3.31). The

66

Chapter 3 A

B

Figure 3.29

The ProteOnTM XPR36 of BioRad with crisscross microfluidics. (A) Ligands can be immobilized in 6 lanes. (B) The analyte can be passed perpendicular in 6 lanes over the surface and detection at the cross sections take place.

Figure 3.30

The LFIRE of Maven BioTechnologies is not an SPR imaging instrument but an ellipsometry imaging system in attenuated total reflection configuration. Biomolecules in the evanescent field will change the polarization of the reflected light.

Figure 3.31

The Proteomic Processor of Lumera. A scalable beam is applied and the reflectivity can be monitored for thousands of spots simultaneously.

SPR Instrumentation

67

reflected light in SPR microscopy changes due to a chemical binding event. The proprietary optics of the instrument allows simultaneous addressing thousands of spots in the reflectivity mode. The Proteomic Processor extends the power of SPR to the analysis of high-density microarrays by use of a novel optical design that includes a microelectromechanical systems (MEMS) mirror for rapid scanning the microarray. The light of a diode laser is precisely oriented onto spots of an array surface. Scanning at 60 Hz, the MEMS mirror optical design permits the simultaneous interrogation and analysis of high-density microarrays. The light beam is scalable to larger surface areas while maintaining uniform beam intensity. Array spot intensity is recorded by a high-definition CCD camera. The device utilizes the Kretschmann configuration and gold-covered, high refractive index glass slides. Lumera combines protein arrays with high-throughput surface plasmon resonance through its proprietary Heterodimer Protein Technology (HPT) peptide tag system. Graffinity Pharmaceuticals GmbH (Heidelberg, Germany) has developed the Plasmon Imager for discovery of small molecular hit and lead compounds. The platform is used to find novel small molecules for drug discovery and chemical genomics approaches. Graffinity has developed high density chemical microarrays for fragment screening consisting of small molecules immobilized on to gold chips in combination with high density spotting [79,80]. High-throughput, label-free fragment screening and screening of displayed-fragment microarrays on the proprietary SPR platform, allows Graffinity to rapidly identify novel drug fragments and lead-like molecules as ligands for a biomolecular target. Synthesized fragment libraries are stored in microtiter plates until spotting on to the sensor fields by using a customized high-throughput and highprecision pintool spotting robot. In Graffinity’s setup, a wavelength shift that corresponds to the increase in mass concentration on the chip surface during binding between the target proteins and the immobilized chemical substances is recorded. Array screening is performed by flowing protein solutions over the chemical microarray, followed by the binding of the target by end-point measurements within 3 hours. Graffinity’s Plasmon Imager allows the parallel readout for up to 9612 sensor fields per array. The measured SPR shift can be visualized in false-colored 2D plots to obtain affinity fingerprints, such as those given in Figure 3.32. The Plasmon Imager instrument has not been marketed, but is used in projects for partners of Graffinity. In the 1990s, the IBIS I and II systems of IBIS Technologies16 (Hengelo, The Netherlands) were introduced to the market. In 2007, IBIS Technologies launched a new advanced imaging SPR instrument with a patent pending for angle scanning imaging technology. The system can be categorized as a new combination of angle scanning SPR (see Section 3.4.3) and an SPR imaging instrument (see Section 3.4.6). The IBIS iSPR instrument (Figure 3.33) allows simultaneous monitoring of multiple binding interactions on microarrays spotted on the sensor surface. The instrument is equipped with a standard planar flow cell but can be used with a cuvette or with a confined wall-jet flow cell as well. 16

Later redesigned by Eco Chemie (Utrecht, The Netherlands) as ESPRIT and SPRINGLE instruments, respectively (see Section 3.4.3).

68

Figure 3.32

Chapter 3

The in-house built Plasmon Imager of Graffinity Pharmaceuticals applies the wavelength interrogation principle (see Section 3.2.5) and a highdensity microarray with thousands of spots will be imaged.

A liquid handler operates the instrument automatically using various vials or microtiter plates (96 or 384 wells), allowing extended stand-alone operation. The Peltier element-controlled sensor compartment is thermostated to better than 0.01 1C and the sample compartment can be cooled separately. The IBIS iSPR system combines the detection of the SPR dip with imaging of the entire sensor surface. In so-called dynamic scan operation, the reflectivity change which is normally applied in fixed-angle imaging instruments is converted into a real SPR dip angle shift and can be used for real-time detection and also calculation of kinetic parameters of binding events at user-defined spots on a microarray, even when spots have SPR dip differences above 3000 millidegrees. Images of the entire microarray can be observed at once from the microscopic view on the monitor. Contrasts of the image for specific regions of interest (ROIs) can be set by the SPR angle or digitally improved by software settings. In the software the SPR dip shifts of reference and control spots can be subtracted from each other. In addition, the user can define hundreds of regions of interest, adding enormous flexibility to the system. In a technical sense, the IBIS iSPR system combines the best of both worlds: real-time imaging of the entire sensor surface is combined with an SPR dip scan (instead of reflectivity) of hundreds of biomolecular interactions (4500). Further, a whole range of (patent-protected) surface chemistries can be delivered by IBIS Technologies. In Chapter 7, results are presented of direct serum detection of autoantibodies from 50 rheumatoid arthritis patients.

SPR Instrumentation

Figure 3.33

69

The SPR imaging instrument IBIS-iSPR of IBIS Technologies. The left computer screen shows the sensorgram and quality of the SPR dip, and the screen on the right shows the microscopic image of the surface. A region of interest (ROI) can be selected in size and position on the surface (red squares) and the user can inspect the quality of the surface before a large series of samples are automatically passed over the surface using the autosampler. The autosampler operation can be inspected through the instrument window.

3.5 Protein Interaction Analysis Systems of Biacore 3.5.1 Introduction Biacore is the market leader in the development and production of SPR-based analytical instruments since their introduction to the market in 1990. The data generated from real-time protein interaction studies using these systems have provided scientists with an unprecedented wealth of information on protein function, within areas as diverse as learning how specific protein domains contribute to biological function and the ability to make informed judgments on the potential of specific proteins as targets in drug development. The Biacore product line includes the Bialite, Biacore J, -X, -1000, -2000, -3000, -C, -S51, –Q, -A100, -T100, -X100 and FLEXChip. Three Biacore systems, the Biacore A100, T100 and FLEXChip, each designed to address a specific need within life science research and the pharmaceutical industry, are reviewed here. SPR-based arrays are now available, with the emphasis on ‘‘information-rich’’ data rather than throughput, delivering information on association, dissociation and strength of interaction while analyzing interactions in parallel. Two interaction array systems are available: FLEXChip, for simultaneous profiling of up to 400 protein interactions in a single run, and Biacore A100, which delivers

70

Table 3.3

Chapter 3

Overview of the Biacore A100, T100 and FLEXChip instrumentation.

Instrument

Application area

Main features

Biacore T100

Research to quality control

Biacore A100

Protein interaction analysis

FLEXChip

Comparative profiling: map, rank and select

Automatic analysis of up to 384 samples per run over the broadest kinetic range Analysis and evaluation of interactions involving low-MW compounds (o1000 Da) Integrated buffer degassing improves robustness when studying interactions at elevated temperatures Integrated sample cooling for temperature-sensitive samples Automated recovery of interaction partners for MS analysis Analysis of up to 3800 interactions in 24 hours Unique flow system for parallel analysis – increased sample throughput and multiplexed analyses Robust data and long unattended run times Software tools designed for array applications Simultaneous profiling of up to 400 protein interactions Efficient comparison and ranking hundreds of interactions Sample recirculation for measuring slow association events

high-quality data on up to 3800 interactions in 24 hours. These systems, together with the Biacore T100, are reviewed in Table 3.3 (see also Figure 3.34A).

3.5.2 Biacore T100 The Biacore T100 is a highly automated system for comprehensive protein interaction analysis from early drug discovery through drug development to quality control. In addition to providing detailed information on kinetics and affinity, software support allows interactions to be thermodynamically characterized. Both the sample compartment and flow cell system of the instrument are temperature-controlled, a feature that extends the potential use of the instrument to include the analysis of interactions at physiological temperatures and above. Although the key features of Biacore’s protein interaction analysis systems have remained consistent over the years – label-free, detailed characterization of protein interactions in real time – they have continued to develop not only in terms of kinetic resolution, ease of use, automation and regulatory compliance, but also in the depth of information revealed about individual interactions.

SPR Instrumentation

Figure 3.34

71

The Biacore instruments as described in this section. (A) Biacore T100; (B) Biacore A100. The FLEXChip instrument is shown in Figure 3.24.

For example, in addition to characterizing kinetic profiles, it is now also quick and simple to characterize the underlying thermodynamic principles that drive interactions, and the possibility of integrating Biacore T100 into an analysis and identification workflow together with mass spectrometry and hence protein identification has firmly placed the system in the field of functional proteomics.

72

Chapter 3

The design of the flow cell system in the Biacore T100 creates optimal conditions for accurate reference subtraction. Four flow cells allow single, paired or serial runs, and paired, on-chip flow cell connections mean that the void volume between flow cells is as small as possible. Kinetic rate constants may be measured over a broad range, from the fastest on-rates encountered in biological systems to the slowest off-rates on-rates from 103 to 107 l mol
1 s
1 (and higher for macromolecular analytes) and off-rates from 10
5 to 0.5 s
1 may be confidently measured. Samples may be maintained from 4 to 45 1C in a temperature-controlled compartment, allowing unattended analysis of temperature-sensitive samples. Assuming an analysis cycle of 7 minutes, up to 384 samples may be processed during 48 hours of unattended operation in a single run. Finally, dedicated software supports kinetic evaluation of low molecular weight compound interactions involving binding partners with molecular weight as low as 100 Da. In addition to a cooling compartment, the entire flow cell system is temperature controlled and is monitored with an integrated buffer degasser, which eliminates the appearance of air bubbles in the flow system, making possible the analysis of samples at elevated temperatures. It is therefore not only possible to analyze temperature-sensitive samples, but also the interactions themselves may be characterized at temperatures ranging from 4 to 45 1C. Consequently, interactions may be studied at physiological temperatures, permitting the behavior of therapeutics in vivo to be predicted more confidently. This is an important advantage of the Biacore T100 over typical ‘‘benchtop’’ assays in applications such as the characterization of monoclonal antibodies as biotherapeutics, where interaction profiles with target proteins may differ radically at ambient and physiological temperatures. A further consequence of the ability to accurately control temperature within the entire flow cell system is the possibility to derive transition state thermodynamic data from kinetic profiles. Kinetic profiling imparts information about the rate of complex association and dissociation (in addition to revealing the affinity of an interaction), but to understand why the interaction proceeds at these rates, it is necessary to define the thermodynamics of the system. Fully understanding molecular recognition by being able to predict binding energetics through thermodynamic analysis may provide the basis for structure-based molecular design of drugs and engineered antibodies. Protein interaction analysis on the Biacore T100 in combination with mass spectrometry (MS) provides the possibility of identifying proteins on the basis of functional binding criteria. In a typical experiment, molecules are isolated from a complex matrix based on their ability to bind specifically an interaction partner immobilized on a sensor surface. The bound material then can be recovered in a non-destructive manner by using a recovery solution that completely dissociates the bound analyte without damaging the immobilized interaction partner on the sensor surface and thus allowing MS analysis of the sample. Several examples of how SPR systems and matrix-assisted laser desorption/ionization time-of-flight MS (MALDI-TOF MS) have been combined to capture (Biacore) and identify (MS) proteins from complex matrices may be

SPR Instrumentation

73

found in the literature [35,36]. Although it has been possible to perform SPR-MS on earlier systems from Biacore, the process has been refined and optimized for Biacore T100, with the entire recovery process defined in the ‘‘Method Builder’’ software in a predefined template. Characteristics of buffers such as ion content, salt concentration and pH are all variables that can radically affect the profile of an interaction. ‘‘Buffer scouting’’ is a novel feature of the Biacore T100 and is intended to help the user rapidly find the optimal buffer conditions to suit a specific interaction; up to four different buffers can be tested at one time. In addition, this flexibility allows the user to study micro-environmental effects on binding properties in mechanistic and stability studies and to define the kinetic properties of protein samples in varied biochemical or biophysical environments. This information may be crucial when selecting candidates intended for use in the complex and variable environment of clinical treatment.

3.5.3 Biacore A100 The Biacore A100 is a protein interaction analysis system delivering high-quality kinetic, affinity, concentration and specificity data. It is designed primarily for applications requiring high sample throughput using small panels of proteins, but can also be used for detailed characterization studies, allowing integration into multiple phases of project pipelines. The open, flexible format permits multiplexed assays that create new possibilities for faster data acquisition. The system offers enhanced productivity in key areas such as antibody selection, biotherapeutic and low molecular weight drug development, immunogenicity studies and proteomics. An optional package is available for work in regulated environments. Hydrodynamic addressing (HA; see Figure 3.35 and Section 3.3.1) is a process by which multiple interactants may be immobilized on detection spots in a single flow cell, allowing simultaneous analysis of interactions. As there is no lag time between interactions, highly accurate reference subtraction allows the measurement of very rapid kinetics. Further, by immobilizing several interactants in one flow cell, comparative binding properties may be directly examined under optimal experimental conditions. By adjusting the relative flow at the two inlets (one for the ligand which should be immobilized and the other for buffer), liquid can be directed to different addressable detection spots. The flow cell design allows rapid and efficient switching of flow between buffer and sample solutions and the transverse arrangement of the detection spots ensures that access of sample to all spots is simultaneous. Although the detection spots are addressed separately during immobilization, the injected sample flows over all spots simultaneously (see Figure 3.35). The flow cell configuration of the Biacore A100 enables up to 3800 interactions to be monitored in a 24 hour run, with a selectable configuration either for maximum number of samples or for maximum information per sample (see Figure 3.36). The capacity for parallel processing and the high quality of kinetic

74

Figure 3.35

Chapter 3

Schematic view of the flow system in the Biacore A100. Planar view; the four injection ports (I1–I4) allow unique interaction conditions and parallel analysis at five measurement spots in each flow cell. Top: side view of flow cell with sensor surface and measurement spots on the bottom and flow cell cartridge (gray) on top.

data make the Biacore A100 an attractive option for applications such as interaction proteomics, drug discovery and biotherapeutics development, where it is vital to be able to handle many samples and have confidence in the data. In biotherapeutics development, for example, the development of monoclonal antibodies is a complex, time-consuming and thus expensive business involving the generation, maintenance and screening of thousands of hybridoma clones. Confident early identification of hybridomas to produce the best candidate antibodies is a critical step in successful, cost-efficient development. In drug discovery programs, the Biacore A100 can provide information-rich data, allowing the identification of high-quality lead compounds that are crucial for progress. Direct binding analysis, offering comprehensive characterization of critical selectivity and kinetic properties, provides data to guide

SPR Instrumentation

Figure 3.36

75

Configuration of detection spots in the Biacore A100 system optimized for (top) sample throughput with parallel analysis of five components in four samples and (bottom) number of analyses per sample with a maximum of 20 components per sample.

key decisions. The identification of highly selective compounds against complex therapeutic targets may benefit strongly from a comprehensive panel approach in order to eliminate potential target-dependent artifacts that may result in false-positive or false-negative leads. Furthermore, the use of multiple control targets to identify non-specific protein binding, binding to recombinant protein tags and possible expression-system artifacts, should be generally applicable to almost any drug discovery program.

3.5.4 FLEXChip The development of SPR-based biosensors has been geared to delivering data of the highest possible quality on a limited number of interactions. There are several reasons, however, to support the design and production of systems with greatly increased capacities for sample throughput. Perhaps the most pressing call for a commercially available protein interaction array is from the proteomics community; with a bewildering amount of novel proteins at hand since the completion of the human genome project, any technology that helps explain their functions is welcome. Additionally, many applications such as antibody screening, hit selection in drug development programs, peptide epitope mapping

76

Chapter 3

and even on-line quality control/safety testing during food production are all activities that could benefit from the increased sample throughput on an array. The development of protein arrays is more complex than the DNA counterpart. Proteins are more difficult to handle, as post-translational modifications, vital for functionality, are seldom preserved in the course of the amplification steps required to obtain reagents in sufficient quantity and purity. Consideration must also be given to variables such as immobilization conditions (see also Chapter 6), orientation and the possibility that the immobilization process may impede or conceal the very binding site of interest. Further complications include the desirability of immobilizing proteins efficiently and at precise concentrations (important for obtaining meaningful association rates) across the entire array. Proteins are also less discriminatory in their choice of binding partners than DNA and so non-specific adsorption to both the sensor surface and to other proteins in a complex mixture, such as clinical samples or hybridoma supernatants, may complicate the interpretation of results from a multiplexed array. The FLEXChip (see Figure 3.24) is an SPR-based array for the simultaneous profiling of up to 400 protein interactions. The flow cell in FLEXChip is a single, broad channel through which a single sample is injected, interacting simultaneously with all the spots on the array. After a panel of interacting partners have been externally spotted on the sensor surface according to experimental requirements, a gasketed window with an inlet and an outlet valve is positioned and hermetically sealed over the array to form the flow cell, which is then inserted into the FLEXChip apparatus. FLEXChip uses a variant of SPR known as grating-coupled SPR, in which incident light from above strikes the entire array, instead of the light from the prism as in the true Kretschmann operated SPR imaging systems. It is important to remember that the information sought from functional protein arrays goes beyond mere detection and that, consequently, technologies delivering this information should not be judged solely in terms of throughput. A technology that profiles entire interactions and delivers data on the association rate constant (ka), dissociation rate constant (kd) and affinity constant (KD) delivers information about the function of a restricted and usually selected population of proteins in a cell or a group of antibodies or peptides. Protein interaction arrays are therefore, of necessity, smaller than DNA arrays. Whereas the proteome itself is finite, the range of protein functions is practically boundless. Cancer specialists, for example, may be interested in identifying binding partners of proteins uniquely expressed in cancer but will be able to answer a more explicit question if they are able to make informed judgments on the consequences of these interaction patterns and ask how they affect disease progression and options for therapy. Functional protein arrays may well prove to be the link between the vast repository of data that has emerged from proteomics initiatives and the world of clinical and biological research. An excellent example illustrating the power of the FLEXChip has been demonstrated, in which its scope for parallel interaction analysis was fully

SPR Instrumentation

77

exploited to integrate secondary screening with a fully automated phage display process [12]. Phage display can generate a tremendous number of potential protein therapeutics, such as antibody Fab fragments. Advances in the automation of the selection (panning a phage display library against an immobilized target molecule) and primary screening (e.g. ELISA) processes are largely responsible for the discovery of a great number of unique antibodies. However, it then becomes a challenge to identify the most promising leads from a large number of candidates. Typically, candidates from a phage display process are selected after several rounds of selection and amplification followed by a primary screen, but the adoption of such a strategy, which aims to reduce demands on the secondary screen, decreases the hit diversity and increases the risk of missing potential candidates. An alternative strategy is to limit the number of rounds of selection so that a more diverse collection of binders makes it through to the secondary screen. Although having such a diverse collection of binders in terms of kinetic profiles increases the likelihood of the discovery of candidates with the desired functional activities, it places unique demands on the performance of any secondary screening process. Wassaf et al. [12] incorporated FLEXChip as a secondary screen into a highly automated selection and screening process in a scheme to find potent human antibody inhibitors to a human serine protease (tissue kallikrein 1, hK1), involved in inflammation. The process was based on a limited number of selection and amplification rounds followed by interaction analysis on hundreds of candidates. This case study showed how an automated procedure, including the secondary screen, was developed to select, purify and concentrate Fabs prior to rapid characterization using FLEXChip. The ability to monitor simultaneously hundreds of interactions in a secondary screen may signal a significant improvement in the speed at which potential therapeutic candidates can be identified. As candidates are fed through the development pipeline, the need for protein interaction analysis continues, for example, for further selection during optimization, for monitoring immunogenic responses and even for batch release testing. The need to fulfill regulatory requirements such as GxP compliance also increases further downstream. These needs can be met with systems such as the Biacore T100 and Biacore A100. FLEXChip is the filter between the vast numbers of candidates generated in a phage display process and ensuring that only the best candidates are advanced along the drug discovery pipeline.

3.6 Conclusion Instruments contain at least three integrated components: (1) SPR instrumental optics, (2) a liquid handling system and (3) the sensor chip. The quality of each of these components reflects the overall performance of the SPR instrument. In this chapter, a short description has been given of SPR products from 25 companies. Gold is still the gold standard for generating the SPR phenomenon in almost all commercial instruments available on the market. The commercial availability of sensor surfaces essentially contribute to accurate and reliable results (see Chapter 6).

78

Chapter 3

Although in the past only Biacore has dominated the market (490%), new players can be identified. However, as explained in this chapter, not all instruments from these manufacturers will generate reliable quantitative kinetic data for kinetic evaluation of rate and affinity constants, but should be regarded as instruments which are able to show qualitatively binding of the analyte with the immobilized ligand. The degree of automation which also contributes to accurate and reliable kinetic data retrieval differs from totally manual to highly automated and can greatly enhance the performance of the SPR instrument. Typically, SPR dip shifts are monitored in the reflectivity mode or angle shift mode. Instruments are categorized in six sections according to their optical configuration. The liquid handling systems mainly comprise flow cells or cuvettes. Special attention has been paid to three Biacore instruments, T100, A100 and FLEXChip, and their application to protein studies, in Section 3.5. As indicated in this timely chapter, the market is now more open than ever before and competition between companies is taking place on several aspects of the SPR systems. The customers of instruments will profit further from this competition, offering more flexibility, innovation and cost effectiveness.

3.7 Questions 1. An instrument for detection of biomolecular interactions can be considered as a total analysis system that consists of three main technology parts. Which three parts are essential for such a biosensor system? 2. A divergent beam SPR instrument shows ‘‘walking of the SPR dip over the sensor surface’’. Explain this effect. 3. In convergent beam instruments, the exact SPR angle is detected without moving parts. A camera is needed to measure, e.g., 20 sensor spots in line. How can one make an image of the sensor surface? 4. In SPR imaging instruments, a parallel beam of light will bring a homogeneous surface in full resonance. How can we make an image of the surface and follow the SPR angle of each region of interest during the biomolecular interaction process in order to calculate the SPR angle position of each spot? How many sensor patches can then be obtained from a sensor surface of, e.g., 5 mm square? 5. Consider an angle scanning instrument. Draw the reflectivity curve as a function of time if the scanner first moves in forward direction and passes the SPR dip and then shifts backwards with the same speed as in forward direction. What happens to this curve after the SPR dip has shifted?

References 1. B. Liedberg, C. Nylander and I. Lundstrom, Biosens. Bioelectron., 1995, 10, i–ix. 2. R. Karlsson, J. Mol. Recognit., 2004, 17, 151–161. 3. E. Stenberg, B. Persson, H. Roos and C. Urbaniczky, J. Colloid Interface Sci., 1991, 143, 513–526.

SPR Instrumentation

79

4. M. Malmqvist, Nature, 1993, 361, 186–187. 5. B. Johnsson, S. Lofas and G. Lindqvist, Anal. Biochem., 1991, 198, 268–277. 6. T. Wink, S.J. van Zuilen, A. Bult and W.P. van Bennekom, Anal. Chem., 1998, 70, 827–832. 7. B.P. Nelson, A.G. Frutos, J.M. Brockman and R.M. Corn, Anal. Chem., 1999, 71, 3928–3934. 8. B.P. Nelson, T.E. Grimsrud, M.R. Liles, R.M. Goodman and R.M. Corn, Anal. Chem., 2001, 73, 1–7. 9. L.-M. Zhang and D. Uttamchandani, Electron. Lett., 1988, 24, 1469–1470. 10. J. Homola, Surface Plasmon Resonance Based Sensors. Springer Series on Chemical Sensors and Biosensors, Vol. 4, Series Editor O.S. Wolfbeis, Springer, Berlin, 2006. 11. E. Kretschmann, Z. Phys., 1971, 241, 313–324. 12. D. Wassaf, G. Kuang, K. Kopacz, Q.L. Wu, Q. Nguyen, M. Toews, J. Cosic, J. Jacques, S. Wiltshire, J. Lambert, C.C. Pazmany, S. Hogan, R.C. Ladner, A.E. Nixon and D.J. Sexton, Anal. Biochem., 2006, 351, 241–253. 13. M. Pe´rez-Moralesa, J.M. Pedrosab, E. Mun˜oza, M.T. Martı´ n-Romeroa, D. Mo¨biusc and L. Camachoa, Thin Solid Films, 2005, 488(1-2), 247–253. 14. J. Ctyrokya, J. Homola and M. Skalskya, Opt. Quantum Electron., 1997, 29, 301–311. 15. C.R. Yonzon, E. Jeoung, S. Zou, G.M. Mrksich and R.P. Van Duyne, J. Am. Chem. Soc., 2004, 126, 12669–12676. 16. J. Homola, S.S. Yee and G. Gauglitz, Sens. Actuators B, 1999, 54, 3–15. 17. A.M.C. Lokate, J.B. Beusink, G.A.J. Besselink, G.J.M. Pruijn and R.B.M. Schasfoort, J. Am. Chem. Soc., 2007; DOI: 101021/ja075103x. 18. P.A. Lowe, T.J.H.A. Clark, R.J. Davies, P.R. Edwards, T. Kinning and D. Yeung, J. Mol. Recogn., 1998, 11, 194–199. 19. B. Rothenha¨usler and W. Knoll, Nature, 1988, 332, 615–617. 20. C.E.H. Berger, T.A.M. Beumer, R.P.H. Kooyman and J. Greve, Anal. Chem., 1998, 70, 703–706. 21. S. Dickopf, M. Frank, H.D. Junker, S. Maier, G. Metz, H. Ottleben, H. Rau, N. Schellhaas, K. Schmidt, R. Sekul, C. Vanier, D. Vetter, J. Czech, M. Lorenz, H. Matter, M. Schudok, H. Schreuder, D.W. Will and H.P. Nestler, Anal. Biochem., 2004, 335, 50–57. 22. J.M. Brockman, B.P. Nelson and R.M. Corn, Annu. Rev. Phys. Chem., 2000, 51, 41–63. 23. P. Schuck, Annu. Rev. Biophys. Biomol. Struct., 1997, 26, 541–566. 24. T.A. Morton, D.G. Myszka and I.M. Chaiken, Anal. Biochem, 1995, 227, 176–185. 25. H.A. Stone, A.D. Stroock and A. Ajdari, Annu. Rev. Fluid Mech., 2004, 36, 381–411. 26. D.E. Davey, D.E. Mulcahy and G.R. O’Connell, Electroanalysis, 2005, 5, 581–588. 27. C.L. Baird and D.G. Myszka, J. Mol. Recognit., 2001, 14, 261–268. 28. Cush, et al., Biosens. Bioelectron., 1993, 8, 347–364. 29. H.J. Watts, D. Yeung and H. Parkes, Anal. Chem., 1995, 67, 4283–4289.

80

Chapter 3

30. L.A. Chtcheglova, M. Vogel, H.J. Gruber, G. Dietler and A. Haeberli, Biopolymers, 2006, 83, 69–82. 31. M.B. Glaubert, J. Fluid. Mech., 1956, 1, 625–643. 32. Internet references, March 2007: www.biacore.com, www.biosensingusa.com, www.nomadics.com, www.dkktoa.net/, www.reichertai.com/, www.plasmonic. de, www.moritex.com, www.biosuplar.de, www.sensia.es, www.k-mac.co.kr/, www.nanofilm.de/, www.bio-rad.com, www.ecochemie. nl, www.thermo.com, www.neosensors.com, www.gwctechnologies.com, www.genoptics-spr.com, www.mavenbiotech.com, www.lumera.com/, www.graffinity.com, www.ibisspr.nl. 33. A. Marquart, Surface Plasmon Resonance, SPR pages http://www. sprpages.nl. 34. H. Jin Lee, T.T. Goodrich and R.M. Corn, Anal. Chem., 2001, 73, 5525–5531. 35. E.A. Smith and R.M. Corn, Appl. Spectroscopy, 2003, 57, 320A–332A. 36. D. Nedelkov and R.W. Nelson, Am. J. Kidney Dis., 2001, 38, 481–487. 37. A. Zhukov, M. Schurenberg, O¨. Jansson, D. Areskoug and J. Buijs, J. Biomol. Tech., 2004, 15, 112–119. 38. R.L. Rich and D.G. Myszka, J. Mol. Recognit., 2006, 19(6), 478–534. 39. R.L. Rich and D.G. Myszka, Anal. Biochem., 2007, 361, 1–6.

CHAPTER 4

Kinetic Models Describing Biomolecular Interactions at Surfaces DAMIEN HALL1,a,b,c a

University Chemical Laboratory, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK; b Institute for Protein Research, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, JAPAN; c Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, B-39, 5249 Nagatsuta, Midori-ku, Yokohama 226-8501, JAPAN

4.1 Introduction To say that the adsorption of molecules from solution to a surface is an important phenomenon in biochemistry is to make a major understatement. Indeed, so much of the fundamental chemistry of life occurs at interfacial regions [1–12] (Table 4.1) that the biochemists or biophysicists who take their subject seriously are required to have both an adequate understanding of how these processes occur and how they might be measured experimentally. The major thrust of this volume is concerned with interrogating the adsorption of biologically important molecules to surfaces by the use of optical biosensor technology. In this chapter, we discuss adsorption events from the perspective of monitoring a measurement signal which provides information, in real time, on the extent of solute adsorption to a surface. As discussion of time-dependent adsorption phenomena necessitates the use of kinetic models, we spend our time reviewing kinetic models that describe a wide range of adsorption behavior. We begin, however, by introducing the terminology of adsorption so that we might have the necessary language with which to conduct a discussion of the subject. 1

Present address: Hall Laboratory. Institute for Basic Medical Sciences. Tsukuba University. 1-1-1 Tennoudai, Tsukuba, Ibaraki, 305-8575, JAPAN.

81

82

Table 4.1

Chapter 4

Role of Adsorption in Fundamental Life Processes.

Biological Function

Role of Adsorption

Intra/Extra-Cellular Trafficking

First step in transfer of individual molecules and molecular complexes between various compartmentalized structures of the cell [1]. Also involved in transport to and from the cell e.g. pinocytosis and exocytosis [2]. Cell to cell communication is important both for single cell [3] and multi-cellular organisms [4]. Communication during such processes as tissue development is mediated by adsorption of specific signalling molecules to receptors located at the cell membrane surface [5]. Initial step in sensory pathways of taste and smell involve adsorption of specific ligands to specific taste or smell receptors located on the cell surface [6,7]. Transmission of nerve impulses across synapses involves release of neurotransmitters from one side of the synaptic cleft and their subsequent adsorption to receptors on the opposing face of the synaptic cleft [8]. Transcription of DNA and translation of RNA involve adsorption of specific protein complexes to the one dimensional adsorptive surface of the nucleic acid polymer chain [9]. The first step in the defence against invading foreign agents involves adsorption of antibodies to the invading agent. The foreign agent constitutes an adsorptive surface composed of matrix sites termed epitopes that are specifically recognised by various immunoglobulin molecules [10]. Material introduced into the body e.g. artificial joints, contact lenses, slow release drug reservoir etc. should have minimal potential to act as an adsorptive surface. Reduction of nonspecific adsorption to various implant devices is an active area of research [11,12].

Cell Signalling

Olfactory/Taste Senses Neuronal Impulse Transmission DNA Transcription RNA Translation Immune Response

Artificial Implant Technology

4.1.1 Terminology of Adsorption In any solution-based adsorption experiment, one wishes to monitor the adsorption of a certain molecule in solution, termed the solute (or alternatively analyte), to the surface of a solid phase. The solid-phase adsorptive surface is composed of a number of sites that exert an attractive force for the solute. Once adsorbed to the surface the solute is termed the adsorbate. However, the term solute and adsorbate refer to any molecule that will bind to the surface specifically or non-specifically. Adsorption experiments are generally preceded by the preparation of a suitable adsorptive surface that is composed of a number of binding sites (alternatively known as matrix sites or immobilized ligand binding sites). The type of adsorptive surface that is prepared is necessarily dependent upon the system that is being studied however we might make a distinction as to the general type of surface in relation to a specific solute as constituting either a distinct array of binding sites or an effective continuum of binding sites (Figure 4.1). As a general rule, we may consider the adsorbing

83

Kinetic Models Describing Biomolecular Interactions at Surfaces

(A)

Type 1

Type 2

Radius = r1

Radius = r2

(B)

z

x Figure 4.1

y

(A) The radius r of idealized spherical solute molecules (shown in green) may be small (r1) or large (r2) in relation to the distance between the adsorption sites on the adsorptive surface. (B) Diagram indicating a twodimensional surface (surface plane). Colors indicate change in potential energy associated with the normal position of the center of the solute molecule from a distance r to r+Dz away from the surface plane. Regions of lowest potential (blue) indicate position of adsorption sites. For adsorption of type 1 solute this two-dimensional adsorptive surface would constitute a distinct array. For adsorption of type 2 solute the surface would constitute a near continuum array. (C) Diagram indicating a threedimensional adsorptive phase positioned at the interfacial boundary of the solid and liquid phases. Regions of low potential energy for solute (adsorption sites) are indicated by blue spheres. The supporting phase (often a polymer gel) is shown by red rods. As with the surface plane shown in (B), this surface phase may exist as either a distinct array or a near continuum array depending upon the dimensions of the adsorbing solute.

84

Chapter 4

(C)

z

x

Figure 4.1

y

Continued

surface as a distinct array of sites if an adsorbate molecule, once bound, does not interfere with the subsequent adsorption of solute to any neighboring sites. Once the adsorptive surface has been constructed, an adsorption experiment may be conducted by exposing this surface to a solution containing solute. Once introduced to the surface the solute will begin to adsorb2 – a process which will continue until an equilibrium position is reached between the concentration of solute and the concentration of adsorbate. At this point, the process of adsorbate desorption may be studied by replacing the liquid phase with a solution not containing solute and recording the dissociation of adsorbate until an equilibrium ratio between solute and adsorbate is achieved. These basic adsorption and desorption experiments can be performed at different concentrations of added solute/initial adsorbate to examine the effect of concentration on the time-dependent (Figure 4.2a) or time-independent (Figure 4.2b) adsorption/desorption processes. Evaluation of the concentration dependence of the kinetic and equilibrium situations allows for the evaluation of the rate and equilibrium constants that define the behavior of the interaction under the particular conditions chosen for the study. These experiments may be repeated under different conditions in which a single environmental variable (e.g. temperature, pressure or composition of the buffer) is systematically altered to examine how the rate and equilibrium constants that define the adsorption event respond to the imposed change. Until relatively recently with the experimental interrogation of solute adsorption from solution was made indirectly, using measurement strategies 2

The particular forces that drive adsorption are outside the scope of this discussion.

Kinetic Models Describing Biomolecular Interactions at Surfaces

Figure 4.2

85

(A) Generalized profile of a time-dependent adsorption experiment: normalized concentration of adsorbate vs. time t. The first arrow shows the point of introduction of the solute to the adsorptive surface. In the association phase solute will continue to adsorb to the surface until an equilibrium is attained. The second arrow shows the point at which the liquid phase is replaced by buffer not containing solute. In this dissociation phase, adsorbate is allowed to dissociate from the surface until a new equilibrium position is reached. The five lines represent experiments carried out at five different concentrations. (B) General profile of a time independent adsorption experiment: the time independent values attained at the finish of each of the kinetic adsorption experiments are plotted against the equilibrium concentration of solute in the liquid phase.

86

Chapter 4

based on difference measurements of the amount of solute depleted from the liquid phase [13]. Such indirect measurement strategies placed a limit on both the type of adsorption processes that could be studied and the temporal resolution with which that process could be resolved. The recent development of new techniques [14,15] allows for direct measurement of the amount of adsorbed solute and has afforded a time resolution that is almost contemporaneous with the adsorption event. Of these new techniques, those based on evanescent wave optical biosensor technology [16–18] have become the most popular, mainly due to the successful commercialization of the technology.

4.1.2 Optical Quantification of Adsorption at an Interface Evanescent wave optical biosensor technology is reliant upon the physics of the total internal reflection (TIR) of light. Light shone at an interface3 below a certain critical angle will enter the second medium and be refracted by it to an extent governed by the refractive indices of the two mediums [19]. At incident angles greater than the critical angle, the light will not enter the second optical medium but will instead be totally internally reflected back into the first medium. The electric and magnetic fields associated with the reflected ray of light, however, do not end abruptly at the point of reflection but will penetrate somewhat into the second medium. Interaction of these decaying electric and magnetic fields (termed evanescent fields) with the second medium beyond the interfacial plane will subtly change the properties of the reflected beam. The degree of non-absorptive interaction of the electrical component of the evanescent light with the matter at the interfacial region will be determined by the electrical polarizability of the material at the interface. Thus, optical biosensor technology can be taken to reflect changes in refractive index at the interfacial layer.4 The field strength associated with the evanescent light at the interface decays exponentially with distance normal to the surface, z, making the observed signal, S, proportional to refractive index changes very close to the surface [eq. (4.1)]. S/

ZN DnðzÞ expð
z=sÞdz

ð4:1Þ

0

The parameter s is the field strength decay constant and has units of distance. For planar surfaces it is typically of the order of 0.3–0.5 times the wavelength of the light used in the TIR experiment [19]. For a wide range of substances5 the change in refractive index in response to varying the weight concentration of

3

An interface existing between two optically transparent mediums. For optically transparent materials, the macroscopic quantity related to the electrical polarizability is the refractive index. 5 The approximate dn/dc values for proteins and nucleic acids are 0.18 and 0.16 ml g
1, respectively. The values for different types of carbohydrate range from 0.10 to 0.18 ml g
1. 4

Kinetic Models Describing Biomolecular Interactions at Surfaces

87

component i, ci, can be approached as a linear function as the derivative dn/dci is approximately constant [eq. (4.2)].
 dn nðci Þ ¼ nðci ¼ 0Þ þ ð4:2Þ ci dci Substitution of eq. (4.2) into eq. (4.1) allows us to relate the change in signal, DS, for situations in which initially no component i is present (ci ¼ 0), with changes in the weight concentration of component i at the interfacial layer [eq. (4.3)]. dn DS / dci

ZN

dci expð
z=sÞdz dz

ð4:3Þ

0

On the basis of eq. (4.3), we can appreciate that the measured signal is, strictly, not a linear descriptor of the concentration of adsorbate but for many cases it will very nearly be so. Figure 4.3 describes the apparent change in signal that would be registered for a constant mass of adsorbate in a number of different modes of adsorption on the surface of an optical biosensor. On the basis of eq. (4.3), the relative signal (normalized to the type 1 case) for each adsorption mode would be type 1, 1; type 2, 1.01; type 3, 0.56; and type 4, 0.23. A variety of ingenious means (of which SPR [20] is perhaps the best known) have been developed to maximize the changes in the reflected beam of light so as to make it a viable means for experimental measurement. Although the physics of each different experimental detection method differ somewhat, all optical biosensor technology share basic characteristics as outlined in eqs. (4.1)
(4.3). More is said about the physics and instrumentation particular to SPR in Chapters 2 and 3. For the remainder of the current chapter we will discuss kinetic models that are relevant to different aspects of the process of adsorption as studied by optical biosensor technology. Although we will not specifically transform the concentration units of the adsorption profile into an equivalent signal as defined by eq. (4.3), we ask the reader to keep in mind the likely consequences of the transformation for some of the more complicated adsorption geometries.

4.2 Defining Factors of the Adsorption Event For adsorption to occur two events must happen in sequence: first the solute must be transported to the surface, and second the solute must successfully interact with the surface to form the adsorbate. We will discuss each of these events in greater detail in subsequent sections; however, a basic understanding of the distinction between the two regimes can be gained by casting the adsorption event in terms of a simple transport/reaction process [21] [eq. (4.4)]. ka

k01

kb

k2

Ci Ð fCi g Ð ðCi Þads

ð4:4Þ

88

Chapter 4 (A) 1

2

3

4

(B)

exp(-z/σ)

1

0.5

0 0

50

100

150

200

250

300

Z (nm)

Figure 4.3

(A) Four different modes of solute adsorption with the adsorptive surface at which the light is totally internally reflected shown on the vertical left axis. Type 1 adsorption: the solute adsorbs directly on the surface plane and maintains its shape. Type 2: the solute adsorbs and undergoes a postadsorption isomerization to maximize its interaction with the surface. Type 3: the solute adsorbs to matrix sites on a supporting polymer molecule in an evenly dispersed manner. Type 4: the solute adsorbs on a supporting polymer molecule preferentially on the bulk solution side. On the basis of eq. (4.3), the relative signal (normalized to the type 1 case) for each adsorption mode would be type 1, 1; type 2, 1.01; type 3, 0.56; and type 4, 0.23. (B) The decrease in evanescent field strength with distance normal to the surface for a decay constant s ¼ 200 nm (approximate case for light used in TIR experiment of wavelength 600 nm).

Kinetic Models Describing Biomolecular Interactions at Surfaces

89

In this formulation, Ci, {Ci} and (Ci)ads represent the concentration of solute of type i, the concentration of solute of type i spatially close to the surface (termed intimate solute) and the concentration of adsorbed solute of type i (the adsorbate), respectively. Equation (4.4) describes the process whereby solute approaches and leaves the surface at a rate defined by phenomenological transport coefficients ka and kb. Once in physical proximity to the surface, the intimate solute may react to form adsorbate according to an association rate function k01 and a reverse rate constant k2, where both k01 and k2 have units of s
1. In the simplest instance the association rate function may be further decomposed [eq. (4.5)] into an intrinsic second-order rate constant k1, the initial total concentration of matrix sites,6 (Cx)tot and some unitless function describing the surface site occupation, f(f), where f ¼ n(Ci)ads/(Cx)tot (n is the average number of sites covered by the adsorbed solute). More will be said about the surface function, but for now we may treat f(f) as describing the fraction of total matrix sites available for participation in further adsorption of solute. k01 ¼ k1 ðCx Þtot f ðfÞ

ð4:5Þ

In the simple scheme set forth in eq. (4.4), the rate of formation of the intimate solute is given by eq. (4.6). dfCi g ¼ ka Ci
kb fCi g
k01 fCi g þ k2 ðCi Þads dt

ð4:6Þ

The so-called transport-limited case represents the situation in which the rate of physical encounter between solute and surface dictates the rate of formation of the adsorbate species. This situation occurs for non-equilibrium cases when [k1 0 {Ci}
k2(Ci)ads] c [kaCi
kb{Ci}], making the rate of formation of the intimate solute effectively zero (d{Ci}/dt - 0). It results in the effective association rate being given by eq. (4.7a). In the limit of zero adsorbate concentration, (Ci)ads E 0, the effective forward rate parameter fi, can be approximated by a ratio of reaction and transfer rate constants with the further approximation for fi allowable if the transport coefficient is much smaller than the reaction rate (kb { k01) [eq. (4.7b)]. In the alternative limit of zero solute concentration, Ci E 0, the effective dissociation rate parameter bi is given by eq. (4.7b), again with the additional approximation valid if kb { k01 (here KR 0 ¼ k01/k2). For a given degree of surface occupation an effective adsorption partition constant

6

To speak of a total concentration of adsorption sites one must assign a reaction volume to the surface and treat the adsorbed layer as a three-dimensional phase. An alternative treatment is to consider the surface layer as a two-dimensional plane with adsorbate and binding site concentrations given in terms of mol m
2. This results in the usual set of units. This results in an unusual set of units for the resulting equilibrium (m3 mol
1) and association rate constants (m3 mol
1 s
1). We prefer the former treatment as it is conceptually similar to solution chemical kinetics. Despite the differences, the two treatments are formally equivalent.

90

Chapter 4

(unit-less) can then be determined by the ratio of effective forward and reverse rate parameters [(Ki)eff ¼ fi/bi] (Eq. 4.7b).   dðCi Þads ka Ci þ k2 ðCi Þads ¼ k01 ð4:7aÞ
k2 ðCi Þads dt kb þ k01 ‘ fi E

k01 ka kb k2 kb k01 ka Ek ; b E E ;ðK Þ ¼ a i i eff kb þ k01 kb þ k01 KR0 k2 kb

ð4:7bÞ

In the antithetical, reaction-limited kinetic regime [k01{Ci}
k2(Ci)ads] { [kaCi
kb{Ci}], the concentration of intimate solute is little perturbed by the reaction component of eq. (4.6), allowing a quasi-equilibrium approximation such that {Ci} E (ka/kb)Ci ¼ KTCi (where KT represents an equilibrium partition constant for formation of the intimate solute) leaving the reaction rate to be written as eq. (4.8a). In this case, the effective association rate parameter, fi, in the limit of zero adsorbate [(Ci)ads E 0] and the effective dissociation rate constant, bi, in the limit of zero solute (Ci E 0) and the effective adsorption partition constant, (Ki)eff ¼ fi/bi, can be described by eq. (4.8b). dðCi Þads ¼ k01 KT Ci
k2 ðCi Þads dt

ð4:8aÞ

k01 ka k2 kb

ð4:8bÞ

‘ fi Ek01 KT ; bi Ek2 ; ðKi Þeff ¼

Although we have presented the two extreme antithetical kinetics regimes the system under study may exhibit ‘‘mixed’’ kinetic behavior, in which case no simplifying relationship can be deduced and the numerical solution of two coupled differential equations [eqs. (4.6) and (4.9)] must be performed with the apparent phenomenological rate constants varying between the two limits set out in eqns. (4.7b) and (4.8b) depending on the relative concentrations of solute, sites for adsorption and adsorbate. dðCi Þads ¼ k01 fCi g
k2 ðCi Þads ‘ ka 4fi 4k01 KT dt

and

kb 4bi 4k2 KR0

ð4:9Þ

4.2.1 Mass Transfer In optical biosensors, the solute is introduced to the adsorptive surface in either a flow-through or a cuvette-type system. In the flow-through biosensor (Figure 4.4), the adsorptive surface constitutes the base of a microfluidic channel through which solution flows at some average velocity v*. As the solute is taken up by adsorption it is replaced anew by solute flowing in from further down the channel. In the cuvette-type biosensor (Figure 4.5), the

Kinetic Models Describing Biomolecular Interactions at Surfaces

91

(A)

Reflected Light

(B)

z

x Figure 4.4

(A) Diagram depicting the general characteristics of a flow-through type biosensor. Solute is introduced as a plug into a microfluidic channel at controlled flow rate. The adsorptive surface at which light is totally internally reflected constitutes one side of the microfluidic channel. As solute is adsorbed to the surface, it is replaced by the solute flowing through behind allowing the liquid phase concentration for certain experimental regimes to be considered a constant. (B) Idealized representation of the velocity field associated with flow in an optical biosensor based on the parabolic flow assumption [eq. (4.12)].

adsorptive surface constitutes the base of a microcuvette into which solute is added. To effect efficient mass transfer, the solution is subjected to shear forces generated by aspiration cycles from a free wall jet device (see Chapter 3) or by a stirring rod suspended near the top of the solution. The most general approach for describing the mass transfer process first involves the spatial discretization of the solution volume comprising the biosensor device, followed by numerical solution of a continuity equation (for the cases of non-turbulent flow [22] describing the diffusion, convection and

92

Chapter 4

(A)

Reflected Light

(B)

z

x Figure 4.5

(A) Diagram depicting the general characteristics of a cuvette-type biosensor. Solute introduced by injection into an open cuvette is subjected to shear forces by an oscillating stirrer. The adsorptive surface at which light is totally internally reflected is positioned at the base of the cuvette. As the solute is adsorbed to the surface it is depleted from solution meaning that except for cases of insignificant depletion the solution phase concentration of solute must be calculated by difference between the total and adsorbed concentrations. (B) An idealized representation of the velocity field associated with stirring in a cuvette type optical biosensor based on the stagnant point flow assumption [eq. (4.13)].

Kinetic Models Describing Biomolecular Interactions at Surfaces

93

adsorption of the solute component) [eq. (4.10)]. Here Di and vi denote the solute’s diffusion coefficient and linear velocity.7,8,9 mass transport reaction
 dCi ¼rðDi rCi Þ
rð½Ci vi Þ
k01 Ci þ k2 ðCi Þads dt dðCi Þads ¼ k01 Ci
k2 ðCi Þads dt

ð4:10aÞ

ð4:10bÞ

For efficient mass transport, we require that, for points in close spatial proximity to the adsorptive surface, the absolute values of the mass transport term in eq. (4.10a) be either greater than the forward reaction term, for association experiments, or greater than, the reverse reaction term, for experiments examining dissociation of adsorbate.10 From inspection of eq. (4.10a), we note that we may influence this ratio by changing the transport parameters Di and vi or alternatively modifying the reaction parameters Ci, (Cx)tot and f(f) (the last two parameters housed in the association rate function k01). For a particular system of interest, one can usually vary Ci and (Cx)tot by careful experimental design leaving the transport parameters Di and vi as the important variables to be considered. With regard to the diffusion of solute, we note that Di for a spherical solute of radius ri existing in an aqueous solvent of viscosity Z at temperature T can be written [23] as eq. (4.11): Di ¼

kT 6pZri

rffiffiffiffiffiffiffiffiffiffiffiffi vi M i 3 3
ri ¼ 4pNA

ð4:11aÞ

ð4:11bÞ

where k is Boltzmann’s constant, NA is Avogadro’s number, v
i is the partial specific volume of the solute and Mi is the solute molecular weight. In the context of achieving efficient mass transfer in biosensor adsorption experiments, we are interested how Di depends on the environmental variables of solution temperature and composition [24–27] (Table 4.2). With respect to temperature, we note that the viscosity of water (and dilute aqueous solutions) between temperatures of 0 and 501C can be described very well by an empirical exponential dependence (where Z01C (LIQUID) ¼ 1.8  10
3 kg s
1 m
1), allowing an estimation11 of the solute diffusion coefficient that shows a significant increase with temperature over this range. 7

Eq. (4.10) must be appropriately modified for spatial boundary conditions, i.e. to account for the motion /reaction of solute at or near physical boundaries. 8 The operator r when applied to a function f, rf, implies the sum of spatial derivatives, e.g. @f/ @x + @f/@y + @f/@z. 9 The reaction component is presented as a simple mechanistically concerted process for simplicity. For sequential based reaction processes this must be modified. 10 That is, (dC/dt)reaction/(dC/dt)transport - 0. 11 Valid for solutes that will not significantly change their average dimensions upon experiencing a change in temperature, e.g. as might a protein upon temperature-induced protein unfolding.

94

Table 4.2

Chapter 4

Dependence of solute diffusion coefficient on solution parameters capable of variation in optical biosensor experimentation.

Particular Dependence Temperature (for Water) [24] Weight Concentration of viscogenic agenti [25] (at 201C) Concentration and dimensions of the supporting gel phase [26,27] i

Relevant equation describing diffusion coefficient kT 6p ½0:0018 expð
0:0236ðT
273:1ÞÞ ri kT Di ¼ 6p ½0:00103 þ Acv þ Bcv ri Di ¼

n  r pffiffiffiffio i j Di ¼ ðDi Þj¼0 exp
Dr

For glycerol, A ¼ 2.18e-6, B ¼ 9.0e-9, for sucrose, A ¼ 1.11e-6, B ¼ 1.96e-8.

With regard to the dependence of the solute diffusion coefficient on solution composition, the most pertinent changes for consideration are those due to the addition of viscogenic agents often required for biochemical stability, most commonly glycerol or sucrose [25], or those brought about by the use of a tethered polymer gel layer [26,27]. Table 4.2 describes the estimated diffusion coefficient for a spherical solute as a function of added glycerol or sucrose as viscogenic agents (designated cv) at 201C [25]. A common feature of optical biosensor experimentation is the use of a chemically inert tethered polymer support such as carboxymethyldextran to act as a point of covalent attachment for biological ligands [17,18]. Over a small distance scale the tethered polymer support has the potential to act as a porous chemical phase acting to retard diffusion of the solute in a size dependent manner. A simple estimate12 of the potential for reduction in diffusion coefficient of a spherical solute’s diffusion coefficient upon entering a porous gel has been provided by Ogston et al. [26,27]. They treated the polymer phase as a randomly arranged collection of very long cylindrical rods having a radius Dr and a fractional volume occupation j. Using this approach, one may estimate the dependence of the diffusion coefficient of a spherical solute on the size of solute and the concentration of rod like polymer13 constituting the gel phase Di(ri, j) as shown in Table 4.2. Although reasonable estimates of the diffusion coefficient14 can be gained from experiment or theory, estimation of the velocity profile for a particular configuration requires detailed device dimensions as input parameters in simulations based on the Navier–Stokes equations [28,29]. However, some general conclusions can be made about the velocity profiles operating in flow-through

12

With some experimental support. Yarmush et al. [32] have additionally raised the possibility that the diffusion coefficient of solute in the gel will be sensitive to changes in the concentration of adsorbate. This will have the effect of decreasing the rate of mass transfer to lower regions of the gel in an occupancy-dependent manner. 14 Diffusion is assumed isotropic. 13

Kinetic Models Describing Biomolecular Interactions at Surfaces

95

biosensor designs and in cuvette-type biosensor designs by assuming idealized velocity profiles that are related to simpler cases.15 For the flow-type biosensor, one approach has been to treat the problem as flow in an open-ended pipe and hence describe the velocity profile [30–34] as the simple parabolic type (Figure 4.4b) where the linear velocity in the x direction, vx, is expressed as a function of position of the height of the channel, z, which has a maximum height of zmax [eq. (4.12)].
 z z2
2 vx ðzÞ ¼ 4vx ðzmax =2Þ ð4:12Þ zmax zmax As can be noted from Figure 4.4b, the linear x component of the velocity of solute i approaches zero at the channel walls and is maximal in the central region. Simple approximate models for predicting the velocity profile are harder to come by for the cuvette-type biosensor because stirring is carried out differently in different machine types. However, some information can be gained from a limiting case analysis in terms of the stagnation point flow assumption in a stirred cuvette [35,36]. In this treatment, a reduced subsection of the flow profile is regarded as emanating from the position of the stirrer bar and is expected to disperse symmetrically as it approaches the surface (leaving a so-called ‘‘stagnant point’’ on the adsorptive surface in the area immediately below the point of stirring and a stagnant flow region next to the surface) [Figure 4.5b and eq. (4.13)]. 3

vx ¼ 0:164B2

3

vz ¼ 0:164B2



1 Z 2 xz r



1 Z 2 2 z r

ð4:13aÞ

ð4:13bÞ

In these equations, B is a multi-term flow constant, dependent on device geometry for which values can be found in Ref. [35] and references therein. With regard to coordinate position, the point of origin is the point on the surface immediately below the stirring device. As with the parabolic flow assumption in the flow-through device, the predicted velocity for the cuvette device approaches zero at the surface. The full numerical solution of the continuity equation ([eq. (4.10)] can be a daunting task. However, nearly all problems involving mass transfer to a surface may, with varying degrees of success, be decomposed into simpler compartment models [34,37,38] (Figure 4.6) reminiscent of the descriptive model outlined in eq. (4.4). The basis of the two-compartment simplification 15

Here we consider convective flow up to the plane of the adsorptive surface. For situations where the adsorption sites are attached to a polymer gel layer we assume that the polymer layer will disturb flow and consider the surface of the tethered polymer layer as the relevant surface for defining the velocity profile. Some authors also have considered the penetration of flow into any such gel layer [39].

96

Chapter 4

lies in the observation (noted above for the stagnant point flow and the parabolic flow cases) that the velocity profile of a fluid flowing over a stationary surface approaches zero at the limit of contact with the surface (i.e. z ¼ 0). This zero flow region or stagnant layer may be considered as extending out a small distance, d, from the surface. A neighboring similarsized region beyond this distance d, termed the bulk, is considered to move

Figure 4.6 (A) Two-compartment model for describing mass transport from bulk solution to a surface. The region of zero flow, v ¼ 0, is assumed to extend out a stagnant layer distance d from the surface, beyond this the bulk solution moves at the average velocity of vBULK. Transport from the bulk to the surface phase across zero flow region occurs predominantly via diffusion. (B) Ratio of steady-state stagnant flow region solute concentration to bulk solute concentration for different values of d and different extents of initial adsorption rate and concentration of matrix adsorption sites, represented as a combined parameter k1(Cx)tot. The value of d has been estimated on the basis of the 5% approximation discussed in the text. The value of (Cx)tot is calculated on the basis of the volume of the stagnant flow region, i.e. (Cx)tot decreases with increasing d. (C) Typical differential rate plots for a system displaying Langmuir adsorption kinetics, which is highly influenced by mass transfer effects [dotted line: simulated using eqs. (4.9) and (4.14a)]. Differential rate plot for adsorption profile with the identical reaction kinetics but without mass transfer effects (solid line). The mass transfer limited data only approaches that of the Langmuir case at near maximal saturation of the adsorption matrix sites.

Kinetic Models Describing Biomolecular Interactions at Surfaces

Figure 4.6

97

Continued

with a common averaged velocity, vBULK. For relatively fast flow rates, transfer of solute from one region to another in the bulk will occur primarily by convection, allowing for the assumption of a single homogeneous compartment having a solute concentration designated CBULK that is determined solely by the flow rate. Due to the assumed zero flow in the stagnant layer region close to the surface, transfer of solute from the bulk solution to the adsorptive surface over the distance d will occur predominantly by diffusion. For such a simplified system the rate of solute transfer into the stagnant layer and the subsequent degree16 of mass transfer limitation for an adsorption reaction (for a given receptor density, concentration of solute and intrinsic reaction kinetics) can be estimated via eq. (4.14). dfCi g Di E 2 ½CBULK
fCi g
k01 fCi g dt d fCi g Di =d2 E CBULK Di =d2 þ k01

ð4:14aÞ

ð4:14bÞ

Values of d for different flow configurations and magnitudes of the stirring/ flow rate have been calculated [28]; typical values17 for current biosensor devices are of the order of micrometers [28,36]. In general, the parameter d representing the thickness of the stagnant layer will become smaller upon Upper limit calculated by assuming irreversible reaction, i.e. k2 ¼ 0 s
1 and assuming steady-state d{Ci}/dt ¼ 0. The assumption of k2 ¼ 0 is equivalent to the case of initial adsorption rate and hence mass transfer limitations will be at their greatest during initial adsorption rates. 17 For solutes with diffusion coefficients of the magnitude of moderately sized globular proteins (D E 1  10
10 m2 s
1). 16

98

Chapter 4 18

increasing the rate of flow. Using eq. (4.14b), we have estimated the effect for the initial adsorption rate for various receptor densities and intrinsic kinetics of the adsorbing system in Figure 4.6a and the results are shown in Figure 4.6b. As can be noted, mass transfer limitations will be greatest for an adsorbing system (solute plus matrix site) having intrinsically rapid adsorption kinetics under experimental conditions where the adsorptive surface comprises a highly concentrated array of matrix sites and the fluid flow/stirring rate and the solute’s diffusion coefficient are both of small magnitude. Such considerations have led to the general recommendation that optical biosensor experiments be conducted under conditions of low matrix concentration and high flow/stirring rates [42–45]. As illustrated in Figure 4.6c, mass transfer-limited adsorption data can be readily identified by the appearance of a downward curvature in plots of the adsorption rate versus the concentration of adsorbate [46]. In experimental studies of adsorption, the best approach is the elimination of any mass transfer effects by good experimental design. When this is technically not possible, a number of approximate strategies based on the two-state model described in eqs. (4.6) and (4.9) have been developed [33,37,38]. More will be said about the practical aspects of this extension to the standard analysis in Chapters 9 and 10. Now, however, having addressed the basics of the mass transfer process in optical biosensor experimentation, we turn our attention to the reaction component of the adsorption event.

4.2.2 Adsorption Mechanisms In this section we formalize the discussion of adsorption mechanisms by deconstructing the subject into smaller component pieces.19 We then use these building blocks to describe a wide range of adsorption behavior. By doing this we hope both to increase awareness of the pool of candidate models available for initial consideration and to educate the reader as to the richness in variety of adsorption behavior that is not confined to just the standard 1:1 binding model or variants thereof. In eq. (4.4) we have essentially pictured the adsorption reaction as the partition of a single solute for which the association rate function k01 is sensitive to the concentration of adsorbate in a manner determined by the true

18

One estimate of the magnitude of d is on the basis of a 5% approximation, i.e. the value of d at which the rate of solute transport into the stagnant region is less than 5% of the rate at which solute would enter via diffusion, i.e. d E 0.05Di/v(d). More detailed approaches to estimate d or a d-like parameter can be found in refs. [35,36] for cuvette systems and [40–41] for flow systems. 19 Standard practice in the selection of a mechanistic model to describe experimental kinetic and equilibrium adsorption data involves selection of a number of likely candidate models followed by non-linear regression to find the model that best fits the data in a statistically relevant manner. There are a number of different strategies for performing such non-linear regression analysis of the data. Depending on the mathematical sophistication of the experimenter, this may be done using user-written software or user-defined models in a non-linear regression software package. Alternatively, a number of semi-‘‘black box’’ routines are provided by the instrument manufacturers or interested third parties. A good starting point for information on the subject of modeling data by non-linear regression analysis can be found in ref. [47]. Discussion of the relative benefits of different fitting strategies is outside the scope of this chapter.

Kinetic Models Describing Biomolecular Interactions at Surfaces

99

second-order adsorption association rate constant, k1, the starting (hypothetical20) total concentration of matrix sites, (Cx)tot, and the surface function, f(f), describing the concentration of available sites for a given extent of surface occupation. By considering the process in this fashion we may neatly delineate our discussion of adsorption mechanisms into a number of separate sections. In the first section we will concern ourselves with general models of idealized partition occurring at a constant concentration of matrix sites [i.e. where f(f) equals 1 and therefore k01 equals k1(Cx)tot]. In the second section we will focus on how competition might affect the partition process of the single solute. In the third section we will describe how the surface function changes with different extents of surface occupation for different modes of adsorption.

4.2.2.1

Idealized Partition Processes

In chemistry, partition refers to the distribution of a solute between two immiscible liquid phases at infinitely dilute concentrations of the partitioning solute. In adsorption experiments, a similar situation is achieved when the solute is dilute and the adsorbate concentration is exceedingly low, thereby allowing us to equate f(f)(Cx)tot with (Cx)tot. By examining the kinetics associated with this low-concentration region we may successfully divorce our analysis of adsorption from complications associated with any changes associated with the surface function. A single molecular adsorption event may occur as a mechanistically concerted adsorption event, observed to be occurring in a single elementary step [eq. (4.15)].21 k01

fC1 g Ð ðC1 Þads k2

ð4:15Þ

Alternatively, a single molecular adsorption event may occur as a mechanistically stepwise process, whereby different parts of the adsorption process occur in a sequential manner with each event itself being defined by a unique characteristic time scale. Although potentially an indefinite number of intermediate forms of adsorbate may exist, e.g. (C1a)ads, (C1b)ads, . . ., we express the general concept in a limited two-state model [eq. (4.16)]. k01

F

k2

k4

fC1 g Ð ðC1a Þads Ð ðC1b Þads

ð4:16Þ

The rate of interconversion between (C1a)ads and (C1b)ads is respectively governed by forward and reverse rate constants F and k4 (units of s
1). If the forward rate of conversion is independent of the local concentration of matrix sites, F takes on a constant value of a simple first-order rate constant. If, 20

By hypothetical we mean that the total concentration of adsorbate at maximal saturation may differ from the total starting concentration of matrix sites if an adsorbate molecule covers multiple potential binding sites. 21 From here on we will make specific reference to the solute type, i.e. in this case we refer to the intimate concentration of solute of type 1 given by {C1}.

100

Chapter 4

however, the rate of interconversion is dependent on the local concentration of matrix sites, F represents a rate function composed, in the simplest case, of a second order rate constant k3 (units of l mol
1 s
1), the total concentration of matrix sites, (Cx)tot (and a stepwise specific surface function, f3(f1a, f1b) that describes the fractional availability of nearby matrix sites and is dependent on the concentration of both types of adsorbed species.) The rate equation [describing the formation of adsorbate for a mechanistically concerted reaction denoted by eq. (4.15)] was given in eqs. (4.7)
(4.9) and is shown in Figure 4.7a.

0.2

(Ci)ads(µM)

(A)

(C1,1)ads

0 0

500

1000

Time (s) 0.2

(Ci)ads(µM)

(B)

(C1a,1)ads

(C1b,1)ads 0 0

500

1000

Time (s)

Figure 4.7

Adsorbate concentration vs. time in different partition processes. (A) Mechanistically concerted homogeneous partition (single class of solute and matrix site resulting in formation of single class of adsorbate (C1,1)ads. (B) Mechanistically stepwise homogeneous isomerization (single class of solute and matrix site, adsorbate formation proceeds by way of a two-step reaction resulting in formation of two classes of adsorbate (C1a,1)ads and (C1b,1)ads.

101

Kinetic Models Describing Biomolecular Interactions at Surfaces

For a stepwise reaction, the rate of formation of the different classes of adsorbate22 is given by eq. (4.17) (Figure 4.7b). dðC1a Þads ¼ k01 fC1 g
k2 ðC1a Þads
FðC1a Þads þ k4 ðC1b Þads dt dðC1b Þads ¼ FðC1a Þads
k4 ðC1b Þads dt

ð4:17aÞ ð4:17bÞ

Assuming that the system is reversible, the most typical diagnostic for identifying adsorbate isomerization23 is from analysis of a dissociation phase adsorption experiment [48]. Depending on the relative extents of occupation of the intermediate states, a bimodal signature of the rate plot of d(C1)ads/dt vs. (C1)ads will be apparent. Another diagnostic sign of adsorbate isomerization will be an apparent incompatibility between the dissociation rate calculated from analysis of the association phase and that calculated from the dissociation phase [48]. Examples of such isomerizing systems have been encountered in experimental studies of multivalent solutes; examples include antibody binding studies [49] and studies of multimeric proteins [50]. A detailed treatise of the theory can be found in ref. [51].

4.2.2.2

Effect of Competing Reactions

Depending on the nature of the experimental system (i.e. the solute, the adsorptive surface and the solvent conditions), there may exist a degree of heterogeneity in the effective strength of the interaction. This apparent heterogeneity may result from heterogeneity of the solute or heterogeneity of the matrix adsorption sites. Whatever the cause, analysis of heterogeneous binding data will result in an apparent distribution of rate constants whose influence will be felt according to the relative concentrations of the different types of solute and matrix adsorption sites. For the case where heterogeneity exists in the solute alone, such that a number p of different intimate solute types [{C1}, . . ., {Cp}] exist, the observed rate of adsorption is given by the summation shown in eq. (4.18a). For the case where heterogeneity is caused by a certain number, q, of different types of matrix sites [(Cx)tot(1), . . ., (Cx)tot(q)], the corresponding summation describing the rate of formation is that shown by eq. (4.18b). For the case where both the solute and the matrix site show heterogeneity, the double summation in eq. (4.18c) is required for an adequate description of the rate of adsorption.24 j¼p  X dðCj;1 Þ j¼1

22

dt

ads

¼

k01ðj;1Þ

 Cj
k2ðj;1Þ ðCj;1 Þads

 ð4:18aÞ

Again here limited to two for convenience. Another potential case of adsorbate isomerization is due to reaction of the adsorbate with other adsorbate molecules. This situation will be dealt with in the section on adsorbate clustering. 24 The heterogeneity reaction can be extended to include conformational change. 23

102

Chapter 4

 k¼q  X dðC1;k Þads 0 ¼ k1ð1;kÞ fC1 g
k2ð1;kÞ ðC1;k Þads dt k¼1 j¼p X k¼q  X dðCj;k Þ j¼1 k¼1

dt

ads

 ¼ k01ðj;kÞ Cj
k2ðj;kÞ ðCj;k Þads

ð4:18bÞ

 ð4:18cÞ

In the above formulation, the subscripted rate function, k 0 1(j, k) and rate constant k2(j, k) refer to the operative rate parameters for interaction between solute of type j and matrix site of type k to produce an adsorbate (Cj, k)ads. Svitel et al. [52] have examined techniques for identifying and quantifying such heterogeneity in biosensor data for interactions obeying simple 1:1 binding models. Phenomenological descriptions of heterogeneity such as those described by eqs. (4.18a
c) also encompass competition reactions (Figure 4.8). One well-known example of a biological competition reaction is that discovered by Vroman et al. [53], in which the heterogeneous proteins of blood plasma were found to adsorb on glass with different characteristic time scales and binding affinities.

4.2.2.3

Surface Functions for Different Modes of Adsorption

For a surface denuded of adsorbate, all potential matrix sites are available for adsorption, making the surface function equal to 1. However, upon adsorption of solute the fraction of available matrix sites will change in a manner dictated by the way in which adsorbate molecules affect the likelihood of subsequent adsorption events. The mathematical form of this dependence is known as the surface function [54–59] and its exact nature will define the kinetic and equilibrium adsorption isotherms [60], i.e. the dependence of the rate and equilibrium extent of adsorption on the concentration of intimate solute for a given adsorbate concentration (and set solution composition and temperature). Here we make use of simple geometric approximations to describe the physical nature of the solute and adsorptive surface and treat attractive and repulsive behavior through simple interaction potentials such as hard particle approximations or square well potentials (mesoscopic models). These mesoscopic models are used here to describe the surface function for various adsorptive behaviors. Such approaches are especially suited to the aims of this chapter, in which kinetic models are used to comment on the phenomenon of adsorption from the viewpoint of a measurement device that provides a single observable experimental parameter related to the total concentration of all species adsorbed at the interfacial layer. With this proviso, we begin our description of surface functions by first more carefully defining the meaning of the quantity f. For the simple case of a spherical adsorbing solute, the surface function is expressed in terms of the quantity fi, which expresses the ratio of the concentration of adsorbate to the total concentration of matrix sites multiplied by ni, a

103

Kinetic Models Describing Biomolecular Interactions at Surfaces 0.2

(Ci)ads(µM)

(A)

(C1,1)ads

(C1,2)ads 0 0

500 Time (s)

1000

0.2

(Ci)ads(µM)

(B)

(C1,1)ads

(C2,1)ads 0 0

500

1000

Time (s)

Figure 4.8

Adsorbate concentration versus time for different competition processes. (A) Mechanistically concerted heterogeneous partition [one class of solute and two classes of matrix site resulting in formation of two classes of adsorbate (C1,1)ads and (C1,2)ads]. (B) Mechanistically concerted heterogeneous partition [two classes of solute and one class of matrix site resulting in formation of two classes of adsorbate (C1,1)ads and (C2,1)ads].

term representing the average number of matrix sites physically covered by adsorbate of type i [eq. (4.19)].25 fi ¼

ni ðCi Þads ðri Þads pr2i E ðCx Þtot Atot

ð4:19Þ

At very low densities of matrix sites, such that adsorption at one site does not influence adsorption at another, ni ¼ 1. At high matrix densities relative to the size of the solute (see Figure 4.1), the matrix will approach the continuum limit 25

The surface function will be a function of the relative orientation for an asymmetric adsorbate but we do not treat that complication here.

104

Chapter 4

(approximation shown in brackets). In this case, one may estimate the value of n on the basis of geometric arguments alone. For planar adsorptive surfaces characterized by an area density rx* of matrix sites (molecules m
2) or three-dimensional adsorptive surfaces characterized by a volume density of matrix sites rx (molecules m
3), we may approximate ni from knowledge of the physical area and volume covered by the adsorbate yielding eqs. (4.20a) and (4.20b), respectively. ni ! rx pr2i ni ! rx ð4=3Þpr3i

at high densities for 2D surfaces at high densities for 3D surfaces

ð4:20aÞ ð4:20bÞ

In discussing surface functions, we recognize that adsorption reactions can be categorized into two general types [60], those capable of being saturated (i.e. a fixed number of adsorption sites) and those incapable of being saturated (non-fixed number of adsorption sites, usually corresponding to multilayer formation) (Figure 4.9). In optical biosensor experimentation involving biological macromolecules, both types of adsorption are encountered [61,62] so we discuss the general characteristics of each in turn. Surface Phases Capable of Being Saturated. In the case of adsorptive surfaces that are capable of being saturated, the simplest form that the surface function can take is that suggested by Langmuir [63], whereby adsorption to any particular matrix site is independent of all others and each binding event decreases the total number of matrix sites by one (such that ni ¼ 1). For a single class of solute adsorbing to a single class of matrix site, we have [eq. (4.21)]. f1ð1;1Þ ðf1;1 Þ ¼ 1
f1;1

ð4:21Þ

Provided that the necessary conditions concerning independence and capacity for saturation described above are met, the Langmuir model may be extended to describe the case of a heterogeneous solute or heterogeneous matrix site. For the most general case of both heterogeneous solute and matrix [Eq. (4.18c)], a separate surface function for each of the k types of matrix sites, f1(j,k), can be written as eq. (4.22), again within the remit of the Langmuir requirements. p   X f1ðj;kÞ f1;k ; . . . ; fp;k ¼ 1
fj;k

ð4:22Þ

j¼1

If the matrix sites are densely packed relative to the matrix site spacing, the adsorption of a solute molecule will influence the potential for adsorption of other solute molecules in the vicinity. This influence may be repulsive, in which case the surface function will decrease more sharply with increasing adsorbate concentration than a corresponding hypothetical Langmuir isotherm26 [60]. 26

By corresponding hypothetical Langmuir isotherm we mean one defined by a surface function in which the total concentration of matrix sites (Cx)tot is not the true concentration of matrix sites but that corresponding to the maximal adsorbate concentration (Ci)ads.

Kinetic Models Describing Biomolecular Interactions at Surfaces

105

(Ci)ads

(A)

Ci (B)

(C)

Figure 4.9

(A) Illustration of two general cases of adsorption behavior. Lower trace: saturation due to a fixed number of sites and monolayer formation (saturation implies that no further adsorption will take place despite increasing the concentration of solute). Upper trace: the type of adsorption that cannot be saturated due to multilayer adsorption. The dotted line represents monolayer coverage. (B) Diagram showing apparent monolayer coverage for saturated adsorption. (C) Diagram showing apparent multilayer formation for adsorption not capable of being saturated.

106

Chapter 4

Alternatively, if the influence is attractive, the extent of decrease in the surface function versus the Langmuir case will lessen. At very high extents of attraction the surface function may even mimic that associated with the Langmuir case [64]. The simplest manner of incorporating symmetrical repulsive interactions into the surface function is to approximate the repulsive interaction by an equivalent interaction of hard circles for planar (two-dimensional) surface phases and hard spheres for three-dimensional surface phases [54,57] (Figure 4.10). Such approaches usually take one of two general forms, depending on the extent of the reversibility of the adsorption event or the degree of mobility of the surface phase. For highly mobile surfaces and/or rapidly dissociating systems, equilibrium fluid models based on scaled particle theory (SPT) expansions [65,66] are particularly useful for describing the surface function. For the case of adsorption of a single class of solute to a 2D continuum of a single class of matrix sites we have eq. (4.23). ( " #) 2f1 f1 f1ð1;1Þ ðf1 Þ ¼ exp

lnð1
f1 Þ þ þ ð4:23Þ 1
f1 ð1
f1 Þ2 The SPT models can easily accommodate a range of different size adsorbates existing on the surface by making the surface function for a given elementary step and a certain class j of solute, a function of the different extents of fractional coverage of all the different adsorbate types on a single class of surface matrix sites f1,1, f2,1, . . . , fn,1, e.g. f1(j,1)(f1,1, f2,1, . . . , fn,1). For a range of different sized adsorbate molecules characterized by radii ri and surface densities ri* (for a two-dimensional adsorptive surface), the surface function describing the probability of finding a free matrix site is approximated by eq. (4.24), for convenience expressed in radii of adsorbate on a near continuum of matrix sites on the surface.
 P   h X i 2p r Ri Pi f1;ðj;1Þ ðr1 ; . . . ; rn Þ ¼ exp

ln 1
p ri R2i þ Rj 1
p ri R2i " P  2 # !) P p2 ri Ri p ri P þ þ R2 P 1
p ri R2i ð1
p ri R2i Þ2 j ð4:24Þ For irreversible adsorption reactions (k2 ¼ 0 s
1), which occur on an adsorptive surface with matrix sites fixed in position (termed immobile), the form of the surface function will be more appropriately described by a modeling approach based on the phenomenon of random sequential adsorption (RSA) [67,68]. At high levels of surface coverage, the surface distribution of adsorbate will be, on average, less efficiently placed and this more ‘‘selfish packing’’ will result in a reduced maximal extent of adsorption, fmax. For monolayer adsorption of one class of solute to a planar continuum surface, the maximal extent of surface coverage for a reversible (or alternatively highly mobile surface) adsorption reaction is fmax ¼ 0.906, and for an irreversible adsorption

Kinetic Models Describing Biomolecular Interactions at Surfaces

Figure 4.10

107

Adsorption behavior showing a saturation limit: repulsive adsorbate– adsorbate interactions. (A) Surface functions for adsorption of a single class of monomeric spherical solute to a single class of adsorptive surface sites versus fractional surface coverage for (1) Langmuir case [eq. (4.21)], (2) 2D – equilibrium fluid calculation [eq. (4.23)], (3) random sequential adsorption calculation [eq. (4.25)] and (4) 3D – equilibrium fluid calculation [eq. (4.26)]. (B) Illustration of the rapid decrease in the surface functions for the particle model based cases [lines 2–4 in (A)]. Suboptimal placement leads to a rapid decrease in the number of available sites. Here black circles represent adsorbate and red halos denote area excluded for subsequent solute adsorption.

108

Chapter 4

reaction occurring on an immobile adsorptive surface the corresponding value of fmax is 0.546 [54]. The surface function from such irreversible random sequential adsorption of a single class of solute to a planar surface has been calculated by Monte Carlo-based computer simulations [55]. A compact polynomial description of the results of such simulations is given by eq. (4.25). f1;ð1;1Þ ðf1 Þ ¼

ð1
ff1 =fmax gÞ3 1
0:812ff1 =fmax g þ 0:2336ff1 =fmax g2 þ0:0845ff1 =fmax g3 ð4:25Þ

At high extents of surface coverage, somewhere between the limits of irreversible adsorption to an immobile surface and reversible adsorption to a highly mobile phase, one may encounter kinetics reminiscent of a glassy state to ordered state phase transition [69]. In this regime the rate at which the surface function changes from the random sequential model to the equilibrium SPT model will be determined by the kinetics of reorganization of the surface phase dictated by the rate constants k1 and k2 and the adsorptive surface phase matrix site diffusion constants [70]. For a three-dimensional surface phase such as that presented by matrix sites existing on a carboxymethyldextran gel layer derivatized with specific matrix sites, the calculation/estimation of fractional site coverage and maximal extent of occupation is more difficult.27 In the volume transcribed by the gel layer a significant volume fraction is already occupied by the gel itself [31]. Additionally, the fractional availability of matrix sites will in part be determined by mass transfer into the gel layer. As solute will enter the surface phase from the top, these surface sites will be preferentially occupied, possibly preventing access to matrix sites below [32]. A possible starting point for the calculation of approximate surface functions for a highly derivatized gel matrix has been suggested [71], which involves treating the gel layer as a 3D phase with evenly dispersed matrix sites. Using such an approach one may calculate28 f1(f1) for a single class of adsorbing solutes using the 3D version of the SPT equation [66] [eq. (4.26)]. ( f1ð1;1Þ ðf1 Þ ¼ exp

lnð1
f1 Þ þ



# " # !)  " 7f1 7:5f21 3f31 þ þ ð1
f1 Þ ð1
f1 Þ2 ð1
f1 Þ3

ð4:26Þ

For the alternative case of attractive surface interactions resulting in preferential cluster formation, the surface function will decrease less sharply than for the purely repulsive surface interaction case with increasing adsorbate 27

For freely accessible insertion of spheres into an open volume, fmax varies between 0.636 and 0.7405 for the equilibrium random and jamming close packing limit. 28 Using eq. (4.20b) for the estimation of n1 and neglecting the presence of the gel layer as a first approximation.

Kinetic Models Describing Biomolecular Interactions at Surfaces

109

concentration and may even approach the behavior defined by the surface function for the Langmuir isotherm [64]. For a highly mobile/reversible surface, clustering may occur via a post-adsorption interaction in a stepwise reaction event (Figure 4.11a) and may be viewed as a type of isomerization reaction as derived in eq. (4.17). Alternatively, clustering of adsorbate may occur as part of the elementary process of adsorption (Figure 4.11b) [72]. With regard to the first mode of cluster formation (monomer adsorption followed by cluster growth), we present a cluster isomerization/growth model based on monomeric adsorbate addition and monomeric adsorbate loss in eq. (4.27) on a single class of matrix sites. In this model, an adsorbate cluster containing i monomers is specified by (C1)ads[i] and the intrinsic association and dissociation rate constants for clusters of size i on the surface as k3[i] and k4[i], respectively. By varying the rate constants for formation of a cluster of i monomers, k3[i] and dissociation of an adsorbate from a cluster of i monomers into monomeric adsorbate and a cluster of size (i
1) monomers, k4[i], between large and low values one can describe the widest possible range of behavior corresponding to a high mobility and low mobility surface phase, for the case where the attractive potential exerted

Figure 4.11

Adsorption behavior showing a saturation limit: attractive adsorbate– adsorbate interactions resulting in cluster formation happening as a result of (A) post-adsorption association due to surface mobility and (B) association occurring in concert with the adsorption event.

110

Chapter 4

by the cluster does not increase the rate of solute adsorption but only effects the rate of adsorbate desorption. dðC1 Þads½1 dt

¼ k01 fC1 g
k2 ðC1 Þads½1 " # Z X þ
k3½i ðC1 Þads½1 ðC1 Þads½i þ k4½i ðC1 Þads½i

ð4:27aÞ

i¼2

dðC1 Þads½i dt

¼ k3½i ðC1 Þads½1 ðC1 Þads½i
1
k4½i ðC1 Þads½i

ð4:27bÞ


k3½iþ1 ðC1 Þads½1 ðC1 Þads½i þk4½iþ1 ðC1 Þads½iþ1 for 2  ioz dðC1 Þads½z dt

¼ k3½z ðC1 Þads½1 ðC1 Þads½z
1
k4½z ðC1 Þads½z for i ¼ z

ð4:27cÞ

For cluster formation occurring in 2D, an initial approximation of the system can be made by modeling each cluster of i monomers as a circular aggregate of radius ri ¼ r1i, thus allowing the use of surface functions described for the 2D continuum surface phase [eq. (4.24)].29 Figure 4.12a describes the change in surface function for the situation when, on the time-scale of adsorption, clustering is rapid. In the alternative case, where cluster growth occurs contemporaneously with the adsorption event, we model cluster formation by defining an attractive potential that contributes a stabilizing energy, DEc, to monomer adsorbed in an area surrounding an already adsorbed monomer or cluster. For simplicity, the stabilizing potential can be considered as a square well that projects a small distance d from the cluster perimeter and the distance of closest approach.30 When comparing monomer adsorption to two equal areas of surface, one existing within the attractive well and the other outside it, the overall affinity of the monomer for the adsorptive surface is modified by a unitless factor KC [eq. (4.28)].

DEC KC ¼ exp ð4:28Þ RT Equation (4.28) represents an equilibrium stabilization factor. However, for a kinetic model one needs to parse out the contributions into the individual forward and reverse rate constants. In principle, the stabilizing effect could be 29

Note that in reality the cluster will be a chain of aggregated adsorbate having a shape that will display fractal-like characteristics. This fractal-like quality will have an effect on the adsorption kinetics. This topic is dealt with in Chapter 5. 30 For reasons of stability of the solution of equations, we actually consider that the zone of attractive potential associated with adsorbate species lies between the distance of closest approach (r1 + ri) and a smaller distance (r1 + ri
d). Such an approximation will not significantly change the form of the effect upon the surface function. Additionally, this approximation will allow for the estimation of an approximate surface function for this clustering mode up to fairly high degrees of surface occupation (however, it will be less reliable as f approaches its maximum value). The distance d should be chosen so that it is less than r1.

Kinetic Models Describing Biomolecular Interactions at Surfaces

Figure 4.12

111

(A) Surface function for the post-adsorption cluster formation case as a function of the extent of cluster formation (line 1, k3/k4 ¼ 1; line 2, k3/ k4 ¼ 100; line 3, k3/k4 ¼ 1000). Inset describes the corresponding size distributions of adsorbate on surface for the three different values of k3/ k4 – note that for the lowest value of k3/k4 the adsorbate exists exclusively as monomer. (B) Surface functions for the case of cluster formation occurring in concert with adsorption. Line 1 describes f1(1,1) and line 2 describes f1(1,2). The interaction distance was set at d ¼ 0.3r1. The additional line represents the calculation of the surface function for the reduced radius [see eq. (4.30a) and footnote 31].

112

Chapter 4

housed either entirely in k1 or entirely in k2. For purposes of discussion the effect is divided equally by modifying the idealized adsorption partition rate constant of monomer to cluster by the factor O(KC) and the adsorption dissociation rate constant of monomer from a cluster by the factor 1/O(KC). In this case, the intrinsic rate constants describing monomer adsorption to a region of surface within the attractive cluster potential are described by eq. (4.29). For an already occupied surface the attractive well around each adsorbate cluster effectively constitutes a different class of matrix sites to that existing outside the zone of attractive potential, similar to the heterogeneous case in eq. (4.18b), for which there are two classes of matrix site. k1ð1;2Þ ¼ k1ð1;1Þ

pffiffiffiffiffiffiffi KC

and

k2ð1;1Þ k2ð1;2Þ ¼ pffiffiffiffiffiffiffi KC

ð4:29Þ

For such a case we may estimate the surface function for the ‘‘first’’ class of sites (matrix alone) using eq. (4.24). The surface function for the ‘‘second’’ class of sites can be calculated as the difference between surface functions calculated on the basis of circular species of true radius and species of modified radius ri – d [see footnote 31 – eq. (4.30a)]. The surface function specifically associated with each cluster composed of i monomers, f1(1,2){i}, can be parsed out by multiplying eq. (4.30a) by the surface area of the potential region around the i-sized cluster and dividing by the total area associated with all regions of attractive potential [Eq. (4.30b)]. f1;ð1;2Þ Ef1;ð1;1Þ ½ðr1
d Þ; . . . ; ðrn
d Þ
f1;ð1;1Þ ðr1 ; . . . ; rn Þ ðC1 Þads½i ðri þ r1
d Þ2 f1;ð1;2Þ½i ¼ f1;ð1;2Þ P z ðC1 Þads½k ðrk þ r1
d Þ2

ð4:30aÞ ð4:30bÞ

k¼1

By modeling each growing cluster as a circle that grows and shortens by either addition or loss of a monomer unit, the rate of formation of each cluster of k monomers can be expressed using eqs. (4.31a–c). Figure 4.12b describes the change in surface functions for this mode of adsorption with varying adsorbate levels. dðC1 Þads½1 ¼ k01ð1;1Þ½1 fC1 g
k2ð1;1Þ½1 ðC1 Þads½1 ð4:31aÞ dt
k01ð1;2Þ½2 fC1 g þ k2ð1;2Þ½2 ðC2 Þads½2 dðC1 Þads½k dt

¼ k01ð1;2Þ½k fC1 g
k2ð1;2Þ½k ðC1 Þads½k


k01ð1;2Þ½kþ1 fC1 g

dðC1 Þads½z dt

ð4:31bÞ

þ k2ð1;2Þ½kþ1 ðCk Þads½kþ1

¼ k01ð1;2Þ½z fC1 g
k2ð1;2Þ½z ðC1 Þads½z

ð4:31cÞ

Kinetic Models Describing Biomolecular Interactions at Surfaces

113

Surface Phases Capable of Supporting Multilayer Growth. The treatment of multilayer growth is a complex problem which has received a number of reviews [73,74] and is important in many areas as diverse as semiconductor preparation [75], gas wetting phenomena [76] and bio-nanotechnology [77]. Here, the aim is to provide a basic introduction to the subject that goes beyond the usual cursory mention of the BET isotherm [78] by using simple models to outline some limiting case behaviors of the major types of multilayer adsorption growth (Figure 4.12). As it vastly decreases the complexity while still providing much chemical insight, we will restrict our kinetic description of multilayer growth that proceeds effectively irreversibly (i.e. k2 - 0 s
1). If the fractional coverage and pertinent rate parameters applicable to the first and subsequent layers are additionally appended by the subscripts {L1}, {L2}, . . . , then a full kinetic description of the system can be made by solving one of the candidate surface functions for the primary adsorption layer {L1} and for each consecutive surface layer above layer one, {L2}, {L3}, . . . , etc. To aid our discussion of multilayer formation we will employ the formalism just described which pictures cluster formation as a form of heterogeneous adsorption. By varying the value of k1(1,1) and KC [eq. (4.29)] for each adsorption layer, one can effectively describe many different modes of multilayer formation. One of the simplest multilayer growth modes is that of the Frank–van der Merwe type, shown in Figure 4.13a [79]. This type of multilayer growth involves sequential deposition of one layer to top of the previous layer. This multilayer formation results from the fact that adsorption occurring in concert with cluster formation is highly favored over adsorption in the absence of cluster formation e.g. for an arbitrary layer i, k1(1,2){Li} c k1(1,1){Li}. In this limit, the overall vertical rate of growth of the multilayer will be much slower than its rate of lateral growth and adsorption of the new layer will generally occur after deposition in the underlying layer has achieved a significant coverage. In analogy with condensation or crystallization phenomena, one may liken the first adsorption event to each new layer as a nucleation event which is followed by a growth event (layer growth proceeding to coverage) [73]. As the layer becomes significantly covered, the chance of another nucleation event becomes greater and the process may begin again. If the horizontal and vertical adsorption rates are approximately equal [k1(1,2){Li} E k1(1,1){Li}] a different type of multilayer adsorption behavior, known as Volmer–Weber growth [80] or island growth, is observed (Figure 4.13b). In this case, adsorption mounds are formed separate from each other. Similar reasoning can be used to describe the phenomena of columnar multilayer growth shown in Figure 4.13c [81]. In this situation, the vertical adsorption rate far outweighs the lateral adsorption rate [k1(1,2){Li} { k1(1,1){Li}], leading to the growth of columns that may or may not be vertical, depending on the initial orientation of the first adsorbing molecules and/or the degree of surface roughness. Although a complete description of the adsorption kinetics for multilayer growth is beyond the scope of this chapter, a fair approximation of the basic

114

Chapter 4 (A)

(B)

(C)

Figure 4.13

Illustration of differences in adsorption growth behavior. (A) Near sequential formation of one layer after another due to preference for lateral (edge to edge) growth classed as Frank–van der Merwe-type growth. (B) Island formation adsorption behavior (also known as Volmer–Weber adsorption) due to similar preferences for lateral and vertical growth modes. (C) Columnar growth behavior resulting from strong preference for vertical growth modes only.

Kinetic Models Describing Biomolecular Interactions at Surfaces

115

behavior of all three preceding types can be arrived at on the back of the following simplifications: 1. The first adsorption layer is allowed to form lateral interactions (cluster formation as described in the previous sections). 2. Deposition of solute to the primary adsorptive surface to form the first surface layer is considered chemically distinct from adsorption to any other surface layer, i.e. (k1){L1} a (k1){LY}, where Y denotes an adsorbate layer greater than 1, i.e. Y41. 3. Deposition on top of the first adsorbed surface layer is {L2} chemically equivalent to deposition on any other higher layer ({L3},{L4} . . . ), i.e. (k1){LY} ¼ (k1){LW}, where Y41 and W41. 4. The total surface area available for adsorption to a particular multilayer, Atot{LY} above the first layer (i.e. {LY} for Y41) at any stage of the experiment is equal to f{L(Y
1)}Atot{L1}. This statement is equivalent to saying that there can be no unsupported growth. The remaining free surface area not covered by adsorbate on the first layer is equal to (1
f{L1})Atot{L1}. 5. Surface functions for each layer of growth are calculated on the basis of an assumed continuous surface area.31 On the back of the preceding postulates the set of rate equations describing the adsorption rate to the first layer can be written using either Eq. set 4.27 or 4.31 and then used again to express the rate of adsorption to each particular layer above the first, using Eq. set 4.31.

4.3 Summary and Conclusions The study of adsorption and interfacial phenomena is indeed a very important subject in biology. Table 4.1 describes some of the essential roles that adsorption phenomena play in the fundamental processes of life. In this chapter, we have examined how adsorption phenomena can be studied using optical biosensor technology. After discussing pertinent features of the optical biosensor measurement technique, we examined some of the physical chemistry behind the process of adsorption. We formalized our discussion of adsorption by breaking it down into its component pieces. We covered the process of mass transfer to the surface and looked at how under some circumstances this could be rate limiting. Under conditions in which mass transfer effects are negligible, we described how the form of the adsorption progress curve would be determined by differences in the adsorption mechanism. We further deconstructed our analysis of adsorption mechanism into two separate discussions. The first

31

Obviously this assumption will be weak when the preceding layer is not clustered and in this case so-called ‘‘edge effects’’ will play some role. However, as the degree of cluster formation becomes greater the assumption will become stronger.

116

Chapter 4

concerned itself with different modes of idealized partition at zero adsorbate concentration. Here, we reviewed several fundamental types of adsorption/ partition including homogeneous concerted, homogeneous stepwise and heterogeneous concerted and stepwise modes. In the second discussion, we examined how each new adsorbate addition would affect the likelihood of the next adsorbate addition. We then introduced and reviewed the different forms that the surface function32 may take for different types of adsorption events. By extension, we examined how both attractive and repulsive interactions between adsorbate molecules or between adsorbate and intimate solute molecules would affect such surface functions. We also examined the effect on the surface function of multilayer growth and introduced some of the basic modes that such multilayer growth might take. Many reviews tend to focus on adsorption mechanisms in the absence of mass transport considerations or alternately put their focus on mass transfer limitations while treating adsorption phenomena with overly simplistic 1:1 binding models. We feel that this review has filled a gap by providing an introduction to the general features associated with the adsorption of macromolecules to surfaces by focusing on both areas. Throughout this chapter we have tried to enter discussions from the viewpoint of a biochemist investigating biologically related adsorption phenomena involving macromolecules. In this vein, we have not concentrated on some of the typical topics more preferred by chemists such as discussions of the energetic differences between physisorption and chemisorption.33 Equally, we presented the discussion of mass transfer effects in non-transformed quantities as opposed to the approaches preferred by the (generally) more mathematically oriented engineering community. However, with these caveats out in front we have unashamedly tried to engage the reader with some of the complexity (and wonders) associated with the biophysical approach to the study of adsorption. Interfacial events form such a part of our everyday living that it is not overly dramatic to finish this review with a line from a short poem by Vroman [82]: ‘‘All we can create and cry is interface’’

4.4 Questions 1. a. With respect to the phenomenon of adsorption define the following terms:  solute  adsorbate  binding/matrix site. 32

A probability function describing the likelihood of solute finding an available site on the surface for adsorption. 33 Distinctions which have less functional meaning when examining the non-covalent adsorption of very large molecules.

Kinetic Models Describing Biomolecular Interactions at Surfaces

117

b. With respect to the natures of both the solute and matrix site, discuss what is meant by the terms:  distinct array of binding sites  continuum of binding sites. 2. a. Optical biosensors measure the amount of adsorbate directly, as opposed to chromatographic procedures, which measure the amount of adsorbate by calculating the difference between the total and free solute concentrations during and after adsorption. Give three advantages of such a direct measurement technique for quantifying the adsorption process. b. If the efficiency of detecting adsorbed solute using an optical biosensor decays exponentially with distance normal to the surface, comment on what type of adsorption reactions and adsorption geometries would be the most straightforward to categorize. 3. a. Starting with the basic transport/reaction scheme outlined in eqn. (4.4), derive the limiting case kinetic behavior for (i) transport-limited and (ii) reaction-limited adsorption. b. Comment on how mass transport might be accounted for in a completely general fashion regardless of tube/cell geometries. c. What approximate forms of the general approach given in your answer to (b) are useful for describing mass transport in a flow-through and cuvette-type biosensor? 4. a. Adsorption reaction mechanisms can be described completely generally as partition events in which the partition rate is a function of the extent and type of surface occupation. Discuss the above statement in terms of the components that constitute the association rate function k01 [eq. (4.5)]. b. Write down kinetic mechanisms for the following types of adsorption behavior where unless specified the adsorption is of a simple concerted type:  adsorption of a homogeneous solute to a homogeneous array of binding sites  multi-step adsorption pathway of a homogeneous solute to a homogeneous array of binding sites  adsorption of heterogeneous solute to a homogeneous array of binding sites  adsorption of homogeneous solute to a heterogeneous array of binding sites. 5. Adsorption reactions can be categorized into two general types, those capable of being saturated (i.e. a fixed number of adsorption sites) and those incapable of being saturated (a non-fixed number of adsorption sites, usually corresponding to multilayer formation).

118

Chapter 4

a. What surface functions are applicable to the following types of saturable adsorption reactions?:  Langmuir adsorption  irreversible adsorption of spherical solute to a continuum array of binding sites  reversible adsorption of spherical solute to a continuum array of binding sites  reversible adsorption of spherical solute to a continuum array of binding sites with adsorbate clustering leading to monolayer formation. b. Describe in general terms the differences between the three types of multilayer adsorption growth with respect to the adsorbate preference for forming lateral or longitudinal contacts with the already adsorbed solute. 6. Mass transport-limited kinetics are beneficial for concentration determination of the analyte in a sample. Why?

4.5 Symbols v
i (C1)ads[i] (Ci)ads (Cx)tot (Ki)eff {Ci} B bi CBULK ci Ci cv Di DEc F DS f fmax f(f)

f3(f1a,f1b) fi

partial specific volume of solute concentration of adsorbed monomer clusters on surface of size i monomers concentration of adsorbed solute of type i initial total concentration of matrix sites effective adsorption partition constant [(Ki)eff ¼ fi/bi] concentration of solute of type i spatially close to the surface (termed intimate solute) multi-term flow parameter, dependent upon device geometry effective dissociation rate parameter solute concentration in the bulk solution weight concentration of component i concentration of solute of type i weight concentration of viscogenic agents solute’s diffusion coefficient stabilizing energy (in cluster formation) forward rate constant (units of s
1) change in signal fractional site/area coverage maximal adsorbate site coverage unitless function describing the surface site occupation, where F ¼ n(Ci)ads/(Cx)tot (here n is the average number of sites covered by the adsorbed solute) stepwise specific surface function, describes the fractional availability of nearby matrix sites effective forward rate parameter

Kinetic Models Describing Biomolecular Interactions at Surfaces

Z i k k1 k01 k2 k3 ka kb k3[i] k4[i] K0R KT Mi n NA ni S T v* vBULK vi vx z zmax rx rx* s Dr j d

119

aqueous solvent of viscosity component i Boltzmann’s constant intrinsic second-order rate constant association rate function (unit: s
1) dissociation rate const (unit: s
1) second-order rate constant (units of l mol
1 s
1) phenomenological transport coefficient phenomenological transport coefficient intrinsic association rate constant for clusters of size i on the surface intrinsic dissociation rate constant for clusters of size i on the surface K0R ¼ k01/k2 partition constant for formation of the intimate solute solute i molecular weight refractive index Avogadro’s number average number of matrix sites occupied by adsorbate i signal (optical) temperature average velocity of liquid in flow cell velocity of the bulk of the liquid linear velocity of solute linear velocity in the x direction axis normal to the sensor surface max height of flow channel volume density of matrix sites (molecules m
3) area density of matrix sites (molecules m
2) decay constant (unit: distance) radius of cylindrical polymer rods fractional volume occupation of polymer small distance, the zero flow region extends out a surface

4.6 Acknowledgements From my time in the U.K. I would like to thank Professor Christopher M. Dobson for providing me with space to work in his laboratory, his keen interest in my research and his continued friendship. During this period I would like to acknowledge the financial support of the Human Frontiers Science Program (HFSP) which financed my stay at the Cambridge University Chemical Laboratories (2003–2007) via the award of a HFSP Long Term Fellowship. From my time in Japan I would like to thank Professor Haruki Nakamura and Assoc. Professor Fumio Arisaka for the amazing support which they have

120

Chapter 4

provided me and for which I am deeply indebted. I would like to acknowledge financial assistance from a RIKEN grant for ‘Research and Development of Next-Generation Integrated Life Simulation Software.’ Finally I am deeply appreciative of the valuable assistance provided by Dr. N. Hirota.

References 1. A.E. Johnson, Traffic, 2005, 6, 1078. 2. I. Mellman and G. Warren, Cell, 2000, 100, 99. 3. M.B. Neiditch, M.J. Federle, A.J. Pompeani, R.C. Kelly, D.L. Swemm, P.D. Jeffrey, B.L. Bassler and F.M. Hughson, Cell, 2006, 126, 1095. 4. K.A. Janes, J.G. Albeck, S. Gaudet, P.K. Sorger, D.A. Lauffenburger and M.B. Yaffe, Science, 2005, 310, 1646. 5. W. Cho and R.V. Stahelin, Annu. Rev. Biophys. Biomol. Struct., 2005, 34, 119. 6. K. Kristiansen, Pharmacol. Ther., 2004, 103, 21. 7. R. Benton, Cell. Mol. Life. Sci., 2006, 63, 1579. 8. J.R. Pugh and I.M. Raman, Biophys. J., 2005, 88, 1740. 9. J. McGhee and P. von Hippel, J. Mol. Biol., 1974, 86, 469. 10. A.A. Aderem and D.M. Underhill, Annu. Rev. Immunol., 1999, 17, 593. 11. S. Abraham, S. Brahim, K. Ishihara and A. Guiseppi-Elie, Biomaterials, 2005, 26, 4767. 12. R.J. Green, M.C. Davies, C.J. Roberts and S.J. Tendler, J. Biomed. Mater. Res., 1998, 42, 165. 13. D.J. Winzor and W.E. Sawyer, Quantitative Characterization of Ligand Binding, Wiley-Liss, New York, 1995. 14. L. Gorton (Ed.), Biosensors and Modern Specific Biospecific Analytical Techniques, in Comprehensive Analytical Chemistry, Series Ed. D. Barcelo, Elsevier, Amsterdam, 2005, p. xliv. 15. J. Fraden, Handbook of Modern Sensors: Physics, Designs and Applications, Springer, New York, 2004. 16. M. Terpstra, Biosens. Bioelectron., 1993, 8, ii. 17. R. Cush, J.M. Cronin and J.M. Stewart, Biosens. Bioelectron., 1993, 8, 347. 18. U. Jonnson, L. Fagerstram and B. Ivarsson, Biotechniques, 1999, 11, 620. 19. F. de Fornel, Evanescent Waves: From Newtonian Optics to Atomic Optics, Springer-Verlag, Berlin, 2001, p. 1. 20. H. Raether, in Physics of Thin Films, G. Haas, M.H. Francombe and R.W. Hoffmann (Eds.), Vol. 9, Academic Press, New York, 1977, p. 145. 21. D. Shoup and A. Szabo, Biophys J., 1982, 40, 33. 22. L. Edelstein-Keshet, Mathematical Models in Biology, SIAM, Philadelphia, PA, 2005. 23. H.C. Berg, Random Walks in Biology, Princeton University Press, New Jersey, 1993, Chap. 3. 24. R.C. Weast (Ed.), CRC Handbook of Chemistry and Physics, 62nd edn., CRC Press, Boca Raton FL, 1981, Table F42.

Kinetic Models Describing Biomolecular Interactions at Surfaces

121

25. S. Uribe and J.G. Sampedro, Biol. Proc. Online, 2003, 5, 103. 26. A.G. Ogston, B.N. Preston and D.J. Wells, Proc. R. Soc. London A, 1973, 333, 297. 27. A. Pluen, P.A. Netti, R.K. Jain and D.K. Berk, Biophys. J., 1999, 77, 542. 28. V.G. Levich, Physicochemical Hydrodynamics, Prentice Hall, Englewood Cliffs, NJ, 1962. 29. R.B. Bird, W.E. Stewart and E.N. Lightfoot, Transport Phenomena, Wiley, New York, 1960. 30. R.W. Glaser, Anal. Biochem., 1993, 213, 152. 31. P. Schuck, Biophys. J., 1996, 70, 1230. 32. M.L. Yarmush, D.B. Patankar and D.M. Yarmush, Mol. Immunol., 1996, 33, 1203. 33. D.G. Myszka, X. He, M. Dembo, T.A. Morton and B. Goldstein, Biophys. J., 1998, 75, 83. 34. C. Wofsy and B. Goldstein, Biophys. J., 2002, 82, 1743. 35. M. Elimelech, X. Jia, J. Gregory and R. Williams, Particle Deposition and Aggregation: Measurement, Modelling and Simulation, Butterworth Heinemann, London, 1998. 36. D.A. Edwards, SIAM J., 2000, 105, 1. 37. P. Schuck and A.P. Minton, Anal. Biochem., 1996, 240, 262. 38. R. Karlsson, H. Roos, L. Fagerstram and B. Persson, Methods Comp. Methods Enzymol., 1994, 6, 99. 39. J. Witz, Anal. Biochem. 1999, 270, 201. 40. B.K. Lok, Y.L. Cheng and C.R. Robertson, J. Colloid Interface Sci., 1983, 91, 104. 41. S. Sjolander and C. Urbaniczky, Anal. Chem., 1991, 63, 2338. 42. D. Hall, Anal. Biochem., 2001, 288, 109. 43. P. Schuck, Annu. Rev. Biophys. Biomol. Struct., 1997, 26, 541. 44. D.G. Myszka, T.A. Morton, M.L. Doyle and I.M. Chaiken, Biophys. Chem., 1997, 64, 127. 45. R. Karlsson and A. Falt, J. Immunol. Methods, 1997, 200, 121. 46. C. Calonder and P.R. Van Tassel, Langmuir, 2001, 17, 4392. 47. W.H. Press, S.A. Teukolsky, W.T. Vetterling and B.P. Flannery, Numerical Recipes in C, Cambridge University Press, Cambridge, 2002. 48. T.M. Morton, D.G. Myszka and I.M. Chaiken, Anal. Biochem., 1995, 227, 176. 49. K.M. Mu¨ller, K.M. Arndt and A. Plu¨ckthun, Anal. Biochem., 1998, 261, 149. 50. N.L. Kalinin, L.D. Ward and D.J. Winzor, Anal. Biochem., 1995, 228, 238. 51. P.R. Van Tassel, L. Guemouri, J.J. Ramsden, G. Tarjus, P. Viot and J. Talbot, J. Colloid Interface Sci., 1998, 207, 317. 52. J. Svitel, A. Balbo, R.A. Mariuzza, N.R. Gonzales and P. Schuck, Biophys. J., 2003, 84, 4062. 53. L. Vroman, A.L. Adams and M. Klings, Fed. Proc., 1971, 30, 1494. 54. P. Schaaf, J. Voegel and B. Senger, J. Phys. Chem. B, 2000, 104, 2204. 55. P. Schaaf and J. Talbot, Phys. Rev. Lett., 1989, 62, 175.

122

Chapter 4

56. 57. 58. 59. 60.

I. Lundstrom, Prog. Colloid Polym. Sci., 1985, 70, 76. R.C. Chatelier and A.P. Minton, Biophys. J., 1996, 71, 2367. P. Wojciechowski and J.L. Brash, J. Colloid Interface Sci., 1989, 140, 239. S. Stankowski, Biochim. Biophys. Acta, 1983, 735, 352. A.W. Adamson, in Physical Chemistry of Surfaces, Wiley, New York, 1982, Chapter 16. T.J. Halthur, A. Bjorklund and U.M. Eloffson, Langmuir, 2006, 22, 2227. D.R. Hall, N.N. Gorgani, J.G. Altin and D.J. Winzor, Anal. Biochem., 1997, 253, 145. I. Langmuir, J. Am. Chem. Soc., 1918, 40, 1361. J.J. Ramsden, G.I. Bachmanova and A.I. Archakov, Phys. Rev. E, 1994, 50, 5072. H. Reiss, H.L. Frisch and J.L. Lebowitz, J. Chem. Phys., 1959, 31, 369. J.L. Lebowitz, E. Helfland and E. Praestgaard, J. Chem. Phys., 1965, 43, 774. M.C. Bartelt and V. Privman, Int. J. Mod. Phys., 1991, 5, 2883. J.W. Evans, Rev. Mod. Phys., 1983, 65, 1281. P.L. Krapivsky and E. Ben-Naim, J. Chem. Phys., 1995, 100, 6778. X. Jin, G. Tarjus and J. Talbot, J. Phys. A, 1994, 27, L195. D. Hall, Optical Biosensor Based Studies of Protein, Adsorption: Theory and Measurement, PhD Thesis, University of Queensland Press, Brisbane, 2000. A.P. Minton, Biophys. J., 1999, 76, 176. P.G. Vekilov and J.I.D. Alexander, Chem. Rev., 2000, 100, 2061. D.D. Vvdensky, A. Zangwill, C.N. Luse and M.R. Wilby, Phys. Rev. E, 1993, 48, 852. F.A. Ponce and D.P. Bour, Nature, 1997, 386, 351. U.G. Volkmann and K. Knorr, Phys. Rev. Lett., 1991, 66, 473. J. Texter and M. Tirrell, AIChE J., 2001, 47, 1706. S. Brunauer, P.H. Emmett and E. Teller, J. Am. Chem. Soc., 1938, 60, 309. F.C. Frank and J.H. van der Merwe, Proc. R. Soc. London, Ser. A, 1949, 198, 205. M. Volmer and A. Weber, Z. Phys. Chem., 1926, 119, 277. G.S. Bales and A. Zangwill, J. Vac. Sci. Technol., 1991, 9, 145. As quoted by J.D. Andrade, in Surface and Interfacial Aspects of Biomedical Polymers, Plenum Press, New York, 1985, Preface to Vol. 1.

61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71.

72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82.

CHAPTER 5

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions: SPR Applications in Drug Development NICO J. DE MOL AND MARCEL J.E. FISCHER Department of Medicinal Chemistry and Chemical Biology, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University, P.O. Box 80082, 3508 TB Utrecht, The Netherlands

5.1 Introduction Increasing evidence can be found that describing receptor ligand interactions in terms of a ‘‘lock-and-key’’ model is no longer adequate. Receptors can be regarded as part of a ‘‘molecular machinery’’, in which ligand binding forms a trigger to activate or deactivate the machinery. According to this view, it is no longer sufficient to know how the key fits into the lock, but we should also find out the mechanism with which the key opens and closes the lock. In other words, in drug design we would be interested not only in the affinity of the ligand for the receptor, but also in the changes of a biological receptor molecule when it forms a complex with a ligand. Such changes may involve conformational adaptation, changes in solvation (i.e. ordering of water molecules) and changes in molecular flexibility. Kinetics is a rather underdeveloped aspect of ligand–receptor interactions. It is readily conceivable that in some cases, such as in dynamic regulation of signal transduction processes, kinetic control prevails rather than affinity control. Rapid onset of formation and an optimum lifetime of the complex can be fine tuned by appropriate association and dissociation kinetics. Explicit references to the biological significance of binding kinetics are rather scarce; some examples are given by Schreiber [1]. Other examples include the serial triggering 123

124

Chapter 5

of T-cell receptors [2] and the activation of the epidermal growth factor receptor ErbB-1 [3]. Elucidation of the molecular architecture, using especially X-ray and NMR techniques has been of crucial importance for understanding how a ligand– protein or protein–protein interaction functions in the molecular machinery. However, for a more complete understanding of the dynamic processes underlying receptor activation, kinetic and thermodynamic studies of ligand– receptor interactions are needed. It is increasingly acknowledged that, to fully appreciate relevant molecular properties of potential drug candidates in a drug design process, there is a need for thermodynamic and kinetic studies [4–8]. Traditionally, van’t Hoff analysis has been used for thermodynamic studies. More recently, the use of sensitive calorimetric techniques in drug research is emerging [9,10]. Stopped-flow has been the method of choice to study kinetics of molecular interactions. With SPR one now can derive kinetic and thermodynamic parameters from a single set of experiments. SPR allows to follow the mass change on the sensor surface in real time, yielding affinity and kinetic data. Thermodynamic and kinetic parameters can be derived from a series of experiments in a temperature range. The combination of kinetic and thermodynamic information from well-designed SPR experiments is unique and offers an added value compared with separate techniques for kinetic and thermodynamic information: it allows even a full transition state analysis of the binding process from a single data set. In this chapter, we describe examples of thermodynamic and kinetic analysis of biomolecular interactions using SPR-based approaches that we have developed in recent years. These examples apply mainly to peptides, interacting monovalently or bivalently with important signal transduction proteins, containing Src Homology 2 (SH2) and SH3 domains. These signal transduction proteins are attractive targets for drug design. The underlying investigations are aimed at validation of the SPR-based approach, at gaining insight into the mechanism of the binding process and finally at using this insight in ligand design. To be able to exploit SPR fully, a few initial problems had to be solved. Part of these problems originated from the fact that in our investigations cuvettebased SPR instruments were used. As discussed in Chapters 3 and 4, in flowbased SPR instruments (e.g. Biacore), a continuous flow of the sample enhances diffusion of analyte towards the sensor surface. In cuvette-based instruments, the hydrodynamic properties are controlled by constant agitation of the bulk solution in contact with the sensor surface. The cuvette-based design offers the following advantages: (1) open architecture allowing manual interventions during a run and (2) long association times without extensive consumption of often precious biological material. A disadvantage might be that during the binding process the concentration of unbound analyte in the cuvette is not constant. In this chapter, a correction for this effect is described. Another complication associated with the cuvette design is that in the dissociation phase the analyte released is not removed from the bulk solution. This problem has been solved by adding competing ligand to prevent rebinding of released analyte during the dissociation phase.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

125

When the association rates are high compared with the diffusion in the bulk solution, mass transport limitation (MTL) occurs.1 Association and dissociation are affected by MTL to the same extent. We describe a simple method to estimate the extent of MTL. As MTL can be easily included in a simple kinetic model, the experimental association curves can be analyzed. Another problem, not related to instrumental design, is that in principle the affinity of the analyte for the immobilized ligand at the sensor surface, as obtained in a direct SPR assay, is not necessarily identical with that in solution. A method is described to obtain thermodynamic binding constants for the interaction in solution, using competition experiments with a concentration range of the ligand of interest. Using this approach, several ligands can be studied using the same sensor surface. We should emphasize that in this chapter the focus is not so much on theory, but rather on application. We would like to give the reader practical tools to obtain reliable kinetic and thermodynamic parameters on the binding processes. In order to achieve this, we need to provide the corresponding conceptual and theoretical background. Following the outlined approach, reliable kinetic and thermodynamic parameters can be obtained, which can greatly increase our knowledge of binding processes. Later in this chapter we show examples of how kinetic and thermodynamic analysis of interactions using SPR can support chemical biology studies in general and rational structure-based drug design in particular.

5.2 Affinity and Kinetics of a Transport-limited Bimolecular2 Interaction at the Sensor Surface In a standard SPR assay, one of the interacting partners (the ligand) is immobilized on the sensor surface. The other component (the analyte) is added in the solution, in our case in a cuvette. In our experiments, the ligand is usually a peptide provided with a linker, to increase the distance between the binding epitope and the matrix on the sensor surface, avoiding steric hindrance between the bound analyte and the sensor matrix (see Figure 5.1). The linker is also provided with a free NH2 terminus, for covalent coupling to the sensor surface using EDC/NHS chemistry.3 This system guarantees a homogeneous surface by uniform coupling of the ligand through the NH2 group. The analyte is a protein with generally a much higher molecular weight than the ligand. This increases the sensitivity of the assay, as the change in SPR angle is proportional to the amount of bound mass. Hydrogel SPR sensor chips are used, containing carboxymethylated dextran chains on a 50 nm gold surface (Figure 5.1), either from Biacore (Uppsala, Sweden) or Xantec (Mu¨nster, Germany). Negatively charged ligands, e.g. peptides 1

For a more detailed treatment of mass transport limitation and diffusion, see Chapter 4, Section 4.2.1. 2 A bimolecular interaction is a biomolecular interaction of only one analyte (A) which binds with one ligand (B) to form complex (AB). 3 For further details, see Chapter 7.

126

Figure 5.1

Chapter 5

Schematic view showing (a) the SPR sensor matrix existing of dextran chains with carboxymethyl groups on a gold surface (50 nm) before coupling, (b) immobilization of the ligand after coupling and (c) binding of analyte to the surface.

containing phosphotyrosines (pY), are more difficult to immobilize, due to lack of preconcentration at the sensor matrix, caused by electrostatic repulsion between the negatively charged peptide and the negatively charged carboxyl groups on the dextran chains. In such cases 1 mol l
1 NaCl is added to the coupling buffer to diminish electrostatic repulsion [11]. Our SPR instruments (IBIS and Autolab ESPRIT) have two cuvette cells: a sample cell and a reference cell. The two cells are treated in an identical way, the only difference being that only the sample cell contains immobilized ligands. The net SPR signal (the signal in the reference cell subtracted from the signal in the sample cell) is used for further analysis. Subtraction of the reference signal allows correction for bulk effects due to addition of the analyte, for transient temperature effects and for non-specific binding which occurs incidentally. In a series of experiments at different analyte concentrations, the affinity of the analyte for the immobilized ligand can be assayed in several ways. The method preferred by us is non-linear fitting of the SPR signal at equilibrium with a Langmuir binding isotherm. Alternative methods are based on the kinetics of the interaction. These methods for determining the affinity of the analyte will be described in the following sections.

5.2.1 Affinity Constants Derived from Equilibrium SPR Signals For a simple bimolecular interaction with molecules A and B forming the complex AB, the equilibrium association constant KA and dissociation constant KD are given by eqs. (5.1a) and (5.1b): KA ¼

½AB ; with KA in l mol
1 ½A½B

ð5:1aÞ

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

KD ¼

½A½B 1 ; with KD ¼ in mol l
1 ½AB KA

127

ð5:1bÞ

where the brackets [A], etc., indicate concentration of the molecules. In a welldesigned affinity experiment, several analyte concentrations are used, which should be in a range around the KD value. In SPR experiments, [AB] and [B] are not approached as concentrations in solution, but as amounts at the surface expressed as SPR signal. The amount of complex AB is proportional to the shift in SPR angle [expressed in millidegrees (m1) or so-called ‘‘response units’’ (RU)]. A conversion factor can be calculated for SPR response to concentrations in the volume of the 100 nm dextran layer at the sensor surface (see, e.g., Box 5.1). The shift in SPR angle is recorded as function of time in a sensorgram. In Figure 5.2, an example of sensorgrams, based on the net SPR signal (Rsample cell – Rreference cell), at different analyte concentrations is shown. Using the kinetic evaluation software supplied with SPR instruments, the shift in SPR angle at equilibrium (Req) is readily determined (see Section 5.2.2.1). Frequently, the data are represented in a Scatchard plot (Req/[analyte] vs. Req) as a straight line. However, large errors can occur in Scatchard plots, especially at low concentrations, with small amounts of binding [13], therefore we prefer non-linear regression using the Langmuir binding isotherm [eq. (5.2)], in which [A] is the free analyte concentration and Bmax is the maximum binding capacity in m1, when all binding sites on the sensor surface are occupied.
 ½A Req ¼ ð5:2Þ Bmax ½A þ KD Examples of plots with fits according to the Langmuir binding isotherm are shown in Figure 5.3.

Figure 5.2

Sensorgrams (net signal) of binding of v-Src SH2 protein to immobilized EPQpYEEIPIYL-peptide. Start of dissociation is indicated by the arrow. v-Src SH2 concentrations form top to bottom: 500, 333, 208, 125, 83.3 and 62.5 nmol l
1. For further details, see ref. [12].

128

Figure 5.3

5.2.1.1

Chapter 5

SPR signal at equilibrium as function of analyte concentration. The lines are the fits with the Langmuir binding isotherm [eq. (5.2)]. Data without depletion correction, open circles; with depletion correction (see Section 5.2.1.1), closed circles. (A) Binding of v-Src SH2 domain (conditions as in Figure 5.2). (B) Binding of Syk kinase tandem SH2 domain. For further details on this interaction, see ref. [14].

Correction for Depletion of Free Analyte Concentration in the Cuvette

In a cuvette, the free analyte concentration decreases due to binding to the sensor. This section describes how depletion of analyte can be quantified and corrected for. The change in SPR angle (in m1) due to a binding process is directly related to the amount4 of bound material per mm2. Under equilibrium conditions the amount of bound analyte is proportional to Req (in m1) and the surface of the sensor S (in mm2) in contact with the bulk solution. To relate the amount of bound analyte to a decrease in the free analyte concentration, the molecular weight (MW) of the analyte and the volume of the bulk solution (Vbulk, in liters) must be known. The depletion correction can be calculated using eq. (5.3). ½Afree ¼ ½A0


Req S  10
9 122  MW  Vbulk

ð5:3Þ

Here [A]0 is the initially added analyte concentration in the bulk and [A]free is the corrected analyte concentration, both in nmol l
1. Two examples of depletion correction are presented in Figure 5.3 in (A) for v-Src SH2 with molecular weight 12 300 Da and in (B) Syk kinase tandem SH2

4

For the IBIS and Autolab ESPRIT instruments used by us, an SPR signal of 122 m1 corresponds to 1 ng mm
2 at 25 1C.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

129

(Syk tSH2) with MW is 29 800 Da. With eq. (5.3), using values for S (6 mm2) and Vbulk (35 ml, the applied volume), the depletion correction is calculated as 0.114 nmol l
1/m1 for v-Src SH2 protein and 0.0472 nmol l
1/m1 for Syk tSH2 protein. It is obvious from Figure 5.3 that for Syk tSH2 the depletion correction has a larger effect: without correction KD is 8.7 nmol l
1 and with correction KD is 5.9 nmol l
1. Although the correction factor for v-Src protein is larger due to the lower molecular weight, the depletion correction has only a limited effect in this case: KD goes from 294 nmol l
1 without correction to 252 nmol l
1 with correction. For Syk tSH2 the effect of correction on KD is much larger. Owing to the high affinity, the Syk tSH2 concentration used in the assay is very low (Figure 5.3). Depletion caused by binding to the sensor surface has a large effect at low concentrations. Another factor with direct effect on depletion correction is the binding capacity Bmax of the sensor surface, as it is directly proportional to Req [see eq. (5.2)]. To minimize depletion and the need for correction, a low binding capacity is advised. In general, a value of Bmax of 100 m1 (or 500 RU in Biacore systems) is more than sufficient for reliable assays.5 In equilibrium affinity assays using, e.g., the Langmuir binding isotherm, the depletion correction can be readily calculated by entering the proper numbers in eq. (5.3) and by using a spreadsheet, the correction can be automatically calculated for every data point. Problems may arise when the sensorgrams are used for kinetic analysis. If the depletion correction is large, the free analyte concentration will substantially diminish during the association phase and second order kinetics might apply [15]. In our experience, as long as the depletion correction is below 10% of the total analyte concentration, firstorder kinetics can be safely used [11]. From eq. (5.3) it can be concluded that if Bmax is below 100 m1, for medium strong interactions (KD E 100 nmol l
1) and analyte molecular weights higher than 10 kDa, depletion corrections will be smaller than 10%. In kinetic analysis of high-affinity interactions as in the case of Syk tSH2 (see Section 5.4.1), one should be aware of the occurrence of second-order kinetics. In these systems, reliable kinetic analysis is possible based on first-order association kinetics, on surfaces with low Bmax (B50 m1) and only at higher concentrations, such that depletion correction remains below B10%.

5.2.2 Affinity Constants and Rate Constants Derived from Kinetic Analysis In the previous section we focused on equilibrium affinity assays based on Req. Alternatives are offered by kinetic analysis based on the shape of the sensorgrams, which can be useful when the association rate is slow. Especially in flowbased SPR instruments lengthy association times to reach equilibrium may cause problems due to large analyte consumption. 5

Or even lower, depending on the sensitivity of the instrument.

130

5.2.2.1

Chapter 5

kobs Kinetic Analysis

Assuming a simple bimolecular interaction with analyte A interacting with immobilized ligand B, forming the complex AB at the sensor surface, ideally the SPR signal vs. time (Rt) is given by eq. (5.4) [16].

Rt ¼

  kon ½ABmax  1
e
ðkon ½Aþkoff Þt kon ½A þ koff

ð5:4Þ

Here kon and koff are the association and dissociation rate constants for formation and dissociation of the complex AB, respectively. Note that Rt is proportional to the amount of complex AB. A new parameter kobs6 is defined as kobs ¼ kon[A] + koff. Using software that is generally supplied with SPR instruments, a fit of the curve of Rt vs. time yields kobs. When kobs is plotted vs. [A], kon can be obtained from the slope of the curve and koff from the intercept. The affinity can be calculated as KA ¼ kon/koff or KD ¼ koff/kon. In Figure 5.4, examples of kobs analysis are shown, using the data sets of Figure 5.3. The parameters of the kobs analysis for v-Src SH2 are included in Table 5.1. Although the plots are linear, as expected from theory, the results deviate from the equilibrium analysis. Now for v-Src SH2 a KD value of 11.9 nmol l
1 is found, which is almost two orders higher than obtained from the Langmuir binding isotherm. For Syk tSH2 no affinity could be determined using this analysis because the intercept is slightly below zero. The reason for these deviations is that these interactions are severely affected by mass transport limitation (MTL), as appears in Sections 5.2.2.2 and 5.3. Under such conditions, eq. (5.4), which forms the base for this analysis, no longer holds. Schuck and Minton [17] showed with theoretical data how MTL influences the kobs vs. [A] plot. Further examples of how MTL affects the outcome of kobs analysis can be found in the literature [17,18]. As shown above, a straight kobs plot by no means indicates that reliable kinetic data can be derived. Unfortunately, a number of examples of erroneous interpretations of kinetic data can be found in the literature, especially using kobs-analysis or closely related methods. To avoid this pitfall, one should be absolutely sure that MTL is not involved. A number of simple selfconsistency tests, e.g. comparing data from equilibrium and kinetic analyses, should be performed before interpreting such kinetic analysis [20]. A simple experiment to test whether MTL is involved is to measure the effect of addition of binding ligand during analyte dissociation to prevent rebinding (see Section 5.3.2.2). The kobs analysis presented in the previous section has severe shortcomings as the outcome is very sensitive to MTL. In the following, an alternative model is offered which also includes MTL. This approach is based on analysis of sensorgrams and curve fitting according to predefined binding models. 6

In earlier publications regarding kinetic evaluations [16], this parameter is denoted ks.

131

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.4

kobs plots for binding of v-Src SH2 (closed circles) and Syk tSH2 (open circles) to immobilized ligands. Analysis of datasets presented in Figure 5.3.

Table 5.1

Kinetic and affinity parameters for v-Src SH2 protein binding to immobilized EPQpYEEIPIYL-peptide (experimental data shown in Figure 5.2), as derived with different approaches (see text). CLAMP global analysis

Method parameter

Binding isotherm

kobs analysis

Model 1

Model 2

KD (nmol l
1) Bmax (m1) kon (l mol
1 s
1) koff (s
1) Lm (m s
1)c Res Ssqd

252  13 323  8 – – – –

11.9a – 9.2 (0.3)  104 1.1 (0.8)  10
3 – –

308a 363  3 3.35  104 0.01 – 4.19

250a 323 7.99  106 2b 6.3  10
6 2.00

a

Calculated from koff/kon. Experimental value from dissociation in the presence of peptide to prevent rebinding. Calculated from ktr with conversion factor (see Box 5.1). d Residual sum of squares [19], indicates quality of the fit. b c

5.2.2.2

Global Kinetic Analysis with a Simple Bimolecular Binding Model

The real-time information on the mass changes resulting from the interaction can be used to study various binding models, also including MTL. In this

132

Chapter 5

1: Bimolecular model A+ B

2: Bimolecular model + transport step

k on k off

A0

AB

ktr k − tr

A+ B

Scheme 5.1

A k on k off

AB

Binding models for a simple bimolecular reaction (1) and a bimolecular reaction including a transport step of analyte from the bulk to the sensor surface (2).

section, the emphasis is on the rate constants of a simple, transport-limited bimolecular reaction, more complicated binding models are presented in Section 5.4. In this chapter, the kinetic analysis is treated using basic chemical kinetics concepts applied to experimental SPR data. In Chapter 4, kinetics is treated with concepts based on physical solute absorption to surfaces.7 In general, differential rate equations for species binding to the sensor surface can be derived from a binding model. Experimental sensorgrams can be fitted to a model and one can analyze how far the experimental data agree with the model. Furthermore, parameters such as rate constants and maximum binding capacity can be derived from the fits. In a global analysis such fit procedures are applied to several curves obtained at different analyte concentrations simultaneously, using the same fit parameters. To explain the procedure we use a simple bimolecular binding model with and without mass transport step (see Scheme 5.1). For model 1, the following differential rate equations can be derived: Association :

d½AB ¼ kon ½A½B
koff ½AB dt

ð5:5aÞ


d½AB ¼ koff ½AB dt

ð5:5bÞ

Dissociation :

The analytical solution for the association phase is eq. (5.4) and for dissociation it is eq. (5.6), where [AB] is directly proportional to Rt, the net SPR signal at time t, and Req the net SPR signal at dissociation time zero. Equation (5.6) describes first-order exponential decay from which koff can be obtained. Dissociation : Rt ¼ Req e
koff t

ð5:6Þ

Model 2 is somewhat more complicated: again in the association phase the time dependence of [AB] is given by the differential rate equation [eq. (5.5a)], but now also the time dependence of diffusion of analyte from the bulk ([A]0) to the sensor surface ([A]) has to be taken into account. The accompanying rate equations are given in eqs. (5.7a) and (5.7b). d½A0 ¼
ktr ½A0 þ ktr ½A dt 7

ð5:7aÞ

Remark: this is also the reason why terminology, symbols, etc., differ in Chapters 4 and 5.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.5

133

Global analysis using the program CLAMP of the association phase of binding of v-Src SH2 protein to a pY-containing peptide (data set as in Figure 5.2). Black lines, experimental curves, red lines, fitted curves. (A) Fit with bimolecular model 1, all fit parameters are left free. (B) Simulation of model 1 with fixed parameters for Bmax (320 m1), kon (8  106 l mol
1 s
1) and koff (2 s
1). (C) Fit with transport model 2, with koff fixed on experimental value. See text for further details.

d½A ¼ ktr ½A0
ktr ½A
kon ½A½B þ koff ½AB dt

ð5:7bÞ

Several programs can be used for solving such differential equations by numerical integration. We used the program CLAMP developed for fitting experimental sensorgrams [19].8 Consistency of the fits is greatly improved by fitting several curves for different analyte concentrations simultaneously with the same kinetic parameters in a so-called global analysis. Examples of global kinetic analysis with CLAMP are shown in Figure 5.5, with the data set of Figure 5.2. To emphasize the kinetic phase, only a relatively short association time interval was allowed. The quality of the fits compared with the experimental data is indicated by the residual sum of squares (res Ssq) parameter [19]. For models 1 and 2 this is rather similar (Table 5.1). However, the initial linear phase observed for the higher concentrations is not very well fitted with model 1, and the fit returns a koff value of 0.01 s
1. This linear phase is indicative for MTL [21]. From experiments in the presence of competing peptide to prevent rebinding of protein during the dissociation phase (see Section 5.3.2.2), a much higher experimental value of 2 s
1 for koff is obtained. Therefore, we conclude that this model does not yield a satisfactory description of the kinetic parameters. The experimental values of KD and Bmax are known from the binding isotherm (Figure 5.3), koff is known from dissociation experiments and kon can be calculated from koff/KD. Therefore, all parameters in model 1 are known, allowing simulation of the sensorgrams based on model 1 (Figure 5.5B). This simulation demonstrates that in practice the association phase proceeds much slower than expected for the high on-rate of 8  106 l mol
1 s
1. This 8

Currently, the features of CLAMP are included in a more extended biosensor data analysis tool named Scrubber2 from the results of David Myszka (see http://www.cores.utah.edu/interaction/ software.html).

134

Chapter 5

underscores MTL: due to the high on-rate, diffusion of analyte to the sensor surface is much slower and becomes rate limiting. In model 2, a step for transport of analyte from the bulk to the sensor surface is included. This model, using the fixed experimental koff value of 2 s
1, gives an excellent fit to the experimental data. The diffusion of analyte to and from the sensor surface is assumed to be equal and is characterized by the rate constant ktr. From eq. (5.7b) it follows that the units of ktr obtained from the CLAMP fit are m1 s
1 l mol
1, as [B] and [AB] are in m1 and the fitted curves are SPR signal (m1) vs. time (s). The value of ktr in m1 s
1 l mol
1 can be converted to the mass transport coefficient [21] (Lm) in m s
1 (see Box 5.1). For v-Src SH2 protein (MW ¼ 12.3 kDa) this conversion factor is 6.66  10
13; applying this conversion yields Lm in m s
1. In Table 5.1, the affinity and kinetic parameters derived from global analysis of the dataset of Figure 5.2, using models 1 and 2, are included. The results illustrate once more that in this severely transport-limited system the outcome of kobs analysis is not reliable. The affinity from MTL model 2 agrees perfectly with that from equilibrium analysis using the binding isotherm (Table 5.1). This is not surprising, as in a considerable part of the fitted curves the signal is at equilibrium (see Figure 5.5C). In model 2, the use of an experimental value for koff is crucial for the outcome. In general, it helps to use in the fits fixed experimental values for, e.g., koff and Bmax.

Box 5.1 Conversion of ktr (in m1 s
1 l mol
1) into the mass transfer coefficient Lm (in m s
1) To calculate Lm two conversions have to be applied: (1) from m1 to mol m
2 and (2) from l mol
1 to m3 mol
1. 1. The SPR signal in m1 corresponds to a fixed amount of material/surface unit. For the IBIS and Autolab ESPRIT instruments used in these studies, 122 m1 corresponds to 1 ng mm
2 or 10
3 g m
2. Taking into account the molecular
3 weight (MW) of the analyte, 1 m1 corresponds to 12210MW mol m
2 . 2. 1 l mol
1 is 1 dm3 mol
1; this corresponds to 10
3 m3 mol
1. Combining 1 and 10
6 2, the conversion factor from m1 s
1 l mol
1 to m s
1 is 122 MW. For v-Src SH2 protein with an MW of 12.3 kDa, the conversion factor is 6.66  10
13. From the fit with model 2, ktr is found to be 9.49  106 m1 s
1 l mol
1. Applying the conversion factor, this corresponds to Lm ¼ 6.3  10
6 m s
1. In Biacore instruments, the SPR signal is expressed in response units (RU); 1 RU corresponds to 1 pg mm
2. As 122 m1 corresponds to 1 ng mm
2 (see above), 1000 RU corresponds to 122 m1, and 1 m1 is 8.2 RU. In this chapter, calculations are based on m1. By using the conversion factor of 8.2, these calculations can be adapted for RU-values.

It is surprising that the experimental data can be described by such relatively simple models. For example, usually not all binding sites are equal: within the

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

135

dextran layer of the sensor the binding sites more close to the gold surface have a higher intrinsic contribution to the SPR signal, due to the exponentially decaying evanescent field (Chapter 2). Furthermore, especially on high binding capacity surfaces approaching saturation of binding, heterogeneity of binding sites is expected. The global kinetic analysis presented here is attractive because it yields thermodynamic and kinetic parameters. However, one should be careful in the interpretation of kinetic parameters as in the fit procedure these can be mutually correlated [22] and in more complicated binding models the separate steps may not be completely kinetically resolved, as described in Section 5.4.

5.3 Detecting Mass Transport Limitation: A Practical Approach Kinetic and affinity analysis with simple models can lead to large errors when MTL is unaccounted for, as shown in Section 5.2. Therefore, it is necessary to detect MTL. In this section, practical methods are described to find transport limitation.9 Furthermore, we describe here two approaches to obtain true off-rates10 from severely MTL-affected dissociation phases.

5.3.1 Effect of Viscosity Change on the Association Phase The essence of MTL is that the on-rate is high and diffusion of analyte from the bulk phase to the biosensor (and partly in the biosensor dextran layer [23]) becomes rate limiting. Viscosity changes of the bulk solution will affect diffusion of the analyte and this should be visible in the sensorgrams of an MTL-controlled interaction. We performed experiments with increasing amounts of glycerol to increase the viscosity. In Figure 5.6, the effect of glycerol on the association of the GST fusion protein of the Lck SH2 domain to immobilized pY-peptide is shown. As expected, increasing the viscosity slows down the association. No effect of glycerol in the applied amounts on the affinity was observed (equilibrium signal not affected). Kinetic analysis of data sets obtained with a range of glycerol concentrations, using model 2 (Scheme 5.1), yields a series of ktr values and Lm transport coefficients. For flow cells, the flux to the sensor surface due to mass transfer (i.e. Lm) was derived to be proportional to D2/3 [24]. The diffusion coefficient D is reciprocally related to the viscosity Z, according to the Stokes– Einstein equation, and therefore Lm should be proportional to Z
2/3. From Figure 5.7, it appears that a plot of Lm vs. Z
2/3 yields a linear relation as predicted for flow systems. For a cuvette instrument, the hydrodynamics might be different, as the bulk solution is subject to continuous agitation. Actually, with this data range it is not possible to discern the flow model from other models, as a plot of Lm vs. Z also has an excellent linear correlation. 9

For a more basic treatment of mass transport phenomena, see Chapter 4. From theoretical considerations by Schuck and Minton [17], it follows that MTL affects association and dissociation to the same extent.

10

136

Chapter 5

Figure 5.6

Effect of glycerol on the association phase of 30 nmol l
1 Lck SH2 GST fusion protein to immobilized Ahx-EPQpYEEIPIYL-peptide. Solid line, no glycerol; dashed line, 5% glycerol; dot-dashed line, 7.5% glycerol. Reprinted from ref. [11], Copyright (2000), with permission from Elsevier.

Figure 5.7

Relation between mass transport coefficient (Lm) from bulk solution to the sensor surface and viscosity (Z) as predicted for the hydrodynamics in a flow cell. Determined in a cuvette based instrument for binding of Lck SH2 GST fusion protein to immobilized EPQpYEEIPIYL-peptide in the presence of 0 to 10% glycerol.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

137

Attempts to correlate Lm values with molecular weight have not been successful. This is probably caused by differences among individual sensors surfaces and the fact that Lm depends not only on diffusion in the bulk solution, but also on diffusion within the sensor surface dextran layer as proposed by Schuck [23].

5.3.2 Transport Limitation in the Dissociation Phase The high on-rate compared with diffusion also affects the apparent dissociation kinetics. If diffusion is slow and the on-rate is high, a considerable amount of dissociated analyte will rebind before there is an opportunity to diffuse away from the sensor surface into the bulk. This implies that if the association phase is transport limited, the dissociation is also transport limited. In cuvette instruments used in a static mode (i.e. released analyte is not removed from the cell), rebinding will always occur due to the equilibrium between free released analyte in the bulk and bound analyte on the sensor surface, even under conditions that transport limitation does not apply! We present two independent methods to assay true dissociation rates, which gave comparable koff values. The first method takes rebinding into account and the second uses added competing ligands during dissociation to prevent rebinding.

5.3.2.1

Rebinding Model for Transport-limited Dissociation

If transport limitation applies, the dissociation phase for a simple bimolecular interaction on a homogenous surface will no longer be described by first-order decay kinetics according to eq. (5.6). A high on-rate compared with diffusion away from the biosensor compartment and the availability of free binding sites on the surface will increase rebinding of released analyte. Schuck and Minton [17] have developed a two-compartment model as an approximate description for the dissociation phase under flow conditions. This model is characterized by the differential eq. (5.8). dRt
koff Rt ¼ kon dt 1 þ ktr ðBmax
Rt Þ

ð5:8Þ

In this model ktr (in m1 s
1 l mol
1) has the same meaning as in model 2 (Scheme 5.1) and represents transport between the bulk and the sensor surface. If ktr c kon no transport limitation will occur and eq. (5.8) then changes to eq. (5.5b). (Bmax – Rt) represents the amount of free binding sites and will be proportional to the amount of rebinding. An example of application of this model using the program REBIND11 is shown in Figure 5.8. Continuous wash steps were performed to remove released analyte. Initially, dissociation proceeds fast (Figure 5.8A), as at the start of the dissociation practically all binding sites are occupied and rebinding is negligible. Soon more binding sites become available, slowing dissociation due to 11

Kindly provided by Dr. Peter Schuck.

138

Figure 5.8

Chapter 5

Sensorgrams of binding of Lck SH2 GST fusion protein to immobilized EPQpYEEIPIYL peptide. (A) Solid line, dissociation without renewal of bulk solution; dashed line, dissociation with continuous wash steps to remove released protein from the cuvette. (B) Upper lines: dotted line, experimental data for dissociation with continuous wash steps; continuous line, fit of the data with the program Rebind based on differential eq. (5.8). Below: residuals of the fit. Reprinted from ref. [11], Copyright (2000), with permission from Elsevier.

rebinding. Continuous renewal of the bulk solution increases the apparent dissociation rate. The dissociation phase with the wash steps is excellently matched by the model in eq. (5.8) (Figure 5.8B) with Bmax fixed at the experimental value derived from the binding isotherm [eq. (5.2)]. Using the sum of squared residuals (SSR) analysis of REBIND, a large interval of 0.06 o koff o 0.95 s
1 falls within 5% of the best SSR value (see Figure 5.9, lower curve). In this interval, koff appears to be strongly correlated with kon/ktr. This especially occurs if (kon/koff)(Bmax
Rt) c 1 [see eq. (5.8)], which is the case under conditions of considerable transport limitation (kon and Bmax – Rt are large). If a surface is used with a lower Bmax, effects of transport limitation can be somewhat diminished; however, in a severe transport-limited system, Bmax should be unrealistically low to prevent transport limitation completely (see Section 5.3.3). Introduction of experimental values for kon and ktr, as derived from global analysis of the association phase with, e.g., model 2 greatly improves the results. As can be seen in Figure 5.9, the SSR analysis indicates a discrete value for koff if kon/ktr is kept at the value from the association phase. The obtained koff-value of 0.38 s
1 is close to the value found for the same interaction (0.6 s
1), in the presence of large amounts of competing peptide to avoid rebinding (see Section 5.3.2.2.). The found value is also in accordance with the rapid dissociation found for other SH2 domains [25]. For several reasons, this result is impressive. First, the slow decay in the dissociation phase in Figure 5.8 in no way suggests such a high koff (see also Figure 5.10). The results confirm that ignoring transport limitation can yield

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.9

139

Sum of squared residuals (SSR) as a function of koff. Broken line, analysis with kon/ktr derived from global kinetic analysis of the association phase based on model 2. Reprinted from ref. [11], Copyright (2000), with permission from Elsevier.

several orders too low koff values, as can be seen in Table 5.1. Second, the fact that the transport parameter, ktr, derived independently from the association phase, gives consistent results in the dissociation phase is a solid experimental confirmation that both the association phase and the dissociation phase are affected to the same extent by transport limitation, as concluded from theoretical considerations [17]. In practice, it appears that dissociation curves that can be fitted well with eq. (5.8) can also be well fitted by a double exponential dissociation equation for two independent binding sites with each their own dissociation rate [17]. In many cases such a fit will result in an artifact and the obtained rates are not meaningful. Before concluding from a double exponential fit that two different binding sites are involved, a simple consistency check should be performed, e.g. by adding competing peptide to diminish/prevent rebinding (see Section 5.3.2.2).

5.3.2.2

Competing Ligand to Prevent Rebinding During Dissociation

As indicated in the previous section, under transport-limited conditions the dissociation phase is considerably influenced by rebinding of released analyte. In principle, this rebinding can be prevented by adding an excess of competing ligand with high affinity for the analyte. In Figure 5.10, examples are shown of dissociation in the presence of increasing amounts of competing ligand for a monovalent Lck protein and bivalent binding Syk protein.

140

Figure 5.10

Chapter 5

Effect of different concentrations of competing peptide ligand on the dissociation rate. (A) 200 nmol l
1 Lck SH2 GST fusion protein with EPQpYEEIPIYL peptide and (B) 5 nmol l
1 Syk tandem SH2 domain with g-ITAM peptide. Reprinted from ref. [11], Copyright (2000), with permission from Elsevier (A) and from ref. [14], with permission from Wiley-VCH (B).

The effect of the ligands is dramatic and illustrates that for a transport-limited interaction, the off-rate can be several orders larger than expected from the dissociation phase without competing ligand. Although the affinity of these proteins for the ligands is rather high, relatively high concentrations are needed to prevent rebinding completely. In control experiments with high concentrations (4200 mmol l
1) of non-binding peptides, the dissociation rate is at maximum, only a bulk effect, a higher baseline is seen, as also occurs in Figure 5.10B, for 4.4  10
4 mol l
1 ITAM peptide. At high concentration of binding peptide the dissociation phase approaches a monophasic exponential decay (see Figure 5.11) and the curve can be fitted with eq. (5.6) to derive the off-rate. In both cases the dissociation rate is very high. For the Lck protein, rebinding (Figure 5.11A) still seems not to be completely suppressed in the presence of even 10
4 mol l
1 peptide. Especially at low R values with more free sites on the surface available for rebinding [see model eq. (5.8)], deviation from first-order decay kinetics is observed; however, the first ca. 80% of the decay can be approximated by the exponential function. The half-lifetime is about 1 s and koff is 0.6  0.1 s
1, close to the value obtained from the rebinding model with fixed experimental value for kon/ktr (0.38 s
1; see Section 5.3.2.1). The dissociation of the Syk protein (Figure 5.11B) is also speeded up in the presence of competing peptide and shows monophasic exponential decay, with koff ¼ 0.13  0.02 s
1, close to that of comparable mono- and bivalent interactions involving SH2 domains [25,26]. The high dissociation rate in combination with the high affinity (KD ¼ 5 nmol l
1) is intriguing. As a rule, high-affinity monovalent interactions have slow dissociation rates, hence the complex has long half-lifetimes, e.g. for the avidin–biotin complex it is over 1 week [27]. The binding of a double phosphorylated ITAM-peptide with the tandem SH2

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.11

141

Dissociation of bound proteins in the presence of competing ligands to prevent rebinding. (A) Lck SH2 GST fusion protein in the presence of 100 mmol l
1 peptide. (B) Syk tandem SH2 protein with 220 mmol l
1 of bivalent binding ITAM-peptide. The monophasic-exponential or firstorder decay fits [eq. (5.6)] are indicated by the dotted lines. Reprinted from ref. [11], Copyright (2000), with permission from Elsevier (A).

domain is bivalent, existing of two weak monovalent interactions (see Section 5.4.1 for more structural details). For such multivalent interaction, the dissociation rate approaches that of a single (much weaker) monovalent interaction contributing to the bivalent binding, when competing (monovalent) ligand is present [28]. Therefore, in the presence of proper ligands, multivalent interactions offer the opportunity to combine high affinity with high off-rates and short lifetime of the complex. In signal transduction processes, the combination of high affinity and short half lifetimes might be decisive for specificity and transiency of protein–protein interactions. It is remarkable that multivalent interactions, e.g. involving tandem-SH2 and tandem-SH3 domains, are abundant in signal transduction processes.

5.3.2.3

Experimental Procedure to Assay High Off-rates

Off-rates approaching 1 s
1, as obtained in the previous section, are at the limit of what can be accurately measured by SPR. For really fast decay kinetics we think that the open cuvette structure is an advance as it allows direct accessibility for manual handling. Our protocol developed for assaying rapid dissociation kinetics in cuvette-based instruments starts with setting the instrumental sampling rate high (5 data points s
1). The instrument is operated in one-channel mode. A 25 ml volume of the analyte protein, preferably with a concentration that saturates 490% of the binding sites,12 is pipetted manually into the sample cell. After reaching equilibrium of binding, the measurement is 12

This analyte concentration is approximately 10  KD.

142

Chapter 5

started and very quickly 10 ml of a high-concentration competing peptide is pipetted manually. The required concentration of peptide to prevent rebinding has to be determined experimentally (Figure 5.10). The data points of the resulting sensorgram can be exported, processed in a spreadsheet and fitted to an exponential function. The data points within 1 s after peptide addition are discarded, as these are often affected by distortions. The experiment is repeated at least in triplicate with ample manual washing steps in between.

5.3.3 Quantitative Considerations on Mass Transport Limitation As explained previously, MTL occurs if the reaction (binding) flux is much higher than the transport flux of analyte from the bulk solution to the sensor. These fluxes are described by the transport coefficient Lm and the Onsager coefficient Lr for the reaction flux [21]. A quantitative measure for MTL is expressed in eq. (5.9). MTL ¼

Lr Lm þ Lr

ð5:9Þ

If the analyte transport is totally rate limiting in the binding kinetics (Lr c Lm), MTL will approach 1. Lm is directly related to ktr as defined in model 2 (Scheme 5.1) for the association phase and eq. (5.8) for the dissociation phase. Lm in m s
1 is obtained by conversion of ktr as indicated in Box 5.1 in Section 5.2.2.2. The Onsager coefficient of reaction flux (Lr) in m s
1 is obtained from eq. (5.10) [21]. Lr ¼ kon ½B 3


1
1

ð5:10Þ

13

Here kon is converted to m mol s units. At the start of the interaction [B] equals Bmax, the maximum binding capacity of the sensor surface, i.e. the concentration of free analyte binding sites on the sensor surface. Bmax can be obtained from the Langmuir binding isotherm [eq. (5.2)] or from global kinetic analysis in m1. Using the same approach as explained in Box 5.1 Bmax is converted to [B] in mol m
2 with eq. (5.11). ½B ¼

Bmax 10
3  ðmol m
2 Þ 122 MW

ð5:11Þ

The extent of MTL allows one to consider whether MTL can be avoided by adaptation of the experimental conditions, e.g. by lowering the binding capacity on the sensor surface or increasing the diffusion rate by increasing flow or agitation of the bulk solution in the cell. The global kinetic analysis with model 2 yields Lr ¼ 1.72  10
3 m s
1 and Lm ¼ 6.3  10
6 m s
1 for the interaction with v-Src SH2 as analyte14

13 14

10
3 times kon in l mol
1 s
1. MW ¼ 12.3 kDa.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

143

(see Figure 5.5). Applying eq. (5.9), under these conditions, MTL is practically 1 for this rather high binding capacity surface (Bmax ¼ 320 m1). To reduce MTL to 0.5 (Lr ¼ Lm), the binding capacity Bmax should be reduced to about 1 m1, which is not a feasible assay condition. Increasing flow and agitation will not be sufficient, either. Assuming that a 5-fold increase in Lm can be obtained, the interaction will still be completely transport limited. In practice, the increase in Lm will be rather limited due to hydrodynamics (stagnant layer) and the dimensions of the (flow) cells and because diffusion within the dextran matrix on the sensor is not sensitive for the flow conditions [23]. In general, for an analyte protein of approximately 40 kDa molecular weight, Lm is around 3  106 m s
1. This number can vary somewhat depending on type of sensor chip. Applying eq. (5.9), this implies that for a surface with Bmax ¼ 100 m1 (corresponding to 2  10
8 mol m
2) that Lm 4 Lr if kono2  102 m3 mol
1 s
1, corresponding to kono2  105 l mol
1 s
1. This number agrees very well with predictions from various MTL models [17,29]. The size of the analyte, of course, influences the diffusion rate and the kon value, but as a rule of thumb, transport limitation will affect binding kinetics to a dextran-based sensor surface, if kon is larger than 105 l mol
1 s
1. In summary, lowering the binding capacity and increasing the flow rate can prevent MTL only in the case of slightly transport-influenced kinetics. In practice, we assume that the Lm/Lr ratio can be improved at most by about a factor 5 on changing the experimental conditions. As a consequence, in moderately and severely transport-limited cases an effect of MTL on the kinetics cannot be avoided.

5.3.3.1

Flow or Cuvette?

One can ask whether a flow or cuvette instrument is to be preferred when it comes to transport limitation. Although differences in hydrodynamic behavior may exist between a flow and a cuvette instrument and detailed hydrodynamic models have been derived for flow cells [23,24], in practice, no significant differences in transport fluxes between flow and cuvette systems have been observed. This is illustrated by the agreement of the Lm value of 9.8  10
6 m s
1 for IL-2 (MW 14 kDa) obtained in a flow-instrument (Biacore) using a model similar to model 2 [30] and the value of Lm ¼ 6.5  10
6 m s
1 for the v-Src SH2 domain (MW 12.3 kDa) obtained in a cuvette instrument (Table 5.1). We conclude that in practice no significant differences exist in the extent of MTL between cuvette and flow instruments.

5.4 Global Kinetic Analysis of Complex Binding Models After describing simple bimolecular interactions, we shift to more complex binding mechanisms with conformational changes, dimerization, multicomponent interactions, multivalent binding, etc. In addition to structural

144

Chapter 5

information as derived from NMR and X-ray analysis, kinetic information and insight into the mechanism is valuable for understanding the binding process and is of special interest for rational drug design. We describe examples of applications of global kinetic analysis with more complex models to illustrate this point.

5.4.1 Global Kinetic Analysis Including Mass Transport and a Conformational Change For better understanding of our first example, the bivalent binding of Syk tandem SH2 domain (Syk tSH2) with a surface loaded with ITAM-peptide, we start with the description of the structural aspects. The interaction of an ITAMderived ligand with Syk tSH2 involves bivalent binding of two phosphotyrosine containing sequences on the ITAM-peptide with the two SH2 domains of the Syk protein. This interaction plays an important role in, amongst others, signal transduction of the IgE receptor (FceRI) and the B-cell antigen receptor [31]. An X-ray structure of Syk tSH2 with an ITAM peptide is shown in Figure 5.12. The linker part in the ITAM, between the two phosphotyrosine-containing sequences, hardly interacts with the protein [32]. The two SH2 domains in the

Figure 5.12

X-ray structure of Syk tandem SH2 domain (ribbon) with doubly tyrosine phosphorylated ITAM (sticks). Both phosphotyrosines are indicated in red. PDB entry 1A81 [32]. Reprinted with adaptation from ref. [14], with permission from Wiley-VCH.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.13

Association phase sensorgrams for the interaction of Syk tandem SH2 domain with an immobilized ITAM peptide. Global kinetic analysis of the experimental data using CLAMP, according to the indicated models in red. Reprinted from ref. [14], with permission from Wiley-VCH.

Model 3: transport

Model 4: transport

Model 5: 1PP-bivalency

conformation model

dimer model

model

k

k

A0

k

A+ B

A

A0

k k

AB

k

AB

k

Scheme 5.2

145

AB*

k

A+ B

A+ B

A k k

AB + A

AB

AB + B

k

AB

k k k

AB2

A2 B

Complex binding models used in global kinetic analysis of association phases.

Syk protein (labeled N-SH2 and C-SH2) are connected by a flexible coiled coil linker, giving some flexibility in the inter-SH2 distance. The association phase of this interaction was subjected to a global kinetic analysis (Figure 5.13). The association phase was assayed at different temperatures as part of a complete thermodynamic analysis (see Section 5.6 for thermodynamic analysis based on SPR). At 11 1C, deviation is observed using model 2 (Scheme 5.1) in global analysis in the association phase as when the signal approaches equilibrium. It is conceivable that a bivalent interaction occurs in two discrete steps [29], as indicated in Box 5.2. After initial monovalent binding, the second step involves a conformational (intramolecular) change, leading to a high-affinity bivalently bound complex AB* (model 3, Scheme 5.2). This interaction is certainly transport limited, as we see a strong effect of added ligand on rebinding in the dissociation phase (Figure 5.10B). Also indicative of transport limitation is the initially linear association phase, especially at higher analyte concentrations [21]. Therefore, a transport step is included in model 3.

146

Chapter 5

Box 5.2 Relation of Kb, Kconf and Kobs in the conformation change model Conformation change model 3 consists of two binding steps (see also Scheme 5.2): an initial binding event occurs, characterized by the equilibrium association constant Kb. Second, a structural change in the bound state occurs, characterized by equilibrium constant Kconf, leading to a higher affinity complex. For a bivalent interaction as in the case of Syk t SH2 binding to doubly phosphorylated ITAM peptides, this last step includes a structural arrangement of the complex with a second intramolecular binding event.

Kb and Kconf are defined by eqs. (5.12) and (5.13): Kb ¼

Kconf ¼

½AB1 ½A½B

kconf ½AB2 ¼ k
conf ½AB1

ð5:12Þ

ð5:13Þ

The observed equilibrium association constant Kobs is defined by eq. (5.14): Kobs ¼

½AB1 þ ½AB2 ½A½B

ð5:14Þ

Note that [AB1] + [AB2] corresponds to the total amount of bound analyte, which is proportional to the change in SPR signal R. Substitution of eqs. (5.12) and (5.13) in eq. (5.14) yields Kobs ¼ Kb ð1 þ Kconf Þ

ð5:15Þ

A plot of Req vs. [A] will also for this case obey a binding isotherm fit as demonstrated in Figure 5.3, and from the fit Bmax and Kobs are obtained.

Applying model 3 gives an excellent fit (Figure 5.13B); in the fits, Bmax and koff were kept at experimental values.15 The fit yields the parameters ktr, kon, kconf and k
conf; the values especially of kon, kconf and k
conf may not be reliable as they are strongly correlated (see Table 5.2). According to model 3, Kobs contains contributions of the initial binding step characterized by association constant Kb 15

Bmax is derived from the Langmuir binding isotherm [eq. (5.2)], which also holds for model 3 (see Box 5.2); koff is derived from the experiment shown in Figure 5.11B.

147

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Table 5.2

Parameters and their correlation from calculations used in Figure 5.13B.

Parameter 1 2 3 4


1

Value
1

ktr (m1 s l mol ) kon (l mol
1 s
1) kconf (s
1) k
conf (s
1)

7

5.4  10 1.19  106 2.81 0.16

Correlation 1

Correlation 2

Correlation 3

–
0.74 0.72 0.72

– –
1.000
1.000

– – – 1.000

and of the second intramolecular binding step or conformation change Kconf (see Box 5.2). Notwithstanding uncertainty in the rate constants, a Monte Carlo run16 with CLAMP gives consistent values for Kb (¼ kon/koff) and Kconf (¼ kconf/ k
conf) of 1.9  107 l mol
1 and 18.5, respectively. This value of Kb is significantly higher than the 7.7  105 l mol
1 found for the monovalent binding of monophosphorylated ITAM peptide [33]. It is likely that the two binding steps in model 3 are not completely resolved. First, no obvious biphasic association phase exists in Figure 5.13; second, by increasing the temperature above 30 1C, we obtain an excellent fit using the bimolecular transport model 2 (Figure 5.13C), indicating that the two binding steps can no longer be discerned. Calculation of Kobs from the fit parameters using eq. (5.15) (Box 5.2) yields a value of 3.6  108 l mol
1, in excellent agreement with Kobs obtained from the binding isotherm at 11 1C (3.4  108 l mol
1). Although the values obtained for Kb and Kconf may not be physically meaningful, they can be used to calculate solid Kobs values with eq. (5.15).

5.4.2 Unusual Kinetics: Intermolecular Bivalent Binding to the Sensor Surface A second example from our work where global kinetic analysis plays a central role in elucidating the binding mechanism to a SPR sensor surface is Grb2 protein binding to an immobilized bivalent polyproline (PP) peptide (2PP). 2PP contains two PP binding epitopes derived from the SOS protein17 separated by a short linker moiety. The Grb2 protein exists of two SH3 domains and one SH2 domain connected by two flexible linkers [34]. The two SH3 domains can each bind to one of the PP epitopes of 2PP in a bivalent mode [34,35]. In Figure 5.14, a model of the bivalent complex of Grb2 protein with 2PP is shown. It was expected that the kinetics of this bivalent interaction could be described by a model similar to that used for the Syk tSH2-ITAM interaction (Section 5.4.1); instead, a different binding model emerged. The binding of Grb2 to immobilized 2PP SPR sensor surfaces with different binding capacities is shown in Figure 5.15. The form of the curve describing the association phase 16

In a Monte Carlo run the fit is repeated for a defined number of cycles with variation of the start parameters within a defined range. 17 The SOS-Grb2 interaction plays an important role in the signal transduction cascade of numerous receptors, controlling among others cell proliferation and differentiation, platelet aggregation and T-cell activation [36].

148

Chapter 5

Figure 5.14

Model based on X-ray structure of Grb2 (PDB entry: 1GRI, ribbons) which a double polyproline peptide docked on the SH3 domains (sticks).

Figure 5.15

Sensorgrams of Grb2 protein (range 330–2000 nmol l
1) binding to immobilized 2PP-peptide with various binding capacities (Table 5.3). (A) and (B) net signal (reference cell subtracted from sample cell); (C) data from the sample cell only due to non-specific binding in the reference cell; lowest curve is the baseline not containing Grb2 protein.

appears to be sensitive to the binding capacity: at high binding capacities, the association phase looks conventional with a steady increase until equilibrium is reached. Lowering the binding capacity yields biphasic association, with a very fast initial increase, taking only a few seconds, followed by a slower increase (Figure 5.15B and C). The SPR signals at equilibrium apparently comply with the Langmuir binding isotherm [eq. (5.2)], as shown in Figure 5.16 for medium binding capacity. At high and medium capacity surfaces the binding isotherm

149

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.16

Table 5.3

Equilibrium analysis with binding isotherm of Grb2 protein to a medium high capacity surface of 2PP peptide (data from Figure 5.15B).

Data from binding isotherm of Grb2 binding to 2PP surfaces.

Binding capacity

Bmax (m1)

KD (nmol l
1)

High Medium Low

1520  55 218  3 73  3

535  40 460  30 NDa

a

ND Not Determined (equilibrium not reached).

gave comparable KD values (Table 5.3). At low binding capacities equilibrium is not reached within 20 min (Figure 5.15C). The data from binding of Grb2 to the 2PP surfaces with various binding capacities have been subjected to global kinetic analysis, exploring several models. As shown in Figure 5.17A, the data from very high binding capacity (Figure 5.15A) could be readily fitted with a conformation change model (model 3, Scheme 5.2). Interestingly, the transport step could be omitted from the model, giving identical results. This suggests that in this case the interaction is not transport limited, in agreement with the calculated value of kon ¼ 5.8  104 l mol
1 s
1, which is below the indicated value for transport limitation (Section 5.3.3). At medium high and low capacity surfaces, model 3 shows systematic deviations from the experimental data as shown in Figure 5.17B: the slope of the slow phase in the fits changes much less with the concentration than experimentally observed. X-ray structures suggest that Grb2 dimers can be formed [37,38] and therefore the data were fitted with dimer model 4 (Scheme 5.2), as can be seen in

150

Figure 5.17

Chapter 5

Global kinetic analysis of Grb2 binding to 2PP surfaces with CLAMP. (A) High binding capacity surface, fitted with conformation change model 3. (B) Medium capacity surface, fitted with conformation change model 3. (C) Medium capacity surface, fitted with dimer model 4. (D) Low binding capacity surface, fitted with dimer model 4. Details of these models are given in Scheme 5.2.

Figure 5.17C. The residual sum of squares parameter from the fit with model 4 was 1.71, compared with 2.50 for model 3. Also for the low capacity surface fitting with model 4 gave good results (Figure 5.17D). According to model 4 binding of the first Grb2 molecule (A in the model) to immobilized 2PP (B in the model) facilitates binding of a second Grb2 molecule. However, we doubt that on the surface such physical Grb2 dimer will be formed, because we cannot demonstrate the formation of dimers in solution upon addition of 2PP-peptide to Grb2, either by chemical cross-linking or dynamic light scattering, in line with published results [37]. An alternative to bivalent intramolecular binding is intermolecular bivalent binding (see Scheme 5.3b). In the flexible dextran matrix of the sensor chip, the distance between 2PP epitopes could easily adapt to facilitate intermolecular

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Scheme 5.3

151

Schematic representation of various binding modes of Grb2 protein to 1PP and 2PP surfaces.

binding. If intermolecular bivalent binding occurs, this should be observed for a monovalent 1PP surface (Scheme 5.3c). In Figure 5.18, the association kinetics of 1PP surfaces with low and high binding capacity are shown. The curves can be readily approximated with a bivalent binding model: first monovalent binding of Grb2 protein to immobilized 1PP, followed by bivalent binding to a second free 1PP ligand (model 5, Scheme 5.2). As expected, the rate of the slow bivalent binding step depends on the binding capacity: with a high capacity it will be easier to find a second 1PP ligand

152

Figure 5.18

Chapter 5

Association phase sensorgrams of Grb2 protein binding to a low (A) and high (B) capacity monovalent 1PP surface. Global kinetic analysis with CLAMP according to bivalent binding model 5 (see Scheme 5.2).

and the rate will be higher. Interestingly, the affinity of the initial fast binding step as derived from the kinetic analysis is approximately 12 mmol l
1, which is similar to the affinity of monovalent binding of 1PP to Grb2 in solution [35]. How can the intermolecular bivalent binding mode to the 2PP surface be reconciled with the observed association kinetics which can be well described by the dimer model 4 (Figure 5.17C and D)? We propose a three-step mechanism as outlined in Scheme 5.4: the first step is monovalent (or bivalent) binding of Grb2 to one 2PP ligand; after this, bivalent binding with another 2PP ligand occurs (model 5). This intramolecular binding prepares a perfect second docking site for another Grb2 molecule as the inter-PP distance is already optimal for bivalent binding. This model also explains why the slow phase is much more delayed in case of low binding capacity (Figure 5.17): it is more difficult to find a second partner for divalent binding. The proposed binding model for the 2PP surface cannot be defined in all detail in the CLAMP program, e.g. no bivalent ligands can be defined and no discrimination between single and double occupation of two 2PP ligands can be made. Actually, model 4 is a simplified approximation of the possible binding modes. It is possible that in the dimer complex (step 3, Scheme 5.4) dimeric interactions, as found in the X-ray structure of Grb2 are involved, as the Grb2 molecules are forced to be together. For the monovalent 1PP surface, such dimer formation cannot take place and the kinetics can be adequately described with model 5.

5.4.3 Global Kinetic Analysis: Concluding Remarks At first sight, in both examples in this section we have a protein that binds bivalently to an immobilized ligand. However, the outcome is surprisingly different. This is illustrative for the use of models in global kinetic analysis: one

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Scheme 5.4

153

Sequential steps in the proposed model for binding of Grb2 to the 2PP surface, ultimately leading to the formation of ‘‘dimers’’.

should be very careful in the interpretation of the applied binding model. In complex situations, good fits of the experimental curves can be derived from more than one model. This raises the question of how far the physical reality is reflected by these models. A model will only provide an approximation: not all reactions can be included in detail, as described above. Unless the kinetic steps are well resolved, the calculated kinetic parameters may not be reliable, as they will also be strongly correlated. Hence it is useful to introduce all possible experimental values in the calculations. Other complications may arise from the fact that not all ligands are equally accessible: the sensor with immobilized ligand comprises a three-dimensional

154

Chapter 5

volume. Especially for high binding capacity sensors, partition into this volume may be hampered near saturation of binding due to crowding [23]. With multivalent binding analytes the hydrogel of the dextran on the sensor may become more ‘‘cross-linked’’, leading to a more compact structure during the binding process. As the SPR signal decreases exponentially with distance from the gold surface (Chapter 2), this might lead to an accounted signal increase vs. the model. In spite of all these considerations, sometimes the quality of the fits with complicated binding models can be stunning (Figures 5.17 and 5.18). As a rule, a proposed binding model should be simple and supported by experimental evidence; additionally, it should include all possible fixed experimental parameters. Global kinetic analysis is a unique tool, providing insight into the binding mechanism, the kinetics of an interaction and the role of protein dynamics. It can inspire new ideas for molecular design and drug development, for example, the length and rigidity of the linker between the two phosphotyrosinecontaining binding epitopes in ITAM-mimetic constructs binding to Syk tSH2 [14]. Ample examples exist using the simple bimolecular models 1 or 2; applications of more complicated models are rather scarce. Such examples are the binding of IL-2 to the heterodimeric IL-2 receptor [30], binding to a heterogeneous surface with two different ligands [39] and the kinetic analysis of amyloid fibril elongation [40]. Deviation from the simple 1:1 model is already indicative of a more complex binding mechanism.

5.5 Affinity in Solution Versus Affinity at the Surface In SPR measurements, interactions take place at the sensor surface, which is not always representative of interactions in solution. This is certainly true for divalent analytes, such as antibodies and GST fusion proteins that form dimers and show an avidity effect when binding to a surface [41]. The amount of analyte binding to the sensor surface in the presence of a competing ligand in solution is influenced by the affinity of the analyte for this ligand. If the affinity is high, a relatively large amount of analyte will be in complex with the ligand in solution and only a small amount of analyte will be available for binding to the surface, resulting in a lower shift in SPR angle. Using this model, Morelock et al. developed a method to obtain thermodynamic binding constants in solution [42]. Based on this, we derived a fitting model for data from competition experiments with constant analyte and varying ligand concentrations in solution (Box 5.3) [43]. An example of competition experiments is shown in Figure 5.19. Experiments were performed at various pH values to determine the shift in pKa of the phosphotyrosine upon binding [43]. The equilibrium dissociation constant at the chip (KC) was determined at each pH and these values were used in the fits. The experimental data was fitted with eq. (5.19) (Box 5.3), using experimental values for [A]tot and KC, while the independent variable in the fit is the ligand concentration [B]tot.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

155

Box 5.3 Thermodynamic binding constants for binding in solution In an SPR competition experiment with ligands for the analyte present both on the sensor surface and in solution, the two binding equilibria are as follows: 1. Interaction between analyte A and immobilized ligand B on the sensor chip (Bc), yielding complex ABc on the sensor. The dissociation constant (KC) is KC ¼

½A½Bc ½ABc

ð5:16Þ

[Bc] and [ABc] are in millidegrees; when all Bc sites are occupied [ABc] ¼ Bmax. 2. Interaction in solution between ligand B in solution with A to form complex AB. The dissociation constant (KS) is ½A½B ð5:17Þ ½AB Note that KS is a thermodynamic binding constant. In analogy with eq. (5.2), the amount of binding onto the surface can be described by a binding isotherm:
 ½Z Bmax ð5:18Þ Req ¼ KC þ ½Z KS ¼

where [Z] is the total concentration of analyte [A]tot minus the amount of analyte in the complex AB ([Z] ¼ [A]tot – [AB]), and eq. (5.18) changes to ! ½Atot
½AB   Bmax ð5:19Þ Req ¼ KC þ ½Atot
½AB The amount of complex AB in solution is a function of the affinity in solution (KS) and eq. (5.17) can be rewritten:    ½Atot
½AB ½Btot
½AB KS ¼ ð5:20Þ ½AB From this equation, it appears that [AB] is a quadratic function of the type ax2 + bx + c ¼ 0, for which the solution is given by the square root equation ffi   qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 KS þ ½Atot þ ½Btot
KS þ ½Atot þ ½Btot þ4½Atot ½Btot ð5:21Þ ½AB ¼ 2 Substituting eq. (5.21) for [AB] into eq. (5.19) yields an equation that fits data from competition experiments. Fitting with [A]tot kept at the experimental value and [B]tot as independent variable provides KS and Bmax. Note that eq. (5.19) contains KC. The value of KC is obtained from a separate experiment. Best fit is expected when [A]tot Z KC. Bmax is the maximum binding capacity upon complete saturation, and not the binding capacity in the absence of competing ligand.

156

Figure 5.19

Chapter 5

Data from SPR competition experiments to determine the binding constant KS in solution. The analyte is Lck-SH2 GST fusion protein (50 nmol l
1), the immobilized ligand and the ligand in solution are identical [a phosphotyrosine 11-mer peptide derived from the hamster middle-T-antigen (hmT)]. Experiments were at different pH: from left to right pH 9, 6.8 and 5. The lines are the fits with the substituted eq. (5.19) (see Box 5.3). Reprinted from ref. [43], Copyright (2002), with permission from Elsevier.

In order to verify the reliability of our approach for obtaining affinity data in solution and to see if affinity at the sensor surface is significantly different from that in solution, KC and KS values are compared in Table 5.4. The data illustrate that the affinity of dimer proteins (the GST fusion protein and not-heated Grb2-SH2; see below) at the surface is larger than in solution. This can be explained by the avidity effect, occurring when the dimer binds bivalently to two ligands at the surface. The case of the Grb2-SH2 protein without GST part is interesting. It has been reported that this protein occurs as a dimer.18 Probably this is an artifact due to the expression as a GST fusion protein, which is known to form dimers through the GST part [44]. From size-exclusion chromatography we estimate that our GST-cleaved Grb2-SH2 protein contains B60% dimer. The dimer is metastable and upon heating to 50 1C the monomer is irreversibly formed [38]. Before heating, the affinity to the sensor surface is higher, due to the large amount of dimer. KS is higher before heating, suggesting that the affinity of the ligand for the dimer in solution is lower than for the monomer. For the pYVNV-peptide binding to monomer Grb2-SH2, consistent values for KS are obtained (230–260 nmol l
1), notwithstanding large differences in KC (7.9 for the GST fusion protein to 790 for full-length Grb2) used in the 18

Dimer formation by domain swapping of an a-helix [38].

157

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Table 5.4

Comparison of affinity at the sensor surface (KC) and in solution (KS), calculated from competition experiments using the substituted eq. (5.19) (see Box 5.3).

Protein+peptide

KC (nmol l
1)

KS (nmol l
1)

Lck-SH2 GST fusion proteina + hmT-peptide Grb2-SH2 GST fusion protein + pYVNV-peptide Grb2-SH2 not heatedb + pYVNV-peptide Grb2-SH2 heated + pYVNV-peptide Full length Grb2 protein + pYVNV-peptide Syk tSH2 + g-ITAM-peptide v-Src SH2 + hmT-peptide

6 7.9 134 220 790 5.4 220

60 260 1800 255 230 5.7 234

a b

If explicitly indicated the dimer forming GST moiety is present. Heating to 50 1C converts the Grb2-SH2 dimer irreversibly to the monomer (see text).

calculations to obtain KS. Moreover, the solution affinities agree with literature values obtained using other techniques. This strengthens our confidence in the competition approach. As a rule, the affinity of monomer proteins in solution is the same as at the sensor surface, even for the bivalent binding Syk tSH2! An exception seems to be the binding of the full-length Grb2 protein to the SH2 domain.19 In the case of a bivalent binding analyte with two (identical) binding sites such as an antibody, the expression for [AB] will be different from eq. (5.21). Now we have to take into account that not all occupied antibody remains in solution, as monovalently occupied antibodies are able to bind to the sensor surface. When a certain fraction of all binding sites are occupied, a statistically determined distribution exists over double-bound, single-bound and unbound antibodies in solution. Unbound antibodies, and also a single-bound antibody with a ligand from solution, can bind to the sensor surface. We have adapted the expression for [AB] (Box 5.3) to the statistical distribution [45] and it appears that this correction has only a modest effect on the resulting KS value (o10%). In summary, the approach derived in Box 5.3 cannot be used for every binding model. Especially when the Langmuir binding isotherm is not suitable for fitting Req as a function of analyte concentration, this approach will not be valid. However, for more complicated binding models obeying the Langmuir binding isotherm, such as the two-step model proposed for Syk tSH2 (Box 5.2), reliable KS values can be obtained. In this case KS will be an apparent binding constant, containing the various contributions to Kobs (see Box 5.2). The competition experiments as described in this section are very attractive in drug research: the affinity of a range of potential drug candidates can be assayed at the same surface! In general the standard error in KS is larger than in KC. Processes in solution may not always be representative for processes at sensor surfaces or in biological systems. We are convinced that in some cases interaction at a surface might be a better model than interaction in solution,

19

A possible explanation for this difference is discussed in ref. [12].

158

Chapter 5

especially with multivalent interactions. For example, the Sos-protein20 contains multiple (six) polyproline sequences to recognize Grb2 SH3 domains [36]. Several Grb2 molecules might bind bivalently to these sequences in one Sos molecule in different combinations. For this a surface loaded with polyproline ligands might be a better model than 1:1 interactions in solution.

5.6 Thermodynamic van’t Hoff Analysis Using SPR Data As described in the Introduction, it is no longer opportune to describe ligand– receptor interactions in terms of a rigid lock-and-key concept. Binding of a receptor by a ligand can influence the dynamics, induce allosteric changes of the receptor or, very importantly, have an effect on bound water molecules. All this can be vital for the biological effects in a biomolecular interaction. In this section we will concentrate on SPR-based assay of thermodynamic parameters, to reveal the biomolecular recognition process, to help understand it and to exploit it for improved rational drug design (see Box 5.4) [5,6,8].

5.6.1 van’t Hoff Thermodynamic Analysis van’t Hoff thermodynamic analysis requires the measurement of the affinity at a range of temperatures. As the SPR signal is extremely sensitive to temperature changes, complications may arise when measuring at temperatures deviating from room temperature, as some time may be needed before complete thermal equilibration is reached. An example is shown in Figure 5.20: in the reference cell a temperature effect is observed in addition to a bulk effect. It takes about 100 s to reach thermal equilibrium. The temperature effect is not visible in the net signal (Figure 5.20). The best approach is to use Req for the affinity assay as by the time Req is reached the system will be in thermal equilibrium. It is important, especially when kinetics are assayed, that the sample is at the correct temperature and that the injection system is well thermostated. The design of newer generations of SPR instruments, e.g. Autolab ESPRIT, is optimized for affinity assays in a temperature range 10–45 1C. The simplest case of binding is a bimolecular interaction, such as that between v-Src SH2 and hmT-peptide as described above. The van’t Hoff thermodynamic analysis of this interaction is shown in Figure 5.21. The data can be readily fitted with eq. (5.26) and the resulting thermodynamic parameters are included in Table 5.5. The data show that the affinity at the sensor surface (KC) matches that in solution (KS), using the method described in Section 5.5. The convex form of the curve indicates a negative value for the heat capacity (DCp) and this interaction appears to be enthalpy driven. A concise description of how to interpret thermodynamic parameters in terms of molecular events during the binding process is given in Box 5.5. A more detailed interpretation can be found in ref. [12]. 20

This is a crucial interaction in the activation of the Ras signaling pathway described in ref. [36].

159

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Box 5.4 Thermodynamics of binding For the simple bimolecular interaction between A and B yielding the complex AB, the change in Gibbs free energy (DG) is related to the standard Gibbs energy change under standard conditions (1 mol l
1 of A and 1 mol l
1 of B, at 25 1C):
 ½AB ¼
RT ln Ka ¼ RT ln Kd ð5:22Þ DG ¼ DG þ RT ln ½A½B where R is the gas constant and T is the absolute temperature. DG1 consists of a heat component released or taken up during the binding process (enthalpy, DH1), and an entropy (DS1) component related to the change in the degree of ‘‘order’’ of the system due to binding: DG ¼ DH 
TDS ¼
RT ln Ka

ð5:23Þ

For protein–protein interactions and other biomolecular interactions DH1 and DS1 change with temperature. The temperature dependence of DH1 and DS1 can be described in terms of the heat capacity (DCp) as given in eqs. (5.24) and (5.25): DH  ¼ DH  ðT  Þ þ DCp ðT
T  Þ

ð5:24Þ

DS ¼ DS ðT  Þ þ DCp lnðT=T  Þ

ð5:25Þ

DCp is assumed constant within the applied temperature range. T1 is the reference temperature, usually 25 1C, and DH1(T1) and DS1(T1) are the values of DH1 and DS1 at this temperature. From eqs. (5.23)–(5.25) follows the integrated van’t Hoff equation, eq. (5.26), which describes the temperature dependence of the affinity constant KA:  


DH  ðT  Þ DS ðT  Þ DCp T
T  T þ þ
ln  ð5:26Þ ln KA ¼ RT R T R T This expression can be used to fit KA values derived at various temperatures versus 1/T, and yields the thermodynamic parameters DH1, DS1 and DCp. A concise description of the interpretation of these parameters in terms of molecular events related to the binding process is given in Box 5.5.

Box 5.5 How to interpret thermodynamic binding parameters? It is important to realize that in a thermodynamic analysis two situations are compared: the situation after the process (e.g. binding) is completed, vs. the situation before the process. This is why we look at the difference in the thermodynamic parameters (indicated as D) including the whole of the process, e.g. also the solvent. The thermodynamic analysis of a single interaction usually tells us whether the binding is entropy or enthalpy driven, but it is not possible to interpret molecular processes more in detail, e.g. whether replacement of water molecules is involved or whether protein dynamics decreases. The power of thermodynamic analysis for drug design lies in the combination of 3D structural information and the study of

160

Chapter 5

structurally related compounds. This can give detailed insight on how specific structural features contribute to binding energetics. The most important thermodynamic parameters in molecular structural events are: DH binding enthalpy represents the heat effects involved in the interaction. It can be directly experimentally determined with calorimetric measurements. The heat effects are caused by the formation and disruption of non-covalent bonds (hydrogen and ionic bonds and van der Waals interactions) and can involve bonds between the reactants, but also bonds of solvent reorganization and conformational rearrangements of the reactants during the binding process. A large part of DH is due to bulk hydration. In drug design, more water molecules at the interaction interface may extend the complementarity of the surfaces and H-bond networks [9]. This is favorable for enthalpy, but disadvantageous due to a loss in entropy, and contributes to the phenomenon of entropy–enthalpy compensation (see text). DS binding entropy can in general be interpreted in terms of degree of order and disorder of the system. This might comprise designed restriction of conformational freedom and rotation of chemical bonds involved in binding. Also, hydration can be a major factor for entropy, e.g. in hydrophobic binding: the burial of water-accessible surfaces and resulting release of water molecules can contribute to binding due to increases in entropy. DCp heat capacity is almost entirely ascribed to solvent effects and is considered of high information content. DCp can be determined directly in calorimetric experiments over a temperature range. DCp can be interpreted in terms of solvent-accessible hydrophobic and polar surface areas, buried in the binding process [5]. A decrease in accessible hydrophobic surface upon binding has a negative effect on DCp; that of polar surface a positive effect. Experimental DCp values have been related to different degrees of success to 3D structural information on these surface changes upon binding. Problems arise due to solvation effects and release or ordering of water molecules during binding. Ordering of water molecules in the binding interface has a large negative contribution to DCp [4].

Figure 5.20

Temperature effect on SPR sensorgrams. Left: signals of sample cell (solid line) and reference cell (broken line). Right: net signal (reference cell minus sample cell). Reprinted from ref. [46], Copyright (2003), with permission from Elsevier.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

161

Figure 5.21

van’t Hoff plot for binding of hmT-peptide with v-Src SH2 domain. (J) Affinity at the sensor surface, KC; (K) in solution, KS. The lines are calculated using eq. (5.26). Reprinted with permission from ref. [12], Copyright (2005) American Chemical Society.

Table 5.5

Thermodynamic parameters derived from van’t Hoff thermodynamic analysis shown in Figure 5.21.

Parameter

v-Src SH2 with 11-mer hmT-peptide
1 a

DH1 (kcal mol ) TDS1 (kcal mol
1)a DCp (cal mol
1 K
1) DG1 (kcal mol
1)a KD (nmol l
1) a


9.4  0.6
0.3  0.6
920  160
9.1 220  30

At reference temperature 25 1C.

5.6.2 Comparison of SPR Thermodynamics with Calorimetry The heat effects of an interaction can be directly measured using calorimetry. Especially the introduction of isothermal titration microcalorimetry (ITC) instruments with improved sensitivity has greatly advanced the use of calorimetry in biomolecular interactions [9]. A debate is ongoing on the equivalence of enthalpy values from van’t Hoff analysis (DH1vH), compared to those from calorimetry (DH1cal): discrepancies between DH1vH, from several techniques for affinity assay and DH1cal have frequently been observed [47–50]. In calorimetric assays the total of the heat effects is assayed, e.g. heat of dilution, of mixing and heat effects due to changes in buffer protonation and solvent equilibria linked with the binding process [47], which go beyond the intrinsic enthalpy contribution of the simple equilibrium A + B " AB. On the other hand, linked equilibria like that of buffers and solvent will also influence the affinity and

162

Chapter 5

DH1vH. The situation may even become more complicated when DCp is not constant with temperature which can occur in multi-step binding processes. Horn et al. demonstrated that, when experimental setup and analysis are performed correctly, there is no statistically significant difference between DH1 values [51]. This holds even for complicated binding models including a conformational equilibrium as shown in Box 5.2. It should be remarked that van’t Hoff analysis is peculiar in its error estimation. Recently, Tellinghuisen demonstrated that the usual way of error estimation in van’t Hoff analysis is actually not correct and that the errors in DG1, DH1, DS1 and DCp are a function of temperature, leading to relatively large errors in DCp [52]. The number of thermodynamic studies using SPR is rather limited [12,14,50,53–55]. In our experience, the thermodynamic parameters from SPR van’t Hoff analysis often compare fairly well with those from ITC [12,14]. As an example, in Figure 5.22 we show the match between our SPR data and ITC data from various studies in an entropy–enthalpy compensation (EEC) plot for a wide range of ligands for the Lck and v-Src SH2 domains [46]. EEC is a universal phenomenon in biomolecular interactions in water and is generally a problem for the medicinal chemist as a gain in enthalpy, e.g. by adding hydrogen bonds to strengthen the binding, will be counteracted by a loss in entropy [56].

Figure 5.22

Entropy–enthalpy plot for binding of various ligands to the Src- or Lck-SH2 domains. Open symbols: data derived from various ITC studies. Closed circles: data derived from SPR competition experiments, with peptide and peptoid ligands. Reprinted from ref. [46], Copyright (2003), with permission from Elsevier.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

163

Combining SPR and calorimetry to explore fully the thermodynamics and kinetics of interacting systems might provide an optimal approach. To explore the information content of especially DCp values fully, ITC is generally a better choice than SPR thermodynamic analysis. A disadvantage of ITC is that it requires much more material (at least 100 times as much as is needed for SPR). The strong point of the SPR technique is the affinity data and kinetics derived from the same data set.

5.6.3 Transition State Analysis Using Eyring Plots SPR analysis has the unique feature that kinetic and affinity information can be obtained from one experiment. This implies that thermodynamic experiments can be performed by analyzing the temperature dependence of kon and koff. Eyring’s transition state theory provides the fundamental conceptual framework for understanding rates of chemical processes [57]. The transition state (AB#) is the high energy state along the pathway of reactants to product (for a binding process, the unbound species to the complex). k1

k2

A þ B Ð AB#
! AB k
1

ð5:27Þ

Based on eq. (5.27), the thermodynamic equilibrium constant for formation of the transition state is defined as K# ¼ [AB#]/[A][B]. Applying statistical mechanics, we obtain the Eyring equation state that holds for a rate constant k: k¼

kB T # K h

ð5:28Þ

where kB is Boltzmann’s constant (1.381  10
23 J K
1) and h is Plank’s constant (6.626  10
34 J s). K# is related to DH# and DS# (the activation enthalpy and entropy, respectively) in the same way as KA is related to DH1 and DS1 [eqs. (5.22), (5.23) and (5.26)]. This implies that for a linear Eyring plot [ln(kh/kBT) vs. 1/T] the data can be fitted with eq. (5.29).
 kh DH # 1 DS# þ ð5:29Þ In ¼
kB T R T R A non-linear Eyring plot can be fitted with eq. (5.26); such fits yield the activation parameters DH#, DC#p and DS#. Transition state analysis using Eyring plots derived from SPR data have been published elsewhere [12,14,53,54]; an example is given in Figure 5.23. The koff values at various temperatures were determined in the presence of high concentrations of competing ligand to prevent rebinding (Section 5.3.2.2). The kon values were derived from kon ¼ koff/KD. It appears that the Eyring plot for koff is linear, indicating that DC#p is zero between the complex and the transition state (vice versa). The plot for kon shows a convex curvature, indicating that DC#p for formation of the transition state from the reactants is not zero. DC#p has the same absolute value as that derived for DCp from the

164

Figure 5.23

Chapter 5

Eyring plots for the interaction of v-Src SH2 domain with hmT 11-mer phosphopeptide. (a) kon, fitted with eq. (5.26); (b) koff, fitted with equation (5.29). Reprinted with permission from ref. [12], Copyright (2005) American Chemical Society.

van’t Hoff analysis of KA for the same interaction (Figure 5.21), because going from the complex to the transition state (koff) DC#p is zero. The Eyring plot for kon (Figure 5.23a) is interesting as it displays non-Arrhenius kinetics above 20 1C, i.e. at higher temperature kon decreases. Non-Arrhenius kinetics have been frequently found for protein folding. A general explanation for this phenomenon is that at higher temperatures a wide region of conformational space is visited and the probability of a flexible ligand or part of a protein having the proper conformation for binding or folding, decreases [58]. Such a model makes sense for the binding of a pYEEI-peptide to an SH2 domain, as the high-affinity binding can be regarded as a ‘‘two-pronged plug into a twoholed socket’’ in need of suitable positioning of the pY and I residue for binding [59]. If, for instance, binding starts with the pY moiety in its binding pocket, at higher temperatures it will be more difficult to have the I residue in the correct position to allow high-affinity binding. A transition state analysis can give additional information, as is also illustrated by the comparison of binding phosphotyrosine ligands to the v-Src SH2 domain vs. Grb2 protein. The affinity and DG1 values are comparable, even the activation energies DG# are nearly identical (Figure 5.24). However, transition state analysis reveals large differences in DH# and DS#. Both DH# values are negative, indicating that upon formation of the transition state, heat is released. The high barrier of activation energy is caused by unfavorable activation entropy contribution. For binding to v-Src SH2 this contribution is about 4 kcal mol
1 more unfavorable. This means that formation of the transition state of the Src SH2 domain from the reactants involves a higher degree of ordering than that of Grb2 SH2. Upon binding, the dynamics of Grb2 and v-Src SH2 domains is decreased to the same extent, leaving the large difference in DS# unexplained [12]. The difference in thermodynamic behavior can probably be

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

Figure 5.24

165

Energy transitions at 25 1C as a function of the binding coordinate for phosphorylated peptide binding to v-Src SH2 domain (solid lines) and to Grb2 SH2 domain (dotted lines). Reprinted with permission from ref. [12], Copyright (2005) American Chemical Society.

attributed to the role of water molecules, which form a hydrogen bonding network at the binding interface between ligand and v-Src SH2 protein [7] upon formation of the transition state, which has an entropy price. On the other hand, the water molecules in the network make a favorable enthalpy contribution to the transition state, explaining the favorable DH#. For binding to Grb2 SH2, such a role of water molecules is not inferred, only direct contact exists between the ligand and the protein. The above described transition state analysis has been criticized, because it is assumed that every activation leads to complex formation [60]. Alternatively, Arrhenius analysis is advocated [60]. In the interpretation of the data, 3D information from X-ray and NMR analysis is essential. However, the 3D structures alone cannot provide information on energetic contributions determining the binding process. Especially in cases where dynamics of ligand and receptor or solvent effects are involved, results of computational chemistry can be expected to be disappointing. The contribution of entropy to the free binding energy can be very large and may influence the affinity by several orders of magnitude. Thermodynamic and kinetic analysis can help to quantify the extent of these contributions and to generate ideas to exploit them in molecular design.

5.7 SPR Applications in Pharma Research: Concluding Remarks and Future Perspectives In this chapter, we have emphasized the role of kinetics and thermodynamics in biomolecular interactions. Notwithstanding the impressive contributions of structural biology and computational chemistry, our understanding and the ability to predict affinities of receptor–ligand interactions remain poor. It is

166

Chapter 5

increasingly acknowledged that for a more accurate notion, thermodynamic and kinetic studies of biomolecular interactions are needed. Modern calorimetric and SPR techniques are the tools to perform such studies and deserve a place in the toolbox of rational design used by the medicinal chemist and chemical biologist. The high information content of SPR data with kinetic and affinity information is unique and allows full thermodynamic and kinetic characterization of an interaction, including transition state analysis, as shown in Section 5.6.3. There are also limitations to the use of van’t Hoff analysis: even with affinity data with relatively small standard errors, DCp has a large error. DCp is important as it can give insight into the nature of the surface area buried upon binding and on the role of water molecules in the binding process (Box 5.5). Using ITC, in general DCp can be more accurately determined by measuring DH at a wide range of temperatures. The best of both worlds is to use SPR and ITC for thermodynamic analysis, with special emphasis on careful experimental design and on the limitations of each method. A concern associated with SPR data is that affinity for a ligand immobilized on a sensor surface might not be identical with that for the ligand in solution. In this respect it is relevant to introduce linkers between the binding epitope and the dextran matrix of the sensor chip, as described in this chapter. In many cases no difference appears, especially for monovalent binding. Using the approach for assays of KS with competition experiments as outlined in Box 5.3, we find comparable affinities at the surface and in solution (see Table 5.4 and Figure 5.21). On the other hand, binding to a surface might be a better model for a biological interaction involving multivalency than (monovalent) interactions in solution. Apart from drug–receptor studies, SPR is also useful for other aspects of drug research. Studies on binding to serum proteins are relevant for distribution properties of drugs [61]. Many important drug targets are membrane-bound proteins, e.g. G-protein coupled receptors (GPCRs). Technology to follow passive and active absorption to membrane interfaces using SPR is under development, as is drug binding to metabolizing enzymes [62]. SPR biosensor systems with supported monolayers and tethered bilayer membranes are under development, but not standard technology yet [63]. In an approach called ligand fishing, crude tissue extracts and cell homogenates are screened for potential ligands or targets using SPR [62]. In such approaches, identification of bound species is crucial, with mass spectrometry (MS) as the ideal platform. In particular, matrixassisted laser desorption/ionization time-of-flight MS (MALDI-TOFMS) and electrospray ionization MS (ESI-MS) are powerful tools for protein identification. It is therefore not surprising that it has been attempted to integrate SPR and MS for proteome analysis, as reviewed in ref. [64]. SPR serves two main purposes in proteome analysis: (1) to confirm and possibly quantify specific binding and (2) to act as a micro purification support for further analysis. MALDI analysis directly on a chip surface is possible [65]. Problems may arise due to the small amount of captured protein on the chip, the many handling steps of the procedure and the

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

167

acidity of the matrix material. In another approach, analytes are eluted and collected in a recovery system. In principle then the whole range of MS techniques, including analysis of digested samples, is available for identification. In drug research, high-throughput screening (HTS) plays an essential role in screening large libraries of compounds. Until recently, the use of SPR technology was hampered by the limited number of surfaces on one sensor chip. In view of the high commercial potential, in the slipstream of the development of array technologies, SPR imaging applications are emerging using microarrays on chips. Examples include Biacore’s Flexchip microarray device [66] and IBIS Technologies IBIS-iSPR instrument (see Chapter 6) and other SPR imaging techniques [67,68] (Genoptics and K-MAC, respectively). This field is developing rapidly and is extremely promising. We expect an increasing impact of SPR technology on drug research. This will be enhanced by further developments of SPR technology for pharma applications, such as high-throughput screening, further integration of SPR and MS and mimicking membrane environments and protein ensembles on SPR surfaces.

5.8 Questions 1. What causes mass transport limitation (MTL) in the kinetics of SPR experiments? 2. How can MTL be diminished by experimental design? 3. How can one perform kinetic analysis under MTL conditions? 4. How can depletion of the analyte in a cuvette-based system be calculated? 5. What is the strength of SPR as a tool in drug development research? 6. Explain why a high loading of the ligand affects the determination of the affinity constant. Describe at least two ways to solve this and determine from the sensorgram given below the affinity constant of an antibody antigen reaction in nmol l
1.

168

Chapter 5

5.9 Symbols [A] [A]0 [A]free [A]tot [B] A AB# B Bmax D DCp DCp# DG DG# DG0 DH# DH0 DS# DS0 h K# KA kb KC kconf Kconf k-conf KD kobs koff kon KS ktr Lm Lr MW

concentration of analyte A at the sensor surface (mol l
1) initially added analyte concentration in the bulk (mol l
1) analyte concentration corrected for depletion (mol l
1) total analyte concentration used in a competition experiment (mol l
1) concentration of free analyte binding sites on the sensor surface (mol m
2 or m1) molecule A (usually analyte) transition state on formation of the AB complex molecule B (usually ligand) maximum binding capacity (in m1) diffusion coefficient (m2 s
1) heat capacity (J mol
1 K
1) activation heat capacity (J mol
1 K
1) Gibbs free energy (of binding) under non-standard conditions (J mol
1) Gibbs activation free energy for formation of the transition state (J mol
1) Gibbs free energy (of binding) under standard conditions (J mol
1) activation enthalpy (J mol
1) change of enthalpy (upon binding) under standard conditions (J mol
1) activation entropy (J mol
1 K
1) change of entropy (upon binding) under standard conditions (J mol
1 K
1) Planck’s constant (6.6262  10
34 J s) equilibrium association constant for formation of the transition state (l mol
1) equilibrium association constant (l mol
1) Boltzman’s constant (1.381  1023 J K
1) equilibrium dissociation constant for affinity at the chip (mol l
1) rate constant for conformation change AB -AB* (s
1) equilibrium constant for conformation change (kconf/k-conf) rate constant for conformation change AB* -AB (s
1) equilibrium dissociation constant (mol l
1) kon[A] + koff (s
1) dissociation rate constant (s
1) association rate constants for formation of the complex AB (l mol
1 s
1) equilibrium dissociation constant in solution (mol l
1) transport rate for diffusion from bulk to sensor surface (in m1 s
1 l mol
1) transport coefficient from bulk solution to sensor surface (m s
1) Onsager coefficient for reaction flux (m s
1) molecular weight of analyte (Da)

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

R Req Rt S T T0 Vbulk Z


1

169


1

gas constant (8.3143 J K mol ) net increase in SPR angle at binding equilibrium (reference cell subtracted from sample cell (m1) net increase in SPR angle at time t (reference cell subtracted from sample cell (m1) surface of the sensor in the cells (mm2) absolute temperature (K) reference temperature (298 K) volume of the bulk solution in the cells (l) viscosity (cP)

5.10 Acknowledgements The examples presented in this chapter originate from work in the Department of Medicinal Chemistry and Chemical Biology, Utrecht Institute of Pharmaceutical Sciences, Utrecht University, The Netherlands. We wish to thank the Head of the Department, Professor Rob M.J. Liskamp, for making it possible to study exciting new compounds in our SPR assays. Many members of the Department have contributed: we especially thank Dr. Frank J. Dekker, Dr. Rob Ruijtenbeek and Dr. Ir. John A.W. Kruijtzer. We are grateful for the cooperation with Dr Isabelle Broutin (Universite´ Rene´ Descartes, Paris) and Professor Albert Heck (Biomolecular Mass Spectrometry, Utrecht University). We thank The Netherlands Organization for Scientific Research (NWO) for financial support.

References 1. G. Schreiber, Curr. Opin. Struct. Biol., 2002, 12, 41. 2. S. Valitutti and A. Lanzavecchia, Immunol. Today, 1997, 18, 299. 3. A.E. Lenferink, E.J. van Zoelen, M.J. van Vugt, S. Grothe, W. van Rotterdam, M.L. van De Poll and M.D. O’Connor-McCourt, J. Biol. Chem., 2000, 275, 26748. 4. G.A. Holdgate, A. Tunnicliffe, W.H. Ward, S.A. Weston, G. Rosenbrock, P.T. Barth, I.W. Taylor, R.A. Pauptit and D. Timms, Biochemistry, 1997, 36, 9663. 5. K.P. Murphy and E. Freire, Pharm. Biotechnol., 1995, 7, 219. 6. H.J. Bo¨hm and G. Klebe, Angew. Chem. Int. Ed. Engl., 1996, 35, 2588. 7. D.A. Henriques and J.E. Ladbury, Arch. Biochem. Biophys., 2001, 390, 158. 8. J.E. Ladbury, Chem. Biol., 1996, 3, 973. 9. R. Perozzo, G. Folkers and L. Scapozza, J. Recept. Signal Transduct. Res., 2004, 24, 1. 10. G.A. Holdgate and W.H. Ward, Drug Discov. Today, 2005, 10, 1543. 11. N.J. de Mol, E. Plomp, M.J. Fischer and R. Ruijtenbeek, Anal. Biochem., 2000, 279, 61.

170

Chapter 5

12. N.J. de Mol, F.J. Dekker, I. Broutin, M.J. Fischer and R.M. Liskamp, J. Med. Chem., 2005, 48, 753. 13. K. Zierler, Trends Biochem. Sci., 1989, 14, 314. 14. N.J. de Mol, M.I. Catalina, F.J. Dekker, M.J. Fischer, A.J. Heck and R.M. Liskamp, Chembiochem, 2005, 6, 2261. 15. P.R. Edwards, C.H. Maule, R.J. Leatherbarrow and D.J. Winzor, Anal. Biochem., 1998, 263, 1. 16. D.J. O’Shannessy, M. Brigham-Burke, K.K. Soneson, P. Hensley and I. Brooks, Anal. Biochem., 1993, 212, 457. 17. P. Schuck and A.P. Minton, Anal. Biochem., 1996, 240, 262. 18. D.R. Hall and D.J. Winzor, Anal. Biochem., 1997, 244, 152. 19. T.A. Morton and D.G. Myszka, Methods Enzymol., 1998, 295, 268. 20. P. Schuck and A.P. Minton, Trends Biochem. Sci., 1996, 21, 458. 21. L.L. Christensen, Anal. Biochem., 1997, 249, 153. 22. D.G. Myszka, T.A. Morton, M.L. Doyle and I.M. Chaiken, Biophys. Chem., 1997, 64, 127. 23. P. Schuck, Biophys. J., 1996, 70, 1230. 24. S. Sjolander and C. Urbaniczky, Anal. Chem., 1991, 63, 2338. 25. S. Felder, M. Zhou, P. Hu, J. Urena, A. Ullrich, M. Chaudhuri, M. White, S.E. Shoelson and J. Schlessinger, Mol. Cell. Biol., 1993, 13, 1449. 26. J.R. Faeder, W.S. Hlavacek, I. Reischl, M.L. Blinov, H. Metzger, A. Redondo, C. Wofsy and B. Goldstein, J. Immunol., 2003, 170, 3769. 27. N.M. Green, Biochem. J., 1963, 89, 585. 28. J. Rao, J. Lahiri, R.M. Weis and G.M. Whitesides, J. Am. Chem. Soc., 2000, 122, 2698. 29. D.G. Myszka, X. He, M. Dembo, T.A. Morton and B. Goldstein, Biophys. J., 1998, 75, 583. 30. D.G. Myszka, P.R. Arulanantham, T. Sana, Z. Wu, T.A. Morton and T.L. Ciardelli, Protein Sci., 1996, 5, 2468. 31. C.S. Navara, Curr. Pharm. Des., 2004, 10, 1739. 32. K. Futterer, J. Wong, R.A. Grucza, A.C. Chan and G. Waksman, J. Mol. Biol., 1998, 281, 523. 33. R.A. Grucza, J.M. Bradshaw, V. Mitaxov and G. Waksman, Biochemistry, 2000, 39, 10072. 34. S. Yuzawa, M. Yokochi, H. Hatanaka, K. Ogura, M. Kataoka, K. Miura, V. Mandiyan, J. Schlessinger and F. Inagaki, J. Mol. Biol., 2001, 306, 527. 35. D. Cussac, M. Vidal, C. Leprince, W.Q. Liu, F. Cornille, G. Tiraboschi, B.P. Roques and C. Garbay, FASEB J., 1999, 13, 31. 36. B.E. Hall, S.S. Yang and D. Bar-Sagi, Front. Biosci., 2002, 7, d288. 37. S. Maignan, J.P. Guilloteau, N. Fromage, B. Arnoux, J. Becquart and A. Ducruix, Science, 1995, 268, 291. 38. N. Schiering, E. Casale, P. Caccia, P. Giordano and C. Battistini, Biochemistry, 2000, 39, 13376. 39. M.B. Khalifa, L. Choulier, H. Lortat-Jacob, D. Altschuh and T. Vernet, Anal. Biochem., 2001, 293, 194.

Kinetic and Thermodynamic Analysis of Ligand–Receptor Interactions

171

40. M.J. Cannon, A.D. Williams, R. Wetzel and D.G. Myszka, Anal. Biochem., 2004, 328, 67. 41. J.E. Ladbury, M.A. Lemmon, M. Zhou, J. Green, M.C. Botfield and J. Schlessinger, Proc. Natl. Acad. Sci. USA, 1995, 92, 3199. 42. M.M. Morelock, R.H. Ingraham, R. Betageri and S. Jakes, J. Med. Chem., 1995, 38, 1309. 43. N.J. de Mol, M.B. Gillies and M.J. Fischer, Bioorg. Med. Chem., 2002, 10, 1477. 44. P. Nioche, W.Q. Liu, I. Broutin, F. Charbonnier, M.T. Latreille, M. Vidal, B. Roques, C. Garbay and A. Ducruix, J. Mol. Biol., 2002, 315, 1167. 45. M.J. Fischer, C. Kuipers, R.P. Hofkes, L.J. Hofmeyer, E.E. Moret and N.J. de Mol, Biochim. Biophys. Acta, 2001, 1568, 205. 46. F.J. Dekker, N.J. de Mol, P. Bultinck, J. Kemmink, H.W. Hilbers and R.M. Liskamp, Bioorg. Med. Chem., 2003, 11, 941. 47. H. Naghibi, A. Tamura and J.M. Sturtevant, Proc. Natl. Acad. Sci. USA, 1995, 92, 5597. 48. P.D. Ross and M.V. Rekharsky, Biophys. J., 1996, 71, 2144. 49. J. Thomson, Y. Liu, J.M. Sturtevant and F.A. Quiocho, Biophys. Chem., 1998, 70, 101. 50. G. Zeder-Lutz, E. Zuber, J. Witz and M.H. Van Regenmortel, Anal. Biochem., 1997, 246, 123. 51. J.R. Horn, D. Russell, E.A. Lewis and K.P. Murphy, Biochemistry, 2001, 40, 1774. 52. J. Tellinghuisen, Biophys. Chem., 2006, 120, 114. 53. A. Baerga-Ortiz, S. Bergqvist, J.G. Mandell and E.A. Komives, Protein Sci., 2004, 13, 166. 54. Y.S. Day, C.L. Baird, R.L. Rich and D.G. Myszka, Protein Sci., 2002, 11, 1017. 55. J. Deinum, L. Gustavsson, E. Gyzander, M. Kullman-Magnusson, A. Edstrom and R. Karlsson, Anal. Biochem., 2002, 300, 152. 56. J.D. Dunitz, Chem. Biol., 1995, 2, 709. 57. G.A. Peterson, Theor. Chem. Acc., 2000, 103, 190. 58. D. Truhlar and A. Kohen, Proc. Natl. Acad. Sci. USA, 2001, 98, 848. 59. J.M. Bradshaw, R.A. Grucza, J.E. Ladbury and G. Waksman, Biochemistry, 1998, 37, 9083. 60. D.J. Winzor and C.M. Jackson, J. Mol. Recognit., 2006, 19, 389. 61. A. Frostell-Karlsson, A. Remaeus, H. Roos, K. Andersson, P. Borg, M. Hamalainen and R. Karlsson, J. Med. Chem., 2000, 43, 1986. 62. M.A. Cooper, Nat. Rev. Drug Discov., 2002, 1, 515. 63. M.A. Cooper, J. Mol. Recognit., 2004, 17, 286. 64. C. Williams and T.A. Addona, Trends Biotechnol., 2000, 18, 45. 65. R.W. Nelson, J.R. Krone and O. Jansson, Anal. Chem., 1997, 69, 4363. 66. D. Wassaf, G. Kuang, K. Kopacz, Q.L. Wu, Q. Nguyen, M. Toews, J. Cosic, J. Jacques, S. Wiltshire, J. Lambert, C.C. Pazmany, S. Hogan,

172

Chapter 5

R.C. Ladner, A.E. Nixon and D.J. Sexton, Anal. Biochem., 2006 351, 241. 67. J.B. Fiche, A. Buhot, R. Calemczuk and T. Livache, Biophys. J., 2007, 92, 935. 68. S.O. Jung, H.S. Ro, B.H. Kho, Y.B. Shin, M.G. Kim and B.H. Chung, Proteomics, 2005, 5, 4427.

CHAPTER 6

Surface Chemistry in SPR Technology ERK T. GEDIG XanTec Bioanalytics GmbH, Mendelstrasse 7, D-48149 Mu¨nster, Germany

6.1 Introduction Whereas the previous chapters focused mainly on the physical basics of surface plasmon resonance, instrumentation and assay design, we will now turn to the heart of SPR and related biosensors: the sensor chip and its surface chemistry. It is here where the biomolecular interaction takes place and although dimensionally extremely small (the coating thickness is measured in nanometers), the sensor chip surface has a tremendous influence on the performance of a biosensor and on the quality of the data retrieved. The nanocomposite coating on a biosensor chip usually consists of the following elements, as depicted schematically in Figure 6.1:
The substrate is a mechanic carrier (e.g. glass) covered with a thin metal layer (e.g. gold; beige layer in Figure 6.1), which enables surface plasmons to be excited. The substrate material and geometry should be compatible with the instrument optics.
A passivation or adhesion linking layer (gray–blue in Figure 6.1) that links the gold surface with the immobilization matrix.
The immobilization matrix, which is a critical element as it is in direct contact with the ligand and the sample and thus influences the specificity and other key characteristics of the biosensor (green brushes in Figure 6.1).
The immobilized ligand (blue Y-shapes in Figure 6.1), usually a biomolecule, is linked to the immobilization matrix and should interact selectively with the analyte (white squares). This chapter focuses on the treatment of the substrate, the adhesion linking layer and the immobilization matrix, including chemistries to couple the ligands 173

174

Chapter 6

Figure 6.1 Nanoarchitecture of a typical SPR sensor chip: the gold-coated (beige) transparent substrate is covered by an adhesion linking layer (gray–blue) to which the immobilization matrix is grafted (here a brush-structured hydrogel in green). Ligands (blue Y-shapes) are covalently coupled to the hydrogel chains to bind analyte molecules specifically (white squares).

to the matrix. The Introduction is followed by a description of the adhesion linking layer and the immobilization matrix. A thorough treatment is given of the immobilization protocols of the ligand to the immobilization matrix, one of which is shown in Figure 6.2. Further, structural features of chip surfaces for different applications are covered and, finally, an overview is provided that should be helpful in selecting the optimal surface for a given experiment.

6.1.1 General Aspects of Surfaces for Biomolecular Interaction Analysis Recognition processes between biomolecules are the key to a thorough understanding of almost all processes in living organisms. To characterize biomolecular recognition, direct optical biosensors are excellent tools as they allow for fast and quantitative analysis without the need for labels. Although direct detection is elegant, label-free detection bears the inherent disadvantage that not only the desired specific components of a biomolecular interaction contribute to the sensor signal, but also non-specific binding (NSB) of other matrix components. This is a major difference in comparison with methods employing labels which detect the labeled component only and are insensitive to other,

175

Surface Chemistry in SPR Technology Sensorgram of ligand immobilization SPR-dip shift

Rligand Time

Phases: baseline - EDC/NHS - baseline - ligand - baseline - ethanolamine - baseline

Figure 6.2

Sensorgram of capturing ligand immobilization to a dextran hydrogel sensor surface by EDC/NHS chemistry (see Protocol 6.3), a commonly used immobilization method. The double-headed arrow reflects the resulting amount of immobilized ligand (Rligand) in the deactivated hydrogel.

unlabeled species. Although NSB can be referenced out by the use of a second channel, the data quality is much better when the ratio of specific vs. nonspecific interactions is high. The second important distinction from traditional methods such as blotting or other solid-phase techniques is the multiple reuse of the chips. Due to significant variations between derivatized chip surfaces and also for cost and time reasons, chip surfaces are reused as long as possible but at least for a series of continuous measurements. Complete regeneration requires that after each measurement cycle the chip surface is brought back to its original state. Proteins and many other biomolecules tend to adsorb irreversibly on untreated metal, glass or plastic surfaces, often losing 90% or more of their activity. The result would be a chip surface with a low density of active ligand, which can only be partially regenerated and is practically unusable after a few cycles, as shown in Figure 6.3. The time-resolved nature of direct optical techniques leads to additional requirements of the chip surface. In contrast to common solid-phase assays yielding one data point per assay, a biosensor delivers the full time course of the interaction. As outlined in Chapters 4 and 5, this high information content can be used to determine either the analyte concentration or – usually with a differently structured surface – the association and dissociation kinetics of the interaction. To understand the underlying processes that occur on the surface of a sensor chip, it is helpful to take a closer look at the basic forces which determine the interaction of biomolecules at the molecular level, listed in Table 6.1. Two of the interactions, the electrostatic (ionic) and the hydrophobic interaction, account for approximately 85% of the overall energy and are therefore the most relevant. If the chip surface and the interacting species are charged, the extent and sign of the electrostatic interaction can be manipulated within a relatively

176

Chapter 6

A

B

nD

t

Figure 6.3

Table 6.1

Unstable immobilization and partial denaturation of ligand (red) and analyte (blue) on an incompatible surface (A) results in incomplete regeneration, yielding in an increasing baseline and decreasing binding capacity from cycle to cycle (B).

Molecular forces contributing to biomolecular interaction processes [1,2].

Force

Energy (kJ mol1)

Distance dependence

Hydrophobic interaction Electrostatic interaction Hydrogen bonding Van der Waals forces

Up Up Up Up

Not applicable r2 r6 r10

to to to to

15 12.5 4 0.4

broad range through modifications of the ionic strength and pH of the buffer. An increasing salt concentration screens charged groups and usually has a practically pH-independent repulsion effect on hydrophilic and charged immobilization matrices because ion pairs are formed which can neutralize the charged domains. At low ionic strengths,1 the pH of the buffer becomes more important, as the electrostatic interactions dominating in this regime are governed by the ligand’s overall charge, which is positive at pH values below 1

I.e. below 0.1 M.

Surface Chemistry in SPR Technology

177

the ligand’s isoelectric point and negative in the more alkaline pH range. Depending on the charge of the sensor chip surface, strong attractive or repulsive forces between a dissolved species and the surface can be the consequence. The hydrophobic interaction cannot be easily controlled, as it is not an attractive force but the exclusion of hydrophobic domains of low surface energy in highly energetic solvents such as water. Hydrophobic interactions are induced by a changed degree of organization of water molecules and are therefore an entropy effect. The entropy of the total system (water and interacting molecules) will increase when a hydrophobic interaction takes place. Generally, substances which strengthen the inner structure of water, such as most salts, increase the hydrophobic contribution (the salting-out effect is caused by a change in entropy); chaotropic substances such as guanidine or ethanol lower it. The influence of the pH is moderate, although a pH close to the pI of the interacting species usually minimizes the extent of electrostatic forces and thus increases the relative influence of hydrophobic interactions. As a consequence, precipitation of protein molecules often occurs at the pI caused by the decreased electrostatic repulsion between molecules. As hydrophobic interactions can lead to partial unfolding and as a consequence to significant activity losses of immobilized proteins, the fraction of hydrophobic domains on the sensor chip surface should – with the exception of surfaces for immobilization of membrane proteins – be kept as low as possible. Keeping the above in mind, it is obvious that especially the surface charge and surface energy of biosensor chips have to be carefully controlled in order to achieve high immobilization yields, to minimize non-specific interaction of the matrix, to retain the biological activity of the immobilized ligand and to achieve high signal-to-noise ratios. Other means are the charge distribution and density, surface structure and functionality, which are key factors to adapting the sensor chip surface to particular applications.

6.1.2 Selection of the Optimal Surface Surface plasmons at the interface between a metal and a dielectric material have a combined electromagnetic wave and surface charge character, as shown in Chapter 2. This combined character results in the electric field component perpendicular to the surface being enhanced near the surface and decaying exponentially with distance (Figure 6.4). The field in this perpendicular direction is said to be evanescent, reflecting the bound, non-radiative nature of surface plasmons. In the sample buffer above the metal, the decay length of the field, dd, is of the order of half of the excitation wavelength of light and is usually defined as the distance over which the intensity of the evanescent field drops to 1/e, i.e. to about 37%. In most commercial instruments the wavelength of light is between 600 and 800 nm and dd is in the range 300–400 nm. One important consequence of the exponential decay of the evanescent field intensity is that a typical SPR biosensor is practically blind at distances beyond 600 nm from its surface. The high signal-to-noise ratios achievable with

178

Figure 6.4

Chapter 6

The intensity of the evanescent field decays exponentially with increasing distance from the metal layer. The SPR angle shift is proportional to the change of effective refractive index close to the surface.

evanescent field sensors are partially due to this insensitivity towards changes in the bulk phase. In addition, identical interactions give rise to different signal shifts, depending on how close to the sensor surface the interaction occurs. A receptor–ligand binding (but also non-specific interaction processes) observed within the first 10 nm from the metal surface result in an almost three times higher response than the same process at a distance of 300 nm. The transport of solutes to the chip surface by convection and diffusion has a profound effect on the signal, in addition to solute binding to the immobilized ligand. The product of these three processes results in a sensorgram which is consequently influenced by factors contributing to any of those processes. In linear flow cells used in many commercial instruments, convection can be controlled simply by adapting the flow rate. At a certain flow rate, a so-called stagnant layer is formed on the sensor, as described in Chapter 4. High flow rates induce small stagnant layers; however, a stagnant layer of less than 2 mm is difficult to achieve. Cuvette geometries, such as those employing the free walljet principle, allow variation of more parameters, as for example the distance of the injector tip from the sensor surface.2 In all of these cases, the analyte is more or less efficiently transported to a distance of a few micrometers from the sensor chip surface, but still relatively far away from the evanescent field and any three-dimensional surface structures. Here, the unstirred diffusion layer begins, through which transport is solely effected by diffusion,3 as depicted schematically in Figure 6.5. 2 3

As described in Chapter 3 Section 3.3.2. Mass transport limitation effects are clearly described in Chapter 5, Section 5.3.

Surface Chemistry in SPR Technology

Figure 6.5

179

Dimensions of the evanescent field and the unstirred diffusion layer. Analyte molecules are transported first mostly by convection and then solely by diffusion.

The diffusion rate through this layer depends largely on the diffusion constant of the analyte molecules, which varies from 3  10–6 to 6  10–6 cm2 s–1 for low molecular weight substances to below 10–7 cm2 s–1 for macromolecules with molecular weights of several hundred kDa. This corresponds to an average diffusion time through the unstirred layer from less than 1 s for small molecules up to several seconds for high molecular weight compounds. The effective concentration of larger molecules in the evanescent field but still not yet bound to the ligands at the sensor surface will then be typically lower than the concentration in the bulk solution. Even at the maximal flow rates, mass transport is often limiting with typical protein molecular weights above 10 kDa. Due to a higher diffusion constant, the relative binding rate of small molecules is usually not affected by diffusion phenomena even at high ligand densities. It is usually not possible to enhance the diffusion rate significantly above a certain upper limit because the maximal flow rate is limited by the volume of the sample, i.e. by the volume of the injection loop. In cases where diffusion limitation should be strictly avoided, as for example in kinetic analysis, the only way to circumvent diffusion-related problems is to decrease the density of the immobilized ligand and/or increase the analyte concentration. On the other hand, if mass transport limited binding is desired, as for concentration determination, then apart from decreasing the flow and thus the diffusion rate, higher immobilization densities of the ligand can be chosen. In both cases – low and high immobilization densities – a correctly selected surface nanoarchitecture can be extremely useful for controlling the amount of immobilized ligand.

180

Chapter 6

The final step in the biomolecular interaction process is the (specific) capture of the analyte by the immobilized ligand, which is governed by the kinetics of the interaction and the properties of the nanoenvironment around the ligand– analyte pair. With kinetic analysis, it is essential that the binding sites are neither sterically hindered nor their affinity affected by the immobilization process, as both can lead to heterogeneously distributed affinities [3,4]. As shown in Figure 6.6, rate

Figure 6.6

Distribution analysis (this method is briefly discussed in Chapter 12) of the complex experimental surface binding kinetics of myoglobin binding to monoclonal antibody immobilized in the carboxymethyldextran matrix of a Biacore CM5 sensor chip. The two-dimensional rate and affinity constant distribution indicates heterogeneous binding sites with different affinities. Reproduced with kind permission from Ref. [3].

Surface Chemistry in SPR Technology

181

and equilibrium constants can be determined based on a novel distribution analysis method. A detailed characterization of the distribution of binding properties provides a useful tool for the optimization of surface immobilization towards the efficient functionalization of biosensor surfaces with uniform highaffinity binding sites and for studying immobilization processes and surface properties. Sufficient spacing between ligands is helpful for decreasing steric problems, but the requirement of surface homogeneity at the nanoscale remains. Therefore, directed immobilization through capture molecules can become necessary to maintain the homogeneity of a binding site population which might otherwise be compromised by the random distribution of the covalent bonds within the immobilization matrix. Finally, the surface charge can also have an effect on the interaction kinetics and on the extent of non-specific interactions, so this feature should also be controlled. Assays with the mere purpose of analyte quantification are generally less critical, as here a certain degree of diffusion limitation is allowable if not desired. In most cases, high immobilization densities are advantageous in order to maximize the specific signal and to deplete the unstirred layer of analyte during the initial phase of the interaction phase. Under these conditions, analyte binding is heavily diffusion controlled and the – then linear – slope of the binding curve is directly proportional to the analyte concentration in the sample, a prerequisite for a precise quantitative assay.

6.2 Adhesion Linking Layers for Gold, Glass and Plastics As outlined in the previous section, it is necessary to protect the sensitive biomolecular ligands from the usually incompatible chip substrate material. Also, suitable functional groups for ligand immobilization have to be introduced. This is mostly achieved by coating the substrate with a bioinert layer that contains functionalizable groups, typically carboxylates. As one of the key characteristics of these layers – their hydrophilicity – makes them water soluble, they would be washed away without the use of an adhesion linking layer, which promotes their adhesion to the substrate material. Therefore (as shown in Figure 6.1), a typical biochip coating contains at least three functional components: an adhesion linking layer, a bioinert matrix and ligands coupled to the matrix. In this section, the adhesion linking layer is described. Ideally, the adhesion linking layer provides a stable link between substrate material and immobilization matrix and also shields the substrate from the sample buffer with a dense and homogeneous film. With respect to the exponentially decaying strength of the evanescent field, thicknesses above 10 nm would significantly decrease the sensitivity of the sensor and can lead to baseline drift caused by swelling effects. On the other hand, thicknesses below 1 nm usually result in unstable and inhomogeneous coatings, so preferably the thickness of the

182

Chapter 6

adhesion linking layer is between 2 and 5 nm. Finally, the refractive index of the adhesion linking layer should be lower than that of the substrate material. Depending on the surface-exposed material of the biochip substrate – typically gold or glass – different routes are chosen to address the above requirements.

6.2.1 Adhesion Linking Layers for Metal Surfaces The cationic surfaces of many transition metals are soft electron pair acceptors and exhibit a strong affinity towards soft electron pair donors such as thiols, disulfides and thioethers. Due to the negative imaginary part of their electromagnetic wavefunction in combination with their chemical inertness, gold, silver or platinum are suited for use in SPR and as electrochemical sensors. Alkyl derivatives of above-mentioned functional groups with a chain length of 410 carbon atoms assemble spontaneously on such substrates and form selfassembled monolayers (SAMs) with high packing densities [5]. Shorter thiols also assemble, but the SAMs are not well defined and relatively unstable [6]. Monofunctional mercaptoalkyls yield hydrophobic surfaces having contact angles higher than 1001. Bifunctional derivatives form monolayers with defined chemistry which are useful intermediates for covalent coupling of ligands or for further derivatization [7]. Typical examples for such compounds are 16-hydroxyhexadecane-1-thiol and the corresponding carboxylated compound 15-carboxypentadecane-1-thiol. The adsorption of these long-chain thiols usually takes place in 1–5 mM ethanolic solutions in 8–24 h. Although the formation of a monolayer is almost complete after a few minutes, the initially formed monolayer is not well ordered and contains many gauche defects within the chains. Over time, the layer becomes more ordered and well packed. In addition to thiols, dithiols and thioethers are also suitable, as all of these groups exhibit sufficiently high adsorption energy on surfaces of Group 7–12 metals, which is typically in the range 40–50 kJ mol–1 – B50% of the strength of a C–C bond. As discussed below in more detail, the stabilizing characteristics of the resulting surface, i.e. its inertness against adsorption of proteins and other sample components, depends critically on the functionality of the surface exposed to the sample. However, the long-term stability of SAMs is limited as they show desorption after a few weeks of exposure to buffer or serum [8]. A less common approach for the derivatization of noble metal surfaces is the adsorption of positively charged or mercapto-derivatized polymers to the negatively charged metal surface. Due to electrostatic attraction and the cooperative effect of several adsorption sites, stable monolayers can be formed.

6.2.2 Adhesion Linking Layers for Inorganic Dielectrics The classical modification route for glass, ceramics and other oxidic surfaces is treatment with silanes which are able to form stable silyl ether links with

Surface Chemistry in SPR Technology

183

exposed hydroxyl groups [9]. A frequently used and versatile silane is GPTMS (glycidylpropyltrimethoxysilane), which is reactive towards amino, sulfhydryl and hydroxyl groups but can alternatively be hydrolyzed to yield a vicinal diol or can be further oxidized to an aldehyde or carboxylate. Another example is APTES (aminopropyltriethoxysilane), which is useful for coupling activated carboxyls, aldehydes or glycidyl moieties. Under optimal conditions the silanes assemble on the surface of a thoroughly cleaned [10] substrate in a uniform monolayer [11]. If properly prepared, such an arrangement allows the attachment of the bioinert matrix or directly of the biomolecular ligand in a similarly uniform fashion.

6.2.3 Adhesion Linking Layers for Plastics Although at present gold and glass dominate as chip surface materials used for direct optical biosensors, it is foreseeable that future low-cost devices will increasingly rely on injection molded consumables made from, e.g., poly(methyl methacrylate) (PMMA), polystyrene, polycarbonate or cycloolefin copolymers (COC). The surface of these materials is a complex, heterogeneous mix of amorphous and crystalline regions consisting of mostly hydrophobic polymer chains which often slowly migrate and rearrange over time. Modification of this kind of substrate usually begins with an oxidative pretreatment either via a wet etch step [12] or – more reproducibly – oxygen plasma treatment [13]. The immobilization matrix can then be coupled either directly or via subsequently adsorbed stabilizing polymer layers [14]. Regardless of the substrate material, adhesion linking layers can be further applied via plasma deposition, allowing fast and simultaneous processing of large batch volumes, and can yield homogeneous coatings with different chemical functionalities at relatively low cost per unit. Typical thick film preparation methods such as dip or spin coating are less suited as the necessary coating thickness of a few nanometers is difficult to control reproducibly with these techniques.

6.3 Bioinert Matrices While the aforementioned adhesion mediators are optimized in terms of maximal interaction with the substrate material, it is this characteristic that induces a significant level of non-specific binding from the sample. Therefore, addition of a bioinert topcoat, or immobilization matrix, becomes necessary.

6.3.1 Non-specific Adsorption of Biomolecules Non-specific adsorption of heterogeneous macromolecules such as proteins or larger aggregates on surfaces is a central aspect in the design of biocompatible materials. However, the complex multistep interaction process depends on the complex composition and concentration of potential adhering components in the

184

Chapter 6

bulk and at the surface, which is still not fully understood. For example, the initial adsorption of plasma proteins is dominated by the small protein (albumin), present at higher concentrations in the bulk, to be replaced later by larger proteins such as immunoglobulin G (IgG) and fibrinogen. This sequential adsorption is called the Vroman sequence [15], which explains the competitive adsorption of plasma proteins for a limited number of surface sites. Reorganization and stapling of biomolecules occur, which depend on several factors and surface parameters. Dynamically, adsorbing species undergo three phases [16]: transport to the surface by convection and diffusion (see Figure 6.5), reversible (labile) attachment on first contact, spreading and conformational rearrangement. Whereas the transport rate is an effective means of controlling the extent of protein adsorption, the initial phase cannot be influenced by the surface functionality and structure. It is possible, however, to tune the degree of protein attachment and subsequent processes by optimizing the nanoarchitecture and molecular-scale functional design of the surface. In this context, it should be stressed that the interaction of surface and proteins with the solvent water plays an major role as the surrounding water molecules compete with potential attachment sites. The high internal energy of water (cohesion) favors the exclusion of low-energy sites, i.e. hydrophobic domains which consequently tend to agglomerate – the already mentioned hydrophobic interaction which is an entropy phenomenon caused by the organization grade of water molecules. This driving force leads to conformational changes after adsorption of proteins to hydrophobic moieties. Proteins attached to poorly hydrated or attracting surfaces – regardless of whether non-specifically attached or immobilized – undergo high deformation into a pancake-like structure [17]. This deformation process is typically irreversible and deactivates the protein. If the adsorbed proteins retained some lateral mobility, clustering can occur [18] as a secondary process, which highlights the difficulty of models based on simple Langmuir isotherms. An essential prerequisite of a bioinert matrix is therefore not just hydrophilicity but also a high degree of hydration. In general, the more hydrophilic the surface, the greater is the surface–water interaction and the higher degree of water molecule organization and hence less chaotic behavior. Hydrophobic molecules will merely irreversibly adsorb on the hydrophilic surface, e.g. by unfolding hydrophobic sites that favor the increase in entropy of the total system upon binding. This primary interaction is the likely basis for two distinct types of response seen towards hydrophobic vs. hydrophilic surfaces [19]. Moreover, at a hydrophobic surface, mutual water molecule hydrogen bonding is disrupted and significant surface dewetting occurs, extending into bulk water. In addition to hydrophilicity, the surface charge is an important factor. It is obvious that proteins will be attracted by oppositely charged structures, although at medium and high ionic buffer strengths and physiological pH ionic groups are usually surrounded by counter-ions, so if the charge density of the surface is not too high, this effect is often less significant than expected. In this context, one should take into account that the hydration sphere of ionized

Surface Chemistry in SPR Technology

185

groups binds water molecules and therefore contributes to the bioinertness of the surface. If not accompanied by irreversible secondary processes, such as hydrophobic adsorption or covalent coupling, the electrostatic interaction is typically fully reversible at high ionic strengths. If the underlying mechanism of a protein adsorption process is unknown, increasing the ionic strength is a useful measure to discriminate between ionic and hydrophobic interactions as the latter become stronger and are irreversible under these conditions. In addition to its reversible character, electrosorption is a relatively welldefined process and can be controlled by pH-induced charge shifts of adsorbates with a defined pI. Furthermore, it is self-terminated when electrostatic neutrality is reached. However, it should be noted that even proteins bearing the same charge as a strongly charged surface can show significant adsorption overriding the net Coulombic force in low ionic strength solution [20]. Oppositely charged domains on the protein interacting with the homogeneously charged surface are likely to be a factor, driving out small counter-ions, an entropy-driven phenomenon referred to as counter-ion evaporation [21]. In accordance with the above, a molecular-level structure–property relationship survey of differently functionalized surfaces [22] revealed that most surfaces that resist the adsorption of proteins incorporate groups that exhibit several characteristics, including hydrophilicity and neutral overall charge. Another factor is the presence of hydrogen bond donors and the absence of hydrogen bond acceptors, indicating the role of hydrogen bonding as a third force contributing significantly to biomolecular interactions. However, carbohydrate-based surfaces, many of which are highly protein resistant, do not follow this rule, indicating that not all aspects of surface–protein interactions are covered by this model. Again, water may be the key to this observation, as uncharged carbohydrates are claimed to orient up to three layers of water [23].

6.3.2 Bioinert Hydrogels In addition to choosing suitable functionalities at the molecular scale, the bioinertness of a surface can be controlled by structural means. Polymers grafted on the surface of a so-called hydrogel are effective in preventing the adsorption of proteins. As the amount of protein adsorbed is the result of the interplay between the bare surface–protein interactions, the competition between the proteins and the polymer chains striving to be in the vicinity of the surface, the polymer–protein interactions and the conformational statistics of both protein and polymer molecules, it is obvious that variations in the coating’s nanoarchitecture result in different adsorption characteristics of the coated surface. Surface stabilization by hydrogels is due to the decreasing degrees of freedom of surface-bound polymer chains when a protein molecule approaches the coated surface (Figure 6.7). As this leads to an energetically unfavorable entropic loss, the system is more stable if dissolved macromolecular species do not interfere with the polymer layer.

186

Chapter 6

Figure 6.7 Entropic stabilization of surfaces: polymer segments in the vicinity of an approaching macromolecule have fewer degrees of freedom than those which can move freely in all directions. Compression of equal charges along the polymer chains adds to the repulsive effect, preventing non-specific binding.

Theoretical studies using the mean-field theory4 [24] demonstrated that branched and loop polymers are more effective in preventing protein adsorption than linear flexible chains [25], as they are more rigid and can provide a denser concentration of desired functional groups (Figure 6.8). Dendritic polyglycerols, for example, have shown good protein resistance and proven superior to dextran coatings used to reduce non-specific binding [26]. However, other features important for evanescent field biochips, such as immobilization capacity, homogeneity at the nanoscale and diffusion characteristics, should also be considered and these usually favor the use of better defined linear structures as the immobilization matrix. An additional feature of hydrogels for SPR-based biosensors is that the hydrogel can be tuned to fit the evanescent field. Ligands can not only be coupled directly to the surface but can also be immobilized in the evanescent volume, making a higher loading of ligands feasible. However, if the hydrogel becomes too thick and dense, ligands and analyte molecules may not penetrate into the evanescent field. The quality of the bioinert hydrogel layer in combination with the type of ligand which should be coupled determines to a great extent the quality of the biomolecular interaction results. A list of common protein-compatible polymers is given in Table 6.2; an in-depth evaluation of the properties of three-dimensional hydrogels can be found in Section 6.4.2. 4

A theory that describes the behavior of surface-grafted polymer chains. At moderate surface concentrations (semi-dilute regime), the statistical mechanical properties of the chains can be evaluated by using an analogy with the classical mechanics of a particle moving in a mean potential field. The analytical solutions of this theory provide a convenient way to explore the conformations and the interactions of such surfaces.

187

Surface Chemistry in SPR Technology

Figure 6.8 Table 6.2

Selected protein-compatible functionalities.

Examples of hydrophilic, protein-compatible polymers.

Class

Polymer

Polysaccharides, natural

Dextran Alginic acid Hyaluronic acid Heparin Chitosan Pectin Carboxymethyldextran Carboxymethylcellulose Poly(vinyl alcohol) Poly(hydroxyethyl methacrylate) Polyglycerol (dendrimer) Poly(ethylene glycol) Poly(propylene glycol) Poly(acrylic acid) Poly(L-lysine)

Polysaccharides, modified Polyalcohols Polyalcohol and polyether Polyethers Polycarboxylates Polyamines

In conclusion, bioinert matrices can practically quantitatively eliminate nonspecific binding even in complex samples such as serum or fermentation broths and signal-to-background ratios can increase by one to two orders of magnitude.

6.4 Choosing the Optimal Nanoarchitecture As outlined above and in Section 6.5, the molecular-scale functionalities of sensor chip surfaces mediate the immobilization of ligands and are further effective means of minimizing the non-specific background. However, their

188

Chapter 6

proper selection alone is usually not sufficient to achieve reliable quality of the sensorgram. Structural characteristics in the submicrometer range must also be considered. In other words, the quality of the nanoarchitecture (e.g. hydrogel) determines efficient coupling for accurate rate and equilibrium constant measurements. A homogeneous distribution of binding sites and homogeneity of their nanoenvironment are vital when it comes to kinetic analysis. Further, depending on the application, different immobilization capacities are required and sometimes extra features, such as a filter functionality, may be advantageous. All these issues are addressed by the design of the nanostructured coating or bioinert matrix. Coatings for bioanalytical and biomedical devices are usually divided into two major groups: two-dimensional (2D) planar coatings and three-dimensional (3D) hydrogels. With SPR biosensor chips, both types are employed. In Table 6.3, a practical overview is given of the application of 2D and 3D nanostructures on sensor surfaces. Table 6.3

Overview of nanostructures for different sensor applications.

Application

Suggested structure

Protein–protein, assay

Hydrogel, 100–500 nm, low–medium density. Linear polycarboxylate or carboxylated polysaccharide Well-stabilized carboxylated 2D surface Hydrogel o50 nm, low density. Linear polycarboxylate Hydrogel, 100–500 nm, low–medium density. Linear polycarboxylate Hydrogel, 4500 nm, high density. Linear polycarboxylate or carboxylated polysaccharide Thin hydrogel, o20 nm, high density. Linear polycarboxylate or carboxylated polysaccharide. Well-stabilized carboxylated 2D surface Hydrogel, 4300 nm, high density. Linear polycarboxylate Hydrogel, 100–500 nm, low–medium density. Carboxylated and streptavidinmodified polysaccharide Thin hydrogel, o20 nm, high density. Linear polycarboxylate or carboxylated polysaccharide, partially alkyl derivatized. 2D mercaptoalkyl SAM Hydrogel, 100–500 nm, low–medium density. Polysaccharide with reduced carboxylation level Hydrogel, 300–500 nm, low–medium density with filter layer Hydrogel, 4100 nm. NHS preactivated linear polycarboxylate

Protein–protein, kinetics Protein–DNA or polysaccharide Protein–peptide or small molecule Protein–cell, virus or particle

DNA–DNA, small molecule or peptide DNA–protein Cell, virus, particle, lipid bilayers–any species Assays in serum or culture medium Ligand fishing from crude samples, assays in whole blood 2D SPR microarrays

Surface Chemistry in SPR Technology

189

6.4.1 Two-dimensional Surfaces Two-dimensional surfaces are well suited for the detection of large, particulate analytes such as viruses or even whole cells, as a hydrogel would be inaccessible for such bulky species, i.e. would keep them outside the evanescent field. Due to the same reasons and the higher field intensity close to the chip surface, 2D structures are also a good choice for the immobilization of high molecular weight ligands, cell fragments or lipid mono- and bilayers. 2D surfaces (Figure 6.9) are often chosen for applications where a low immobilization capacity is important, such as the kinetic analysis of medium to high molecular weight compounds. As the density of immobilized ligand is limited to a maximum of one monolayer (1–2 ng mm–2 for a typical protein), depletion of analyte from the diffusion layer during the interaction phase can hardly occur. Also, the diffusion of the analyte to and from the ligand is not hindered by an extended immobilization matrix, both contributing to kinetically controlled binding. This free accessibility also minimizes rebinding of dissociated analyte to free binding sites during the dissociation phase, which can happen in extended hydrogels and results in the measured dissociation rate constant (apparent koff) being slower than the true koff. In kinetic studies, homogeneity of 2D coatings is essential, as a narrow distribution of the immobilized ligand’s activity and accessibility is a prerequisite for a correct fit of the resulting sensorgrams with the appropriate kinetic model.

Figure 6.9

Schematic illustration of 2D surface architecture. Ligand molecules are immobilized on a few nanometers thick, planar immobilization matrix and are readily accessible to analyte molecules.

190

Figure 6.10

Chapter 6

STM image of a sputtered gold surface. Crystalline 111 domains create terrace-like nanostructures with roughness of around 30 nm. Scale bar for STM in picoamps.

Here, problems can occur, as at the nanoscale level the untreated surface of a sensor chip is not a well-defined, homogeneous, two-dimensional plane. In fact, most materials, regardless of whether glasses, noble metals or plastics, show an irregular nanotopology with a roughness well above 20 nm, which corresponds to 410 times the diameter of the immobilized ligand. The structure of gold surfaces used in SPR sensors depends on the deposition method and, due to the higher deposition energy, is denser and more homogeneous if the gold is sputtered (Figure 6.10) compared with vapor-deposited films, which consist of loosely adhering gold clusters. More problematic are molecular-scale inhomogeneities, although they are sometimes useful as they increase the surface available for immobilization. These steps, gaps and tips represent discontinuities which can lead to pinhole defects of coatings, thus inducing local non-specific interactions or deactivation of sensitive immobilized ligands (Figure 6.11). Non-specific binding sites are of particular relevance with high molecular weight molecules and aggregates, as due to cooperative effects even a low fraction of such attachment points can be sufficient to induce irreversible adsorption of larger species. For these reasons, the frequently employed SAMs of long-chain mercaptoalkyls can be problematic and it is advantageous to coat them with a second, smoothing – ideally thin polymeric – layer. The latter further increases the stability of the coating and can provide a spacer functionality which is useful for increasing the accessibility of the immobilized ligand. An alternative and popular approach is the use of short polyether – usually PEG – chains terminally linked to the SAM layer, which has a similar effect.

Surface Chemistry in SPR Technology

Figure 6.11

191

Schematic illustration of possible defects of SAMs adsorbed on a molecular scale ‘‘nanorough’’ gold surface. Such defects can cause local non-specific interactions.

6.4.2 Three-dimensional Hydrogels Whereas with a sufficiently sensitive detector 2D surfaces form a good basis for many biomolecular interaction studies, numerous applications remain which require a ligand density far above one monolayer, as either the analyte is relatively small (o10 kDa) or the optics are not sufficiently sensitive. For this purpose, surface-grafted 3D hydrogels have been developed with thicknesses from below 10 up to more than 1000 nm (see Figure 6.1). These structures provide an increased density of attachment sites and – what is more important – use a higher volume fraction of the evanescent field for the specific interaction than 2D coatings. A welcome side-effect of this approach is that the absolute influence of non-specific factors such as bulk shifts decreases proportionally with increasing occupation of the evanescent field volume by the ligand–analyte pair. Further advantages of hydrogels are the improved stabilization against non-specific interactions, their spacer functionality that keeps the immobilized ligand away from the surface, thus improving its accessibility, and the protection of sensitive ligands against denaturation, especially when the surface is being dried. Hydrogels can also help to solubilize hydrophobic ligands which would otherwise tend to form insoluble aggregates on the chip surface. In principle, surface-grafted hydrogels can be made from any water-soluble polymer, but in practice polycarboxylates, polyethers and polyols are preferred as they show significantly lower background than, for example, polyamines. As in the early 1990s the first commercial sensor chip coatings were derived from well-characterized, dextran-based solid phases for affinity chromatography, the most popular matrix material today is carboxymethylated dextran. Other carboxylated polysaccharides such as alginate, pectin, carboxymethylcellulose or hyaluronic acid can also be used, but are less common. Because these polymers are of natural origin, their structure is often irregular, i.e. they can be more or less branched or form hyperstructures, such as helices or suprafibers, thus contributing to heterogeneous binding site populations [3]. Especially in the hydrated state, carbohydrates are relatively bulky molecules, occupying a considerable fraction of the evanescent field volume and rendering it unavailable for ligand immobilization. The bulky structure can also

192

Figure 6.12

Chapter 6

Structural differences between polysaccharide- (left) and synthetic polycarboxylate (right)-based surface-grafted hydrogels. See text for details.

hinder free diffusion of the analyte. For these and other reasons, efforts have been made to replace polysaccharides by better defined synthetic polycarboxylates with a smaller molecular footprint, as illustrated in Figure 6.12. In addition to these steric issues, pure polycarboxylate coatings have the advantage that they cannot form ester crosslinks upon EDC/NHS activation, a common side-reaction with carboxymethylated carbohydrates. After activation, the NHS polyesters are moderately hydrophobic, which is a prerequisite for spotting ligand microarrays used for parallel detection with 2D SPR optics: on a hydrophilic surface the spots run and merge. On the other hand, hydrophobicity can make a polycarboxylate hydrogel insoluble and ‘‘collapse’’, i.e. precipitate on the chip surface, if too activated, so optimization of the activation level is important. Regardless of whether a natural polysaccharide or synthetic polymer, two main parameters can be varied to control immobilization capacity and diffusion characteristics of a 3D surface: the thickness and the density of a hydrogel. The thickness depends on the molecular weight and structure of the polymer. The thickness of surface-grafted polymer monolayers ranges from a few nm up to 2 mm with immobilization capacities between 1 and 4100 ng mm–2. Typically, due to the limited penetration depth of the evanescent field, thicknesses between 20 and 200 nm are employed; thicker hydrogels can be useful to achieve ultrahigh immobilization densities for the detection of low molecular weight analytes. Hydrogels with a thickness above 1 mm are also useful to keep particulate contaminations or air bubbles outside the evanescent field, resulting in a very robust surface. In such structures, heavy diffusion limitation is often observed with a mean diffusion time of several seconds across the hydrogel. Immobilization capacity can also be controlled by the hydrogel density. In addition, variation of the chain density can increase the selectivity as it can

Surface Chemistry in SPR Technology

193

result in a filtering effect excluding macromolecules above a certain molecular weight. For mere detection or analyte quantification purposes it is advisable to choose relatively high densities as this not only maximizes the signal but also minimizes non-specific effects. However, care must be taken that the hydrogel density is not too high, as analytes with a molecular weight in excess of some tens of kDa can agglomerate in the upper layer of dense hydrogel matrices, clog the pores and prevent other analyte molecules from diffusing into the lower parts of the sensing layer. If the analyte carries hydrophobic domains, agglomeration and in the worst case even (partial) collapse of the hydrogel can also occur. Irregularly shaped binding curves are a typical indication for this phenomenon. In addition, steric hindrance can result in the aforementioned heterogeneous accessibility and affinities of the immobilized binding sites. It should be stressed that the above is correct for medium and high molecular weight analytes only. For low molecular weight compounds (below 2 kDa), hydrogels with a maximum immobilization capacity should be chosen, as these molecules are too small to be affected by high ligand densities and also diffuse fast enough to avoid mass transport limitations. Signal maximization and thus a maximum binding site density are the first priority here, so the hydrogel of choice for these applications should have a high thickness and medium to high density, allowing a high loading of ligands on the hydrogel. Composite structures, i.e. hydrogels with additional filter layers, can be advantageous for samples containing particulate matter, such as blood or crude fermentation broths. An inert, relatively thick hydrogel on top of the ligandderivatized layer excludes particles and cells which might be present in the sample, as shown in Figure 6.13. These surfaces can also be used for assaying because such a layer acts as a diffusion barrier, so that most of the interaction takes place in a diffusion-controlled manner. The result is an extended linear slope of the binding curve which is easy to fit and facilitates the calculation of analyte concentrations. Charge should also be considered when selecting chip coatings for a particular application. Generally, the surface should be uncharged to minimize ionic non-specific interactions. Although most biomolecules are negatively charged and consequently repelled by the usually anionic chip surfaces, numerous applications exist where at least partially positively charged species are present which show a significant adsorption to polycarboxylate surfaces. Biomedical samples (e.g. serum) are typical examples of such analyte matrices and especially here a low background is crucial as very often the analyte concentrations are low. As a certain amount of COOH groups is usually required for electrostatic preconcentration of the ligand during the immobilization process and its covalent attachment, the carboxyl density has to be optimized in these cases. With dextran-based hydrogels it could be shown that a reduction of the carboxymethylation level from one COOH group per anhydroglucose unit to one group per 4–8 units effectively reduces ionic non-specific interactions with a still sufficient density of attachment sites. Situations may occur, however, where the charge density or the strength of the carboxyl functions is not sufficient. An example is the immobilization of

194

Figure 6.13

Chapter 6

Schematic view of a hydrogel with an additional filter layer. Only the short polymer chains within the evanescent field are carboxyl functionalized and can immobilize the ligand. The larger polymers carry no ligands. Their function is to prevent cells, cell debris and particulate contamination from entering the sensitive volume close to the chip surface.

compounds with a pI below 4. Within the pH range of the usually applied preconcentration buffers, these molecules are still negatively charged, rendering their electrostatic accumulation in the hydrogel matrix difficult. Below pH 4, an increasing percentage of the hydrogel-bound carboxylate groups is protonated, i.e. uncharged, further enhancing this difficulty. Introduction of strongly acidic groups such as sulfonate groups into the matrix is helpful as they stay unprotonated at sufficiently low pH and can attract even relatively acidic species. A way to introduce sulfonate groups is activation with sulfo-NHS instead of the non-sulfonated compound.

6.5 Coupling Procedures for Ligand Immobilization Coupling the ligand to the matrix on the sensor chip surface is a critical step when pursuing biomolecular interaction studies. The immobilization strategy should be chosen such that the immobilization level is sufficient and can be controlled. Also, the activity and steric accessibility of the ligand has to be preserved. In some cases, e.g. when the regeneration of the ligand–analyte pair is impossible or results in unacceptable activity losses of the ligand, repeated removal and re-immobilization of the ligand are advantageous, as this renders fresh ligand for each analyte binding cycle. Generally, immobilization methods can be divided into adsorptive, covalent, ionic and capture molecule-mediated coupling.

Surface Chemistry in SPR Technology

195

6.5.1 Adsorptive Immobilization Adsorptive methods, typically the immobilization to low-energy surfaces such as plastics via hydrophobic interactions, are the simplest and most popular approach in solid-phase assays as this process occurs spontaneously upon contact of a protein-containing buffer with a hydrophobic surface [27,28]. Less common is adsorption on noble metals, such as gold or silver, through strong thiol–metal bonds, although the method is easy to perform and – except for thorough cleaning – requires no prior surface preparation. Due to unavoidable structural changes, usually a large part of the adsorbed ligand is denatured [29]. Typically, less than 10% of the immobilized binding sites remain active and sterically accessible after adsorptive immobilization. Another disadvantage is the poor stabilization of the resulting surface against non-specific interactions and the impossibility of complete regeneration after analyte binding. Therefore, such surfaces usually have to be blocked with suitable blocker proteins5 and cannot be regenerated. As stable adsorption requires cooperative forces of several surfaceaffine functional groups, the adsorbate must have a sufficient number of such residues and a molecular weight of at least 10 kDa. This method is generally not suitable for non-derivatized nucleotides, peptides or small molecules.

6.5.2 Preconcentration Methods Prior to Covalent Immobilization Covalent methods are relatively straightforward, can give high coupling yields and form stable covalent bonds between the ligand and a suitable biocompatible sensor chip coating. Therefore, they are usually the first choice when evaluating different ways to immobilize biomolecules on a chip surface. A disadvantage is the randomly oriented coupling which occurs equally at active and not active sites of the ligand and hence can affect the affinity of at least a fraction of the immobilizate. In extreme cases – especially with small ligands, for example peptides – complete deactivation can even occur. Non-uniform coupling of the ligand may lead to distribution of rate and equilibrium constants, which can be determined with the distribution analysis method described in Chapter 12. Regardless which of the covalent coupling methods described further below is chosen, care must be taken that the ligand molecules approach close enough to the reactive groups as otherwise no reaction can take place. This basic rule of ‘‘interaction before reaction’’ is an important prerequisite for good coupling yields, but it can be difficult to realize on well-stabilized surfaces which are designed to suppress any interaction. This obvious dilemma can be addressed in two ways, as follows.

6.5.2.1

Electrostatic Preconcentration

If the surface is – usually negatively – charged, electrostatic preconcentration can take place by adjusting the pH of the coupling buffer so that the net charge 5

Typical examples are BSA or casein; other ready-made cocktails are commercially available.

196

Chapter 6

of the ligand is opposite to the surface charge. If the ionic strength of the coupling buffer is sufficiently low, typically below 20 mM, and the pI of the ligand is close to or optimally above the pI of the surface, the electrostatic attraction between surface and ligand will override the hydrophilic or steric stabilization. Under these conditions, the ligand accumulates at the surface until electrostatic neutrality is reached – the so-called preconcentration. Interestingly, due to the aforementioned counter-ion evaporation effect, sometimes electrosorption also occurs, when the overall charge of the ligand is neutral or even the same as the surface.

Protocol 6.1 Electrostatic preconcentration Preparation of the coupling buffers. The pH of the buffer should be at least 0.5 o pIProtein to ensure a positive net charge of the protein, which is required for electrostatic interaction with the negatively charged surface. Also, a very low ionic strength is essential. Prepare 5 mM buffers from formic and acetic maleic acid carefully titrated with 0.1 M NaOH and avoid overtitration as this will increase the salt concentration. Buffer ranges: pH 3.0–4.0: sodium formate pH 4.0–5.5: sodium acetate pH 5.5–6.0: sodium maleate These buffers are subject to rapid microbial growth and after sterile filtration should therefore be aliquoted, stored frozen until use and always used fresh. Do not add preservatives such as sodium azide, as azide interferes with later activation steps. Procedure to check the electrostatic preconcentration: 1. Mount a carboxylated sensor chip in the instrument. Make sure that the liquid handling system is free from any protein contamination, since even minor amounts of desorbed proteins will concentrate on the charged sensor surface. 2. Optional: to remove hydrophobically bound proteins from tubes, etc., clean the flow system with an easily desorbable detergent solution followed by doubly distilled water. If SDS is used, wash with a 10 min pulse 50 mM glycine  HCl pH 8.5 afterwards. 3. Elute electrostatically adsorbed contaminants from the surface for 10 min with 2 M NaCl, 10 mM NaOH. 4. Check the baseline with coupling buffer. After 10–15 min, almost no drift should be observed. 5. Inject protein solution to verify the preconcentration conditions. If no interaction occurs, the ionic strength or pH should be lowered. Increasing the protein concentration is optional but usually not helpful. Repeat the last three steps until a sufficient preconcentration of the protein is observed on the sensor surface. Note: some ligands come with significant salt contamination from previous purification steps or as preservative (for example, ammonia salts, Tris, sodium azide). These additives are often not stated on the product data sheet and can interfere severely with the preconcentration process and also quench the active groups. This results in a significantly reduced immobilization yield and therefore the ligand should generally be microdialyzed into the coupling buffer.

Surface Chemistry in SPR Technology

197

6. Elute the protein from the surface for 2–5 min with elution buffer. 7. Wash with water until the baseline is stable. Procedure for coupling ligand using the electrostatic preconcentration effect: 8. Prepare the buffer for covalent activation of choice and activate the surface. See paragraphs below for details. 9. Wash as quickly as possible with coupling buffer. 10. Inject the protein solution for 5–50 min. Since some of the carboxylate functions have been converted into non-charged groups by the matrix activation procedure, the preconcentration will usually be somewhat slower compared with the unactivated surface. The immobilization yield can be increased by repeated reinjection of the protein solution. 11. When working at low pH, inject water for 10 min to complete the coupling reaction, which proceeds slowly under these conditions. 12. Quench remaining active groups as described below. 13. Optional: remove physisorbed proteins with regeneration buffer. 14. Wash with coupling buffer. The amount of covalently bound protein can now be determined by comparison with step 4. 15. Switch to suitable running buffer and start interaction experiments. Note: At least two analysis cycles (see next chapter) are needed to equilibrate and stabilize the coating for reliable and accurate measurements of the rate and affinity constants.

The preconcentration effect is independent of the ligand concentration, it works even at concentrations as low as a few mg ligand ml–1 until the coupling buffer is practically quantitatively depleted of ligand. As the electrostatic forces work in a cooperative manner, the efficiency of the preconcentration is proportional to the molecular weight of the ligand. Achievable ligand densities depend mainly on the surface nanoarchitecture, i.e. 2D or 3D coatings, and on the thickness and density of the latter. With hydrogels as thick as 1000 nm, high protein densities of well above 100 ng mm–2 (and 4106 RU or 4101 SPR angle shift) can be reached. Apart from changing the ligand concentration and contact time, the ligand density is controlled by adjusting the ionic strength of the coupling buffer or varying the activation level. The latter is the preferred variable as lowering the density of activated groups also prevents multisite coupling, which can lead to unwanted crosslinking and ligand deactivation. In this context it should be noted that the preconcentration after activation is typically lower than that of an underivatized surface. This is caused by partial conversion of charged groups into neutral and insoluble moieties, which results in a decrease of the overall charge density and also in a lower hydrogel hydration, which in turn can lead to shrinking, and thus reduction of the available volume for ligand immobilization.

6.5.2.2

Dry Immobilization

This less frequently employed method is useful when it is not possible to preconcentrate a ligand electrostatically on to the sensor chip surface. This is

198

Chapter 6

the case with ligands having a pI below 4, such as DNA, with relatively small and acidic compounds such as some peptides or if the surface is not charged. In addition to bearing functionalities which are able to provide covalent coupling, the ligand to be immobilized must be robust enough to survive the drying which is necessary to bring it close to the reactive groups on the sensor surface. Suitable sensor surfaces have to be preactivated offline (see Protocol 6.2). Alternatively, NHS preactivated chips are commercially available. Care should be taken that the surface functional groups present on the surface cannot react with each other, as such self-quenching drastically reduces the coupling yield. Commonly used carboxymethyldextran hydrogel surfaces, for example, are not suitable for dry immobilization, as the NHS esters form esters with abundant hydroxyl functionalities of the polysaccharide matrix, leading to deactivation and unwanted crosslinking, a side-reaction that occurs – although to a lesser extent – also during the standard wet activation process. The advantages of dry immobilization are high coupling yields and the possibility of using less reactive coupling chemistries, as for example epoxide activation, which require harsher reaction conditions, such as elevated temperatures.

Protocol 6.2 Dry immobilization (off-line) 1. Place chips in a suitable receptacle and incubate the surface for 10 min. with 2 M NaCl, 10 mM NaOH. 2. Wash thoroughly with doubly distilled (dd) water. 3. Prepare the buffer for covalent activation and activate the surface. See paragraphs below for details. 4. Wash thoroughly with dd water and remove liquid traces with a quick centrifuge spin or a sharp jet of clean compressed air/nitrogen. For reproducible drying the direction of the jet of air/nitrogen is important, e.g. 451 to the surface. It is essential that no droplets dry on the surface as even dd water leaves contamination behind which might interfere with later processing steps. 5. Place chips in a dry and clean receptacle that is small enough to be placed in a vacuum desiccator. 6. Spot 10–30 ml of a solution of at least 0.3 mg ligand ml–1 low ionic strength coupling buffer manually on to the chip surface so that the sensing area is entirely covered. Commercial spotters, preferably contactless spotters, may be used to produce a microarray to be measured with an SPR imaging instrument.* Dry as fast as possible in a vacuum desiccator but without desiccative to avoid potential hydrolysis. Note: some ligands come with significant salt contamination from previous purification steps or as preservative (for example ammonia salts, Tris, sodium azide). These additives are often not stated on the product data sheet and can quench the active groups. As this results in a significantly reduced immobilization yield, the ligand should generally be microdialysed into coupling buffer before spotting. 7. Incubate for 1–6 h depending on the activation employed. With less reactive chemistry and robust ligands, additional heating to elevated temperatures up to 90 1C might be necessary.

Surface Chemistry in SPR Technology

199

8. If the chips are not used immediately, put them in a freezer at this stage of the process. 9. Dissolve unbound ligand over 1 h with running buffer, wash with dd water and dry as described above. 10. The chips are now ready for interaction analysis. *Commercial spotters usually spot 0.1–1 nl and the spot diameter may vary between 50 and 300 mm, which depends on the hydrophilicity of the surface. Be aware that the ligand concentration in nanograms per square mm should be recalculated from the spot size, concentration of ligand in the spotting liquid and the coupling yield which can vary between a few and 50%.

6.5.3 Covalent Activation Chemistries Some common activation chemistries to covalently immobilize the ligand to the chip surface are listed in the following sections. Most are based on carboxylic acid residues as this functional group is present on many common immobilization matrices.

6.5.3.1

Amine Coupling via Reactive Esters

Due to its flexibility, relative ease of use, high coupling yields and robustness, amine coupling via reactive esters is the most frequently employed immobilization method. The reaction conditions for coupling proteins, peptides and small molecules to carboxylated chip surfaces are well characterized and extensive optimization studies have been performed [30]. Typically, the surface-bound carboxyl groups are first activated by a carbodiimide and converted into active ester intermediates, which are then aminolyzed by lysines or the N-terminal NH2 group of the ligand. For more than 50 years, carbodiimides have been used to mediate the formation of amide bonds between a carboxylate and a primary or secondary amine [31,32]. Over time a number of different reagents has been developed, all forming a reactive O-acylisourea intermediate with the COOH function, which is reactive towards nucleophiles, such as primary and secondary amines, hydrazides, primary and some secondary alcohols, thiols and – an important side-reaction – water. Whereas many carbodiimides, for example dicyclohexylcarbodiimide (DCC), are water-insoluble and of limited use for bioconjugation, others are readily soluble in water and therefore the reagents of choice for water-based activation methods [33]. 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) is most frequently employed for conjugating biomolecules. Not only the reagent itself but also the isourea reaction byproducts are water soluble, facilitating activation and chip processing. EDC is usually supplied as the hydrochloride, which is both hygroscopic and labile in the presence of water. It should be stored in a dry container at –20 1C and allowed to warm to room temperature before opening in order to prevent water condensation, which will cause decomposition over time. Ideally, the EDC container should be opened under dry gas, such as

200

Chapter 6

N2. For the same reason, EDC-containing activation buffers should always be prepared fresh and not from stock solutions. Note that inactive EDC is among the most frequent reasons for failed immobilizations. EDC activation proceeds effectively from pH 4.5 to 6.5 in 2-(N-morpholino)ethanesulfonic acid (MES) or phosphate buffer. Amine- or carboxylate-containing buffers should be avoided as they would react with the carbodiimide. As the O-acylisourea intermediate hydrolyses within seconds, it would not survive the time from activation until injection of the ligand. Therefore, it has to be converted into a more stable active ester. This is achieved by adding the water-soluble compound N-hydroxysuccinimide or its sulfonated form N-hydroxysulfosuccinimide to the activation mix. In contrast to O-acylisourea, the thus formed (sulfo-)NHS ester intermediate has a half-life from minutes at pH 8–9 up to several hours at pH 4–5. EDC/(sulfo)NHS coupling (Figure 6.14) is highly efficient but one should be aware that side-reactions can occur, for example esterification with abundant hydroxyl groups of polysaccharide hydrogel coatings. The resulting esters, however, are subject to hydrolysis, as are also temporarily formed thioesters or acylated histidine side-chain nitrogens [34]. The high efficiency of the reaction can result in multisite coupling and crosslinking if larger ligands with many lysines are immobilized. Finally, the transesterification tendency of NHS esters

Figure 6.14

Reaction scheme of COOH activation with EDC/NHS and ligand coupling.

Surface Chemistry in SPR Technology

201

should be noted as it can lead to unwanted cross-activation of ligand carboxyl groups or carboxylate buffer components. The aminolysis of (sulfo-)NHS esters occurs smoothly at slightly alkaline pH, i.e. in carbonate or borate buffers (pH 8–10) where most amines are deprotonated. This is therefore the reaction condition of choice for immobilization of small ligands and many peptides. For the immobilization of larger species, such as most proteins, it is advisable to create conditions where only a small fraction of the lysine residues is reactive in order to prevent multisite coupling and preserve the protein’s activity. A low pH between 4 and 6 is therefore recommended in these cases and also required in order to achieve the above-described electrosorption effect. As an additional advantage, coupling the amine terminus of proteins is favored over random immobilization through lysine residues at lower pH values. The reason for this effect is the lower pI of N-terminal amines compared with lysine NH2 groups. Such optimization of the pH can result in a significantly increased activity, especially of sensitive proteins. In some cases, especially with labile ligands or when preparing high-capacity surfaces for low molecular weight compound analysis, slight crosslinking of the immobilized ligand by injecting a short pulse of low concentration EDC/NHS over the surface-bound protein can result in additional stability of the immobilized ligand. However, the applicability of such an additional crosslinking step should be tested empirically.

Protocol 6.3 The EDC–NHS coupling procedure 1. Prepare a solution of 0.5 M N-hydroxysuccinimide (NHS) in 0.5 M 2-(N-morpholino)ethanesulfonic acid (MES) buffer, pH 5.5. Optionally, 50 mM sulfo-NHS in 50 mM MES, pH 5.3 may be used. 2. Prime a COOH-modified sensor chip according to the procedure to check the electrostatic preconcentration as described above. Elute electrostatically adsorbed contaminants from the surface for 10 min with 2 M NaCl, 10 mM NaOH. 3. Check the baseline with coupling buffer. After 10–15 min, almost no drift should be observed. 4. Make a 5–100 mM solution of solid N-ethyl-N 0 -(dimethylaminopropyl)carbodiimide (EDC) in the NHS/MES buffer from step 1. The optimal EDC concentration depends on the coating and the ligand. A low EDC concentration (around 10 mM) is recommended for linear polycarboxylate coatings and for the immobilization of larger ligands with abundant lysine residues. Planar coatings and CMD hydrogels can be activated with higher EDC concentrations, especially if the ligand is small. 5. Inject this activation mix over the sensor chip for 1–20 min. Variation of the incubation time allows control of the activation level. To achieve a lower activation level, which is recommended, for example, for polycarboxylate hydrogels, an activation time o10 min should be chosen. 6. For the standard wet immobilization protocol, wash briefly with water or slightly acidic coupling buffer and continue with injection of ligand. For dry immobilization, wash with 5 mM acetic acid and dry according to Protocol 6.2.

202

6.5.3.2

Chapter 6

Amine Coupling Through Reductive Amination

Carbonyl groups such as aldehydes, ketones and glyoxals can react with amines to form labile Schiff base intermediates, which can be reduced to yield stable secondary amines (Figure 6.15). The Schiff base formation is efficient at both low and high pH and sodium borohydride and sodium cyanoborohydride are the most commonly used reduction reagents. The difference is that cyanoborohydride is about five times milder [35] and – in contrast to sodium borohydride – does not reduce free aldehyde groups, i.e. leads to higher coupling yields by shifting the reaction equilibrium towards the product side. Reductive amination of Schiff bases is a highly selective reaction that proceeds smoothly under mild conditions and – unlike coupling through amide

Figure 6.15

Oxidation of a polysaccharide hydrogel and immobilization of an aminecontaining ligand through reductive amination.

Surface Chemistry in SPR Technology

203

bonds – preserves the positive charge of the reacted amine. It is therefore an interesting alternative for the immobilization of sensitive ligands to polysaccharide surfaces which, after periodate oxidation, display a sufficient density of aldehyde groups in the form of cyclic hemiacetal structures [36]. Useful chip substrates for this method are either coated with non-derivatized dextran or carboxymethyldextran with a low degree of carboxymethylation. The latter has the advantage that it allows electrosorptive preconcentration of protein, i.e. working with lower ligand concentrations in the immobilization buffer. In addition to coupling via active esters and reductive amination, several other amine reactive chemistries exist but are not further described here as their use with evanescent field-based biosensors is limited.

Protocol 6.4 Amine coupling through reductive amination 1. Prime the chip surface as described above (see points 2 and 3 in the EDC/NHS coupling Protocol 6.3). 2. Inject 10 mM sodium periodate in water over 30 min. To control the density of aldehyde groups, the concentration and contact time may be varied. 3. If working without applying the electrosorption effect, inject the ligand at a concentration of at least 1 mg ml–1 in in phosphate, borate or carbonate buffer (pH 7–10) plus 10 mM sodium cyanoborohydride (Caution: extremely toxic!). pH 8–9 is optimal and other buffers may be used provided that they do not contain competing amines (such as Tris). If protein ligands are electrostatically preconcentrated, a lower pH and lower ionic strength are required. As the reaction proceeds slowly under these conditions, long contact times of several hours should be chosen. 4. After completion of the coupling reaction, inject 0.5 M Tris or ethanolamine  HCl (pH 8.0) plus 10 mM sodium cyanoborohydride for 1 h. Alternatively, the excess aldehydes may be reduced by injection of 50 mM sodium borohydride in 0.1 M sodium carbonate (pH 9). 5. Wash with running buffer of choice until the baseline is stable and start the interaction analysis cycle.

6.5.3.3

Thiol Coupling

An alternative to coupling of proteins through their amine functionalities is the use of thiol groups after reduction of pendant disulfide bridges (Figure 6.16). Choosing this immobilization strategy makes sense if amines are not present or a sufficient number of disulfide groups are available which are located far enough

Figure 6.16

Reduction of disulfide-containing compounds.

204

Chapter 6

from the protein’s active site. In such cases a higher activity of the immobilizate can be achieved compared with the more randomly oriented amine coupling [37]. Immobilization of the reduced proteins can occur either by disulfide exchange using activated disulfides, by nucleophilic substitution of haloacetyls or by alkylation of maleimides. The first method – formation of disulfides – is reversible under reducing conditions, a relevant parameter for full regeneration of chip surfaces [38]. The last two chemistries yield stable bonds which cannot be cleaved again. Whereas the site-directed coupling can result in a higher activity of the immobilizate, the additionally introduced disulfide, acetyl or maleimide residues can lead to slightly increased non-specific interactions. Prior to thiol coupling, the protein of interest has to be reduced. This is usually done using thiol-containing compounds such as dithiothreitol (DTT), 2-mercaptoethanol, 2-mercaptoacetic acid or 2-mercaptoethylamine. For a more convenient purification of the reaction mixture after the reduction step, immobilized reducing agents may be used. The use of complexing additives is generally recommended to prevent reoxidation that is catalyzed by traces of heavy metal ions. If using alternative protocols it should be ensured that no reducing reagents remain in the reduced protein solution.

Protocol 6.5 Protein disulfide reduction procedure 1. Add 0.5 ml of thiolated agarose to a column and equilibrate with 0.1 M phosphate buffer pH 8.0. 2. Activate the column by adding 1 ml of 10 mM DTT in 0.1 M phosphate buffer pH 8.0 containing 1 mM EDTA. 3. Wash the column with 20 column volumes of 0.1 M phosphate buffer pH 8.0. 4. Add the protein to be reduced in 0.1 M phosphate buffer pH 8.0 containing 1 mM EDTA. Recover fractions and collect all samples. Read the absorbance at 280 nm to determine which fractions contain the eluted protein. Pool these fractions, aliquot and freeze at –20 1C until required.

During the thiol-disulfide exchange, the free thiol of the reduced ligand reacts with an activated disulfide on the chip surface and forms a new mixed disulfide with release of a leaving group – usually pyridyl disulfide (Figure 6.17). The latter is easily transformed into pyridine-2-thione, a non-reactive compound not capable of participating in further mixed disulfide formation. Thus only the surface bound end of the mixed pyridyl disulfide has potential for becoming attached to the sulfhydryl-containing ligand. This disulfide exchange reaction occurs over a broad pH range and in a variety of buffers, including acidic, low ionic strength regimes used for electrostatic ligand preconcentration. As already mentioned, the resulting disulfide bond is cleaved under reducing conditions and is also affected by thiol buffer additives.

Surface Chemistry in SPR Technology

Figure 6.17

205

Pyridyl disulfide-mediated disulfide exchange.

Protocol 6.6 Thiol-disulfide exchange procedure 1. Prime the disulfide-derivatized chip (see points 2 and 3 in the EDC/NHS coupling Protocol 3 as described above). 2. Inject 100 mM DTT or other reducing agents in 0.1 M phosphate buffer (pH 8.0) for 20 min. 3. Wash for 5 min with running buffer. 4. Inject 10 mM pyridyl disulfide in 0.1 M phosphate buffer +20% ethanol for 20 min. (Note: dissolve pyridyl disulfide in 100% ethanol before diluting to 10 mM with phosphate buffer–ethanol mixture). 5. Wash for 15 min with running buffer. 6. Switch to water as running buffer and wash the system for 10 min until the baseline is stable. 7. Inject 10–100 mg reduced protein ml–1 coupling buffer plus 1 mM EDTA for 10–20 min. Significant preconcentration should occur in this step. 8. Quench excess active disulfides with 1 mM mercaptoethanol in 0.1 M acetate buffer (pH 4.2), 1 M NaCl for 30 min (prepare fresh). 9. Wash with running buffer of choice until the baseline is stable and start the interaction analysis cycle.

Thiol Coupling via Maleimides. For irreversible immobilization via thiol groups, maleimide or haloacetyl coupling is the method of choice (Figure 6.18). The alkene group of the maleimides undergoes an alkylation reaction with the sulfhydryls and forms a stable thioether bond. The specificity of this coupling for sulfhydryl groups is high at neutral pH [39], whereas at higher pH some cross-reactivity with amino groups may occur [40]. At pH 7 the reaction of maleimides with sulfhydryls is about

206

Figure 6.18

Chapter 6

Maleimide coupling of sulfhydryl groups.

1000 times faster than its reaction with amines [41]. As a further sidereaction – especially at higher pH – the maleimide group may undergo hydrolysis to an open maleamic acid form, before but also after the reaction with a sulfhydryl. In addition to the maleimide method described here, sulfhydryl-containing ligands can also be immobilized through haloacetyl or epoxy activation (see Section 6.5.3.5).

Protocol 6.7 Maleimide coupling of sulfhydryl groups 1. Activate a carboxylated surface with EDC/NHS as described in Protocol 6.3. Use of a 10-fold diluted activation mixture should be considered as such a reduced activation level is sufficient to obtain a good immobilization yield with most proteins. 2. Convert the active NHS esters into amino groups by injection of 1 M ethylenediamine hydrochloride (pH 6.0) for 10 min. 3. Quench remaining reactive groups with 1 M ethanolamine hydrochloride (pH 8.5) for 15 min. 4. Wash for 10 min with 0.1 M HCl. 5. Inject 20–50 mM heterobifunctional reagent, N-g-maleimidobutyryloxysuccinimide, in HBS for 10 min. 6. Wash for 5 min with dd water. 7. Inject the reduced protein (Protocol 6.5) in suitable coupling buffer (see Protocol 6.1) for 10 min. 8. Wash for 5 min with HBS. 9. Hydrolyze excess unreacted maleimido groups by a 10 min exposure to 0.1 M NaOH. Other, milder solutions may be used for inactivation but require a longer contact time. 10. Wash with running buffer of choice until the baseline is stable and start the interaction analysis cycle. Alternatively, maleimide preactivated chips, which are commercially available, may be used. In this case, start with step 6. Note: as with most other protocols, at least two analysis cycles (see the next chapter) are needed to equilibrate and stabilize the coating for reliable and accurate measurements of the rate and affinity constants.

Surface Chemistry in SPR Technology

Figure 6.19

6.5.3.4

207

Hydrazide coupling of a carbohydrate through its anomeric aldehyde end group.

Immobilization of Aldehydes Through Hydrazide Groups

Aldehyde groups can be valuable reactive sites when immobilizing carbohydrates or glycoproteins [42]. In the case of carbohydrates, often the anomeric aldehyde, i.e. the product of the mutarotation, is used (Figure 6.19); they can also be created by mild oxidation of sugar residues by sodium periodate. Alternatively, amino groups can be converted with glutaraldehyde. As carbohydrate groups are typically located at some distance from the ligand’s binding site, immobilization through them might be considered a useful method to retain the activity of immobilized glycoproteins. However, as immobilization using the EDC/ NHS method can also be performed under very mild conditions, the difference between the two methods is often smaller then expected; a comparative study of antigen binding capacities of immobilized monoclonal antibodies showed almost no difference between EDC/NHS- and hydrazide-mediated coupling [43]. Hydrazide-derivatized sensor chips can be generated in situ by EDC/NHS activation of carboxylated surfaces followed by reaction with hydrazine or the bifunctional and less toxic adipic acid hydrazide solution. The activation level and thus the density of immobilized ligand can be controlled through concentration of the EDC/NHS mixture and by the reaction time. To avoid crosslinking of 3D surface structures, the concentration of the hydrazine and dihydrazide should be chosen high enough. The coupling reaction itself proceeds best at slightly acidic pH and is therefore compatible with the conditions chosen for electrostatic preconcentration. Small amounts of sulfate catalyze the reaction. The hydrazone bond is relatively stable at neutral and alkaline pH but somewhat labile in acidic buffers. To avoid ligand leakage, it should be stabilized by reduction with cyanoborohydride. Due to their typically neutral to acidic nature, carbohydrates cannot be immobilized using the electrostatic preconcentration effect (Protocol 6.1). The method to follow is either injection at high concentrations (several tens of mg ml–1) or offline immobilization using the dry method according to Protocol 6.2.

208

Chapter 6

Protocol 6.8 Glycoprotein oxidation 1. Dissolve the protein at a concentration of 3–10 mg ml–1 in 10 mm sodium phosphate, 0.1 M NaCl (pH 6.2). 2. From a freshly prepared sodium periodate stock solution add a sufficient volume to reach a final concentration of 5–10 mM. 3. Carry out reaction for 30 min in the dark. 4. Immediately purify the oxidized protein into slightly acidic immobilization buffer (see above) using a suitable column or spin column. Ensure that the purified protein contains no periodate contamination. 5. Pool protein fractions, eventually dilute with immobilization buffer to a concentration of 30–300 mg ml–1 and inject immediately (point 5 in Protocol 6.9), as due to intermolecular Schiff base formation self-polymerization, deactivation and precipitation may occur over time. If solutions have to be stored, freeze at –20 1C.

Protocol 6.9 Hydrazide activation of a carboxylated surface and coupling of aldehyde-derivatized ligands 1. Prime the surface according to Protocol 3 as described above. 2. Activate for 3 min with 0.2 M EDC, 0.05 M NHS. For lower ligand densities dilute this activation mixture. 3. React for 15 min with 0.1 M adipic acid hydrazide hydrochloride (pH 8.0). 4. Quench remaining NHS esters over 30 min with 1 M ethanolamine hydrochloride (pH 8.5). 5. Inject oxidized ligand according to Protocol 6.8 and react for at least 20 min. 6. Stabilize hydrazone bonds by reduction with 50 mM sodium cyanoborohydride (Caution: toxic! – prepare in a fume hood) in 0.1 M acetate buffer (pH 4.0) for 20 min. 7. Wash with running buffer of choice until the baseline is stable and start interaction analysis.

6.5.3.5

Coupling Through Epoxy Groups

Epoxy-mediated immobilization is a robust method with a long tradition, especially for immobilization of carbohydrate ligands in affinity chromatography [44]. With optical biosensors this technique is relatively seldom employed, as fully carboxymethylated dextran hydrogels are difficult to activate and the previously described alternative methods work under milder reaction conditions. However, for coupling of carbohydrates, but also for selective reaction with amine and sulfhydryl nucleophiles, epoxy coupling can

Surface Chemistry in SPR Technology

Figure 6.20

209

Epoxy activation of a dextran matrix and coupling of a carbohydrate.

be an interesting alternative (Figure 6.20). Note that crosslinking can occur as a side-reaction. Protocol 6.10 describes epoxy sensor chip activation and dry immobilization of carbohydrates. It can be applied if the ligand is robust enough to survive drying and heating at strongly alkaline pH.

210

Chapter 6

Protocol 6.10 Epoxy activation of chip surfaces and coupling of carbohydrates 1. Prime dextran or partially carboxymethylated dextran-coated chip according to Protocol 6.3 as described above. Fully carboxymethylated dextran is difficult to epoxy activate and should not be used. 2. Activate for 15 min with 2% epichlorohydrin (Caution: toxic!) in 0.1 M potassium hydroxide. For a lower activation level and a lower ligand density, the pH and reaction time should be decreased. 3. Wash chip thoroughly with water and dry according to Protocol 6.2 as described above. 4. Prepare 100 ml of a solution of 10–100 mg ligand ml–1 10 mM NaOH and cover the sensing area of the chip with a few ml of this solution so that it is totally covered by the liquid. 5. Dry thoroughly. 6. Heat for 30 min at 70–90 1C. Lowering the reaction temperature is another possiblity to decrease the coupling yield. 7. Wash for 2 h with running buffer. 8. Wash with water, dry as described above (Protocol 6.2) and install the chip to start an analysis cycle.

In addition to using the above-described methods, ligands can be covalently immobilized via radical substitution, photocoupling or other crosslinking chemistries. Generally, most methods employed in bioconjugate synthesis or for the preparation of supports for affinity chromatography are also suitable for the derivatization of sensor chip coatings.

6.5.4 Electrostatic Methods Electrostatic/ionic immobilization is a technique frequently used for the attachment of oligonucleotides to polyamine-coated microarray substrates. As it is non-selective and incurs considerable electrostatic forces towards the immobilized ligand, the drawbacks of this method are similar to those of adsorptive techniques (see Sections 6.3.1 and 6.5.1). It is therefore suitable for robust ligands only. A specific feature of this technique is its potential reversibility by changing the ionic strength and/or pH or through the addition of chelating additives. However, this results in labile immobilization complexes, significant leakage of ligand and, as a consequence, an unstable baseline. Therefore, this method is still seldom employed for biosensors. An exemption is the use of immobilized nitrilotriacetic acid (NTA) or iminodiacetic acid derivatives or similar chelating groups, which allow reversible immobilization of His6-tagged molecules through complex formation with transition metal ions, preferably Ni21 [45–47].6 Such complexes are relatively stable, although interestingly more sensitive to changes in conditions than NTA affinity columns. For total regeneration of the chip surface they can be cleaved by chelating agents such as EDTA. 6

Protocol 6.12.

Surface Chemistry in SPR Technology

211

Protocol 6.11 Electrostatic adsorption of DNA on (poly)amine coated sensorchips 1. Prime aminomodified chip surface for 10 min with 0.1 M HCl. 2. Wash for 2 min with degassed 5 mM sodium acetate (pH 5.0). 3. Inject 1–100 mM DNA in degassed 5 mM sodium acetate (pH 5.0) for 5 min. A baseline increase between 3000 and 6000 mRIU (about 2000–4000 RU or 200– 300 mdeg) should be observed. 4. Equilibrate for at least 15 min with running buffer as used for the interaction experiment. Depending on the ionic strength and pH, a certain fraction of the electrosorbed DNA is desorbed. 2 M NaCl plus 10 mM NaOH results in quantitative desorption. 5. For a more stable immobilization, activate the surface with epichlorohydrin as described in Protocol 6.10 and use amino- or sulfhydryl-derivatized DNA.

Protocol 6.12 Ni21-mediated immobilization of His6-tagged ligands [47] 1. Mount the sensor chip derivatized with NTA or another suitable chelator. 2. Prepare the immobilization buffer: 10 mM HEPES, 0.15 M NaCl, 0.005% Tween 20 (pH 7.5). Higher NaCl concentrations and a higher pH lower the immobilization yield; lowering the pH to 6.9 increases it. 3. Optional: microdialyze a 50–200 nM solution of ligand into immobilization buffer. 4. Condition the surface for 5 min with 0.5 M Na EDTA (pH 8.5). 5. Wash for 2 min with immobilization buffer. 6. Inject 0.3 M NiCl2 or NiSO4 for 2 min. Depending on the density of chelating groups, a 20–60 RU baseline increase should be observed. 7. Wash for 2 min with immobilization buffer. 8. Inject 50–200 nM ligand in immobilization buffer for 2–5 min. Through pH and ionic strength of the immobilization buffer the amount of immobilized ligand can be controlled via the ligand concentration and contact time and should be relatively low as a certain number of unoccupied Ni complexes are required for stabilization, i.e. continuous rebinding of the weakly bound (KDo10–6) His6-tagged immobilizate. The binding is improved at very low flow rates and at high pH. 9. Equilibrate with sample buffer until the baseline is stable. HEPES was found to give the best results but PBS and Tris also work. Low concentrations (maximum 50 mM) of EDTA in the sample buffer stabilize the assay as it scavenges contaminating metal ions. 10. Inject analyte in sample buffer and analyze interaction. 11. Instead of regenerating the ligand–analyte interaction, the undissociated ligand– analyte pair can be quantitatively removed with a 3 min injection of 0.5 M Na EDTA (pH 8.5). If the surface is of good quality, about 100 such regenerations can be carried out without significantly affecting the surface capacity. This also means that the same chip can be used for many different proteins.

212

Chapter 6

6.5.5 Directed Immobilization Oriented coupling with retention of the ligand’s activity and structure can be achieved via indirect immobilization through site-specific capture molecules on the chip surface. These can be proteins such as antibodies or protein A [48], but also smaller, more robust moieties such as biotin [37,49]. While such indirect methods seem to be less problematic and – at least theoretically – should render a higher percentage of ligands still active after the immobilization process, one should be aware that bulky (protein) linkers occupy a significant fraction of the evanescent field volume, which is then not available for ligand molecules and can result in smaller signals and – especially in the case of high molecular weight analytes and hydrogel matrices – might cause steric hindrance. Furthermore, larger capture molecules can alter the affinity profile of the immobilized ligand or induce non-specific interactions. An interesting feature of regenerable affinity-based ligand surfaces such as protein A is that the ligand can be removed together with the analyte during regeneration. This is advantageous if it is difficult to break the ligand–analyte interaction. When using biotinylated molecules, one should take into account that biotinylation is usually carried out with NHS-activated biotin derivatives. However, covalent coupling of unbiotinylated ligand using the EDC/NHS method as the coupling chemistry is identical but usually gives higher immobilization yields.

Protocol 6.13 Immobilization of biotinylated ligands on streptavidin-modified surfaces 1. Prepare a 1–20 mg ml–1 solution of ligand in compatible buffer. The choice of buffer is not critical as the biotin–streptavidin interaction is strong (KDc10–15) and takes place under a wide range of conditions. More important are minor contaminations of free biotin often present with biotinylated biomolecules. As biotin is usually much smaller then the biotinylated ligand, it diffuses faster to the yet unoccupied binding sites and can drastically reduce the amount of bound ligand. As a general precaution one should therefore thoroughly microdialyze (or otherwise purify) the ligand into the corresponding buffer. 2. Equilibrate a streptavidin-modified sensor chip with immobilization buffer. 3. Inject ligand solution for 1–30 min. The immobilization level can be controlled through the ligand concentration and contact time. Although the binding itself proceeds rapidly, larger ligands generally require more time to reach the biotin binding pockets than do smaller ones. 4. Run 2–3 test analysis cycles. Depending on the strength of the regeneration agent, it might happen that a more or less significant fraction of the immobilized ligand is desorbed again. This is due to a few low-affinity binding sites but is also caused by dissociation of the tetrameric streptavidin into monomeric subunits. In this case re-inject ligand solution over the chip and repeat analysis cycles until a stable baseline is reached.

Surface Chemistry in SPR Technology

213

Protocol 6.14 Immobilization of antibodies on protein Amodified surfaces 1. Equilibrate a protein A-modified sensor chip with sample buffer for 5 min until the baseline is stable. Alternatively, protein A can be immobilized on a carboxylated chip surface following Protocol 6.3. 2. Inject 5–50 mg of IgG in physiological sample buffer for 3 min. 3. Wash for 2 min with sample buffer. 4. Inject the analyte sample and record the sensorgram. 5. Optional: inject sample buffer and monitor dissociation. 6. Strip the analyte together with ligand IgG with two 1 min pulses of 0.1 M HCl. The protein A surface is now ready for immobilization of fresh IgG. Protein A is robust and survives several hundred regeneration cycles.

6.5.6 Immobilization of Membrane Proteins Although direct adsorptive immobilization of protein ligands is not optimal for specific biomolecular interaction analysis (see Section 6.5.1), hydrophobically driven adsorption can be a valuable method for indirect immobilization of membrane-bound receptor molecules. Usually, these proteins denature when removed from the lipid bilayer membrane and – if not stabilized with detergents – require a more or less intact membrane environment to display their normal functionality. This can be achieved by either first integrating t he transmembrane species into vesicles and liposomes, which are then fused on a hydrophobic surface or on an amphiphilic hydrogel [50,51], or by first immobilizing the membrane protein on a partially alkyl-derivatized hydrogel sensor chip followed by on-chip reconstitution with a lipid–detergent mixture [52]. The approach using a hydrogel layer has the advantages that the lipid bilayer is supported by a hydrated structure, better resembling the natural environment, and that the preparation process is more robust, i.e. not so easily affected by impurities as is the case with strongly hydrophobic surfaces.

Protocol 6.15 Preparation of mixed micelles [50] 1. Prepare a lipid film by pipetting a small volume of 10 mM lipid in chloroform solution into chloroform-washed round-bottomed glass tubes and drying this solution under a nitrogen stream (fume hood). Slight heating in a 30 1C waterbath accelerates the drying process. 2. Evacuate for at least 2 h in a vacuum desiccator to remove traces of solvent. 3. Add 25 mM of the detergent octylglucoside (OG) in 9 mM HEPES (pH 6.4) and 135 mM NaCl to yield a 3.3 mM lipid solution. If other detergents are used, the volume of the detergent solution should be adjusted such that the ([detergent]– cmcdetergent)/[lipid] ratio is between 2 and 3.5. Note that the optimal ratio depends on the lipid and the detergent and must be identified empirically from case to case.

214

Chapter 6

Protocol 6.16 On surface reconstitution of lipid bilayers without and with immobilized receptor protein [50] Priming the surface: 1. Mount a partially alkyl-derivatized carboxylated hydrogel coated sensor chip. 2. Wash the surface for 5 min with 20 mM Chaps. Optional immobilization of receptor protein: 3. Activate the surface for 7 min with 0.2 M N-ethyl-N-dimethylaminopropylcarbodiimide and 50 mM N-hydroxysuccinimide. 4. Inject 10–100 mg of receptor protein in suitable coupling buffer (see Protocol 6.1) plus 20 mM octyl glucoside for 10–20 min. 5. Quench remaining active groups for 15 min with 1 M ethanolamine pH 8.5 plus 20 mM octylglucoside. Formation of lipid bilayer and reconstitution of receptor protein: 6. Inject the mixed micelles (prepared as described in Protocol 6.15) for 2 min and slow flow. 7. Wash with running buffer and start interaction experiments. Note: The surface can be completely regenerated by two 2 min injections of 20 mM Chaps or 50 mM octyl glucoside.

Alternatively, liposomes or cell fragments can be modified with suitable tags such as biotin, oligonucleotides or His6, which are then specifically bound by the corresponding capture molecules immobilized on a hydrophilic surface. The ligand (receptor protein) density can be controlled either via the lipid/ transmembrane protein ratio during the vesicle preparation or – when using the on-chip reconstitution protocol – by varying the NHS activation level (Protocol 6.12, step 3). In summary, there is no ideal, general-purpose immobilization strategy and the optimal choice depends on the experimental approach, and the nature of the ligand, analyte and sample buffer. Generally, it is desirable that the immobilization chemistry does not interfere with the activity and binding characteristics of the ligand by linking to the sensor chip surface. In most cases, covalent methods give good results. As a rule of thumb, it is advisable to begin experiments with standard amino coupling chemistry, combined with low- to medium-density hydrogels (see Chapter 6.4.2), and, if unsuccessful, to try alternative, site-directed immobilization strategies. As can be seen in Figure 6.21, the standard amino coupling (Protocol 6.3), due to its robustness and good immobilization yields is the most popular method, followed by the direct immobilization of biotinylated ligands. However, one should keep in mind that the popularity of different immobilization methods is significantly predetermined by the commercial availability of suitable sensor chip surfaces. Table 6.4 might be helpful in identifying

215

Surface Chemistry in SPR Technology

Figure 6.21

Table 6.4

Relative use of different immobilization techniques in 2003 [53].

Immobilization strategies for different ligands with protocol numbers in parentheses (the recommended methods are listed first).

Ligand

Immobilization method

Remarks

Proteins, general

EDC/NHS (6.3) Thiol exchange, maleimide (6.6, 6.7) Reductive amination (6.4) Ni21/ His6 (6.12) Biotin/ streptavidin (6.13) Direct adsorption On-surface reconstitution (6.16) Incorporation into labeled micelles EDC/NHS (6.3) Protein A (6.14) Thiol exchange, maleimide (6.6, 6.7) Ni21/His6 (6.12) Hydrazide (6.9) Biotin/ streptavidin (6.13) EDC/NHS (6.3) Thiol exchange, maleimide (6.6, 6.7) EDC/NHS (6.3) Hydrazide (6.9) Thiol exchange, maleimide (6.6, 6.7) EDC/NHS (6.3) (6.2) Reductive amination (6.4) EDC/NHSactivated peptides to

Dry coupling may be necessary if pI o 4, as preconcentration becomes difficult

Membrane proteins Antibodies

Antibody fragments, affibodies Glycoproteins Peptides

Requires labeled lipids Obey different affinities After oxidation of glycosyl residues Critical for steric reasons

Dry coupling may be necessary as preconcentration is

216

Table 6.4

Chapter 6

(continued )

Ligand

Small molecules

DNA, oligonucleotides

DNA, native PCR products Carbohydrates Cells, cell fragments

Viruses, fragments

Immobilization method amines Biotin/ streptavidin (6.13) Ni21/ His6 (6.12) Epoxide coupling (6.10) EDC/NHS (6.3) (6.2) Reductive amination (6.4) Biotin/streptavidin (6.13) Hydrazide (6.9) Reverse EDC/NHS to amino surface Epoxide coupling (6.10) EDC/NHS, dry (6.3) (6.2) Epoxy coupling (6.10) Biotin/streptavidin (6.13) Electrosorption (6.11) EDC/NHS, dry (6.3) (6.2) Hydrazide (6.9) Epoxy coupling (6.10) Specific capture antibodies Electrosorption to 2D amine surfaces Lectins Streptavidin (6.13) Specific capture antibodies Electrosorption to amine surfaces

Remarks difficult. Dilute sample to avoid crosslinking Chemistry depends on functional groups of ligand. Dry coupling may be necessary as no preconcentration effect EDC/NHS and epoxy coupling require aminomodified oligos but give higher immobilization yields Combine with epoxy activation

If biotinylation possible

immobilization methods for the most common ligands, although each method is not suitable in each specific case.

6.6 Conclusions and Outlook In this chapter, an introduction to surface chemistry for SPR and similar biosensors has been presented. A few key factors, namely the composition and nanostructure of the immobilization matrix, determine the performance and characteristics of the sensor chip and can be used to optimize the system for a particular application. As a direct consequence of the increasing variety of ligand– analyte interactions which can be analyzed by SPR, a choice of immobilization methods has been developed, the most important of which have been described. It can be concluded that the data quality delivered by optical biosensors is only as good as the surface of the biochip, hence careful selection of the most optimal surface for the corresponding experiment is necessary. Because SPR is a very sensitive yet non-selective method, artifacts caused by defective or

Surface Chemistry in SPR Technology

217

suboptimal surfaces become immediately visible, stressing the requirement for rigorous optimization. Despite the enormous progress in the fields of bionanotechnology and surface science during recent years, surface design, especially for analytical and biomedical devices, remains a demanding challenge. The interplay of multifunctional macromolecules, ions and other dissolved species with water molecules at interfaces results in a complex scenario which is difficult to describe with theoretical models. Therefore, several phenomena and processes at the molecular scale remain not fully understood. It is predictable, however, that the last white spots on the map of surface science will probably disappear within the next 10 years, contributing to sensor chips with increased signalto-noise ratio and a more homogeneous distribution of binding sites. In this context, it is likely that today’s frequently employed polysaccharide matrices of microbial origin will be gradually replaced by better defined synthetic polymers with optimized functionality and structure. In addition to optimization of the surface homogeneity and minimizing nonspecific background, a clear demand exists for universal and easy to perform immobilization techniques which are standardized and can be used for any ligand irrespective of its chemistry, size or structure. On the way to such universal solutions we will probably see the development of individual, application-related chip surfaces with corresponding kits allowing reliable and even fully automated immobilization of the most frequently employed ligand classes. Surfaces for SPR microarrays to be used with 2D detectors are another area of further development as the sheer number of proteins in the proteome creates an increasing demand for parallel characterization in multiplex analysis.

6.7 Questions 1. Try to scale the dimensions roughly by drawing perpendicular to the surface (1) the gold layer thickness of an SPR device, (2) the antibody coating, (3) the evanescent field and (4) the stagnant layer in mass transport-controlled kinetics if we use a flow cell height of (5) two white blood cells. 2. Why is it important to check the quality of the sensor surface in an SPR image or SPR dip shape check? What is the meaning of a shallow and wide SPR curve? 3. The most popular immobilization chemistry is the EDC/NHS protein coupling to a carboxymethylated dextran chip. Predict the sensorgram of the coupling of a protein (molecular weight 10 kDa) with high isoelectric point including ethanolamine deactivation. 4. Draw a typical sensorgram of three analysis cycles if the ligand is decoupled after each regeneration step (e.g. 50% loss of ligand after a regeneration step). 5. Contact of the sensor surface with air may result in serious effects on the performance of the sensor chip. Which coatings show the most dramatic

218

Chapter 6

effects and should be prevented from any contact with air? And which coating is robust? 6. A common observation with affinity biosensors is the dependence of the electrostatic adsorption (preconcentration) and thus the immobilization capacity from the flow rate of the liquid handling system. Would you expect a higher or lower capacity with increasing flow rate? Explain the phenomenon. 7. a. Small ligands or oligonucleotides are frequently immobilized via biotinstreptavidin interaction. Why? b. An alternative can be the direct immobilization through reactive groups. What should be obeyed in this case? c. Discuss the advantages and disadvantages of indirect vs direct immobilization.

References 1. B. Alberts, D. Bray, J. Lewis, M. Raff, K. Roberts and J.D. Watson, Molkularbiologie der Zelle, 2nd edn., VCH, Weinheim, 1990. 2. J. Israelachvili, Intermolecular and Surface Forces, 2nd edn., Academic Press, San Diego, 1992. 3. P. Schuck, L.F. Boyd and P.S. Andersen, Measuring protein interactions by optical biosensors. In Current Protocols in Protein Science, John Wiley & Sons, New York, 1999, 20.2.1-20.2.21. (Annual Report 1 Z01 OD 010485). 4. J. Svitel, A. Balbo, R.A. Mariuzza, N.R. Gonzales and P. Schuck, Biophys. J., 2003, 84, 4062–4077. 5. R.G. Nuzzo and D.L. Allara, J. Am. Chem. Soc., 1983, 105, 4481–4483. 6. C. Pale-Grosdemange, et al., J. Am. Chem. Soc., 1991, 113, 12–20. 7. C.D. Bain, J. Evall and G.M. Whitesides, J. Am. Chem. Soc., 1989, 111, 7155–7164. 8. N.T. Flynn, T.N.T. Tran, M.J. Cima and R. Langer, Langmuir, 2003, 19, 10909–10915. 9. P. Van der Voort and E.F. Vansant, J. Liq. Chromatogr. Relat. Technol., 1996, 19, 2723–2752. 10. J.J. Cras, C.A. Rowe-Taitt, D.A. Nivens and F.S. Ligler, Biosens. Bioelectron., 1999, 14, 683–688. 11. P. Silberzan, L. Leger, D. Aussere and J. J. Benattar, Langmuir, 1991, 7, 1647–1651. 12. S.R. Holmes-Farley, R.H. Reamey, T.J. McCarthy, J. Deutch and G.M. Whitesides, Langmuir, 1985, 1, 725–740. 13. U. Lappan, H.-M. Buchhammer and K. Lunkwitz, Polymer, 1999, 40, 4087–4091. 14. A. Delcorte, P. Bertrand, E. Wischerhoff and A. Laschewsky, Langmuir, 1996, 13, 5125–5136. 15. L. Vroman, J.S. Mattson and C.A. Smith, Science, 1974, 184, 585–586. 16. M.A.C. Stuart, Surfact. Sci. Ser., 1998, 75, 1–25.

Surface Chemistry in SPR Technology

219

17. P. Billsten, M. Wahlgren, T. Arnebrandt, J. McGuire and H. Elwing, J. Colloid Interface Sci., 1995, 175, 77–82. 18. S. Ravichandran and J. Talbot, Biophys. J., 2000, 78, 110–120. 19. P. Vadgama, Annu. Rep. Prog. Chem., Sect. C, 2005, 101, 14–52. 20. A. Wittemann, B. Haupt and M. Ballauff, Phys. Chem. Chem. Phys., 2003, 5, 1671–1676. 21. C. Czeslik, R. Jansen, M. Ballauff, A. Wittemann, C.A. Royer, E. Gratton and T. Hazlett, Phys. Rev. E, 2004, 69, 021401. 22. E. Ostuni, R.G. Chapman, R.E. Holmlin, S. Takayama and G.M. Whitesides, Langmuir, 2001, 17, 5605–5620. 23. P.M. Claesson, in Biopolymers at Interfaces, ed. M. Malmsten, Marcel Dekker, New York, 1998, Vol. 75, pp. 281–320. 24. S.I. Milner, T.A. Witten and M.E. Cates, Europhys. Lett., 1988, 5, 413–418. 25. I. Szleifer, Physica A, 1996, 244, 370–388. 26. C. Siegers, M. Biesalski and R. Haag, Chem. Eur. J., 2004, 10, 2831–2838. 27. B.W. Morrisey and C.C. Han, J. Colloid Interface Sci., 1978, 65, 423–431. 28. C.J. Van Oss and J.M. Singer, J. Reticoendothelial Soc., 1966, 3, 29–40. 29. J.E. Butler, L. Ni, R. Nessler, K.S. Joshi, M. Suter, B. Rosenberg, J. Chang, W.R. Brown and L.A. Cantarero, J. Immunol. Methods, 1992, 150, 77–90. 30. B. Johnsson, S. Lo¨fas and G. Lindquist, Anal. Biochem., 1991, 198, 268–277. 31. D.G. Hoare and D.E. Koshland, J. Am. Chem. Soc., 1966, 88, 2057–2058. 32. J.C. Sheehan and J.J. Hlavka, J. Org. Chem., 1956, 21, 439–441. 33. J.C. Sheehan, J. Preston and P.A. Cruickshank, J. Am. Chem. Soc., 1965, 87, 2492–2493. 34. P. Cuatrecasaes and I. Parikh, Biochemistry, 1972, 11, 2291–2299. 35. L. Peng, G.J. Calton and J.W. Burnett, Appl. Biochem. Biotechnol., 1986, 14, 91–99. 36. S.P. Massia and J. Stark, J. Biomed. Mater. Res., 2001, 56, 390–399. 37. B. Renberg, I. Shiroyama, T. Engfeldt, P. Nygren and A.E. Karlstro¨m, Anal. Biochem., 2005, 341, 334–343. 38. D.J. O’shanessy, M. Brigham-Burke and K. Peck, Anal. Biochem., 1992, 205, 132–136. 39. D.G. Smyth, O.O. Blumenfeld and W. Konigsberg, Biochem. J., 1964, 91, 589–595. 40. C.F. Brewer and J.P. Riehm, Anal. Biochem., 1966, 18, 248–255. 41. G. Gorin, P.A. Martin and G. Doughty, Arch. Biochem. Biophys., 1966, 115, 593–597. 42. D.J. O’shannessy and R.H. Quarles, J. Appl. Biochem., 1985, 7, 347–355. 43. B. Johnsson, S. Lo¨fas, G. Lindquist, A. Edstro¨m, R.-M. Mu¨ller Hillgren and A. Hansson, J. Mol. Recognit., 1995, 8, 125–131. 44. L. Sundberg and J. Porath, J. Chromatogr., 1974, 90, 87–98. 45. D.J. O’Shannessy, K.C. O’Donnell, J. Martin and M. Brigham-Burke, Anal. Biochem., 1995, 229, 119–124. 46. P.D. Gershon and S. Khilko, J. Immunol. Methods, 1995, 183, 65–76.

220

Chapter 6

47. L. Nieba, S.E. Nieba-Axmann, A. Persson, M. Ha¨ma¨la¨inen, F. Edebratt, A. Hansson, J. Lidholm, K. Magnusson, A.F. Karlsson and A. Plu¨ckthun, Anal. Biochem., 1996, 252, 217–228. 48. J. Quinn, P. Patel, B. Fitzpatrick, B. Manning, P. Dillon, S. Daly, R. O’Kennedy, M. Alcocer, H. Lee, M. Morgan and K. Lang, Biosens. Bioelectron., 1999, 14, 587–595. 49. P. Nilsson, B. Persson, M. Uhle´n and P.A. Nygren, Anal. Biochem., 1995, 224, 400–408. 50. P.A. Ohlsson, T. Tja¨rnhage, E. Herbai, S. Lo¨fas and G. Puu, Bioelectrochem. Bioenerg., 1996, 38, 137–148. 51. M.A. Cooper, A. Hansson, S. Lo¨fas and D.H. Williams, Anal. Biochem., 2000, 277, 196–205. 52. O.P. Karlsson and S. Lo¨fas, Anal. Biochem., 2002, 300, 132–138. 53. R.L. Rich and G. Myszka, J. Mol. Recognit., 2005, 18, 1–39.

CHAPTER 7

Measurement of the Analysis Cycle: Scanning SPR Microarray Imaging of Autoimmune Diseases RICHARD B.M. SCHASFOORT,a ANGELIQUE M.C. LOKATE,b J. BIANCA BEUSINK,a GER J.M. PRUIJNb AND GERARD H.M. ENGBERSc a

Biochip Group, MESA+ Institute for Nanotechnology, Biomedical Technology Institute (BMTI), Faculty of Science and Engineering, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands; b Department of Biomolecular Chemistry, Nijmegen Centre for Molecular Life Sciences, Institute for Molecules and Materials, Radboud University Nijmegen, P.O. Box 9101, NL-6500, HB Nijmegen, The Netherlands; c IBIS Technologies B.V. P.O. Box 1242, 7550 BE Hengelo, The Netherlands, Internet: www.ibis-spr.nl

7.1 Introduction Surface plasmon resonance (SPR) [1] is a direct sensing analysis technique with a unique combination of features. Since it is a label-free technique it does not involve changes of the structure of the (bio)molecules that may be involved in the biomolecular interaction and the technique yields information on the concentration (how many analyte molecules are present in a complex sample), the affinity (how strong an interaction is) and the kinetics (how fast an interaction is) of biomolecules and their interactions [2]. Almost every SPR-experiment applies the same basic procedure in which the first step, the so-called immobilization (see also Chapter 6), involves covalent binding of a ligand to the surface of the sensor followed by the so-called analysis cycle [3]. In order to detect the presence or a specific interaction of a 221

222

Chapter 7

certain analyte in a sample, prior to analysis, a capturing entity (the ligand) should be immobilized on the sensor surface [4]. Although the ligand can be any type of (bio)molecule, more than 80% of the capturing agents used in SPR analysis are related to proteins (e.g. antibody, antigen, receptor, peptide). In the previous chapter, various chemical protocols are given for the attachment of ligands to the sensor surface. The present chapter covers the analysis cycle for measuring biomolecular interactions with SPR detection. After a discussion of the analysis cycle, it focuses on the use of SPR in immunoassays, including the latest developments of new SPR instrumentation: scanning SPR microarray imaging for monitoring autoimmune diseases. SPR imaging in combination with microarrays for genomics and proteomics applications is rapidly evolving. To demonstrate the features and benefits of multianalyte microarray-based biomolecular interactions, the analysis cycle of autoimmune antibody interactions in serum monitored with SPR imaging is discussed.

7.2 The Analysis Cycle After immobilization of a preferably pure ligand, the analysis cycle starts with the introduction of a buffer solution in the measurement cell, in order to generate a (stable) baseline. Subsequently, a sample, containing the ligand’s binding partner (the analyte), is introduced into the measurement cell and the analyte binds to the ligand, leading to the selective accumulation of the analyte on the sensor surface. This causes an increase in the refractive index near the sensor surface. This change in refractive index can be measured in real time. The selectively accumulated mass on the sensor chip surface (which is generally expressed in pg
mm2) correlates linearly with the change in the refractive index near the sensor surface measured by the SPR instrument [5]. A rule of thumb is that for an instrument that uses light with a wavelength of 670 nm, 1 ng
mm2 protein gives a signal of about 100 millidegrees (m1) angle shift. In Biacore instruments, 10 resonance units (RU) correspond to approximately 1 m1 SPR angle shift. When light of another wavelength is used then the angle shift should be calibrated. The response in SPR experiments of which a typical example is shown in Figure 7.1 directly reflects the change in the refractive index of the liquid phase near the surface of the sensor chip. The observed response is the sum of the response caused by association of analyte with the ligand and that caused by the difference in refractive index between the solution and the baseline buffer. The latter is called the ‘‘bulk effect’’ [6] and is caused by the presence of dissolved material including buffer components, biomolecules and salt. The bulk effect needs to be subtracted from the observed response in order to measure the true binding of the analyte molecules. If a reference channel is available in the SPR instrument, then the bulk effect can be determined accurately. This reference channel may also correct for non-specific adsorption of sample components to the surface of the sensor chip. Using baseline buffer as the solvent of the analyte will minimize the bulk effect. In experiments for

223

Measurement of the Analysis Cycle

Bulk shift Injection

Phases: baseline,

Figure 7.1

Time

association,

dissociation,

regeneration, baseline

A typical sensorgram obtained from an instrument with a reference channel showing the phases of an analysis cycle. Two channels/spots are measured; one with ligand and the other without ligand as a reference channel. The refractive index bulk shifts can be clearly observed from the reference channel. Double arrow represents the specific response. (1) Baseline phase: initially, baseline buffer is in contact with the sensor surface to establish the baseline. For sensor calibration purposes, the injection of a calibration liquid (e.g. 1% glycerol spiked in baseline buffer) can be incorporated in this phase (not shown). (2) Association phase: sample containing the target component is injected, the capturing elements on the sensor surface bind the target component resulting in complex formation. (3) Dissociation phase: On injection of baseline buffer, target components (and also non-specifically bound molecules) dissociate from the surface. Note: in this sensorgram, dissociation as an exponential decay is very low. (4) Regeneration phase: the regeneration solution (e.g. low-pH buffer) is injected to remove the remaining bound analyte. (5) After this phase, the cycle is completed and a new experiment can start by establishing the baseline again. If remaining accumulated mass is present, the baseline level will increase.

qualitative determination of whether the analyte will bind or not, careful adjustment of the buffer is not critical. However, for quantitative kinetic determinations in which a series of different analyte concentrations are used, it is important that buffer exchange is performed reliably, for instance by dialysis or gel filtration. Tuning the refractive index of the background buffer to mimic the sample and using the baseline buffer for dilution of the sample is a way to minimize the bulk effect. After the association process, buffer solution is introduced into the system, which initiates the dissociation phase. During the dissociation phase, the specifically and non-specifically bound compounds may detach from the surface, resulting in a decrease in response. Because in the dissociation phase the background buffer is usually the same as the buffer used during the baseline measurement, the exact amount of associated material can be

224

Chapter 7

calculated. It should be mentioned that in order to prevent rebinding effects in the dissociation phase, it is sometimes preferred to add ligand to the dissociation buffer (see Chapter 5, Section 5.3.2.2). This may induce a bulk shift which should be subtracted using the response obtained in the reference channel. Subsequently, a regeneration solution that, for instance, induces a strong pH shift (including a strong bulk refractive index shift to low SPR angles) is injected and the ligand–analyte complex dissociates (preferably) completely. Injection of the baseline buffer closes the analysis cycle and brings the system into its starting situation and the completed sensorgram is now available for analysis. Exactly timed liquid handling procedures are required to determine accurate kinetic rate constants of biomolecular interactions. In experiments where the sensorgram is compensated for bulk shift effects (for instance, when multichannel instruments are used in which one channel serves as reference), the association and dissociation phase of the sensorgram allow the calculation of the reaction rate and the thermodynamic equilibrium constants. If this is available, it is performed using instrument-specific software. Alternatively, it can be performed with general kinetic evaluation software. Furthermore, from the sensorgram the effectiveness of the regeneration procedure can be determined. Incomplete regeneration will result in residual analyte and/or non-specifically bound components on the sensor surface, which will increase the baseline, whereas too harsh sensor regeneration will result in a decreased binding capacity of the sensor in subsequent analysis cycles1.

7.3 Buffer Solutions for Measuring the Analysis Cycle 7.3.1 Baseline Buffer The running, background or baseline buffer should create optimal conditions for the binding of the ligand to the analyte. For biomolecular interactions the baseline buffer is usually a physiological buffer with sufficient salt and (near-) neutral pH. Phosphate-buffered saline (PBS) is a standard baseline buffer. Alternatively, 10 mM HEPES–KOH (pH 7.4), 0.15 M NaCl is often used. The addition of a surfactant, e.g. Tween 20 (0.03–0.05%) to the buffer is favored in order to minimize non-specific binding. The surfactant not only enhances the ratio between specific and non-specific binding but also helps to prevent air bubble adsorption on the surface. However, surfactants should not be added if hydrophobic surfaces are used for non-covalent attachment of membranebound components. Moreover, a surfactant should not be used during the immobilization cycle (see Chapter 6). Sometimes 3 mM EDTA is added to trap remaining bivalent cations (e.g. Mg21 or Ca21), which may interfere with the carboxylate groups in the hydrogel layer. Blocking components may help reducing non-specific binding, e.g. 3% BSA or HSA if patient serum is used. 1

See Figure 6.3 in Chapter 6 for a sensorgram depicting unstable immobilization and partial denaturation of the ligand.

Measurement of the Analysis Cycle

225

In the experiments on samples from rheumatoid arthritis (RA) patients described in Section 7.6, normal sheep serum was applied. All buffer solutions should be prepared with Milli-Q water and preferably filtered through a 0.22 mm filter and degassed before use.

7.3.2 Regeneration Solution For repeated use of the same sensor chip, the surface should be regenerated by removal of analyte and any other non-covalently bound material. However, the ligand should be kept intact and should not be inactivated or denatured during the regeneration phase. Commonly used solutions for regeneration include lowpH buffers, e.g. 10 mM glycine–HCl (pH 1.5). Optimal sensor regeneration includes a pH shock and regeneration is preferably performed as two consecutive steps of, for instance, 30 s each, rather than one step of 60 s. If a negatively charged gel-type sensor is used, not only the specific bond between ligand and analyte will break during low pH regeneration, but also the hydrogel will collapse and the analyte will be more or less squeezed out of the hydrogel. If the ligand or analyte cannot withstand low pH, sometimes a high alkaline pH is used, e.g. 10 mM NaOH (pH 4 11). Alternative regeneration solutions have a high salt concentration and the salts used include chaotropic agents that are chemicals, such as urea and guanidine hydrochloride, that disrupt hydrogen bonding in aqueous solutions. It should be noted that concentrated solutions of these agents may denature proteins, because they also interfere with hydrophobic interactions.

7.4 SPR-based Immunoassays In general, an immunoassay is a laboratory technique based on the binding between an antigen and its homologous antibody in order to identify and quantify the specific antigen or antibody in a sample. In classical immunoassays, the detection of the concentration of an analyte relies on signals generated by various labels (fluorescent dyes, enzymes or radioisotopes) attached to antigens or antibodies [7]. Labeling may disrupt the binding sites involved in the interaction. Moreover, labeling induces heterogeneity of the biomolecular interaction because in most cases labeling of a specific molecule (e.g. an antibody) is not homogeneously distributed. In addition, the label itself might interact with the capturing ligands, leading to false positives. A way to circumvent some of the problems is to use a labeled secondary binding molecule, but this extra step requires an additional, high-affinity binding compound and it will also increase the required analysis time. SPR-based biosensors measure protein–protein binding directly as a shift in surface-bound masses. Since SPR is a label-free analysis technique, it is preferred for studying biomolecular interactions. The observed binding rate of analyte to the ligand at the sensor surface may be limited either by the rate of interaction (kinetics) or the rate of supply of

226

Chapter 7

analyte to the surface (mass transport). Large amounts of ligand on the surface lead to higher interaction rates and low sample flow rates lead to a slower supply of analyte to the surface. Both of these factors contribute to mass transport-limited binding. These considerations are important for kinetic evaluation of association rate constants ka and dissociation rate constants kd. If mass transport limits the observed binding rates, then the apparent association rate constant from the simple model A + B # AB will be lower than the true value. A high ligand loading leads to rebinding effects during the dissociation phase and lower apparent dissociation rate constants will be measured for AB # A + B. In order to prevent rebinding, the ligand which is immobilized to the surface may be added to the dissociation buffer too. Rebinding has been demonstrated at high ligand loadings in hydrogels. It affects the calculation of the affinity constant dramatically and deviations of as high as two orders of magnitude have been reported [8]. Concentration measurements require a high loading of ligand on the surface and constant hydrodynamic conditions (flow rate or mixing speed). If the analyte dissociates from the surface it should not rebind at free sites while diffusing out of the hydrogel. Kinetic rate experiments require a low surface loading of the ligand, ideally the lowest ligand loading that still gives a measurable response after analyte binding. Mass transport limitation and other kinetic evaluation considerations are treated in detail in Chapters 4 and 5. Figure 7.2 shows different assay formats that are appropriate for SPR including direct, competitive, inhibition and sandwich assays.

7.4.1 Direct Assay The direct assay is described by A + B # AB. In this type of assay, antibodies directed against the antigen are immobilized on the sensor surface (¼ ligand). Sample solution containing the antigen (analyte) is then incubated with the sensitized sensor surface. The signal increase resulting from antigen binding correlates with the amount of antigen in the sample. Direct assays can also be designed with the antigen coupled to the surface and detection of the binding of the specific antibody. This is the case in the example shown in Section 7.6 of an immunoassay for the monitoring of autoimmune autoantibodies in the serum of rheumatoid arthritis patients.

7.4.2 Competition Assay This type of assay is typically applied for the detection of low molecular weight antigens that do not generate sufficient signal, while accumulating at the sensor surface. In this assay format specific antibodies are immobilized on the sensor surface and sample solution that contains the antigen is mixed with an antigen conjugate. Because of its high molecular weight, the conjugate enhances the SPR angle shift. The antigen–conjugated antigen mixture is incubated with the sensor surface. The difference in signal between a reference sample containing

Measurement of the Analysis Cycle

227

Figure 7.2 Immunoassay formats commonly used in SPR measurements. (a) Direct assay: the ligand (antibody) is immobilized on the sensor surface interaction with the analyte (here antigen) yields a detectable refractive index shift. (b) Competition assay for measuring small molecules where direct capturing of the antigen yields insufficient refractive index shift, while the conjugated antigen is large enough for a measurable refractive index shift (see also Figure 11.4). (c) Inhibition assay where the analyte is the same molecule as the immobilized ligand. Antibody is added to the sample in excess. The analyte forms conjugates with the antibody, inhibiting the binding to the sensor surface (see also Figure 11.2). (d) Sandwich assay with secondary antibody.

only conjugated antigen and the sample solution indicates the amount of antigen in the sample. In this assay, high antigen concentrations in the sample will result in low signals (less conjugated antigen can be bound). Competition assays are often used for the detection of toxic compounds (see also examples in Chapter 11). The maximum signal is generated when no free (toxic) analyte is present. When the signal is too low, possibly the analyte is present in the sample or the ligands are denatured or poisoned by the sample and no longer active while the analyte is not present. Both outcomes are harmful and will require action. If the competition assay shows a response of the unconjugated antigen as in the direct assay, then severe calibration procedures have to be performed. Preferably in SPR competition immunoassays the conjugated antigen is a large refractive index label (e.g. a latex bead or gold nanoparticle) loaded with the antigen.

7.4.3 Inhibition Assay In this assay format, the target antigen is immobilized on the sensor surface. Sample solution containing the antigen is mixed with specific antibodies in

228

Chapter 7

excess and incubated with the sensor surface. Antibodies will bind both to antigen in solution and to antigen that is bound to the sensor surface. The difference in signal between a blank sample that does not contain the antigen and the sample solution indicates the amount of antigen in the sample. In this assay, high antigen concentrations in the sample result in low signals (less antibodies remain to bind to the surface). Because antibodies have high molecular weight, their binding is directly detected.

7.4.4 Sandwich Assay In sandwich assays, antibody molecules against the analyte are immobilized on the sensor surface, capturing the analyte molecules when sample solution is incubated with the sensor surface. In the next step, a secondary antibody binds specifically with either the antigen or the antigen-bound antibody. The antigen is captured by a sandwich of two antibodies. Only very high affinity capture antibodies should be used, in order to avoid a mixture of affinities of each component in the sandwich. Several steps can be build in, but this complicates the analysis. Often, a highly specific goat/rabbit/sheep anti-mouse IgG is immobilized as the first capturing agent which traps a monoclonal (mouse) antibody for the antigen. Streptavidin–biotin linkers are often used because of the high affinity constant, which will not interfere with the rate and equilibrium constants of the analyte–ligand pair. The increase in signal (as a result of antigen binding) is proportional to the amount of antigen in the sample. Washing the surface with buffer is followed by the injection of a secondary antibody. The high molecular weight of the secondary antibody is usually sufficient for monitoring the binding process. For further signal enhancement, antibody conjugates with colloidal gold or latex particles as refractive index label can be used [9].

7.5 Detection of Multiplex Analysis Cycles Using Scanning SPR Imaging SPR imaging can be considered as a new trend (see Chapter 12), although the principle was published decades ago [10,11]. Currently computers are fast enough to process digital images allowing kinetic measurements of biomolecular interactions in real time. This allows the monitoring of multiple analysis cycles on a microarray by an SPR imaging instrument. In our studies, the instrument for biomolecular interaction sensing by imaging SPR (IBIS iSPR, IBIS Technologies, Hengelo, The Netherlands) is used for this purpose and is represented schematically in Figures 7.3 and 7.4. Fixed-angle instruments in the Kretschmann configuration (Chapter 3) are based on the relationship between a small change in the intensity of the reflected light and the mass of bound analyte, i.e. a fixed incident light angle is employed and mass changes are estimated from the intensity of the reflected light. However, the applicable dynamic range and linear relationship of this

Measurement of the Analysis Cycle

229

Figure 7.3 Schematic representation of the optical configuration of the IBIS iSPR instrument. An 840 nm light beam is passed through a p-polarizer and a focusing lens before reaching the hemispherical prism and gold sensor. The incident light is reflected by the angle operated mirror with a maximum scanning angle of 81. The light beam permits imaging of a 50 mm2 sensor surface. The light reflects from the sensor and passes through another lens before the CCD camera transfers the uncompressed images to the software.

Figure 7.4

Schematic representation of the liquid handling system of the IBIS iSPR in flow cell configuration. Samples are injected from the thermostated sample rack, using the needle of the liquid handler robot arm connected to the flow cell.

experimental setup are limited [12] and the optimal incident angles of each spot of a microarray differ considerably when ligand or analyte panels with different molecular weights are monitored, as explained in Chapter 3. Therefore, choosing one fixed angle to monitor many biomolecular interactions in a microarray will

230

Chapter 7

result in only qualitative data. Kinetic rate constants, however, require accurate quantitative data including subtraction of ‘‘bulk effects’’ for all the spots of the microarray simultaneously.

7.5.1 Dynamic Range of Scanning SPR Imaging The continuous SPR dip angle scanning mode of the IBIS iSPR provides a large dynamic range, thereby facilitating accurate, simultaneous measurements of, for instance, the presence and concentration of large proteins and small peptides in the same sample. In Figure 7.5, the difference in refractive index dynamic range of angle scanning and fixed-angle measurements is shown. The actual SPR angle depends on the refractive index in the evanescent field of the liquid–gold interface and microarrays spotted with different ligands show an initial distribution of shifts resulting in baselines at different angle positions. A problem exists when the biomolecular interaction is followed at fixed angle caused by the non-linear shape of the SPR curve as the baselines from spots of different ligands and immobilized biomolecular masses show different levels of reflectivity. In SPR imaging instruments, the SPR angles of many regions of interest (ROIs) should be monitored simultaneously. Comparison/subtracting baselines of different reflectivities is essentially incorrect. In fact, the slope of the reflectivity vs. angle curve is not constant and every spot requires its own calibration. The only reliable parameter that directly reflects the concomitant mass change on an SPR sensor is the SPR angle [13]. Thus, only when the SPR angle is monitored for all spots on a microarray independently can the Linearity scanning angle and fixed angle SPR 2500 200

fixed angle (right Y-axis)

1500

150

1000

100

500

50

0 0

Figure 7.5

0.005

0.01

0.015 0.02 ∆ refractive index

0.025

0.03

(∆R%)

Angle shift in m°

scanning angle (left Y-axis) 2000

0 0.035

Relationship of the angle shift of the SPR dip (m1) (left axis) and the change in refractive index (Dn) of glycerol dilution series in PBS in the scanning angle mode. On the right axis the reflectivity change (DR%) is shown for this range of refractive indices at fixed angle settings (the value of the fixed angle is in the inflection point of the left-hand flank of the SPR curve; see Chapter 3, Figure 3.2). Starting angle is at absolute reflectivity of about 40%.

Measurement of the Analysis Cycle

231

magnitude and affinity of biomolecular interactions be reliably compared among all spots of the microarray. In the IBIS instrument, the SPR angle of all spots of a microarray can be determined continuously. Thanks to its novel angle-scanning principle with fitting algorithms to calculate the resonance angle of the SPR dip for each ROI, this instrument quantifies the protein mass distribution on the microarray directly. The unique, innovative character of the IBIS imaging SPR instrument lies in the accuracy of SPR angle determination of multiple ROIs in combination with real-time imaging of the complete sensor surface. Inspection of the sensor surface using the microscopic SPR image shows the quality of spots (in terms of coating homogeneity), which is an important feature for generating reliable data (see Figures 7.7, 7.8 and 7.9). SPR angles of microarrays of more than 500 ROIs can be determined simultaneously, with the maximum number of ROIs being limited by the computing power and the lateral resolution of the SPR technique. In Figure 7.6, an image of a microarray is shown, which is spotted, in situ, in the IBIS iSPR instrument. Under scanning conditions, only homogeneous spots with equally distributed pixel intensities give reliable SPR results. A special needle in the liquid handling system automatically spots about 50 nl on a defined location creating the microarray. Drying of the spots can be prevented if the humidity is close to 100%. Evaporation can be delayed by applying non-interfering compounds in the spotting buffer, e.g. glycerol. However, this may impair the reproducibility of the results. In combination with angle scanning and inspection of the SPR dip in terms of depth and width, an indication of the quality of the sensing surface can be obtained. If, e.g., bare gold surfaces are used for ligand immobilization, often cauliflower images can be observed caused by irreproducible drying effects which can result in unreliable sensorgrams. Preferably hydrophilic coatings such as hydrogels are

Figure 7.6

SPR image of 5  6 spots generated in situ with the liquid handling system of the IBIS iSPR instrument. Right image: ROIs (up to 500) can be distributed automatically or manually by the operator of the instrument.

232

Chapter 7

applied for optimal results. The best contrast can be obtained when homogeneous intensities in one spot are at the left flank of the SPR curve. If the light is not homogeneously distributed over the entire surface, it has an effect on the shape (width) of the reflectivity vs. angle plot or SPR dip, but will not affect the minimum resonance condition.

7.5.2 Liquid Handling Procedures For the IBIS iSPR instrument, different types of flow cells and cuvettes are available and alternatively the instrument can be combined with ‘‘lab-on-achip’’ devices for controlling the sample flow over a microarray. Flow cells proved to be the best with regard to performance, accuracy and precision, providing the lowest limit of detection and best response stability. Cuvettebased systems offer the user maximum flexibility. An advantage of a cuvettebased systems for studies of kinetic rate constants is that higher mass transport can be obtained towards the interaction site [14]. A disadvantage is that the mixing parameters such as injection, needle height and mixing speed need to be optimized and controlled in order to prevent shear stress at the sensor surface. For example, particles in the sample can hit the sensor surface, leading to damage of the microarray.

7.5.3 Determination of the Limit of Detection Using Multiplex Analysis Cycles In order to obtain an indication of the limit of detection (LOD) a microarray containing various densities of two different ligands was spotted on a sensor chip. Regeneration of the sensor with 10 mM glycine–HCl (pH 1.5) provided the ability to test several antibody concentrations on one sensor chip. Figure 7.7 shows the sensorgram [angle shift (m1) as a function of time (s)] of eight analysis cycles of the interaction of the biotinylated peptides and -proteins with an anti-biotin antibody. The different curves represent the interaction with the various amounts of biotinylated peptides spotted using Top Spot [15,16] as indicated in the figure caption. The curves varying over time (x-axis) represent the different concentrations of the analyte, anti-biotin IgG. From the analysis cycles of every spot, in principle overlay plots of the interactions with the different analyte concentrations can be generated in order to calculate the onand off-rate constants and affinity equilibrium constants. The affinity constants can be compared with a ‘‘one-shot’’ or single injection analysis, a term recently introduced by Bio-Rad’s Proteon system. When a serially diluted ligand is immobilized on the surface, the LOD can be expressed as the smallest change in refractive index of the bulk solution or refractive index change in the evanescent field caused by accumulated mass. As indicated before, if compounds with an increased refractive index replace the background electrolyte buffer solution, a linear relationship exists between the surface coverage (in %) and the molecular weight of the adsorbate.

233

Measurement of the Analysis Cycle Response curves of immobilized peptide in femtomole

Angle shift (mdeg)

300 250 417 208 104 52 26 0

200 150 100 50 0 0.1

1

10

100

1000

[IgG] (pM)

Figure 7.7

Sensorgram of eight analysis cycles (raw data) of a biotin–antibody interaction microarray (shown in Figure 7.8) at different ligand and analyte concentrations for the determination of the on and off rate constants. Injection of a single analyte was done, which is similar to the one-shot analysis approach as described in Chapter 12. An injection of the highest analyte mouse anti-biotin mouse IgG (150 kDa) concentration (133 pM) resulted in a gradual decrease in responses of the spots loaded with different ligand biotinylated peptide (2.4 kDa) concentrations obtained by spotting using Top Spot [14,15] of 417, 208, 104, 52 and 26 fmol and two controls, non-biotinylated peptide and the background. After this first analysis cycle, a lower mouse (53 pM) anti-biotin mouse IgG (150 kDa) analyte concentration was injected. Only non-specific dissociation can be observed at higher concentrations of the analyte. In total eight analysis cycles were carried out over the range of 133, 53, 13, 5.3, 1.3, 0.53, 0.13 and 0 pM anti-biotin mouse IgG. Each curve shows the transient 2 m1 response as a result of the change in flow direction. The few spikes are caused by air bubbles disturbing the SPR signal. The inset shows the dose–response curves of the spots as obtained from this sensorgram after a fixed interaction time just after the association phase. In Figure 7.8, the SPR view of this microarray is shown.

Therefore, a low molecular weight compound (e.g. a small peptide) will contribute less to the accumulated mass per surface area than a high molecular weight compound (e.g. an immunoglobulin). The LOD can be used to characterize an instrument; however, in Chapters 4 and 5 it has been shown that the LOD also depends strongly on the affinity constant of biomolecular

234

Figure 7.8

Chapter 7

Imaging SPR vs. fluorescence microscopy. Microarray of the spotted ligands using Top Spot [15,16] after interaction with 133 pM Alexa Fluor 488-labeled mouse anti-biotin IgG. The top two rows contain various peptide concentrations spotted in duplicate: from left to right, 417 fmol of a non-biotinylated peptide as a negative control followed by 26, 52, 104, 208 and 417 fmol of the biotinylated peptide. The bottom two rows contain various IgG concentrations spotted in duplicate: 6.67 fmol of a non-biotinylated IgG as a negative control followed by 0.42, 0.83, 1.67, 3.33 and 6.67 fmol of biotinylated IgG. In the SPR imaging view (a) the non-biotinylated ligand (left column; peptide and IgG) can be clearly seen, whereas in the fluorescence image (b) these spots are not observed.

interactions [14]. In our test system using the biotin–streptavidin pair, the specifically bound analyte will not dissociate significantly because of the high affinity constant of this couple. At high concentrations of the analyte only a low non-specific dissociation effect occurs, as shown in Figure 7.7. The molecular mass of the analyte (in this case the anti-biotin IgG, B150 kDa) gives an indication of how many molecules are needed to obtain a significant response compared with the signal-to-noise ratio. The amount of analyte bound to the immobilized ligand can be derived from the measured angle shift; 1 m1 ¼ 10.8 pg mm2, resulting in 1.62 amol IgG in the case of an ROI of 150  150 mm and 65 zmol or 4.1  104 IgG molecules bound to the minimal ROI of 30  30 mm. Figure 7.7 shows the importance of both a balanced combination of the ligand and analyte concentrations for concentration measurements (maximum angle shift) and kinetic parameter determination (low rebinding, low mass transport limitation). Figure 7.8 shows the corresponding fluorescence image after removing the sensor chip from the system. More on combining fluorescence detection and SPR excitation can be found in Chapter 9. Secondary antibodies can be used not only to introduce fluorescent labels to check the system but also for linking to refractive index labels such as latex particles and colloidal gold to improve sensitivity (see Section 7.4). The secondary antibody linked to the mass label will induce a measurable shift, thereby improving the LOD if the mass label fits into the evanescent field and in the hydrogel. The use of secondary antibodies can even increase the specificity of a reaction; however, an additional interaction step has to be included which overrules the favorable label-free one-step measurement. The application of SPR imaging for multi-analyte detection in one step is shown in the following section.

Measurement of the Analysis Cycle

235

7.6 Monitoring of Autoantibodies in Serum of Rheumatoid Arthritis Patients The potency of SPR imaging is demonstrated by monitoring the analysis cycles of serum autoantibodies of rheumatoid arthritis (RA) patients that bind specifically to citrullinated peptides in a single step. Autoimmune diseases are characterized by the presence of high-affinity autoantibodies directed against self-proteins, such as rheumatoid factor for RA [17], Sm for systemic lupus erythemathosus (SLE) [18] and Ro/SS-A and La/SS-B for Sjo¨gren’s syndrome [18]. Although at least some autoantibodies are known to be involved in cell and tissue damage, their mechanistic role in the pathogenesis of the disease is generally not known [19]. Nevertheless, the specificity of autoantibody responses highlights their potential as important tools for improved diagnosis, disease classification and prognosis. Miniaturized multiplex assays can deliver a fingerprint of a patient’s autoantibody repertoire requiring only a limited amount of patient material. During the last decade, various research groups have made important contributions to the application of multiplex assays for autoantibody detection. In 2002, Robinson et al. employed protein and peptide ligand arrays, representing candidate autoantigens, to survey autoantibody binding [20]. Arrays of in situ synthesized peptides can also be generated with photolithography to perform antibody characterization [21]. Another approach is to apply ‘‘virtual arrays’’ in a homogeneous assay system with addressable beads [22]. However, in these systems at least one of the interactants has to be labeled, which may disrupt the binding sites involved in the interaction, leading to false negatives. In addition, the label itself might interact with the immobilized proteins, leading to false positives [23]. A way to circumvent these problems is to use a labeled secondary binding molecule, which might result in improved assay sensitivity. In this section, scanning SPR microarray imaging is used to measure the presence of anti-citrullinated protein antibodies (ACPA) in the sera of RA patients. Recently, it was shown that the so-called citrulline amino acid (which can be generated post-translationally from arginine) is a critically important moiety of the antigenic determinants targeted by RA-specific autoantibodies [24]. Cyclic citrullinated peptides (CCPs) are widely used as antigenic targets in ELISA-based diagnostic tests for RA. The sensitivity and specificity are 71 and 99%, respectively [25].

7.6.1 Experimental Conditions for Serum Measurements 7.6.1.1

Serum Samples

The sera were obtained from the Department of Rheumatology, University Hospital Nijmegen. Sera were collected from patients visiting the outpatient clinic who had been diagnosed as having RA according to the revised criteria of the American College of Rheumatology. To assess specificity further, we

236

Chapter 7

analyzed a group of serum samples from healthy individuals and groups of sera from patients with osteoarthritis and SLE, obtained from various clinics and hospitals. Sera were stored at 80 1C until used. Anti-CCP2 ELISA was performed by IMMUNOSCAN RA (Euro-Diagnostica, Arnhem, The Netherlands), in accordance with the manufacturer’s instructions with the recommended 25 U ml1 cut-off.

7.6.1.2

SPR Microarray Interaction Studies

SPR detects changes in refractive index in the hydrogel (200 nm) which is linked to the gold surface. Due to the small molecular weight of the synthetic peptides used (B1500 Da), the contrast of the immobilized array to the background is not high. To visualize the array for determining the ROIs, human IgG was also spotted. Incubation, washing and regeneration were performed in an automated way using liquid handling procedures (LHPs) in the instrument for biomolecular interaction sensing (IBIS iSPR, IBIS Technologies Hengelo, The Netherlands). A serum sample plug of 400 ml (diluted 1:50 in PBS, 0.03% Tween 20) was guided backwards and forwards over the array in a flow cell at a rate of 1 ml s1. The serum sample plug was surrounded by two air plugs to prevent the diffusion of serum components into the buffer. Between all steps the flow cell was rinsed with PBS, 0.03% Tween 20. The array was regenerated by injection of 400 ml of 10 mM glycine–HCl (pH 1.5) twice for 30 s. Two incubation– regeneration cycles were completed before applying the sera in order to block non-specific binding sites and create an optimal reactive sensor surface. Analysis of the data were done using the software supplied.

7.6.2 Results and Discussion of Monitoring Analysis Cycles for Autoantibody Screening In our proof-of-principle experiments, a 24-spot microarray containing human IgG, and also two different linear citrullinated peptides I and II and the corresponding control peptides (containing arginine instead of citrulline), were spotted (1 nl per spot) on the sensor surface. Because none of the peptides contained lysine residues, coupling is expected to occur exclusively via the N-terminal primary amino groups, thereby ensuring oriented, end-on immobilization of the peptides. One set of peptides was synthesized with a C-terminal biotin tag. These peptides were used to investigate possible differences in immobilization efficiency by assessing the SPR angle shift that resulted from incubation of the array with an anti-biotin antibody. The corresponding SPR angle shifts were 211  10 m1 and 212  12 m1 (m1 ¼ milli degree) for the citrullinated and arginine-containing peptide I, respectively. From these data, it can be concluded that there was no difference in immobilization efficiency between the two peptides. After placing the spotted sensor chip on top of the hemisphere prism in the flow cell-based instrument, self-defined liquid handling procedures (LHPs) were

Measurement of the Analysis Cycle

237

used to increase the inter-experiment reproducibility. Serum incubation, washing and regeneration were done in an automated manner. In the instrument an angle scan including the capture of up to 80 images is carried out in 8 s. When a spot meets the optimal SPR conditions, the reflected light at a certain ROI within this spot will reach a minimum value resulting in a dark spot. Images of the reflected light at three different scan angles after incubation with serum for 1 h are shown in Figure 7.9. The SPR dip of each ROI can be visualized in a reflectivity versus angle plot as shown. The 0 m1 setting is an arbitrary value, which can be set before each individual experiment by manual adjustment over a range of 101 to obtain the best angle range. At an incident angle of 700 m1 the background area near the array is in resonance (left image), at –200 m1 the spots coated with the arginine control peptides are in resonance (middle image) and at 100 m1 the spots coated with citrullinated peptide I are in resonance (right image). By curve fitting of the reflected light intensities as a function of scan angles for each spot of the microarray, the exact value of the SPR dip angle in m1 can be calculated and plotted in a sensorgram (Figure 7.10). As described above, the y-axis of the sensorgram does not represent an arbitrary reflectivity parameter, but contains the exact values for the SPR angle at maximum resonance. These exact SPR angles can be normalized for easy comparison of all the different sensorgrams of all the spots and are linearly correlated with the refractive index, corresponding to the mass of protein bound to the sensor surface. Figure 7.10 shows the sensorgram of an analysis cycle obtained during incubation of the microarray with a RA serum. ROIs in a background region nearby the array and ROIs within the spots of immobilized human IgG and of two arginine-containing peptide I and II controls showed relatively small angle shifts. Binding to the two corresponding citrullinated peptides I and II, however, resulted in angle shifts of 400 and 250 m1, respectively. The intra-array variation was very low, as illustrated by the sensorgrams of the quadruplicate interactions shown in Figure 7.10. The noise, i.e. the baseline difference in resonance angle between the highest and the lowest values, measured over 100 time points (1000 s) in one individual ROI, was 1.35 m1, making angle shifts of 3 m1 significant. In Figure 7.11, the repeatability of the serum interactions with the microarray by an automated liquid handling procedure is shown. At the start of the sensorgram, two injections of a 1% glycerol solution were made for sensor calibration purposes. The sensor chips could be used for up to 50 analysis cycles by treatment of the sensor surface with two repetitive incubations with 10 mM glycine–HCl (pH 1.5) for 30 s after each serum incubation step. Furthermore, alternating injections of normal sheep serum and three RA patient sera were performed. After completing these six serum incubations, the total protocol was repeated with the same injection sequence to determine the stability of the array. Sequential measurements of patient sera on a single spot showed variations of less than 5% in binding to the citrullinated peptide, even when six interaction–regeneration cycles were performed between the two sequential measurements. At the end the sensor surface was again calibrated with two

238

Chapter 7

Figure 7.9 SPR curves obtained from the microarray by scanning the angles in a range of 3.41. The three reflectivity images are from the same array, but obtained at different angle positions as shown by the arrows. After the procedure of angle scanning in steps of 50 m1 a reflectivity image of the surface is obtained and gray values of the region of interests are stored in the database. By viewing the image at different angle positions after, e.g., a spotting procedure, the quality of the image can be checked for optimal resonance conditions. If, e.g., a spot is not homogeneous then the depth and width of the SPR curve becomes higher and broader. Here three images of the microarrays are shown where in the first image the background is in resonance. The middle image shows the spots of peptide II in resonance after interaction with serum antibodies of an RA patient and the image on the right shows the spots at the angle position where spots with peptide I are in resonance after interaction with serum compounds. For details of the spots and specific interaction with autoantibodies, see Section 7.7. Gray values of ROIs are averaged and determined for each angle position. The SPR curves from each ROI are fitted in order to calculate the exact minimum.

Figure 7.10

Sensorgram of 24 multiple analysis cycles of RA patient antibodies with peptides immobilized on the SPR sensor chip. After the association phase (A), the array was washed with PBS, 0.03% Tween 20 (B) and regenerated with two sequential steps of 10 mM glycine–HCl (pH 1.5) for 30 s (C). Between the sequential glycine–HCl steps the array was rinsed with PBS, 0.03% Tween 20 (D). Blue (upper) and dark green lines (lowest) are the quadruplicates of citrullinated peptide I and the corresponding arginine control, respectively; light green (second upper) and purple lines are the quadruplicates of citrullinated peptide II and the corresponding arginine control, respectively. Red lines are from the human IgG (n ¼ 8) and the brown lines are from regions close to the array (blank controls) (n ¼ 8). The enlarged section shows the four individual curves from the quadruplicates of citrullinated peptides I and II.

Measurement of the Analysis Cycle 239

240

Figure 7.11

Chapter 7

Sensorgram (raw data) is shown showing the repeatability of serum interactions with a sensor array using an automated liquid handling procedure. At the beginning and at the end of the sensorgram a calibration cycle (Z) is performed for sensor calibration purposes by duplicate injections of a 1% glycerol solution (delta 160 m1). Then 12 analysis cycles from A to L were performed with subsequent alternating injections of a negative sheep serum (samples A, C, E, G, I and K) and positive RA patient samples (B, D, F, H, J and L). The positive sample was in three concentrations in serial dilution [analysis cycle (B, H); 1/100 and (D, J) 1/200 diluted and (F, L); 1/400 times diluted]. One can observe that the baseline increased by 20 m1 after 12 injections.

injections of a 1% glycerol solution. One can observe that the baseline increases by 20 m1 after 12 injections. In Figure 7.12, a typical sensorgram (raw data) is shown of seven analysis cycles of different RA sera of three RA patients and control sera without antiCCP antibodies and normal control serum. In addition, the responses of an SLE patient serum and an osteoarthritis patient serum are shown. We tested 50 RA sera and 29 control sera (9 SLE patient sera; 10 osteoarthritis patient sera; 10 normal healthy control sera). The interaction of sera with the microarray was quantified by calculating the ratio between the angle shifts observed for the citrullinated peptide I and the corresponding arginine control peptide (hereafter designated C/R ratio) upon binding of serum antibodies. The mean C/R ratio obtained with scanning SPR microarray imaging for the RA sera which were tested positive in a CCP2-ELISA (RA CCP+) is 8.7. The C/R ratio for the RA-sera that were tested negative in the CCP2-ELISA (RA CCP), normal

Measurement of the Analysis Cycle

Figure 7.12

241

SPR sensorgram (raw data) of seven analysis cycles of different RA sera and control sera with CCP and its negative control. Between different sera the microarray was regenerated with 10 mM glycine–HCl (pH 1.5). (A) RA patient serum without anti-CCP antibodies; (B) SLE patient serum; (C) RA patient 1 serum; (D) normal sheep serum control; (E) RA patient 2 serum; (F) osteoarthritis patient serum; (G) RA patient 3 serum.

controls, SLE patients and OA patients where between 0.9 and 1.2. A detailed study of these 79 patient sera with ELISA and SPR test results were published recently [27]. The currently most widely applied target for the detection of anti-citrullinated protein antibodies, cyclic citrullinated peptides, are recognized by only about 70% of RA patients [23]. Due to the heterogeneity of the anti-citrullinated protein response in RA, the use of additional citrullinated peptides may allow the detection of such antibodies in patients that are not reactive with the peptides used in the CCP2 ELISA. The use of microarrays monitored by SPR imaging will facilitate the simultaneous detection of the various anti-citrullinated protein antibodies.

7.7 Features and Benefits of Monitoring Analysis Cycles with SPR Imaging Compared with fixed-angle based SPR imaging systems, the scanning mode of the IBIS system measures the relevant angle of the SPR dip, permitting valid comparison of ROIs and correction for bulk effects. Due to the integrated liquid handling system in combination with the X–Y–Z robot, it is possible to run the experiments automatically.

242

Chapter 7

Although in the past many autoantigens have been identified and characterized, to date most of the assays that have been developed to detect autoantibodies are ELISA based, thus allowing only separate analyses for each type of autoantibodies. The need for multiplex analysis systems has been emphasized by recent observations that the specificity of the detection of anti-CCP antibodies in RA patient sera can be increased by monitoring reactivities with both the citrulline- and arginine-containing peptides in parallel [26]. In all of the multianalyte autoantibody studies, a secondary antibody conjugate was necessary to visualize bound antibodies. Here, we show that SPR imaging of protein/peptide microarrays provides a method that allows one-step multi-analyte detection of autoantibodies in patient sera, which does not require additional reagents to visualize antibody binding. Additional advantages of this system are that the SPR dip angle scanning principle allows accurate and simultaneous monitoring of biomolecular interactions between molecules of varying masses and that the ligand-containing sensor chips can be efficiently regenerated and re-used. Although the sensitivity of detecting RA-specific autoantibodies by the scanning SPR microarray imaging system under the conditions applied in this study is slightly lower than that of the ELISA systems, we expect that further optimization will lead to similar sensitivities. Moreover, for the low-titered sera a signal amplification step, e.g. by using a (gold-labeled) secondary antibody, may raise the sensitivity to levels that allow positive signals for all anti-CCP autoantibody-containing sera. The initial results obtained with 50 rheumatoid arthritis patient sera (and 29 control sera) proves that SPR imaging is a promising, multi-analyte label-free technique for the development of diagnostic tests. Although kinetic evaluation of the analysis cycle including the determination of the thermodynamic equilibria is beyond the scope of this chapter (see Chapters 4 and 5), the application of measuring analysis cycles of microarrays has been demonstrated for RA patient autoantibody detection. In order to quantify the response of RA patient sera, the ratio of the binding to the citrulline- and arginine-containing peptides is used. An accurate calculation of this ratio is obtained only by comparison of the SPR angle shifts (and not reflectivity) measured in the scanning SPR microarray imaging system.

7.8 Conclusion In this chapter, the analysis cycle of a biomolecular interaction has been discussed, and exemplified by experiments using an SPR imaging instrument. The preferred assay format is a direct assay described by A + B # AB, while other formats can be used for specific cases generally where the molecular weight of the analyte is too low for direct detection. The challenge of the small intrinsic refractive index label of the analyte can be circumvented by using alternative assay formats: competition, inhibition or sandwich. Real-time monitoring of the binding of biomolecules using SPR imaging allows the analysis of association and dissociation rate constants for determination of

Measurement of the Analysis Cycle

243

the affinity constants of many biomolecular interactions. The imaging feature of the scanning SPR microarray technology is of importance to check the quality of the sensor surface and to carefully identify the regions of interest in order to obtain high reproducibility. In addition, the accurate measurement of the SPR dip angle for monitoring the presence of autoantibodies in sera of autoimmune patients allows the comparison of each curve (e.g. citrulline vs. arginine), including subtraction of the common mode effect (i.e. bulk shift jumps). Scanning SPR microarray imaging will be of great use in any field that requires multi-analyte detection in an automated analysis cycle.

7.9 Questions 1. After spotting a microarray of peptide ligands on a sensor surface for SPR imaging, it is wise to also immobilize a high molecular weight ligand. Why? 2. If a spot or selected ROI is not homogeneous, it will affect the shape of the SPR curve. How? Draw possible SPR dip curves. 3. Ligands can be spotted on the sensor surface in different concentrations. However, the degree of immobilization caused by diffusion of the ligand to the surface is mass transport controlled. How is spotting of ligands to the surface in exact serial dilution possible? 4. Inspection of the surface at several angles is a strong feature of scanning angle imaging instruments. Describe the benefits. 5. In SPRI instruments, ligands can be diluted over the sensor surface, reference spots can be defined, positive and negative controls can be spotted and the ratios can be determined. Baseline correction, subtraction and ratio determinations can be applied in the kinetic evaluation process as shown in this chapter. Explain why scanning angle SPR imaging instruments show in principle higher reproducibility than fixed-angle SPR imaging instruments. Explain why reflectivity data (y-axis in a sensorgram) plotted as a function of time introduces artifacts. How can these artifacts be corrected?

7.10 Acknowledgement The authors thank The Dutch Technology Foundation STW for funding projects TMM 6209, Proteomics on a Chip, and TMM 6635, Screenchip.

References 1. B. Liedberg, C. Nylander and I. Lundstro¨m, Sens. Actuators, 1983, 4, 299. 2. J.M. McDonnell, Curr. Opin. Chem. Biol., 2001, 5, 572.

244

Chapter 7

3. 4. 5. 6.

R. Karlsson and A. Fa¨lt, J. Immunol. Methods, 1997, 200, 121. B. Johnsson, S. Lo¨fa˚s and G. Lindquist, Anal. Biochem., 1991, 198, 168. D.G. Myszka, J. Mol. Recognit., 1999, 12, 279. K. Nagata, Significance of the real-time analysis of biological interactions, in Real-Time Analysis of Biomolecular Interactions, K. Nagata, H. Handa, (Eds.), Springer, 2000, pp. 3–9. D. Wild (Ed.), The Immunoassay Book, Stockton Press, New York, 1994. N. Murai and M. Yoshida, Molecular Chaperones in Part 4 Chapter 1, in Real-Time Analysis of Biomolecular Interactions, K. Nagata, H. Handa (Eds.), Springer, 2000, pp. 87–95. G.A.J. Besselink, R.P.H. Kooyman, P.J.H.J. van Os, G.H.M. Engbers and R.B.M. Schasfoort, Anal. Biochem., 2004, 333, 165–167. B. Rothenha¨usler and W. Knoll, Nature, 1988, 332, 615. C.E.H. Berger, T.A.M. Beumer, R.P.H. Kooyman and J. Greve, Anal. Chem., 1998, 70, 703–706. B.P. Nelson, A.G. Frutos, J.M. Brockman and R.M. Corn, Anal. Chem., 1999, 71, 3928. E. Stenberg, B. Persson, H. Roos and C. Urbaniczky, J. Colloid Interface Sci., 1991, 143, 513–526. N.J. de Mol, E. Plomp, M.J. Fischer and R. Ruijtenbeek, Anal. Biochem., 2000, 279, 61. O. Gutmann, R. Niekrawietz, R. Kuehlewein, C.P. Steinert, S. Reinbold, B. de Heij, M. Daub and R. Zengerle, Analyst, 2004, 129, 835–840. B. de Heij, M. Daub, O. Gutmann, R. Niekrawietz, H. Sandmaier and R. Zengerle, Anal. Bioanal. Chem., 2004, 387, 119–122. R. A. Mageed, in Manual of Biological Markers of Disease, W. J. van Venrooij, R. N. Maini (Eds.), Kluwer, Dordrecht, 1994, Chapter B1.1, pp. 1–27. J.G. Routsias, A.G. Tzioufas and H.M. Moutsopoulos, Clin. Chim. Acta, 2004, 340, 1. F.A. van Gaalen, J. van Aken, T.W.J. Huizinga, G.M.Th. Schreuder, F.C. Breedveld, E. Zanelli, W.J. van Venrooij, C.L. Verweij, R.E.M. Toes and R.R.P. De Vries, Arthritis & Rheum., 2004, 50, 2113–2121. W.H. Robinson, C. DiGennaro, W. Hueber, B.B. Haab, M. Kamachi, E.J. Dean, S. Fournel, D. Fong, M.C. Genovese, H.E. Neuman de Vegvar, K. Skriner, D.L. Hirschberg, R.I. Morris, S. Muller, G.J. Pruijn, W.J. van Venrooij, J.S. Smolen, P.O. Brown, L. Steinman and P.J. Utz, Nat. Med., 2002, 8, 295. S.P. Fodor, J.L. Read, M.C. Pirrung, L. Stryer, A.T. Lu and D. Solas, Science, 1991, 251, 767–773. R.J. Fulton, R.L. McDade, P.L. Smith, L.J. Kienker and J.R. Kettman Jr., Clin. Chem., 1997, 43, 1749–1756. M.A. Cooper, Anal. Bioanal. Chem., 2003, 377, 834. G.A. Schellekens, B.A. de Jong, F.H. Van den Hoogen, L.B. Van de Putte and W.J. van Venrooij, J. Clin. Invest., 1998, 101, 273–281.

7. 8.

9. 10. 11. 12. 13. 14. 15. 16. 17.

18. 19.

20.

21. 22. 23. 24.

Measurement of the Analysis Cycle

245

25. W.J. van Venrooij, A.J. Zendman and G.J.M. Pruijn, Autoimmun. Rev., 2007, 6, 37–41. 26. A. Vannini, K. Cheung, M. Fusconi, J. Stammen-Vogelzangs, J.P. Drenth, A.C. Dall’Aglio, F.B. Bianchi, L.E. Bakker-Jonges, W.J. van Venrooij, G.J.M. Pruijn and A.J.W. Zendman, Ann. Rheum. Dis., 2007, 66, 511–516. 27. A.M.C. Lokate, J.B. Beusink, G.A.J. Besselink, G.J.M. Pruijn, R.B.M. Schasfoort, J. Am. Chem. Soc., 2007; DOI: 10.1021/ja075103X.

CHAPTER 8

Advanced Methods for SPR Imaging Biosensing ALASTAIR W. WARK, HYE JIN LEE AND ROBERT M. CORN Department of Chemistry, University of California–Irvine, Irvine, CA 92697, USA

8.1 Introduction Microarray biosensors have become an invaluable biotechnological tool for the rapid, multiplexed detection of surface bioaffinity interactions. For example, nucleic acid microarrays are currently being applied to the areas of genomics [1,2], genetic testing [3], gene expression [4,5], and single nucleotide polymorphism (SNP) genotyping [6,7]. Many researchers are also interested in developing protein microarrays for application in the areas of proteomics [8–10] and drug discovery [11,12]. In addition, the detection and profiling of multiple protein biomarkers in biological fluids (e.g. blood, serum, urine) by antibody microarrays is a potentially powerful method for the diagnosis of diseases and the monitoring of subsequent therapeutic treatments [13,14]. An attractive alternative to traditional fluorescence-based microarray detection methods is the surface-sensitive optical technique of surface plasmon resonance imaging (SPRI). SPRI, also denoted SPR microscopy [15–17], was originally applied to the study of surface morphology in phospholipid monolayers and other thin surface films [15]. Since those initial efforts, SPRI has evolved into a primary method for the measurement of bioaffinity adsorption onto biopolymer microarrays by detecting changes in the local refraction index upon binding [18–31]. This is a significant advantage in the analysis of biological samples where labeling multiple biomarkers with fluorophores or nanoparticles is often not possible. Over the last decade, a number of improvements have been made in SPRI in terms of instrumentation, surface microarray fabrication and microfluidic formats for the multiplexed measurement of surface adsorption kinetics with microarrays. A brief review of some of these advances is given in Section 8.2. 246

Advanced Methods for SPR Imaging Biosensing

247

While multiplexed microarray analysis of bioaffinity interactions is a valuable research tool, even more specificity and sensitivity can be obtained by coupling the bioaffinity process to an enzymatic transformation. This coupling is often employed in solution-phase biotechnological processes; for example, coupling of DNA hybridization with polymerase chain reaction (PCR) leads to the process of PCR amplification in genomic DNA samples [32,33]. However, the direct incorporation of solution enzymatic methods such as PCR into a parallel microarray format is difficult, because any intermediate solution species will diffuse on to neighboring array elements. Instead of using solution enzyme chemistry, it is better to utilize a surface enzyme reaction on biomolecules attached to the microarray surface. The use of surface enzyme chemistry ensures that the parallel nature of the multiplexed microarray assay remains intact. In Section 8.3, we describe some of the basic equations that we have developed for the use of surface enzyme kinetics in SPRI, and a new enzymatically amplified SPRI methodology to detect DNA down to femtomolar concentrations. Finally, in Section 8.4 we describe how the incorporation of DNA-coated gold nanoparticles into the enzymatically amplified SPRI methodologies can further increase the sensitivity of these array bioaffinity measurements. DNA-coated gold nanoparticles were originally employed with SPRI by He et al. to detect DNA in a sandwich assay format at a concentration of 10 pM [30]. Using a nanoparticleenhanced surface ligation strategy, we demonstrate SNP genotyping with SPRI at concentrations as low as 1 pM, and with a combination of poly(A) polymerase and DNA-coated gold nanoparticles, we can detect microRNA in biological samples down to a concentration of 10 fM [34].

8.2 Advances in SPRI Instrumentation and Surface Chemistry Since the first demonstration in 1997 of the use of SPRI and microarrays for monitoring DNA–DNA interactions [35,36], our group has continued to develop this technology focusing in particular on (i) optimizing the SPRI instrumental set-up and (ii) designing new chemical strategies for the surface attachment of different biomolecular probes in an array format. A schematic diagram of an SPRI apparatus is shown in the middle inset in Figure 8.1, where the output from a collimated white light source is passed through a polarizer and then directed onto the prism/chip assembly at a fixed optimal angle. The reflected p-polarized light is then collected via a narrow-band interference filter centered at 830 nm onto a CCD camera. We have found that this white light– NIR filter combination has several advantages compared with the use of a visible monochromatic laser beam such as improved sensitivity and the removal of interference fringes from the SPR image that are often problematic when using coherent laser excitation [37]. A number of commercial SPR imaging instruments are now available from GWC Technologies, Biacore (HTS Biosystems), IBIS Technologies and GenOptics; for more on instrumental aspects, see Chapter 3.

248

Figure 8.1

Chapter 8

A simplified SPRI set-up (middle inset) and representative SPRI difference images obtained for the detection of bioaffinity interactions with biopolymer microarrays. From top center and clockwise, protein biomarker binding to an antibody microarray, antibody binding to a peptide microarray, protein interactions with a His-tagged protein microarray, lectin binding to a carbohydrate line array, response regulator protein binding to a dsDNA microarray, 16mer ssDNA hybridization adsorption on an ssRNA microarray and 16mer ssDNA hybridization adsorption on an ssDNA microarray. The middle inset shows a simplified SPRI set-up. Briefly, the output from a collimated white light source (L) is passed through a polarizer (P) with the resulting p-polarized light directed on to a high index prism/sample assembly at an optimal incident angle. The reflected light from the sample assembly is then collected via a narrow band pass filter (F) on to a CCD camera (C).

In conjunction with SPRI instrumentation, the use of well-characterized and robust surface chemistries to tether biological molecules to gold surfaces in an array format is extremely important for successful SPRI biosensing measurements. Consequently, we have developed a variety of surface immobilization strategies for covalently attaching different biomolecular probes to chemically modified gold surfaces [38–41]. These attachment chemistries can be used along with either UV photopatterning [39] or microfluidic techniques [42] to create arrays of multiple, independently addressable elements on a single surface. Figure 8.1 shows a set of representative SPRI images obtained from various studies that our group has performed [18,34,43–52]. A detailed summary of the biopolymer microarrays (e.g. DNA, RNA, peptide, protein, carbohydrate) prepared for different biomolecular targets is listed in Table 8.1.

249

Advanced Methods for SPR Imaging Biosensing

Table 8.1

Summary of different surface bioaffinity measurements performed using SPRI and biopolymer microarrays.

Microarray probes (on surface)

Target biomolecules (in solution)

DNA ssDNA DNA ssDNA ssDNA

DNA ssDNA (16–18 bases) RNA ssRNA (18 bases) 16S ribosomal RNA Messenger RNA

DNA Biotinylated DNA ssDNA dsDNA dsDNA dsDNA

Kads (M
1)

Ref.

2.0  107

45,53

1.8  107 N

53 86 42

Proteins Streptavidin ssDNA-binding protein Mismatch-binding proteins Phosphorylated OmpR Phosphorylated VanR

N N N 1.6  108 6.6  106

35 87 87 50 50

Peptides Flag-peptide S-peptide

Antibodies and proteins Anti-Flag S-protein

1.5  108 1.7  107

51 52

Carbohydrates Mannose Galactose

Lectins Concanavalin A Jacalin

2.2  107 5.6  106

88 88

Proteins His-tagged ubiquitin His-tagged Flag His-tagged RFP His-tagged TATA boxbinding protein

Antibodies and DNA Anti-ubiquitin Anti-Flag Anti-RFP dsDNA

N N N N

89 89 89 89

Antibodies Anti-b2-microglobulin Anti-cystatin C

Protein biomarkers b2-Microglobulin Cystatin C

1.4  108 1.0  108

43 43

RNA RNA aptamer

Proteins factor IXa

1.6  107

71

RFP, red fluorescent protein; ssDNA, single-stranded DNA; dsDNA, double-stranded DNA; N, not measured with SPRI.

A third area in which SPRI biosensing has improved is in examining the thermodynamic and kinetic parameters of surface bioaffinity interactions. A simple one-step surface bioaffinity adsorption process involving the specific binding of a target biomolecule (T) to a surface-attached probe (P) or ligand can be described by the reaction ka

T þ P Ð TP kd

ð8:1Þ

250

Figure 8.2

Chapter 8

(a) Schematic showing target ssDNA (T) hybridization adsorption on ssDNA microarray elements (P). (b) SPRI difference image obtained for sequence-specific 16mer target ssDNA hybridization adsorption on ssDNA microarray elements. (c) A representative plot of relative surface coverage (y) as a function of target DNA concentration. The solid line is a Langmuir isotherm fit to the data, from which a value of Kads ¼ 2  107 M
1 was determined.

where TP is the surface-bound target–probe complex and the rates of T adsorption and desorption are defined by ka and kd, respectively (see Figure 8.2a). An example of such a reaction would be DNA hybridization adsorption on a DNA microarray. Figure 8.2b shows an SPRI difference image of a four-component ssDNA microarray following exposure to two complementary 16mer DNA target sequences. A positive increase in percentage reflectivity (D%R) is observed at only the perfectly matched array elements due to the formation of duplexes via hybridization adsorption. Fresnel calculations and experimental evidence show that if the SPRI response remains below 10%, then it is directly proportional to the relative surface coverage (y) of complementary DNA [53], where y ¼ GTP/Gtot and G represents the molecular surface density. At equilibrium, the fractional surface coverage reaches a steady state and this equilibrium surface coverage (yeq) is given by the Langmuir adsorption isotherm:

yeq ¼

Kads ½T 1 þ Kads ½T

ð8:2Þ

Advanced Methods for SPR Imaging Biosensing

251

where the Langmuir adsorption coefficient (Kads) is defined as Kads ¼ ka/kd. Figure 8.2c shows a typical Langmuir isotherm plot of y versus concentration for 16mer DNA from which a value of Kads ¼ 2107 M
1 was obtained [45,53]. At low surface coverages, the Langmuir isotherm depends linearly on DNA concentration and eq. (8.2) becomes yeq ¼ Kads[T]. In DNA hybridization adsorption, the lowest DNA concentration observed with SPRI was 1 nM, which corresponds to a yeq of 0.02. In addition to equilibrium measurements, further information on surface bioaffinity interactions can be obtained by measuring the kinetic parameters ka and kd. For the case of adsorption from a solution of concentration [T] on an unoccupied surface, the time-dependent fractional surface coverage, y(t), is given by yðtÞ ¼ yeq ½1
e
ðka ½Tþkd Þt 

ð8:3Þ

where yeq is the equilibrium value for y at a particular bulk concentration, as given using eq. (8.2). For the case when y ¼ 1 at t ¼ 0, the desorption rate can be described by yðtÞ ¼ yeq e
ðkd Þt

ð8:4Þ

Equations (8.3) and (8.4) have been used frequently to analyze the adsorption of biomolecules on surfaces, especially with SPR [54] and SPRI [52]. Compared with single-channel SPR measurements, real-time SPRI has a great advantage in being able to determine simultaneously the rates of target adsorption/desorption at multiple different probe elements on a single chip. Figure 8.3a shows the design of a poly(dimethylsiloxane) (PDMS) microchannel flow cell that facilitates well-controlled and reproducible sample delivery to each array element in addition to significantly reducing the sample volume (B10 ml). SPR imaging kinetic experiments are performed using a continuous flow-through microchannel to prevent mass transport limitations, whereas equilibrium measurements are obtained under stopped-flow conditions using a larger volume flow cell (B100 ml). Additionally, a specially designed water-jacketed flow cell, which controls the system temperature to within 0.1 1C, is used to perform temperature-dependent studies. We have applied real-time SPRI to determine accurately the ka and kd of S protein binding to an array composed of various S peptide analogues [52]. For this study, customwritten software that rapidly acquires and organizes large data sets during realtime acquisition was used (see Figure 8.3b). These real-time SPRI methods have also been employed to characterize quantitatively our surface enzyme methods described in Section 8.3. Finally, we have also continued to pursue the development of improved SPRI biosensing methods; one example is the design of novel multiple-layered chip structures that support the generation of long-range surface plasmons (LRSPs) [45,55–58]. Essential to the LRSPR chip design shown in Figure 8.4 is the use of Cytop as an inert optically transparent material whose refractive index is very close to that of water, with the depth and position of the LRSPR

252

Figure 8.3

Chapter 8

(a) SPRI raw image of a peptide array created from parallel PDMS microfluidic channels to deliver heterobifunctional cross-linker and probe molecules before being replaced by a second serpentine PDMS channel to create a continuous flow cell for use in kinetics measurements. The dotted lines indicate regions of interest (ROIs) on the array where the change in percent reflectivity is measured as a function of time. (b) Simultaneous real-time SPRI measurements obtained for each ROI [peptide elements and poly(ethylene glycol) background] when the array was first exposed to a 150 nM solution of S protein and then rinsed with phosphate buffer only. The SPRI signal increases then decreases in response to protein adsorption/desorption. The adjacent poly(ethylene glycol) regions are used to correct the SPRI signal for background effects. Reprinted with permission from reference 52.

Advanced Methods for SPR Imaging Biosensing

Figure 8.4

253

(a) SPR reflectivity curves of p-polarized light as a function of incident angle for a long-range SF10 glass/Cytop (1180 nm)/Au (32 nm)/water configuration (J) and a conventional SF10 glass/Au (45 nm)/water SPRI configuration (B) at excitation wavelength 814 nm. The solid lines show the results of a theoretical fit to the data using a multiple phase Fresnel calculation. Refractive indices (n) used in the calculation: n(SF10) ¼ 1.711, n(Cytop) ¼ 1.336, n(Cr) ¼ 3.186 + 3.47i, n(Au) ¼ 0.185 + 5.11i and n(H2O) ¼ 1.328. (b) Calculated relative optical field intensities for the LRSPR and conventional SPR assemblies whose schematics are shown as insets. The conventional SPR field intensity is multiplied by 100 for visibility. Reprinted with permission from reference 45.

254

Chapter 8

mode strongly dependent on the thickness of both the Cytop and gold films. Compared with conventional surface plasmons, LRSPs possess longer surface propagation lengths, higher electric field strengths and narrower resonance curves. This is demonstrated in Figure 8.4a, which shows in situ scanning angle reflectivity measurements obtained for both a conventional SPRI chip design [SF10/gold (45 nm)/water] and that of a LRSPR chip [SF10 prism/Cytop (1180 nm)/gold (32 nm)/water] [45]. To demonstrate the advantages of LRSPR, DNA microarrays were prepared on both conventional and long-range chips. Repeated measurements of DNA hybridization adsorption showed an B20% increase for the LRSPR D%R signal compared with regular SPRI chips. Perhaps of more interest is the much higher surface electric field strengths associated with the generation of LRSPs. Figure 8.4b displays the results of n-phase Fresnel calculations [59,60] which demonstrate that the LRSP electric field intensity at the gold sensing surface is higher by as much as 103 compared with that associated with conventional surface plasmons. These enhanced fields should lead to further research advances in surface plasmon fluorescence spectroscopy (SPFS) studies [61].

8.3 Surface Enzymatic Transformations for Enhanced SPRI Biosensing To enhance further the biosensing capabilities of SPRI and overcome some of the difficulties associated with detecting very low target concentrations (o1 nM), we have developed a number of advanced methodologies which couple surface enzyme reactions and bioaffinity interactions on biopolymer microarrays [18,34,46,48,49]. In this section, we first outline a simple theoretical framework that quantitatively describes the catalysis reaction of an enzyme in bulk solution with a surface immobilized substrate. The application of this model to analyze real-time SPRI and SPFS measurements of RNase H surface activity is also discussed. Next, we provide two examples of surface enzymatic processes on nucleic acid microarrays that are coupled with surface bioaffinity interactions to enhance the sensitivity and selectivity of the biosensor. These processes are (i) the enzymatically amplified detection of genomic DNA using RNase H and RNA microarrays and (ii) the application of surface ligation chemistry for the fabrication of RNA microarrays from DNA microarray templates for the study of protein–RNA aptamer interactions.

8.3.1 Measuring Surface Enzyme Kinetics The quantitative characterization of the surface enzyme reactions utilized in advanced SPRI biosensing methods is extremely important, as an enzyme can react orders of magnitude slower on a surface as compared with in solution. Here, we describe a simple kinetic model that includes enzyme adsorption, desorption and the surface enzymatic reaction and apply it to the analysis of realtime SPRI and SPFS measurements of the surface hydrolysis of RNA–DNA

Advanced Methods for SPR Imaging Biosensing

Figure 8.5

255

A series of real-time SPRI data curves obtained for the RNase H hydrolysis of surface RNA–DNA (R1-D1) array elements at different concentrations of RNase H solutions [0.5 nM (K), 1.0 nM (’), 2.0 nM (m) and 4.0 nM (E)]. The experimental curves were globally fitted using eqs. (8.8)– (8.10) to obtain the parameter values ka ¼ 3.4 ( 0.2)  106 M
1  s
1, kd ¼ 0.10 (0.05) s
1, kcat ¼ 1.0 (0.1) s
1 and b ¼ 180 ( 20). The solid lines are the fitted theoretical kinetic curves. The inset is a schematic illustration of the RNase H hydrolysis of surface RNA–DNA heteroduplexes. Reprinted with permission from reference 64.

heteroduplexes1 by RNase H. This surface reaction (see Figure 8.5 inset) forms the basis of our enzymatic amplification methodology described in Section 8.3.2. For the case of a 1:1 binding of an enzyme molecule (E) to a surface-immobilized substrate (S), we have identified three processes that control the overall reaction rate: km

Eðx¼NÞ
! Eðx¼0Þ ka

S þ Eðx¼0Þ Ð ES kd

kcat

ES
! S þ Eðx¼0Þ 1

ð8:5Þ

ð8:6Þ

ð8:7Þ

A heteroduplex is a double-stranded molecule of nucleic acid composed of two single complementary strands derived from different sources.

256

Chapter 8

where E(x¼N) and E(x¼0) are the bulk and surface enzyme concentrations, respectively, ES is the surface enzyme–substrate complex, S* is the surfacebound product and kcat is the surface reaction rate for the enzyme complex. When microfluidics are used for solution delivery, the enzyme diffusion can be described by a steady-state mass transport coefficient (km) that can also be written as D/d, where D is the diffusion coefficient for the enzyme and d is the steady-state diffusion layer thickness [62,63]. The kinetic equations for this reaction scheme can be expressed in terms of the relative fractional surface coverages of each of the three surface species (denoted yx ¼ Gx/Gtot, where x ¼ S, ES or S*): yS þ yES þ yS ¼ 1

ð8:8Þ

dyES ka ½Eð1
yES
yS Þ
ðkd þ kcat ÞyES ¼ dt 1 þ bð1
yES
yS Þ

ð8:9Þ

dyS ¼ kcat yES dt

ð8:10Þ

In eq. (8.9), [E] is the bulk enzyme concentration and b is the dimensionless diffusion parameter [62,63], defined by b¼

ka Gtot ka Gtot d ¼ D km

ð8:11Þ

These equations were derived in a series of recent papers [18,64,65] and can be solved using simple Euler integration methods with the initial conditions yS ¼ 1 and yES ¼ yS* ¼ 0 at time t ¼ 0 to yield three time-dependent surface coverages yES(t), yS*(t) and yS(t) that can be separately profiled over the course of the surface reaction. The kinetic model described above has been applied to the quantitative analysis of time-resolved SPRI and SPFS measurements of the catalytic behavior of RNase H on surface-immobilized RNA–DNA heteroduplexes. Figure 8.5 shows the D%R loss observed in real-time SPRI measurements due to the selective removal of RNA from the surface at enzyme concentrations ranging from 0.5 to 4 nM. The analysis of the SPRI kinetic curves (shown as solid lines in Figure 8.5) using eqs. (8.8)–(8.10) yielded best-fit values of 3.4  106 M
1  s
1, 0.1 s
1, 1.0 s
1 and 180 for ka, kd, kcat and b, respectively. Also, the analysis showed yES(t) remained very small (B10
3) throughout the reaction with kcat c ka[E]. This means that once adsorbed, RNase H reacts very quickly and is immediately released from the surface. The observed kcat value of 1.0 s
1 is significantly faster than kcat measured in a similar study [65] for the surface hydrolysis of double-stranded DNA by Exonuclease III (kcat ¼ 0.01 s
1). In this surface reaction, the time-dependent SPRI signal initially increased before an eventual overall decrease due to the loss of the complementary strand in the surface DNA duplex. This behavior is

Advanced Methods for SPR Imaging Biosensing

257

qualitatively very different from that of the RNase H reaction in Figure 8.5, where at no point was a net increase in SPRI signal observed. Since enzyme adsorption and surface duplex hydrolysis both contribute to the measured SPRI response, some assumptions on the relative contributions of both reaction steps are required when using our kinetic model. Therefore, to characterize the surface enzyme reaction completely it is important to obtain an independent set of in situ surface kinetic measurements. In the case of RNase H, this was achieved using fluorescently labeled RNA and applying the technique of SPFS [64]. The loss in SPFS signal due to enzymatic hydrolysis of the surface attached RNA is a direct measure of yS(t) and was analyzed in a similar manner to the SPRI data for various enzyme concentrations. The ka, kd and kcat best-fit values obtained from the SPFS measurements were almost identical with the values obtained from the SPRI data described above, indicating that fluorescence labeling of RNA does not significantly affect kcat.

8.3.2 RNase H–Amplified Detection of DNA The first major breakthrough in the application of surface enzyme processes to greatly improve the sensitivity of SPRI bioaffinity measurements was the use of the RNase H surface process for the amplified detection of DNA [48,49]. Figure 8.6a shows how this enzymatic amplification methodology works. An RNA probe microarray is exposed to a solution containing both target DNA and RNase H. When a target DNA molecule binds to a complementary RNA probe array element, RNase H recognizes the formed RNA–DNA heteroduplex and selectively hydrolyzes the RNA strand (Step 1). The target DNA is released back into solution and is free to bind to another surface RNA probe and RNase H will again hydrolyze the RNA in the surface heteroduplex (Step 2). This repeated target binding–enzymatic hydrolysis–target release will continue until all the RNA probe molecules are removed from the surface (Step 3). Amplification is achieved because only a very small number of target DNA molecules are required to produce a large change in SPRI signal. The sensitivity of RNase H amplified SPRI was found to be sufficient for the detection of DNA at concentrations as low as 1 fM. Consequently, we have been able to directly detect target sequences in a human genomic DNA sample without PCR amplification. Figure 8.6b shows an SPRI difference image of a three-component RNA microarray following 4 hours exposure to a genomic DNA sample. At the R1 and R2 elements, which contain sequences designed to bind specifically to the TSPY gene on the Y chromosome, a decrease in SPRI signal (–0.7%) was observed2, whereas no change in SPRI signal was observed at either the R3 elements or the array background. The concentration of the TSPY gene sequences in commercially available male genomic DNA was estimated to be 7 fM [48]. In addition to human genomic DNA, this enzymatically amplified SPRI method can be applied to the ultrasensitive detection and identification of DNA and RNA from viruses and bacteria. 2

Note that the minimum change in SPRI reflectivity that can be measured is around 0.08%.

258

Figure 8.6

Chapter 8

(a) Schematic outlining enzymatically amplified SPRI methodology using an RNA microarray for the detection of target DNA. (b) An SPRI difference image obtained for the detection of male genomic DNA in the presence of RNase H. A schematic of the three-component RNA microarray is also shown where R1 and R2 are specifically designed to bind to the TSPY gene on the Y chromosome and R3 is a negative control sequence. Reprinted with permission from reference 48.

Advanced Methods for SPR Imaging Biosensing

259

8.3.3 Fabrication of RNA Microarrays with RNA-DNA Surface Ligation Chemistry A second example of how surface enzyme processes can be applied to enhance the biosensing capabilities of SPRI is the use of ligation chemistry to create RNA microarrays from DNA microarrays. The ability to fabricate stable and active single-stranded RNA (ssRNA) microarrays is essential for a successful RNase H amplified SPRI measurement and will also promote the use of RNA microarrays for the study of RNA–protein, RNA–RNA and other bioaffinity interactions. Despite the many potential benefits of RNA microarrays, at present only a handful of reports are available on the fabrication of RNA microarrays in the literature [48,49,66–70]. The use of modified RNA sequences such as biotinylated RNA [66,69,70] or thiol-modified RNA [48,49] can be time consuming and costly with also the increased possibility of RNA degradation during both the modification and surface attachment procedures. To address these problems, we have recently developed two novel RNA fabrication strategies utilizing two different enzymes: (i) T4 DNA ligase and (ii) T4 RNA ligase [47,71]. Both of these enzymes catalyze the ligation of unmodified RNA onto a DNA microarray, but only T4 DNA ligase requires the use of a complementary DNA template sequence [47]. Figure 8.7 outlines a simplified schematic for the creation of an ssRNA microarray using T4 DNA ligase. A single-stranded (ssDNA) microarray is first prepared by chemically attaching 3 0 -thiolated, 5 0 -phosphorylated ssDNA. These anchor DNA array elements (DA) are then exposed to a solution containing T4 DNA ligase and both probe RNA (RP) and template ssDNA (DT), which will hybridize to the DA surface. Following the formation of a phosphodiester bond between the 5 0 -phosphate of DA and the 3 0 -hydroxyl group of RP the array surface is thoroughly rinsed with 8 M urea in order to denature and remove the DNA template and any T4 DNA ligase, resulting in the creation of biologically active ssRNA array elements. The SPRI difference image [b – a] in Figure 8.7 shows a D%R increase of 2.2  0.3% following the ligation of 24mer RNA [47]. We also found that the original ssDNA microarray could be regenerated with RNase H hydrolysis allowing the surface ligation process to be repeated to create a new RNA microarray. This is because RNase H specifically cleaves the phosphodiester bonds in the RNA component of the DNA-RNA heteroduplex (formed by complementary DNA, DC, hybridization–adsorption on RP) to produce 5 0 -phosphate and 3 0 -OH termini. This ligation-hydrolysis cycle could be repeated up to three times using the same ssDNA microarray without any degradation in SPRI signal. In addition, when the ligated microarray was used for the detection of 1 pM DNA via RNase H amplification, the initial rate of change in D%R observed was over 20 times faster than that observed with microarrays prepared using thiol-modified RNA. This is attributed to an increase in RNase H activity due to the anchor DNA sequence increasing the distance of RP from the gold surface. A second example of the application of surface ligation chemistry for the fabrication of ssRNA microarrays is the use of the enzyme T4 RNA ligase,

260

Figure 8.7

Chapter 8

Simplified schematic of the fabrication of a renewable ssRNA microarray via RNA–DNA ligation chemistry followed by RNase H hydrolysis. The ssRNA microarray was created by the selective ligation of RNA (RP) to DA elements of a two-component DNA microarray in the presence of a DNA template (DT). Hybridization of complementary DNA (DC) on to the ligated ssRNA followed by the selective hydrolysis of RP using RNase H regenerates the original 5 0 -phosphorylated ssDNA surface. The right inset shows a representative in situ SPRI difference image [b – a] obtained by subtracting images acquired before and after RNA ligation. Reprinted with permission from reference 47.

which does not require a complementary DNA template sequence [71]. To create a multi-component RNA microarray using this approach, solutions containing both enzyme and RNA probe were delivered onto individual array DNA elements via spotting. Following completion of the ligation reaction, the microarray was rinsed with 8 M urea to remove any enzyme and non-ligated ssRNA from the surface. As outlined in Figure 8.8a, the prepared bioactive ssRNA array elements can then be directly applied for the study of RNA– protein bioaffinity interactions. Figure 8.8b shows an SPRI difference image obtained for the adsorption of 20 nM human factor IXa (fIXa) on a threecomponent RNA aptamer microarray. The array pattern is shown in Figure 8.8c, where RF has a strong binding affinity for fIXa with both RC and RR aptamer variants acting as negative controls. In addition to the screening of different aptamer sequences, these ligated RNA microarrays can also be applied to the ultrasensitive detection of protein biomarkers [71].

8.4 Nanoparticle-amplified SPRI Biosensing A second method for increasing the sensitivity of SPRI bioaffinity measurements by several orders of magnitude is the use of functionalized gold

Advanced Methods for SPR Imaging Biosensing

Figure 8.8

261

(a) Schematic showing the creation of an RNA aptamer microarray via RNA–DNA surface ligation chemistry using T4 RNA ligase. DNA and RNA bases are shown as open circles and filled circles, respectively. Step (i): microarray elements composed of 5 0 -phosphorylated ssDNA are individually exposed to solutions containing ssRNA aptamer molecules and T4 RNA ligase. This results in the formation of a phosphodiester bond between the 3 0 -hydroxyl of the ssRNA and the 5 0 -phosphate of the ssDNA. Step (ii): after ligation, the surface is rinsed with 8 M urea to remove the enzyme and non-ligated ssRNA. The ssRNA aptamer microarray is then used for the study of protein–aptamer binding events. (b) SPRI difference image obtained for the specific adsorption of 20 nM fIXa on the RF array elements of a three-component microarray. (c) A pattern of a three-component aptamer array where RF has a high binding affinity towards fIXa and RT and RR are negative control aptamer sequences.

nanoparticles, an approach which is also completely compatible with the use of surface enzyme chemistry. Nanoparticle-amplified SPRI detection of DNA was originally demonstrated by He et al. who used DNA-modified nanoparticles to achieve a detection limit of 10 pM in a sandwich assay format [30]. Although the dependence of the SPR response as a function of both nanoparticle size and the nanoparticle
planar film separation distance has been investigated by several groups [72–74], nanoparticles are not yet routinely applied for enhanced SPRI bioaffinity sensing. In this section, two novel methodologies combining surface enzyme reactions and nanoparticle-enhanced SPRI measurements are

262

Chapter 8

described for (i) the analysis of single nucleotide polymorphisms (SNPs) in genomic DNA and (ii) the femtomolar detection of microRNAs (miRNAs3) in a total RNA sample. In both studies, DNA-modified gold nanoparticles (B13 nm diameter) were utilized with a maximum absorbance at 525 nm. The surface plasmon excitation wavelength in our SPRI measurements is 830 nm. Therefore, the amplification in SPRI signal is primarily due to changes in refractive index and not absorptive coupling between the gold nanoparticles and thin gold film. A theoretical description of nanoparticle SPR can be found in Chapter 2.

8.4.1 Single Nucleotide Polymorphism Genotyping The development of new methods for the rapid, multiplexed detection and identification of single nucleotide polymorphisms (SNPs) in human genomic DNA samples is attracting major interest as these methods can be used to accelerate the discovery of specific disease-related mutations and also for largescale human genetic variation studies. In principle, SPRI measurement with DNA microarrays is an ideal candidate technique for SNP genotyping. Various researchers have employed both conventional SPRI and single-channel SPR formats to detect single base mismatches in DNA by hybridization adsorption [24,77–80]. However, the SPRI detection limit of 1 nM described in Section 8.2 is insufficient for SNP genotyping of unamplified genomic DNA samples where the detection of single base pairs in a DNA sequence at a concentration of 1 pM or better is required. In order to demonstrate that SPRI can be used for multiplexed SNP genotyping, we developed a novel approach that involves the sequence-specific ligation of target DNA to an ssDNA microarray followed by the nanoparticle-amplified SPRI detection of the surface ligated product. As outlined in Figure 8.9, a solution containing the 36mer target (T), a 5 0 -phosphorylated ssDNA ligation probe (L) and the enzyme Taq DNA ligase (E) is introduced to a two-component ssDNA microarray containing the probes PG and PA [44]. The two array probe DNA sequences differ only at their 3 0 -terminal nucleotide (G for PG and A for PA). The target DNA molecules hybridize simultaneously to the ligation probes and the array probes, resulting in the formation of two different surface complexes, L–T–PG and L–T–PA. Single base pair selectivity is achieved because Taq DNA ligase will only catalyze the formation of a phosphodiester bond between the juxtaposed P and L probes when they are both perfectly complementary to the hybridized target. Here, the ligation probe will be specifically ligated to PG but not to PA. Following an 8 M urea wash to remove any target, non-ligated probe and enzyme from the microarray, SPRI detection of the ligated L–PG array elements is achieved through the hybridization adsorption of gold nanoparticles modified with ssDNA 16mers (LC) which are complementary to the ligation probe sequence. 3

MicroRNAs are a new class of small, non-coding RNA molecules (19–23mers) that can regulate gene expression in both plants and animals [75,76].

Advanced Methods for SPR Imaging Biosensing

Figure 8.9

263

Schematic of SNP genotyping method based on the combined use of surface ligation chemistry and nanoparticle enhanced SPRI. Reprinted with permission from reference 44.

Prior to performing experiments using real genomic DNA samples, the sensitivity and single base specificity of this advanced SPRI methodology was established using synthetic target oligonucleotides. A four-component array was designed consisting of three 20mer probe sequences (PA, PC and PG) that are identical except for the last nucleotide at their 3 0 -termini (A, C and G, respectively) and also a poly T sequence (PN), which serves as a negative control. Specific ligation of the 16mer L probe to only one of the array components could be detected for 36mer T concentrations as low as 1 pM [44]. Having established the detection limit of this SNP genotyping method, we then applied the same microarray design to screen the PCR products of human genomic DNA samples for a possible point mutation in the BRCA1 gene that is associated with breast cancer. In this case, probe PG is the perfect complement to the wild-type allele, probe PC is the perfect complement to the mutant allele in NA13710, while probe PA is the perfect complement to the mutant allele in NA14637. Both mutant samples were obtained from the Coriell Institute [44]. The SNP genotyping results for the wild-type and NA14637 samples are shown in Figures 8.10a and b, respectively. The SPRI difference image in Figure 8.10a shows an increase in reflectivity only at the PG array elements, indicating that the genotype of the wild-type DNA sample is a C–C homozygote. However, in Figure 8.10(b), a D%R increase was observed at both the PG and PA array elements, identifying the NA14637 sample as a C–T heterozygote. The NA13710 sample (data not shown) was identified as a C–G heterozygote. These experiments successfully demonstrate for the first time that SPRI can be

264

Figure 8.10

Chapter 8

SNP genotyping of 1 nM PCR amplicons of two genomic DNA samples (a) wild-type and (b) NA14637. Both SPRI difference images measure the hybridization adsorption of LC-modified nanoparticles on a fourcomponent DNA microarray (PA, PC, PG and PN) after the surface ligation and denaturation steps are completed. The SPRI difference image (left) and corresponding line profiles (middle) are shown next to the array pattern (right). An increase in SPRI signal was observed only at the PG array elements for the (a) wild-type sample identifying it as being a C–C homozygote. For the (b) NA14637 mutant sample, an SPRI signal increase was observed at both the PG and PA (boxed) elements identifying it as a C–T heterozygote. Reprinted with permission from reference 44.

applied for multiplexed SNP genotyping achieving both high specificity and sensitivity. The detection limit of 1 pM is comparable to that typically reported for fluorescence imaging measurements of DNA microarrays [81,82] and we expect that this detection limit will further improve with the incorporation of enzymatic amplification methods.

8.4.2 MicroRNA Detection A final example of an advanced SPRI biosensing methodology is the use of both a surface poly(A) polymerase reaction and nanoparticle amplification for the ultrasensitive microarray detection of microRNAs (miRNAs) down to a concentration of 10 fM. In the previous example, improved sensitivity was obtained from nanoparticle adsorption only. However, in this case, both the enzyme reaction and DNA-coated nanoparticles contribute to the amplification of the

Advanced Methods for SPR Imaging Biosensing

Figure 8.11

265

SPRI difference image (top middle) and corresponding line profile (bottom) obtained after a surface polyadenylation reaction (left inset) on a microarray composed of array elements with relative ssRNA surface coverages ranging from 10
1 to 10
4. The array elements were created by diluting a 1 mM solution of thiol-modified ssRNA with 1 mM thiol-modified ssDNA solution prior to surface immobilization and surface polyadenylation reaction at the 3 0 -end of the surface-attached ssRNA substrate.

SPRI response. As indicated in the left inset in Figure 8.11, poly(A) polymerase selectively catalyzes the multiple addition of adenosine residues to the 3 0 -OH end of ssRNA molecules resulting in the formation of long poly(A) tails. To characterize the surface polymerase reaction, 5 0 -thiol-modified, 3 0 -OH ssRNA was covalently attached to the gold surface via a surface thiol–maleimide reaction [34,49] to create an ssRNA monolayer with free 3 0 -hydroxyl groups that are accessible to the enzyme. Figure 8.11 shows an SPRI difference image along with the corresponding line profile following the completion of the polyadenylation reaction on a microarray where the fractional surface coverage of ssRNA ranged from a full monolayer to 10
4. The diluted array elements were prepared by mixing a 1 mM solution of thiol-modified ssRNA with a 1 mM solution of a noninteracting thiol-modified ssDNA sequence of similar length in a ratio that varied from 1:10 to 1:104. An hour was sufficient for completion of the polyadenylation reaction to gather the data in Figure 8.11, clearly showing that SPRI reflectivity changes could be detected at ssRNA surface coverages of around 10
3. A further gain in SPRI sensitivity can be achieved by the hybridization adsorption of T30 DNA-coated nanoparticles on the surface polyadenylated ssRNA. Figure 8.12 compares the total change in SPRI reflectivity measured after both polyadenylation and nanoparticle adsorption with the SPRI signal

266

Figure 8.12

Chapter 8

SPR reflectivity changes as a function of ssRNA surface coverage after surface polyadenylation reaction (K) and after nanoparticle (NP) adsorption on a surface where the polyadenylation reaction has already occurred (’). The inset shows the scheme of the hybridization–adsorption of T30-coated gold nanoparticles (Au-NP) on the poly(A) tail formed at the 3 0 -end of the surface-attached ssRNA molecules. The dashed line indicates the minimum detectable SPRI response signal. Reprinted with permission from reference 34.

obtained after the polyadenylation step only [34]. When a 10 nM nanoparticle solution was employed, the SPRI signal responded linearly at ssRNA fractional surface coverages ranging from 10
6 to 10
4. At higher coverages, the SPRI responsivity decreases with signal saturation observed at a DR of B25%, as expected from theory [53]. The remarkably low value of 10
6 corresponds to an ssRNA surface density of 5  106 molecules  cm
2 or just B10,000 RNA molecules on a 500 mm square array element! These measurements can be used to estimate what concentration of target miRNA in solution would be required to achieve a similar surface coverage following hybridization adsorption onto a complementary nucleic acid microarray. At low miRNA concentrations (C), the fractional surface coverage (y) can be described using the equation y ¼ KadsC. Given a Langmuir adsorption coefficient (Kads) of 108 M
1, a y of 10
6 corresponds to an miRNA concentration of 10 fM. Figure 8.13 shows how this polyadenylation
nanoparticle amplification scheme can be applied to the ultrasensitive detection of miRNAs using SPRI [34]. The recent surge in interest in miRNAs has led to an increased need for new methods that can perform multiplexed, quantitative analyses at very low target concentrations [83–85]. The first step (i) in our miRNA detection

Advanced Methods for SPR Imaging Biosensing

Figure 8.13

267

Detection of microRNAs using a combination of surface polyadenylation chemistry and nanoparticle-amplified SPRI. Step (i): hybridization adsorption of miRNA onto a complementary LNA array element. Step (ii): poly(A) tail addition at the 3 0 -end of surface-bound miRNAs using poly(A) polymerase. Step (iii): hybridization–adsorption of T30-coated Au nanoparticles to poly(A) tails. Reprinted with permission from reference 34.

methodology involves the sequence-specific hybridization adsorption of the miRNA target on a single-stranded locked nucleic acid (LNA) microarray. LNAs are nucleic acid analogues containing one or more nucleotides modified with an extra bridge connecting the 2 0 -O and 4 0 -C atoms of the ribose moiety; the RNA–LNA binding strength (B108 M
1) is at least 10 times greater than that associated with RNA–DNA heteroduplex formation [83]. The presence of the surface-bound miRNA is then detected with SPRI following polyadenylation (step ii) and T30-coated nanoparticle adsorption (step iii). Initial measurements were performed using synthetic analogues of the target miRNA molecules with the SPRI response was found to increase linearly over a concentration range of 10–500 fM. Both the detection limit and linear response range could be adjusted by varying the nanoparticle concentration. At miRNA concentrations of 100 pM and above, polyadenylation was sufficient to detect the presence of miRNA without nanoparticle amplification. As a final demonstration, the surface polyadenylation-nanoparticle amplification methodology was applied to the multiplexed detection of three different miRNAs present in a total RNA sample extracted from mouse liver tissue. A four-component microarray was constructed containing three LNA probes designed to bind to the known miRNA sequences [34,84], miR-16, miR-122b and miR-23b, with a DNA probe used as a negative control. A 250 ng RNA

268

Chapter 8

sample in a volume of 500 ml was circulated over the microarray surface repeatedly for 4 h followed by surface amplification. Analysis of the resulting SPRI difference image (Figure 8.14a) and corresponding line profile (Figure 8.14c) clearly shows the miR-122b sequence as the most abundant with a measured DR of 9.7% [34]. Having already calibrated the linear SPRI response range using synthetic analogues of the target sequences we were able to estimate miRNA concentrations of 20 fM, 50 fM and 2 pM for miR-16, miR-23b and miR-122b, respectively. Further verification of the concentration of the least abundant miRNA (miR-16) was obtained by repeating the measurement with the addition of 100 fM synthetic miR-16. As shown in Figures 8.14b and c, a 5-fold increase in SPRI signal was observed only at the miR-16 probe array elements. These experiments demonstrate unequivocally that polyadenylation–nanoparticle amplified SPRI measurements can be used to profile miRNA sequences quantitatively in biological samples. The current detection limit of 5 amol (10 fM in 500 ml)

Figure 8.14

Quantitative analysis of miRNAs from 250 ng of mouse liver total RNA using polyadenylation–nanoparticle amplified SPRI measurements. (a) SPRI difference image obtained by subtracting images acquired before and after the nanoparticle amplification step. (b) An SPRI difference image obtained from a separate chip using the same total RNA concentration as the top image plus the addition of 100 fM synthetic miR-16. (c) Comparison of line profiles taken from both SPRI difference images with the solid and dashed lines corresponding to top and bottom images respectively. (d) Schematic of the four-component LNA probe microarray and the line profile location. Reprinted with permission from reference 34.

Advanced Methods for SPR Imaging Biosensing

269

opens up the possibility of complete miRNA profiling with SPRI using very small amounts of sample (o250 ng).

8.5 Summary and Outlook This chapter highlights some of the recent advances that we have made in improving the microarray-based biosensing capabilities of SPRI. The continual development of new surface attachment chemistries and array fabrication methods on gold film surfaces has resulted in a large variety of biomolecular interactions that have been studied in a multiplexed format using SPRI. Furthermore, the selectivity, sensitivity and applicability of SPRI bioaffinity measurements can be greatly enhanced by incorporating surface enzyme reactions into the detection scheme. The ability to detect unlabeled DNA sequences at concentrations as low as 1 fM using RNase H amplified SPRI is a remarkable improvement compared with the previous nanomolar detection limit of SPRI based on hybridization absorption only. Also, we found the combined use of surface enzyme reactions and nanoparticle-enhanced SPRI to be particularly powerful. This allowed us to demonstrate for the first time that SPRI can be applied for SNP genotyping; the surface Taq DNA ligase reaction enhanced the specificity of the SNP detection while DNA-coated nanoparticles amplified the SPRI response. Even greater improvements in sensitivity were achieved through the combined use of a surface poly(A) polymerase reaction and DNAcoated nanoparticles, which we used to simultaneously detect multiple miRNAs present at concentrations ranging from 20 fM to 2 pM in a total RNA sample. The level of sensitivity obtained using these combined enzyme–nanoparticle amplified SPRI methods is equal to or better than that of fluorescence imaging measurements of DNA microarrays, where a detection limit of 1 pM is typically reported [79,80]. Another strength of SPRI versus fluorescence is the ability to monitor directly multi-affinity interactions involving the adsorption of two or more species on a single surface probe, which is important for the development of more complex multi-step assays. Furthermore, the nucleic acid enzymes described in this chapter are only some of the many possible enzymes (e.g. proteases, kinases) that can be utilized on biomolecules attached to microarray surfaces. In the case of more complex surface enzymatic processes, the combined use of SPFS and SPRI measurements will support the development of more complicated surface kinetic models and also help optimize the surface enzyme reaction through changes in surface density and vertical spacing of the immobilized substrate. The surface enzyme reactions at gold surfaces described in this chapter can also be coupled with fluorescence or electroactive labeled molecules; this opens up the possibility of creating new enzymatic methods that combine SPRI with other surface-sensitive detection techniques. We expect that, as more advanced surface amplification methods are developed for ultrasensitive microarray bioaffinity measurements, the technique of SPRI will continue to grow in importance as an invaluable research tool that can be applied to many areas of modern biotechnology research.

270

Chapter 8

8.6 Questions 1. A significant improvement in SPRI instrumentation is obtained by using a non-coherent light source instead of coherent laser excitation. (A) Suggest some reasons why and (B) outline other features of the optical set-up required for SPRI measurements. 2. Most surface bioaffinity measurements utilize the specific adsorption of target biomolecules (T) from solution onto a surface that has been chemically modified with probe biomolecules (P). If the target and probe interact in a simple 1:1 ratio, then in the absence of bulk transport the surface reaction can be represented in the form ka

T þ P Ð TP kd

where TP is the surface bound target–probe complex. Both P and TP are surface species, and in the Langmuir approximation their surface concentrations GP and GTP are linked to the total concentration of surface sites Gtot by Gtot ¼ GP þ GTP A. If we define y as the fraction of occupied surface sites, y ¼ GTP/Gtot, write down the differential equation for the time evolution of y (i.e. dy/ dt ¼   ) using ka, kd and the bulk concentration [T ]. B. Find the steady-state equilibrium surface coverage yeq, which is obtained in the steady-state approximation (dy/dt ¼ 0). This equation defines the Langmuir adsorption coefficient Kads ¼ ka/kd. What is yeq at a bulk concentration equal to 1/Kads? 3. One of the main factors that determine the SPRI detection limit for DNA hybridization adsorption is the lowest fractional surface coverage of target DNA on the gold chip surface that can actually be measured. At low target concentrations the solution to question 2B can be approximated as yeq ¼ Kads[T]. In the text, the DNA detection limit reported for long-range SPRI measurements was 1 nM, with a value of 1 pM described for nanoparticle-enhanced SPRI detection. In the case of RNase H amplified SPRI, a detection limit of 1 fM was reported. For the detection limits of 1 nM, 1 pM and 1 fM, calculate yeq, GTP and estimate the number of target DNA molecules contributing to a change in signal within an array element that has an area of 500  500 mm. Assume Kads ¼ 2  107 M
1 and a surface density of DNA probe molecules of 5  1012 molecules  cm
2 with all probes available for target binding. 4. In this chapter, several examples where surface enzyme reactions form the basis of advanced SPRI biosensing measurements are demonstrated. Equations (8.8)–(8.11) describe the reaction of an enzyme in solution with a surface-immobilized substrate that couples Langmuir adsorption/desorption kinetics with the rate of the surface enzyme

Advanced Methods for SPR Imaging Biosensing

271

reaction. The time-dependent SPRI signal D%R(t) responds to both enzyme adsorption and the loss of DNA or RNA substrate from the gold surface (see Figure 8.5, inset) and can be represented by D%R(t) ¼ yES–yS*. A. Using simple Euler integration methods (which can be easily applied in Excel, IGOR Pro or any mathematics software) with the initial conditions ys ¼ 1 and yES ¼ yS* ¼ 0 at time t ¼ 0, try to simulate the surface kinetic reaction described in eqs. (8.8)–(8.11). Generate two sets of plots showing yES, yS*, yS and D%R(t) versus time using the following parameter values: Parameter Set 1 where kcat 4 ka[E]: ka ¼ 1  106 M
1  s
1 kd ¼ 0.025 s
1 kcat ¼ 2.5 s
1 [E] ¼ 2.5  10
7 M b ¼ 0. Parameter Set 2 where kcat o ka [E]: ka ¼ 1  106 M
1  s
1 kd ¼ 0.025 s
1 kcat ¼ 0.025 s
1 [E] ¼ 2.5  10
7 M b ¼ 0. B. To demonstrate the impact of mass transport effects on the kinetics of the surface enzyme reaction, replot Parameter Set 2 using b ¼ 50. 5. In both this chapter and Chapter 2, gold nanoparticles are used for the sensitive detection of biomolecular interactions. What is the fundamental difference in applying these particles in SPR instruments? 6. In Sections 8.3.2 and 8.4.2, two very different enzymatic amplification methods are applied. Describe and compare both these methods. 7. A detection limit of 5 amol (10 fM in 500 ml) of miRNA with SPRI is shown for very small amounts of sample (o250 ng). Which two tricks were applied to achieve this limit of detection?

8.7 Acknowledgements This research was funded by the National Institutes of Health (2RO1 GM059622-04) and the National Science Foundation (CHE-0551935). The authors would like to thank Dr. S. Fang for obtaining the SPRI difference image shown in Figure 8.11.

References 1. D.J. Lockhart and E.A. Winzeler, Nature, 2000, 405, 827. 2. P.O. Brown and D. Botstein, Nat. Genet., 1999, 21, 33. 3. R.B. Stoughton, Annu. Rev. Biochem., 2005, 74, 53.

272

Chapter 8

4. J.M. Thomson, J. Parker, C.M. Perou and S.M. Hammond, Nat. Methods, 2004, 1, 47. 5. R.L. Stears, T. Martinsky and M. Schena, Nat. Med., 2003, 9, 140. 6. Y.P. Bao, M. Martin Huber, T. Wei, S.S. Marla, J.J. Storhoff and U.R. Mu¨ller, Nucleic Acids Res., 2005, 33, e15. 7. D. Gerion, F. Chen, B. Kannan, A. Fu, W.J. Parak, D.J. Chen, A. Majumdar and A.P. Alivisatos, Anal. Chem., 2003, 75, 4766. 8. P.F. Predki, Curr. Opin. Chem. Biol., 2004, 8, 8. 9. H. Zhu and M. Snyder, Curr. Opin. Chem. Biol., 2003, 7, 55. 10. G. Macbeath and S.L. Schreiber, Science, 2000, 289, 1760. 11. S.F. Kingsmore, Nat. Rev. Drug Discov., 2006, 5, 310. 12. P. Bertone and M. Snyder, FEBS J., 2005, 272, 5400. 13. M. Srivastava, O. Eidelman, C. Jozwik, C. Paweletz, W. Huang, P.L. Zeitlin and H.B. Pollard, Mol. Genet. Metab., 2006, 87, 303. 14. W. Gao, R. Kuick, R.P. Orchekowski, D.E. Misek, J. Qiu, A.K. Greenberg, W.N. Rom, D.E. Brenner, G.S.S. Omenn, B.B. Haab and S.M. Hanash, BMC Cancer, 2005, 5, 110. 15. W. Hickel, D. Kamp and W. Knoll, Nature, 1989, 339, 186. 16. B. Rothenha¨usler and W. Knoll, Nature, 1988, 332, 615. 17. E. Yeatman and E.A. Ash, Electron. Lett., 1987, 23, 1091. 18. H.J. Lee, A.W. Wark and R.M. Corn, Langmuir, 2006, 22, 5241. 19. K.S. Phillips, T. Wilkop, J.-J. Wu, R.O. Al-Kaysi and Q. Cheng, J. Am. Chem. Soc., 2006, 128, 9590. 20. L.K. Wolf, D.E. Fullenkamp and R.M. Georgiadis, J. Am. Chem. Soc., 2005, 127, 17453. 21. T. Wilkop, Z. Wang and Q. Cheng, Langmuir, 2004, 20, 11141. 22. M. Kyo, K. Usui-Aoki and H. Koga, Anal. Chem., 2005, 77, 7115. 23. V. Kanda, P. Kitov, D.R. Bundle and M.T. McDermott, Anal. Chem., 2005, 77, 7497. 24. A. Okumura, Y. Sato, M. Kyo and H. Kawaguchi, Anal. Biochem., 2005, 339, 328. 25. M. Kyo, T. Yamamoto, H. Motohashi, T. Kamiya, T. Kuroita, T. Tanaka, J.D. Engel, B. Kawakami and M. Yamamoto, Genes Cells, 2004 9, 153. 26. J.S. Shumaker-Parry and C.T. Campbell, Anal. Chem., 2004, 76, 907. 27. J.S. Shumaker-Parry, M.H. Zareie, R. Aebersold and C.T. Campbell, Anal. Chem., 2004, 76, 918. 28. V. Kanda, J.K. Kariuki, D.J. Harrison and M.T. McDermott, Anal. Chem., 2004, 76, 7257. 29. E.A. Smith and R.M. Corn, Appl. Spectrosc., 2003, 57, 320A. 30. L. He, M.D. Musick, S.R. Nicewarner, F.G. Salinas, S.J. Benkovic, M.J. Natan and C.D. Keating, J. Am. Chem. Soc., 2000, 122, 9071. 31. L.A. Lyon, W.D. Holliway and M.J. Natan, Rev. Sci. Instrum., 1999, 70, 2076. 32. S.F. Gonzalez, M.J. Krug, M.E. Nielsen, Y. Santos and D.R. Call, J. Clin. Microbiol., 2004, 42, 1414.

Advanced Methods for SPR Imaging Biosensing

273

33. S. Sengupta, K. Onodera, A. Lai and U. Melcher, J. Clin. Microbiol., 2003, 41, 4542. 34. S. Fang, H.J. Lee, A.W. Wark and R.M. Corn, J. Am. Chem. Soc., 2006, 128, 14044. 35. C.E. Jordan, A.G. Frutos, A.J. Thiel and R.M. Corn, Anal. Chem., 1997, 69, 4939. 36. A.J. Thiel, A.G. Frutos, C.E. Jordan, R.M. Corn and L.M. Smith, Anal. Chem., 1997, 69, 4948. 37. B.P. Nelson, A.G. Frutos, J.M. Brockman and R.M. Corn, Anal. Chem., 1999, 71, 3928. 38. B. Frey and R.M. Corn, Anal. Chem., 1996, 68, 3187. 39. J.M. Brockman, A.G. Frutos and R.M. Corn, J. Am. Chem. Soc., 1999, 121, 8044. 40. A.G. Frutos, J.M. Brockman and R.M. Corn, Langmuir, 2000, 16, 2192. 41. E. Smith, M.J. Wanat, Y. Cheng, S.V.P. Barreira, A.G. Frutos and R.M. Corn, Langmuir, 2001, 17, 2502. 42. H.J. Lee, T.T. Goodrich and R.M. Corn, Anal. Chem., 2001, 73, 5525. 43. H.J. Lee, D. Nedelkov and R.M. Corn, Anal. Chem., 2006, 78, 6504. 44. Y. Li, A.W. Wark, H.J. Lee and R.M. Corn, Anal. Chem., 2006, 78, 3158. 45. A.W. Wark, H.J. Lee and R.M. Corn, Anal. Chem., 2005, 77, 3904. 46. H.J. Lee, Y. Li, A.W. Wark and R.M. Corn, Anal. Chem., 2005, 77, 5096. 47. H.J. Lee, A.W. Wark, Y. Li and R.M. Corn, Anal. Chem., 2005, 77, 7832. 48. T.T. Goodrich, H.J. Lee and R.M. Corn, J. Am. Chem. Soc., 2004, 126, 4086. 49. T.T. Goodrich, H.J. Lee and R.M. Corn, Anal. Chem., 2004, 76, 6173. 50. E.A. Smith, M.G. Erickson, A.T. Ulijasz, B. Weisblum and R.M. Corn, Langmuir, 2003, 19, 1486. 51. G.J. Wegner, H.J. Lee and R.M. Corn, Anal. Chem., 2002, 74, 5161. 52. G.J. Wegner, A.W. Wark, H.J. Lee, E. Codner, T. Saeki, S. Fang and R.M. Corn, Anal. Chem., 2004, 76, 5677. 53. B.P. Nelson, T.E. Grimsrud, M.R. Liles, R.M. Goodman and R.M. Corn, Anal. Chem., 2001, 73, 1. 54. D.J. O’Shannessy, M. Brigham-Burke, K.K. Soneson, P. Hensley and I. Brooks, Anal. Biochem., 1993, 212, 457. 55. G.G. Neminger, P. Tobiska, J. Homola and S.S. Yee, Sens. Actuators B, 2001, 74, 145. 56. F. Yang, G.W. Bradberry and J.R. Sambles, Phys. Rev. Lett., 1991, 66, 2030. 57. J.C. Quail, J.G. Rako and H.J. Simon, Opt. Lett., 1983, 8, 377. 58. D. Sarid, Phys. Rev. Lett., 1981, 47, 1927. 59. W.N. Hansen, J. Opt. Soc. Am., 1968, 58, 380. 60. http://www.corninfo.ps.uci.edu/calculations.html. 61. A. Kasry and W. Knoll, Appl. Phys. Lett., 2006, 89, 101106. 62. C. Bourdillon, C. Demaille, J. Moiroux and J. Saveant, J. Am. Chem. Soc., 1999, 121, 2401. 63. P. Schuck and A.P. Minton, Anal. Biochem., 1996, 240, 262.

274

Chapter 8

64. S. Fang, H.J. Lee, A.W. Wark, H.M. Kim and R.M. Corn, Anal. Chem., 2005, 77, 6528. 65. H.J. Lee, A.W. Wark, T.T. Goodrich, S. Fang and R.M. Corn, Langmuir, 2005, 21, 4050. 66. E.J. Cho, J.R. Collet, A.E. Szafranska and A.D. Ellington, Anal. Chim. Acta, 2006, 564, 82. 67. J.R. Collett, E.J. Cho, J.F. Lee, M. Levy, A.J. Hood, C. Wan and A.D. Ellington, Anal. Biochem., 2005, 338, 113. 68. T. Mori, A. Oguro, T. Ohtsu and Y. Nakamura, Nucleic Acids Res., 2004, 32, 6120. 69. T.G. McCauley, N. Hamaguchi and M. Stanton, Anal. Biochem., 2003, 319, 244. 70. M.B. Murphy, S.T. Fuller, P.M. Richardson and S.A. Doyle, Nucleic Acids Res., 2003, 31, e110. 71. Y. Li, H.J. Lee and R.M. Corn, Nucleic Acids Res., 2006, 34, 6416. 72. L. He, E.A. Smith, M.J. Natan and C.D. Keating, J. Phys. Chem. B, 2004, 108, 10973. 73. L.A. Lyon, D.J. Pena and M.J. Natan, J. Phys. Chem. B, 1999, 103, 5826. 74. L.A. Lyon, M.D. Musick and M.J. Natan, Anal. Chem., 1998, 70, 5177. 75. R.W. Carthew, Curr. Opin. Genet. Dev., 2006, 16, 203. 76. D. Bartel, Cell, 2004, 116, 281. 77. T. Endo, K. Kerman, N. Nagatani, Y. Takamura and E. Tamiya, Anal. Chem., 2005, 77, 6976. 78. J. Liu, S. Tian, L. Tiefenauer, P.E. Nielsen and W. Knoll, Anal. Chem., 2005, 77, 2756. 79. A.W. Peterson, L.K. Wolf and R.M. Georgiadis, J. Am. Chem. Soc., 2002, 124, 14601. 80. K. Nakatani, S. Sando and I. Saito, Nat. Biotechnol., 2001, 19, 51. 81. H.-P. Lehr, M. Reimann, A. Brandenburg, G. Sulz and H. Klapproth, Anal. Chem., 2003, 75, 2414. 82. T. Livache, E. Maillart, N. Lassalle, P. Mailley, B. Corso, P. Guedon, A. Roget and Y. Levy, J. Pharm. Biomed. Anal., 2003, 32, 687. 83. M. Castoldi, S. Schmidt, V. Benes, M. Noerholm, A.E. Kulozik, M.W. Hentze and M.U. Muckenthaler, RNA, 2006, 12, 913. 84. T. Babak, W. Zhang, Q. Morris, B.J. Blencowe and T.R. Hughes, RNA, 2004, 10, 1813. 85. P.T. Nelson, D.A. Baldwin, L.M. Scearce, J.C. Oberholtzer, J.W. Tobias and Z. Mourelatos, Nat. Methods, 2004, 1, 155. 86. B.P. Nelson, M.R. Liles, K. Frederick, R.M. Goodman and R.M. Corn, Environ. Microbiol., 2002, 4, 735. 87. J.M. Brockman, B.P. Nelson and R.M. Corn, Annu. Rev. Phys. Chem., 2000, 51, 41. 88. E.A. Smith, W.D. Thomas, L.L. Kiessling and R.M. Corn, J. Am. Chem. Soc., 2003, 125, 6140. 89. G.J. Wegner, H.J. Lee, G. Marriott and R.M. Corn, Anal. Chem., 2003, 75, 4740.

CHAPTER 9

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies WOLFGANG KNOLL, AMAL KASRY, JING LIU, THOMAS NEUMANN,1 LIFANG NIU, HYEYOUNG PARK,2 HARALD PAULSEN,3 RUDOLF ROBELEK, DANFENG YAO AND FANG YU4 Max Planck Institute for Polymer Research, Ackermannweg 10, D-55128 Mainz, Germany

9.1 Introduction Among the various sensing principles proposed for bioaffinity studies, optical evanescent wave techniques have gained the lead in popularity. Next to evanescent ellipsometry [1] and the various optical waveguide platforms [2,3], surface plasmon resonance (SPR) spectroscopy [4–6], in particular, has found widespread applications and has demonstrated its potential for the sensitive detection of bioanalytes in numerous examples [7–9]. Since its introduction as a method for bioaffinity studies in 1983 by Liedberg et al. [10] and since the presentation of the first commercial instrument by Biacore in 1990 [11], the number of papers published has grown exponentially to currently more than 1500 contributions each year (Figure 9.1). This success story is largely based on the fact that SPR represents a label-free detection principle – the mere presence of the bound analyte slightly changes the optical architecture at the sensor surface which is probed by the surface plasmon mode propagating along this 1

Present address: Graffinity, Heidelberg, Germany. Present address: Samsung, Seoul, Korea. 3 Johannes Gutenberg University of Mainz, FB Biology, Gresemundweg 2, D-55099 Mainz, Germany. 4 Present address: Stanford University, Stanford, CA, USA. 2

275

276

Chapter 9 1600

140 Surface plasmon fluorescence

Number of publications/year

1400

120

Surface plasmons

1200

100

1000

80

800

60

600

40

400

20

200 1990

1995

2000

0 2005

Year

Figure 9.1

Number of publications per year dealing with surface plasmon resonance spectroscopy (red curve) and with surface plasmon fluorescence spectroscopy (blue curve).

metal/dielectric interface. Moreover, these SPR-based detection principles offer very attractive sensitivities for in situ and real-time monitoring of bio-analytes. Another reason for the rapid growth of SPR biosensing applications is directly linked to the fact that the required functionalization of the sensor surface can be achieved fairly easily using a variety of surface modification protocols ([12–14], and including Chapter 6). One particularly simple strategy is based on the self-assembly of a monolayer of thiol derivatives on Au surfaces [15]. Figure 9.2 shows the schematics of such a multilayer architecture developed for the functionalization of SPR Au chips with a two-dimensional binding matrix for protein- and DNA-sensing processes. First, a biotinylated thiol derivative is co-assembled together with OH-terminated diluent molecules (in a molar ratio of 1:9) in a binary mixture, allowing for the specific binding of a monolayer of streptavidin, a tetrameric protein with an extremely high affinity (KA ¼ 1015 M
1) for biotin. Two of the four binding sites face the aqueous phase and thus can be used to couple other biotinylated molecules or small objects to the sensor surface generating the designed architecture with the specificity for the analyte of interest. Even though this two-dimensional binding matrix fulfils many criteria required for biosensor coatings, e.g. the efficient exposure of binding sites which are specific for the particular analyte while simultaneously minimizing non-specific adsorption [16], it is wasting sensitivity by monitoring binding

277

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies Target DNA

Biotinylated probe

Streptavidin

Biotin

SAM

Au LaSFN9

Figure 9.2

Biofunctional interfacial architecture based on the self-assembled monolayer (SAM) of a binary mixture of thiol derivatives (i.e. 10% of a biotinylated thiol in 90% of OH-terminated thiols) designed for the stable attachment of a monolayer of streptavidin as a generic binding matrix. In this example, this layer is employed for the coupling of biotinylated DNA oligonucleotide catcher probes (blue strands) for surface hybridization studies with target strands from solution (red strands).

events only within a thin two-dimensional slice of the analyte solution that is probed by the evanescent surface plasmon wave reaching some 150 nm out into the buffer [6]. Hence it was a natural next step to use a quasi-three-dimensional binding matrix based on a polymer brush that is partially coupled to the substrate and reaches out into the analyte solution some 100 nm in the case of Biacore’s dextran brush, matching the extent of the evanescent surface plasmon field. This way, binding events away from the interface, but still within the evanescent tail of the surface plasmon mode probing the brush, also contribute to the sensor signal (although with an exponentially decreasing weighting factor). Roughly 3–5 times more protein compared with the 2D matrix bound per unit area of the sensor surface is monitored, thus enhancing the sensitivity of the technique considerably. However, if very small analyte molecules with a low molecular mass are to be detected or if only very low coverage on the sensor surface (even for large proteins) can be obtained, the resulting changes of the effective refractive index probed by the surface plasmon are too minute to be detected. An example is given in Figure 9.3; the angular scans taken in the Kretschmann configuration

278 1.0

Chapter 9 R

F / cps

1

F / cps

R

6

2.0x10 0.8 0.6 0.4

reflectivity: before after

6

1.5x10 reflectivity fluorescence

fluorescence: before after hybridization

6

1.0x10 injection of target solution

5

5.0x10

0.2 0.0 46

rinsing

(b)

(a)

0.0

0 48

50

Figure 9.3

52 54 56 58 angle θ/deg

60

62

0

200 400 600 800 1000 1200 1400 1600 time t/sec

Hybridization experiments on small amounts of chromophore-labeled target strands to the probe matrix in the Kretschmann configuration. (a) Recorded reflectivities, R, and fluorescence intensities, F, as a function of the angle of incidence before (full squares and dotted line) and after the binding of the target strands to the surface immobilized catcher probes (open circles and solid line). (b) Kinetic experiment with injection of the target solution and a rinsing step, as indicated by the arrows. While reflectivity remains unchanged, the fluorescence increase demonstrates successful binding.

of an SPR surface hybridization experiment with sensor architecture as shown in Figure 9.2 before and after binding of the target analyte are virtually superimposed (a). Consequently, also the kinetic mode (b) shows no change in the reflected intensity monitored as a function of time after the injection of the target solution. Hence neither of these modes of operation allows for labelfree detection of this hybridization event. We should point out, however, that this complete lack of sensitivity is a direct consequence of the specific design of this experiment: in order to reduce the Coulombic cross-talk between neighboring binding sites, especially for studies in low ionic strength buffers, the individual catcher strands were separated from each other by coupling them via a biotin linker to the streptavidin monolayer, thus generating a very dilute binding matrix and, hence, a low analyte coverage even at 100% hybridization efficiency. It has been shown that other catcher binding matrices based, e.g., on thiolated probe strands do allow label-free detection of target hybridization, but at the expense of a dense binding matrix resulting in significant cross-talk between neighboring hybridization sites [17].

9.2 Surface Plasmon Fluorescence Spectroscopy (SPFS) An obvious way to enhance sensitivity and to push the limit of detection (LOD) to lower surface coverage is the use of fluorescent chromophores covalently attached to the analyte molecules. In this approach, the resonantly excited

279

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies photodiode laser-shutter laser 632.8 nm

chopper

2θ θ prism

polarizers flow cell

goniometer lens attenuator r filter PMT

PC

shutter controller

Figure 9.4

motorsteering

photoncounter

lock-in amplifier

Schematic of a Kretschmann surface plasmon spectrometer with added fluorescence detection unit consisting of a collection lens, an attenuator (optional), a set of filters for the separation of scattered light and a photomultiplier tube (PMT) or a (color) CCD camera for the microscopic mode of operation (cf. Figure 9.22).

surface plasmon waves excite the fluorophores and their emitted photons can be monitored by a simple detection unit attached to a conventional SPR setup [18–21], shown in Figure 9.4 for the Kretschmann configuration, but can also be realized with grating couplers.5 Just as in surface-enhanced Raman spectroscopy [22], here too one leverages on the strong optical field enhancements that can be achieved at resonant excitation of surface plasmons. This is shown in Figure 9.5, comparing attenuated total internal reflection (ATR) results with mere total internal reflection (TIR) data. The coupling prism was Ag metal coated on the lower half of the base area only and then spincoated with a thin polymer layer that was doped with a low concentration of fluorescent chromophores [23]. In this way, a mere vertical shift of the prism relative to the (horizontal) plane of incidence in the reflection setup of the Kretschmann configuration allows for recording of the reflected intensities as a function of incidence angle for either TIR or ATR mode of operation. Simultaneously, the emitted fluorescence can be monitored for both cases. While the intensity monitored for TIR shows only a flat background signal (originating mostly from the intrinsic fluorescence excited by the laser beam passing through the high-index prism), the intensity recorded in the case of the excitation of a surface plasmon mode exhibits the strong angle dependence expected for this resonance behavior. The peak intensity shows an enhancement of more than a factor of 100 for this Ag substrate. Even with an Au coating, which is typically used for chemical stability 5

As shown in Figure 9.19a.

Chapter 9 1.0

1x106

0.8

8x105

SPS Ag TIR Glass

0.6

SPFS Ag TIRF Glass

0.4

6x105

4x105

fluorescence/cps

reflectivity R

280

2x105

0.2

0.0 20

Figure 9.5

θc

25

30 angle θ/deg

35

0 40

Comparison of the reflectivities in the (attenuated) total reflection mode of excitation of a surface plasmon wave in an angular scan at a Ag/(thin) polymer film/air interface (filled circles) with just total internal reflection at a glass/polymer/air interface (open circles). Given are also the fluorescence intensities recorded as a function of time for the two modes of operation [dashed line, total internal reflection fluorescence (TIRF); solid line, surface plasmon fluorescence spectroscopy (SPFS)]. The polymer film was doped with Cy5, a commonly used fluorophore that can be excited at l ¼ 633 nm.

reasons when using aqueous buffer solutions in biosensor studies, the enhancement can account for more than an order of magnitude. The slight angular displacement between the minimum angle in the reflectivity scan and the fluorescence intensity peak position is a consequence of the resonant excitation of the surface plasmon waves interfering with the directly reflected laser beam: the re-radiated and out-coupled plasmon light destructively interferes with the plane wave of the laser reflected at the prism/metal interface provided their relative phase shift corresponds to p or 1801. For the surface plasmon, being a damped oscillator, this phase shift relative to the reflected laser is reached just slightly above the angle for maximum excitation strength at resonance. Strictly, the fluorescence gives the correct angular position of the maximum field intensity upon excitation of the surface plasmon. The intensity profile normal to the metal/dielectric interface at the correct resonance angle decays exponentially in both directions into the metal as well as into the dielectric phase. This is shown in Figure 9.6a calculated for a multilayer configuration consisting of a high-index glass prism (LaSFN9, n ¼ 1.85 at l ¼ 633 nm)/50 nm Au/water). The intensity distribution normal to the interface

281

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies (a) Prism Au layer

Water 2

12 1 8

4

0 200

0 0

100 z/nm

(b)

Quenching profile

Optical intensity Is/Io

16

AF-Ab Dextran brush

Figure 9.6

(a) Intensity profile of a surface plasmon mode at resonance (solid line) and the relative fluorescence intensity emitted by a chromophore as a function of its separation from the metal surface (dashed line); (b) schematic representation of a polymer (dextran) brush used to assemble a quasi three-dimensional binding matrix functionalized by binding sites (circles) for the chromophore (Alexa)-labeled antibody (AF-Ab).

suggests trying to bring the analyte molecules as close to the metal surface as possible in order to place their chromophores into the highest possible optical field. One has to take into account other effects on chromophores in the excited state near a metal substrate, which can lead to, e.g., quenching of fluorescence. Qualitatively, the various distance regimes are summarized in Figure 9.7. As shown in Figure 9.7a, in the immediate proximity of the metal, the resonant excitation energy (Fo¨rster) transfer from the excited state of the donor chromophore to the acceptor states of the metal substrate leads to strong quenching of the fluorescence emission, largely compromising the field enhancement obtainable in this configuration. At intermediate distances, shown in Figure 9.7b, an efficient back-coupling of the excitation energy from the vibrationally relaxed excited state of the chromophore to the metal substrate leads to the excitation of a red-shifted plasmon mode that can re-radiate via the prism at its respective resonance angle. This back-coupling effect can be used to enhance further the probability of fluorescence emission; however, the recording of this fluorescence light through he prism is inconvenient from a practical point of view. By far the easiest mode of operation is to monitor the fluorescence emitted directly from chromophores sufficiently separated from the substrate surface illustrated in Figure 9.7c. The dye molecules are still within the substantially

282

Chapter 9 Dipole-dipole coupling

Surface plasmon back-coupling

Free emission

prism

metal

dye (c) (a)

dielectric

Figure 9.7

(b)

Schematic representation of the various coupling schemes for chromophores near a metal surface. (a) If the dye is within the range of the Fo¨rster resonance energy transfer regime (B5–7 nm), the (dipolar) coupling between the donor chromophore and the acceptor (states of the) metal substrate results in efficient quenching of fluorescence; (b) at intermediate distances, efficient back-coupling leads to excitation of a red-shifted surface plasmon mode that can re-radiate via the prism and only at sufficient distances free emission of the chromophore fluorescence is observed (c).

enhanced optical field of the surface plasmon mode, but they are not quenched at all. This combination of field enhancement and fluorescence detection forms the basis for the largely enhanced sensitivity applied for a wide range of bioaffinity studies shown on selected examples in this chapter.

9.3 Interface Kinetics Based on the Langmuir Adsorption Model Evanescent wave biosensors offer an easy way to measure the kinetics of the reversible binding of a biomolecule from solution to a binding site (typically another biomolecule) immobilized on the sensor surface. Although theoretical aspects are treated in depth in Chapters 4 and 5, a brief analysis of kinetics is described here, which is relevant for Section 9.4. The conventional treatment starts with a simple 1:1 interaction model [24], equivalent to the Langmuir adsorption model [25], which is the simplest physically plausible isotherm based

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

283

on three assumptions:  the adsorption cannot proceed beyond monolayer coverage;  all sites are equivalent and the surface is uniform;  the ability of a molecule to adsorb to a given site is independent of the degree of occupation of neighboring sites. If these conditions are met, the dynamic equilibrium is given by kon

A þ B Ð AB koff

ð9:1Þ

assuming that A is the molecule binding from solution (analyte) and B is the species immobilized on the sensor surface (ligand). The forward and reverse reaction rates are described by the adsorption (association) rate constant kon and the desorption (dissociation) rate constant koff, respectively. The association process results in the formation of the complex [AB] and is described by d½AB ¼ kon ½A½B dt

ð9:2Þ

and the dissociation rate of the complex [AB] is given by


d½AB ¼ koff ½AB dt

ð9:3Þ

Once a dynamic equilibrium is established, the rates of both processes are equal, i.e. kon ½A½B ¼ koff ½AB

ð9:4Þ

Hence the equilibrium constants can be expressed by the rate constants according to Ka ¼

½AB kon ¼ ½A½B koff

ð9:5Þ

Kd ¼

½A½B koff ¼ ½AB kon

ð9:6Þ

and

where Ka and Kd are the affinity constant and the dissociation constant, respectively. This formalism is mathematically identical with that of the treatment of the interaction in the homogeneous phase. However, at the solid–liquid interface the transport (diffusion and convection) of A from the bulk solution to the interface must be taken into account (cf. Chapter 5).

284

Chapter 9

9.3.1 Mass Transport-controlled Kinetics If the surface concentration of B (ligand) is very large and the mass transport rate km is small compared with the association rate constant kon, i.e. km { kon[B], the interaction is controlled by the mass transport rate. Then the complex formation rate is solely dependent on the bulk concentration of analyte A and the binding signal increases linearly with time: d½AB ¼ km ½Abulk  dt

ð9:7Þ

This can be used for the concentration analysis of analyte A, since the slope of the initial stage of the binding curve is proportional to the analyte concentration. Theoretically, the linear range of the dose-response curve has no limitation at the lower concentration side. If the reaction rate is fully masstransport limited, the sensor surface acts like an infinite sink and [Asurface] ¼ 0. In this case, km for all practical situations can be described by [26] 2

1

km ¼ 0:98ðD=hÞ3 ðv=bxÞ3

ð9:8Þ

where D is the diffusion coefficient, h and b are the height and the width of the flow cell, respectively, v is the volumetric flow rate and x is the distance from the flow cell entrance.

9.3.2 Interaction-controlled Kinetics If the mass transport rate is much larger than the association rate constant or if the surface concentration of the immobilized species is low, i.e. km c kon[B], then [Asurface] ¼ [Abulk] and the binding rate can be expressed as d½AB ¼ kon ½A½B
koff ½AB dt

ð9:9Þ

The surface concentration of the free binding site, [B], is the difference between the concentration of the complex at saturation, [ABmax], and the current complex concentration, [AB]: ½B ¼ ½ABmax 
½AB

ð9:10Þ

Combining eqs. (9.9) and (9.10) and considering that the response R scales linearly with the complex concentration [AB], one obtains dR ¼ kon c0 ðRmax
RÞ
koff R dt

ð9:11Þ

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

285

(a)

Response

Dissociation

Association

Time

Response

1

0.5 (b)

Kd

0

Concentration

Figure 9.8

Schematics of a typical interaction analysis involving a kinetic association and dissociation process (a) and an equilibrium analysis based on a 1:1 Langmuir model.6 (b) Both methods allow for the determination of the intrinsic affinity constant of the interaction partners.

where c0 is the concentration of the analyte and Rmax is the saturation signal at sufficiently high analyte concentration. The solution of eq. (9.11) yields R¼

kon c0 Rmax ½1
e
ðkon c0 þkoff Þt  ¼ R0 ½1
e
ðkon c0 þkoff Þt  kon c0 þ koff

ð9:12Þ

This is shown schematically in Figure 9.8a as the association phase. For the dissociation phase, c0 ¼ 0, hence dR ¼
koff R dt

ð9:13Þ

and the solution becomes (cf. Figure 9.8a, dissociation phase) R ¼ R0 e
koff t

6

Note the logarithmic scale of concentration.

ð9:14Þ

286

Chapter 9

Equations (9.12) and (9.13) can be used to give kon and koff from a single set of association/dissociation experiments using non-linear curve fitting.

9.3.3 Equilibrium Analysis Once a dynamic equilibrium has been reached, the net effect of the association and dissociation process is zero, i.e.
dR ¼ kon c0 Rmax
Req
koff Req ¼ 0 dt

ð9:15Þ

where Req is the equilibrium response at a given analyte concentration c0. Therefore, the equilibrium signal reflects the affinity constant Ka and dissociation constant Kd of the interaction couple. This can be converted to a format which resembles the 1:1 Langmuir isotherm: c0 KA Rmax c0 KA þ 1 c0 Rmax ¼ c0 þ K d

Req ¼

ð9:16Þ

The isotherm is an S-shaped curve if using logarithmic axis for the concentration, as shown in Figures 9.8b and 9.14. On the application of an analyte concentration of c0 ¼ Kd, Req is half of the saturation response Rmax.

9.4 Applications of the Kinetic Model Although this pseudo-first-order kinetic model has been used very successfully in qualitative studies (such as demonstration of interactions between biomolecules), the determination of the kinetic rate constants of binding is often complicated by the fact that most binding curves deviate from the single exponential time course expected for a simple pseudo-first-order reaction. Apart from the experimental causes (e.g. the sample depletion, noise, drift, impurity), major concerns for the deviation are focused on mass transport/ rebinding effects, multivalent interactions/avidity effects, heterogeneity in the immobilized ligands/matrix effects and complex binding mechanisms. On improving the experimental design (e.g. by using high flow rates and low surface capacities) and applying advanced analysis algorithms (e.g. the Global Analysis [27], fitting association and dissociation phase data for a series of concentrations simultaneously), the contribution of most of these effects can be minimized or even corrected [28].

9.4.1 Surface Hybridization Reactions of Oligonucleotides The first set of examples that we discuss concern hybridization studies between surface-attached catcher probe oligonucleotides and chromophore-labeled

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

Figure 9.9

287

Various detection schemes for hybridization reactions between surfaceattached catcher probes and oligonucleotide targets from solution: (a) the targets carry the chromophores that emit fluorescence photons; (b) the catchers are labeled with a fluorophore which – upon hybridization – are placed further away from the substrate surface; (c) the probes are labeled with donor and the targets with acceptor dyes, leading upon hybridization to efficient energy transfer.

targets from solution. From among the feasible experimental schemes [29], a few typical ones are summarized in Figure 9.9. The most direct approach is given in (a): the sensor surface is functionalized by a single-stranded DNA oligonucleotide catcher probe with a base sequence specific for a target strand from solution which carries the fluorophore. Upon binding, the chromophores are placed in the resonantly enhanced evanescent field of the surface plasmon mode and emit strong fluorescence light. The number of emitted photons is directly related to the number of bound target strands. This scheme will be discussed in detail below. An alternative strategy is presented in Figure 9.9b with the corresponding experimental results given in Figure 9.10. In this case, the probe strand at the sensor surface carries the chromophore which emits fluorescence light at a level that reflects the compromise between the evanescent character of the excitation field and the quenching profile for energy transfer, as discussed above. Upon the binding of the target analyte, the resulting double strand stiffens and thus stretches, thereby pushing the chromophore at the end of the probe strand further away from the (quenching) sensor metal surface. The result is a net increase in the fluorescence intensity because the (slight) decrease in the optical excitation is by far overcompensated by the increase in emission intensity owing to the reduced quenching upon the growth in chromophore–metal distance. This principle for the recording of a hybridization reaction combines the best of

288

Chapter 9 140000 rinsing 120000

+ labeled probe + 75 %

fluorescence/cps

100000 80000 60000

+ target (MM0, unlabeled) 40000 20000 0 0

Figure 9.10

1000

2000 time/s

3000

4000

5000

Binding and hybridization experiment between chromophore-labeled probe strands at the sensor surface and unlabeled targets in solution. On binding, stretching of the double strand relative to the single probe strand leads to a net increase in fluorescence intensity.

both worlds: the sensitivity of fluorescence spectroscopy with the attraction of a label-free analyte molecule. The last concept that is outlined in the schematic cartoon in Figure 9.9c is based on the popular resonant energy transfer reaction between (the excited state of) a donor chromophore, attached, e.g., to the probe strand and (the ground state of) an acceptor dye coupled to the analyte target molecule [30]. Upon hybridization, the two chromophores come sufficiently close to each other to allow for efficient energy transfer, resulting in a variety of spectral changes of the observed emission that range from donor emission quenching to sensitized acceptor fluorescence emission to the appearance of novel spectral bands. The strong distance dependence of the energy transfer being efficient only within the Fo¨rster radius of typically 5–7 nm allows for a number of detailed investigations, e.g. at the single molecule level, of the structural and dynamical aspects of DNA hybridization reactions in solution and at surfaces. According to the theoretical treatment of surface hybridization reactions within the Langmuir model, one has a number of experimental options for the determination of the relevant parameters, i.e. kon and koff and from there the affinity constant Ka. Starting from an unoccupied, bare probe matrix, one can follow the association phase of the surface hybrid formation after injection of a target solution at a concentration that should be, at least, in the range of the half-saturation value c1/2 given by the affinity value for this hybrid (c1/2 ¼ Kd ¼ Ka
1). Once the saturation coverage (for this given bulk

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

289

concentration) is reached, one can record the dissociation process by continuously rinsing the flow cell with pure buffer solution, thus triggering the dissociation of the hybrids and the free target strands being washed out. If one measures these association and dissociation phases systematically at different concentrations of the injected target solution but for a very limited time only (e.g. for 10 min), and with regeneration steps between such that each time association starts from a bare probe, one can derive reliable values for kon and koff because they will be averaged over the concentration range investigated. This global analysis allows one, in particular, to check whether the dissociation rate constant, koff, is concentration (and coverage) independent as predicted by the Langmuir model, and it shows whether the association process, indeed, speeds up linearly with increasing bulk concentration c0. Finally, by monitoring equilibrium coverage as a function of c0, e.g. by recording angular fluorescence scans after a plateau has been reached in the association phase, one can perform titration experiments which yield the affinity constant directly (cf. Figure 9.8b). By comparing the various parameters determined by the different techniques, one can check for internal consistency of the prediction of the Langmuir model and, can test its applicability to surface hybridization reactions. The application of kinetic SPFS measurements for the quantification of hybridization reactions and, in particular, for the discrimination of single nucleotide polymorphisms (SNPs) between the catcher probe P2 and different target strands is illustrated in Figure 9.11a–c. The architecture of the sensor coating was identical in all three experiments, the only difference being the sequence of the various 15mer oligonucleotide target strands used [20]. The fully complementary strand T2 (MM0) (cf. Figure 9.11d) binds very fast at the target concentration employed (c0 ¼ 1 mM), reaches then a stable level of the emitted fluorescence intensity and comes off the surface only very gradually with a barely detectable loss of fluorescence intensity upon rinsing (Figure 9.11a). The kinetic behavior changes dramatically if a target solution is injected with a strand sequence that differs by only a single nucleotide within the recognition sequence of 15 bases (T1, cf. Figure 9.11d): after the injection of the target solution a slower fluorescence increase reflects the reduced association rate constant although the final intensity measured after about 15 min reaches almost the same level as in the MM0 case (Figure 9.11b). The most pronounced difference is seen during the dissociation phase induced by flushing the flow cell with buffer: it takes 2–3 h, but then the surface-bound hybrids are completely dissociated and the targets rinsed out of the flow cell. Continuous flow of the buffer solution through the cell prevents rebinding of target compounds to the sensor surface. As indicated by the full black curve, the whole process is described by a single exponential decay as predicted by the Langmuir model [cf. eq. (9.14)]. A further significant change in the kinetic response is seen if a target strand representing a mismatch 2 (MM2) situation is flowing through the cell: the fluorescence intensity barely increases. The full black curves are fits to the

290

Chapter 9 (a)

1.5 x 106

Ifl /cps

1.0 x 106

5.0 x 105 P2/T2: MM 0

0.0 0

1000

2000

3000

4000

5000

6000

7000

Time / sec (b)

1.5 x 106

Ifl /cps

1.0 x 106

5.0 x 105

P2/ T1: MM 1

0.0 0

5000

10000

15000

Time / sec (c)

1.5 x106

Ifl /cps

1.0 x106

5.0 x 105 P2/T3: MM 2 0.0 0

(d)

1000

2000 3000 Time / sec

4000

5000

Probes: 5′-biotin-T15-TGT ACA TCA CAA CTA-3′

DNA-P2: O

PNA-P2: Targets: T1: T2: T3:

O

biotinNH2

N H

O

TGT ACA TCA CAA CTA-NH2 O 6

3′-ACA TGC AGT GTT GAT - cy5 - 5′ 3′-ACATGT AGT GTT GAT - cy5 - 5′

3′-ACA TGC ACT GTT GAT - cy5 - 5′

Figure 9.11

Kinetic experiments with association and dissociation between surfaceattached probe strands P2 and target strands from solution with different complementarity, i.e. MM0, T2 (a); MM1, T1 (b); MM2, T3 (c). Target concentration in each case: c0 ¼ 1 mM. (d) The base sequences for the probes and targets, respectively, used in Figures 9.11–9.13 and 9.19.

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

Table 9.1

Rate constants kon and koff and the affinity constant Ka ¼ kon/koff, derived from the kinetic experiments shown in Figure 9.11.

Parameter
1
1

291

kon (M s ) koff (s
1) KA (M
1)

T2 (MM0) 4

3.7  10 7  10
6 5.3  109

T1 (MM1) 3

8.9  10 3.7  10
4 2.4  107

T3 (MM2) 10 7.7  10
4 1.3  104

experimental data based on the Langmuir model with rate constants summarized in Table 9.1. A global analysis with the P2/T2 system cannot yield reliable kon values because the dissociation is so slow, as shown in Figure 9.11a, that no meaningful fit to the nearly flat intensity curve (measured typically for a few minutes only) would be obtained. However, upon the introduction of a single mismatch in the base sequence of the target (T1, in Figure 9.11d) the duplex is considerably destabilized, thus the dissociation is enhanced and the loss of fluorescence intensity can be measured within the 10 min rinsing phase of the global analysis. This is shown in Figure 9.12a for a probe matrix assembled from biotinylated peptide nucleic acids (PNA-P2), the neutral mimic of the corresponding DNA-T2 (cf. Figure 9.11d). The recognition sequence was identical with the DNA P2; however, the spacer to the biotin group was a stretch of six ethylene oxide-containing units [30] instead of the 15 thymines in the case of the DNA probe strands. The rate constants ka ¼ (konc0 + koff) [cf. eq. (9.12)] derived for the PNA-P2–T1 hybridization is plotted in Figure 9.12b. From the slope of the linear dependence on the bulk concentration c0 the association rate constant is obtained as kon ¼ 3.1  103 M
1 s
1, slightly lower than that for the DNA-P2–T1 case (cf. Table 9.1). Upon flushing the cell with buffer and monitoring the dissociation phase, one can determine the koff rate constant(s) as a function of coverage according to eq. (9.14). The corresponding measurements are also shown in Figure 9.12a. Note that after starting the buffer solution an instant decrease in the fluorescence intensity can be observed, indicating the starting point of the dissociation phase. This drop originates from removal of free fluorophores (targets) in the solution near the surface flushed out by the buffer. From the decrease in the fluorescence intensity during the rinsing process the dissociation rate constant was determined as koff ¼ 2.5  10
4 s
1, which is the average of all values obtained by fitting each dissociation part of the measurement (during rinsing) using eq. (9.14). Thus, the affinity constant, Ka (Ka ¼ kon/koff), is found to be Ka ¼ 1.2  107 M
1 for the PNA-P2–T1 MM1 hybrid, which is of the same order as for the DNA–DNA hybrid. For a titration experiment, a series of angular scans were taken after target solutions of 1, 5, 10, 20, 50 and 100 nM had been injected and equilibrium for each new bulk concentration was reached. The corresponding series of angular scans is shown in Figure 9.13a. Several features are noteworthy. (1) No

292

Chapter 9 1.5

Fluorescence / 105 cps

(a) c0 / nM 1.0 500

0.5

200 100 50 20 10

0 0

400

800 Time/sec

1200

1600

0.0036 (b)

ka /sec-1

0.0030

0.0024

0.0018

0.0012 0

100

200

300

400

500

600

Concentration c0 /nM

Figure 9.12

(a) Global analysis of the association and dissociation phase of DNA target T1 hybridization to PNA-T2 probe matrix in a solution containing 10 mM phosphate buffer. (b) ka ¼ konc0 + koff obtained from fitting the data (open squares) of (a) as a function of target concentration c0.

significant shift of the surface plasmon minimum angle was observed, the various reflectivity curves are virtually superimposed, indicating negligible increase in the optical thickness upon forming the PNA–DNA duplex. (2) The bulk solution fluorescence excited by light transmitted through the 50 nm Au substrate at 451 (below the critical angle yc ¼ 47.31) was measured at each concentration. As demonstrated in Figure 9.13b, this intensity is a linear function of the target concentration (from 1 to 100 nM), due to the direct excitation of the fluorophores in the bulk solution. (3) In Figure 13c the open squares are the data from angular scans maximum intensity (after rinsing

293

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies (a) 1.0

Transmission Exp.

Kinetic Exp. 2.4

Reflectivity R

n 100 nM n 50 nM n 20 nM n 10 nM n 5 nM n 1 nM backgroun background

0.6

0.4

0.2

1.6

0.8

(2

Fluorescence/105 cps

c0

0.8

(1 0.0

0.0 45

50

55

60

65

Angle /deg (c) Fluorescence @ 56.6°/cps

Fluorescence @45°/cps

(b) 1.5x105 1.0x105 5.0x104 0.0 0

Figure 9.13

20 40 60 80 100 Concentration c 0 /nM

2.4x105 1.6x105 8.0x104 0.0 0

20

40

60

80

100

Concentration c0 /nM

(a) Angular scans taken after saturation was reached for T2 target solutions (1, 5, 10, 20, 50 and 100 nM), hybridizing to a PNA-P2 matrix, together with the background fluorescence intensity. Open squares are reflectivities. Angular fluorescence curves before (open symbols) and after rinsing (full symbols) the flow cell with buffer following the equilibrium binding of targets from solution. (b) Fluorescence intensity measured at y ¼ 451, (below critical angle), giving information of the bulk concentration c0. (c) Plots (open squares) of the maximum fluorescence intensity (y ¼ 56.61) taken from (a) versus target concentration c0. Solid line: fit by a Langmuir isotherm with Ka ¼ 1.7  108 M
1.

for 3 min) and the solid curve is a simulated Langmuir fit using eq. (9.16) with an affinity constant of Ka ¼ 1.7  108 M
1, more than an order of magnitude lower than the affinity constant for the DNA–DNA hybrid (cf. Table 9.1). The observed behavior of binding of the targets from solution to the surfaceimmobilized probe strands can be essentially understood from the Langmuir model. The isotherms describing the three different match/mismatch situations in Figure 9.11 are displayed in Figure 9.14 by the three different S-shaped binding curves (solid, dashed and dotted black curves). They are separated

294

Chapter 9 100

Surface coverage φ/%

80

Probe P2/Target:

60

T3 (mismatch 2)

40

T1 (mismatch 1) 20

T2 (mismatch 0)

0 1E-13 1E-12 1E-11 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0.01

Concentration cO /M

Figure 9.14

Langmuir isotherms of three targets, MM0, MM1 and MM2, with significantly different c1/2 (Kd) values (cf. the dashed red line at F ¼ 50%). The concentration of the kinetic experiments displayed in Figure 9.11 is given by the blue line at c0 ¼ 1 mM.

from each other on the (bulk) concentration axes by the 2–3 orders of magnitude difference in their respective affinity constants, Ka, resulting in a shift of the curves according to the respective half-saturation constants Kd (cf. the red dashed line at F ¼ 50% and the corresponding red dotted lines to the abscissa). Based on these plots one expects, for the kinetic experiments performed at a bulk concentration of c0 ¼ 1 mM (cf. the blue dashed line in Figure 9.14), that the intensities for both the MM0 and the MM1 target strands reach almost the same level near saturation because this concentration is way above the respective half-saturation concentrations for both targets. On the other hand, the MM2 strands at this concentration are barely hybridizing to the interfacial matrix, because for this low-affinity pair the bulk concentration is not high enough to induce any significant binding. This experiment also proves that unspecific adsorption does not play any role in the observation of these binding curves: most of the molecular details of the three analytes are identical, e.g. the length of the strand, the charge (number and density) and almost all bases. Hence any non-specific interactions are virtually the same for all targets. This points to one important design principle for optimized SNP detection: by quantifying the affinities for targets of different mismatches to the surface probe, one can identify the bulk concentration best suited to differentiate between the two sequences in a (static end-point) titration experiment. In order to discriminate MM1 from MM0 in the above example, a bulk concentration of c0 ¼ 3 nM would be ideal because this

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

295

concentration is right in between the c1/2 values of the two strands. The concentration of 1 mM used in the kinetic measurements shown in Figure 9.11 obviously was too high. At this concentration both targets hybridize to the extent that almost a complete monolayer is formed, and hence would give about the same intensity when measured with a fluorescence scanner after binding. Of course, in a sensor platform such as the one described here, i.e. with the full in situ and real-time recording of the whole association and dissociation process by surface plasmon fluorescence spectroscopy, one could easily discriminate between different mismatches by the significantly different rate constants. Already after a few minutes or even seconds of recording the early phase of the association process a clear discrimination of the MM0 from the MM1 targets is feasible. When dealing with extremely dilute solutions, one needs to take into account limitations given by the diffusion of the analyte from the bulk of the flow cell across the unstirred layer to the sensor/liquid interface. This mass transferlimited regime is sketched schematically in Figure 9.15. With the closed loop of the flow cell one can easily switch, e.g. from buffer to the injection of a solution with a particular (low) analyte concentration. As was outlined above, the analyte molecules, A0, then have to diffuse across the unstirred layer and bind to the functional groups, B, at the sensor surface. This leads to a linear increase in the fluorescence intensity as a function of time after injection of the analyte solution [cf. eq. (9.7)] with a slope that is a linear function of the bulk concentration of A0.

Flow cell and fluidic system

Peristaltic pump Sample

Figure 9.15

The sensor flow cell with closed loop of analyte solution being pumped from the sample container through the measuring cell passing along the sensor surface (left). For ultra-low analyte concentrations this mass transfer-limited regime leads to a linear increase in the sensor (fluorescence) signal proportional to the bulk analyte concentration c0.

296

Chapter 9

(b) 100 pM

4

1 pM 0

105 10

0

50

100 150 Time /min

200

250

(c)

4

103 100 fM baseline stability limit -

1

0.1

-

102 10

Figure 9.16

2 pM

-

Slope/cps • min-1

10 pM

4

-

(a)

1 10 100 Concentration/pM

-

Fluorescence /10 cps

8

1000

(a) Schematics of the supramolecular architecture with catcher oligonucleotides (blue) assembled at the sensor surface for the study of hybridization reactions with PCR amplicons (black/red); (b) time-dependent recordings of the diffusion and binding of amplicons from lowconcentration solutions to the surface-attached PNA probes; (c) calibration curve obtained by plotting the slopes of the linear increase in the fluorescence intensity in (b) as a function of the respective bulk analyte concentration. The limit of detection [as defined by the intersection of the calibration curve (red line) with the baseline stability limit (blue dashed line)] is at c0 ¼ 100 fM.

An example of mass transfer-limited binding curves is given in Figure 9.16b. The analyte in this case was an amplicon of 125 bases shown schematically in Figure 9.16a, prepared by heating the double-stranded PCR products to 96 1C, followed by injecting the solution into a low ionic strength buffer at 0 1C [31,32]. This procedure was shown to separate the duplexes and to stabilize the single strands via Coulombic repulsion, thus preventing rapid re-hybridization in the bulk. Since the catcher strands at the sensor surface were PNAs, the surface hybridization of the PCR amplicons to the catcher sequences at the detector was not influenced by the low ionic strength. The bulk concentrations used in these experiments (1–100 pM) were much lower than the half-saturation value (about 3 nM [30]) for this hybridization reaction. Moreover, only the very

297

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

early stage of the diffusion/adsorption process was monitored. The linear increase in the fluorescence intensity was recorded for a maximum of 30 min in the case of the lowest concentration to only a few seconds for the highest concentration (c0 ¼ 100 pM). In each case this was followed by a rinsing step and regeneration induced by 10 mM NaOH to break all hybrids. By plotting the slope of the binding curves as a function of the corresponding bulk concentration, one finds a linear relation as shown in Figure 9.16c, which further proves the validity of the treatment of the data by the mass transferlimited regime. The intersection of this calibration plot with the baseline stability limit, which indicates the background drift of the fluorescence intensity when measured in the absence of any target molecules (see also Figure 17b), yields a limit of detection of cLOD ¼ 100 fM, far below that of label-free detection with SPR.

3 33 pM

4

Dye-labeled antibody

Evanescent field

(a)

Fluorescence / 10 cps

Fluorescence

(c)

Antigen Dextran SAM Gold

buffer

2 3.3 pM 333 fM 333 fM 1

Prism

67 fM

0

200

400

600

Gold Time / minutes 2x104 (d)

(b) Binding signal/cps min-1

1x104 Fluorescence/cps

1x104 1x104 1x104 1x104 9x103 8x103 7x103

106 Detection limit LOD≈ 500 aM 103

Baseline deviation level

6x103 5x103 -10

Figure 9.17

100 0

10

20 30 40 Time /minutes

50

60

10-2

100

102 104 Concentration / fM

106

Protein binding studies in the mass transfer-limited regime. (a) Interfacial binding matrix based on a dextran brush. (b) Control of the baseline stability: solid lines give drift stability, the average value of which (plus three times the standard deviation) defines the limit of detection (LOD). (c) Time-dependent fluorescence increase upon binding of chromophorelabeled antibodies from solutions of different (low) concentrations to the dextran matrix-immobilized antigens. (d) Calibration curve obtained by plotting the slopes of the linear fluorescence increase taken from (c) as a function of the bulk concentration. Note the linear dependence (slope ¼ 1), covering a concentration range of six orders of magnitude.

298

Chapter 9

9.4.2 Protein Binding Studies All examples discussed so far were based on the recording of fluorescence from chromophores covalently bound to the analyte molecule of interest. This does not constitute a major limitation for the detection of PCR amplicons because the use of fluorophore-labeled primers is well established. However, there might be other situations where the attachment of a fluorescent label is not possible or bears the risk of changing the characteristics of the interaction with affinity partners in a significant way, e.g. when dealing with proteins. In this case one is interested in developing detection schemes that offer the sensitivity provided by fluorescence spectroscopy, but do not require the analyte to carry the fluorophore directly. Competitive replacement assays, obviously, fulfill that requirement: the analyte of interest replaces a chromophore-labeled ligand that was preadsorbed on the binding sites at the sensor surface. Much of the optical principles for fluorescence-based detection of hybridization reactions between a surface-attached probe oligonucleotide and a target strand from solution applies to the study of protein binding reactions at surfaces in very much the same way [33]. However, generally, proteins are significantly more delicate in their behavior and, in particular, are more sensitive to the proper control of their interfacial interaction potentials. A direct consequence is the need for a significantly more complex interfacial architecture for the functionalization of sensor surfaces which allows for the attachment of ligands or other binding partners in the proper orientation, flexibility, without the loss of specificity or even the risk of partial denaturing and with the control of any unwanted non-specific binding (NSB). A number of different strategies for the attachment of binding partners for the recognition and binding reaction with proteins from solution have been reported. For the case of bioaffinity studies using surface plasmon fluorescence spectroscopy, the dextran brush introduced by Biacore turned out to be ideally suited for this purpose [34]. As can also be seen in Figure 9.6, the extent of the brush with its roughly 100 nm thickness matches very well the extent of the evanescent field of the surface plasmon. The functionalization of the brush via its –COOH groups along each dextran chain allows for the covalent attachment of antigens. If the antigens are recognized by the corresponding chromophorelabeled antibody, binding with high affinity from solution, strong fluorescence emission occurs when excited by a resonant surface plasmon field. This is shown schematically in Figure 9.17a. This matrix also minimizes intensity losses due to quenching, because most of the binding events occur sufficiently far away from the metal surface. Figure 9.17b shows a series of stability tests, i.e. recordings of the baseline fluorescence as a function of time without any analyte injected. The average value of the slope of these curves (plus three times the standard deviation) gives the baseline stability limit needed for the determination of the limit of detection. The mass transfer-limited linear increase in the fluorescence intensity seen after the injection of antibody solutions can be sensitively monitored down to extremely low concentrations, as shown in Figure 9.17c. The plot of the slope of the intensity increase as a function of

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

299

the bulk protein concentration again gives a straight line (Figure 9.17d), confirming the validity of the diffusion model. The intersection of the calibration line with the baseline stability limit again defines the LOD of this detection method and is found to be as low as cLOD ¼ 500 aM (5  10
16 M), far below LODs ever observed with SPR [21]. The highest concentration used in the above measurements was 1 nM, barely reaching the half-saturation concentration of this antibody–antigen pair with its affinity in the range of KA ¼ 2  108 M
1. However, when working with a 6.7 nM solution one can monitor the fluorescence signal at various stages of the antibody adsorption and correlate it with the resulting angular shift of the SPR signal at this coverage. In this way, one can calibrate the observed fluorescence intensity in terms of the number of proteins binding to the matrix. For the LOD of 500 aM one finds that the monitored fluorescence increase corresponds to the diffusion and binding of about 10 proteins per mm2 per minute, approaching the range of single molecule detection. All fluorescence data presented so far were based on the recording of quasimonochromatic fluorescence intensity by photon counting with the emitted light passing through a narrow bandpass filter for stray light suppression. However, an alternative means of light detection, this time with the full spectral information of the emitted fluorescence, is given by the use of a spectrometer that disperses the light into its spectral components. In this way, not only the existence of an analyte can be monitored through its fluorescence emission, but also additional processes at the sensor surface which lead to spectral changes of the emission can be recorded. The principle of this approach is illustrated in Figure 9.18 for a protein layer coupled to the sensor matrix by the well-established His-tag strategy. A generic streptavidin monolayer was introduced to couple a layer of biotinylated nitrilotriacetic acid molecules which – after activation by exposure to 500 mM NiCl2 solution – can be used to attach a monolayer of a recombinant protein/ chromophore aggregate, i.e. the light-harvesting complex LHCII known from chloroplasts, that was modified by a stretch of six histidines at the C-terminus. Figure 9.18a illustrates the final architecture, and Figure 9.18b shows the SPR protocol of the assembly process: after binding of the Ni21 ions, the injection of the protein (dark arrows) leads to the formation of an oriented LHCII monolayer of a density which is controlled by the streptavidin matrix. After rinsing the multilayer with pure buffer (white arrows) followed by a 0.35 M solution of EDTA, a high-affinity chelator for divalent ions, all Ni21 ions are released and the aggregates can be specifically disassembled between the NTA on the surface and the His-tag at the protein, which is then rinsed out. This process is fully reversible, as demonstrated by the virtually identical SPR responses of the repeated assembly and disassembly steps. After binding the unmodified LHCII bands found for the protein in solution can be identified also for the surface-immobilized protein layer (Figure 9.18c, red curve) in the fluorescence spectrum of wavelength-resolved SPFS. Injecting the unmodified protein mixed with protein labeled with an acceptor dye (DY-730 from Dyomics, Jena, Germany), the blue curve in Figure 9.18c

300

Chapter 9

(a) Intensity / 104 cps

6

4

(c)

2

0 600

700 800 Wavelength/nm

900

0.35M EDTA 0.30 Reflectivity

(b) 500µM NiCl2

0.20 0

40

80

120

Time/min

Figure 9.18

(a) Interfacial architecture assembled from a binary thiol SAM, a layer of streptavidin, biotinylated nitrilotriacetic acid (NTA), Ni21 ions and a layer of His-tagged light harvesting complex (LHCII) proteins, some with attached acceptor chromophore. (b) Assembly protocol with injection of LHCII after binding of Ni21 ions (dark arrows), rinsing step (light arrows) and injection of EDTA for the removal of Ni21. (c) Wavelength-resolved fluorescence spectra of a monolayer of LHCII (red curve) and of a similar monolayer in which a fraction of the protein complexes was chemically modified by an acceptor dye (DY-730 from Dyomics) emitting at l ¼ 758 nm (blue curve).

indicates the occurrence of an efficient intermolecular energy transfer between the proteins in the monolayer. This is documented by the appearance of the red-shifted emission shoulder that originates from the energy transfer from the unmodified donor protein (with its 12 chlorophyll and three carotenoid molecules in the complex) and the lower-energy acceptor chromophore chemically attached to some of the proteins. The degree of energy transfer depends strongly on the molar ratio of the two proteins in the immobilized monolayer (not shown).

9.5 Novel Approaches to SPFS In most cases, the Kretschmann configuration will be the coupling scheme of choice not only for SPR but also for SPFS because the evanescent character of

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

301

the surface wave being excited from the back of the prism allows for easy attachment of the flow cell from the front side and the recording of the resulting fluorescence emission off the base plane of the prism. However, the alternative scheme for surface plasmon excitation by laser light based on a surface grating structure offers a number of advantages for biosensing in general and for surface plasmon fluorescence spectroscopy in particular. For example, the need for a relatively high index prism for the momentum matching condition when working in aqueous buffer solutions can barely be met by cheap plastic prisms and requires expensive specialty glasses. Even then, relatively high angles of incidence are needed. In the following, novel instrumental solutions are given for the SPFS challenge.

9.5.1 Grating Coupling for SPFS The use of a grating coupler offers an important element of flexibility in that the choice of the grating periodicity, L, and hence the modulus of the grating vector, G ¼ 2p/L, allows for the tuning of the angle of resonant excitation, y, for a given laser wavelength to any desired value from normal incidence to essentially y ¼ 901, according to the momentum matching condition for grating coupling: kpsp ¼ kphoton siny þ mG

ð9:17Þ

The grating coupler is illustrated in Figure 9.19. Since the laser is incident from the front side, as shown in Figure 9.19a, the Au metal coating can be of any thickness provided that it remains opaque and no light is reflected from the back side (an advantage in the context of quality control for the mass production of such sensor chips). Moreover, the substrate can be made of any material, e.g. a plastic chip that can be surface structured very easily and at low cost by hot embossing with a master grating (which offers the additional advantage that all gratings are virtually identical in terms of their grating constant and amplitude, Fourier components, roughness, etc.). As in this case the laser beam passes through the flow cell before exciting the surface plasmon wave at the sensor (grating) surface, at higher analyte concentrations, this may lead to an additional fluorescence contribution, i.e. from chromophore-labeled analyte molecules flowing through the cell, resulting in a so-called bulk-jump upon injection of the analyte solution. However, for very dilute concentrations this is of no concern. Figure 9.19b gives an example of the angular dependence of the reflectivity and of the corresponding fluorescence for both, p- and s-polarized laser excitation. At higher angles the reflectivity curves show the appearance of the
1st-order diffraction intensity, resulting in a slight decrease of the specularly reflected intensity (not seen, of course, in the fluorescence intensity which only reflects the interfacial intensity of the surface plasmon mode). Note also the pronounced angular shift between the reflectivity minimum angle and the angle for maximum fluorescence intensity.

302

Chapter 9 O-ring

Au coating

5x105

substrate

4x105

Fluorescence/cps

Cover glass

(a)

PMT θout

θin flow cell

3x105 2x105

250 nM 100 nM Rinse 50 nM 20 nM 10 nM

(c)

10 nM (without t-20)

1x105 0 0

100 200 300 400 500 600 700 Time/minutes

p-light SPS s-light SPS p-light fluorescence s-light fluorescence

0.5

4

(b)

3 2 1

0

Fluorescence/106 cps

Reflecivity R

1.0 4 3 2 (d)

1 0

0 2

4

Figure 9.19

6

8 10 12 14 16 18 20 22 Incident angle/deg

0

50 100 150 200 Concentration/nM

250

(a) Schematics of the excitation of surface plasmons by a grating on the back of a flow cell, with fluorescence detection through the cell. (b) Angular reflectivity (open black symbols) and fluorescence scans (closed blue symbols) in grating coupling. The Au gating was functionalized with a probe layer to which a chromophore-labeled target strand was hybridized. (c) Titration of surface attached P2 probe matrix with DNA target T1 at different concentrations. The association and dissociation rate constants were derived from the fits to the data points (red curves) and are given in Table 9.2. (d) Langmuir isotherm constructed from the fluorescence intensities from (c) after the equilibrium coverage at each new concentration was reached, with data fit using KA ¼ 7  107 M
1 (red curve).

A typical concentration titration experiment between surface bound DNA-P2 probes and T1 targets (MM1) in solution is given in Figure 9.19c. The interfacial architecture on the grating surface was the binary SAM/ streptavidin-based matrix (shown in Figure 2) used also for the hybridization experiments in the Kretschmann configuration. The open squares are the experimental fluorescence data points and the red curves are calculations based on the Langmuir model with kon and koff parameters summarized in Table 9.2. The kon values were obtained by fitting the association phases recorded after injecting bulk solutions of different concentrations, while the final dissociation experiment (at c0 ¼ 0 nM) with its exponential decay of the fluorescence intensity could be fitted by a single koff rate constant (cf. Table 9.2). Using the various kon values and the koff value of the dissociation experiment for the determination of the KA values, one finds, in agreement with the Langmuir

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

Table 9.2

303

Rate constants, kon and koff, and the affinity constant, KA ¼ kon/koff, determined from the kinetic titration experiment presented in Figure 9.19.

Concentration (nM)

kon (M
1 s
1)

10, no Tween 20 50 100 250 0

3.4  104 4.7  104 2.3  104 2.4  104 2.4  104

koff (s
1)

2.5  10–4

KA (M
1) 14  107 19  107 9.2  107 9.6  107 9.6  107

model, that the affinity constant for the MM0 hybrid does not depend on the bulk concentration. The values obtained vary within a factor of two, well within the range of accuracy achievable in these experiments. As shown in Figure 9.19d, by plotting the fluorescence intensities recorded at the end of each association phase after injection of a new bulk solution, representing the newly established equilibrium surface coverage, one obtains again a Langmuir adsorption isotherm similar to that constructed from angular fluorescence scans presented for a Kretschmann experiment in Figure 9.13. The red curve is a fit with an affinity constant of KA ¼ 6.7  107 M
1. Given the fact that such an equilibrium titration experiment is a fundamentally different approach for the determination of affinity constants, the agreement with the kinetically determined KA values adds strong support to the application of the Langmuir model for the analysis of hybridization reactions between surfaceattached oligonucleotide probes and target strands in solution. Deviations are only seen for very long (highly charged) PCR amplicons in low ionic strength buffers or for high target densities on the sensor surface. These deviations are well explained by Coulomb interactions between probes and targets and between neighboring hybridization sites as we are dealing with highly charged interfaces and relatively dense oligo-electrolyte brushes.

9.5.2 Long-range Surface Plasmons for SPFS Before the introduction of another very promising mode of operation in SPFS, i.e. the use of long-range surface plasmons (LRSP) as the excitation light source, we briefly review a few basics of ‘‘normal’’ surface plasmons. These non-radiative, surface-bound electromagnetic modes propagate along a (noble) metal/dielectric interface with an optical field that peaks at the interface and decays exponentially both into the metal and into the dielectric. The classical Kretschmann configuration with a prism as the coupling element allows for the required energy and momentum matching between the exciting (laser) photons and the surface plasmon modes, provided that the refractive index of the prism is sufficiently high compared with the dielectric medium in contact with the metal surface. For (bio-)sensor applications with the transducer operating typically in water (n ¼ 1.33 at l ¼ 633 nm) this requires that the thin metal

304

Chapter 9

layer that guides the surface wave is in optical contact with a high index glass, e.g. LaSFN9 (Schott) prism base. This coupling geometry is shown schematically in Figure 9.20a. A typical angular reflectivity scan simulated for this configuration at a laser wavelength of 633 nm and for a 50 nm Ag layer in contact with water is given in Figure 9.20b. One important feature of the resonantly excited bound wave seen as the narrow dip in the reflectivity spectrum is the extent of its evanescent field (at resonance) into the dielectric medium. This is given in Figure 9.20c for the architecture in Figure 9.20a, but simulated for 40 nm of Au in order to match the experiment described below. One finds the typical penetration depth of PSP modes in the region of Lz ¼ 178 nm (red dashed lines in Figure 9.20c). 2.5 Field Intensity (Hy)

1.0

(a) n =1.33

Reflectivity

0.8 0.6 (b) 0.4 0.2 0.0 40

metal

2.0 1.5 (c)

1.0 0.5

50 nm Ag

40 nm Au 0.0 −500

45 50 55 60 65 70 Incident angle / deg

0

500 1000 z / nm

1500

2000

2.5

n =1.33 600 nm 1.33

0.8

Figure 9.20

(e)

0.6 0.4 0.2

metal

Field Intensity (Hy)

(d )

Reflectivity

1.0

0.0 40

2.0 1.5

(f)

1.0 0.5

50 nm Ag 45 50 55 60 65 Incident angle / deg

40 nm Au 70

0.0 -500

0

500 1000 z / nm

1500

2000

(a) Schematics of the Kretschmann configuration for surface plasmon excitation. (b) Angular reflectivity scan simulated for a high refractive index prism (n ¼ 1.85 at l ¼ 633 nm, 50 nm Ag with e ¼
17 + 0.7i, in contact water of n ¼ 1.33). (c) Optical field distribution, normal to the interface for the architecture given in (a) but calculated for a 40 nm Au layer (e ¼
12.3 + 1.29i) (dashed blue lines). For easier comparison the field is normalized to its value at the metal/water interface. (d) Kretschmann configuration for the excitation of LRSP: a low refractive index cladding layer is deposited on the high refractive index prism (n ¼ 1.33), followed by the metal coating (Ag with d ¼ 50 nm), operating in water. (e) Reflectivity scan simulation for the architecture given in (d). Note the two minima at lower and at higher angles than in ‘‘normal’’ SPR in (b). (f) Optical field distribution for a layer architecture as in (d), but calculated for a thin Au layer (d ¼ 40 nm and e ¼
12.3 + 1.29i, with a cladding layer n ¼ 1.29) to simulate the experimental conditions. Note the substantial asymmetry in the field distribution with the significantly enhanced decay length into the analyte solution {in order to show this difference from normal SPR better [cf. (c)], we scaled both fields to the value at the metal/water interface}.

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

305

The evanescent character of this surface plasmon mode guarantees the surface sensitivity and selectivity of the technique which does not suffer from bulk contributions to the detected (fluorescence) signal, e.g. by scattering from larger objects such as cells. If a very thin metal film is sandwiched between two dielectric media of (nearly) identical refractive indices, nd, plasmon modes excited at the two opposite interfaces will interact with each other provided that the metal layer is sufficiently thin (do40 nm), hence the optical fields within the metal start to overlap, establishing a transverse standing wave. This interaction lifts the dispersion degeneracy of the two identical evanescent waves and two new, coupled modes appear, a symmetrical and an antisymmetric wave (referring to their transverse electric field distribution) [35]. The latter has attracted considerable interest, because its electric field across the metal film, responsible for the energy dissipation by the lossy metal, is largely reduced and the propagation length of the mode is considerably increased. Hence this mode is also called long-range surface plasmon (LRSP) as opposed to the short-range surface plasmon (SRSP) mode which is subject to enhanced dissipation. A suitable way to excite LRSPR is the prism-coupled Kretschmann configuration given in Figure 9.20d [36,37]. In this case, a high refractive index glass prism is first coated with a low refractive index cladding layer (e.g. PTFE, n E 1.33 and a thickness of d E 600 nm), followed by the deposition of a 50 nm thin Ag layer. The simulated angular reflectivity spectrum of such a sandwich sample in water (n ¼ 1.33 at l ¼ 633 nm) is given in Figure 9.20e. The LRSP and SRSP modes can be seen as the sharp dip in the reflectivity spectrum at small angles and the much shallower and broader feature at higher angles, respectively, in roughly symmetrical angular positions relative to normal SPR (cf. Figure 9.20b and e). The magnetic field profile, Hz, of the LRSP normal to the planes of the sandwich layers, scaled to its value at the metal/analyte solution (water) interface, is given in Figure 9.20f. For this simulation we chose 40 nm gold as the metal layer and a refractive index for the PTFE cladding of n ¼ 1.29 in order to match the actual experimental conditions (see below). The asymmetric refractive index profile of the cladding layer as substrate and water as superstrate with their slightly different refractive index and also with their different thicknesses (500 nm vs. infinite, respectively) results in a rather asymmetric field distribution on both sides of the metal film. Reduced damping is responsible for the extended propagation of LRSPs and also for the narrow angular resonance seen in the reflectivity spectrum in Figure 9.20e. This has triggered interest in using these modes also in optical biosensors for detecting thin coatings covering the metal film, e.g. by an adsorbed protein or DNA analyte layer [8]. The sharp dip in the resonance with its angular position shifting up the binding of these biopolymers to slightly higher values should result in a change in reflectivity (fixed-angle mode) which should be substantially larger for an LRSP mode than for a classical SPR wave. However, the extension of the LRSP optical field reaching out much further into the buffer solution reduces the effect of a thin adsorbed coating on the

306

Chapter 9

dispersion of the plasmon mode, hence the shift of the mode, Dy, is considerably lower than in SPR. When using LRSP in conjunction with surface-plasmon field-enhanced fluorescence detection, the analyte molecules that carry a chromophore label and bind from solution to a correspondingly functionalized sensor surface will be excited by the evanescent modes of the surface plasmons with the optical intensity of the LRSP field being significantly enhanced at the interface and extending much further out into the solution compared with the situation in normal SPR (compare the red dashed lines in Figure 9.20c and f ). The enhancement for LRSP-excited fluorescence becomes particularly obvious if one places the chromophore layer with the analyte further away from the Au surface [38]. The weaker decay of the LRSP field out into the analyte solution should be seen in the difference of fluorescence excitation by the two modes of operation if the chromophores are placed at a certain distance away from the interface. To this end, two samples were prepared, one for normal SPR and another one suited for LRSP excitation. In both cases the test analyte, i.e. the chromophore-labeled antibodies, were adsorbed on a 500 nm thick PTFE layer on top of the Au substrate in order to probe the differences in the decay lengths of the two plasmon fields. As expected, the measured fluorescence excited by the normal SPR almost completely decayed to a barely detectable value (Figure 9.21, full triangles), whereas the fluorescence intensity of the identical protein layer probed by the LRSP field shows a high value (Figure 9.21, full circles): the peak intensity ratio amounts to a factor of 33.

9.5.3 Fluorescence Imaging and Color Multiplexing As an example of the use of surface plasmon fluorescence microscopy for hybridization studies in the format of an m  n sensor array, we present data obtained from experiments with quantum dots (QDs) as fluorescent probes. QDs are small, inorganic, semiconducting nanocrystals that possess unique optical properties, the most significant being (i) their broad absorption spectra and (ii) the composition- and/or size-engineered emission properties, making it possible to excite different QDs simultaneously with a single wavelength light source, but monitoring their luminescence light emitted at different wavelengths in a color-multiplexed recording mode [39]. The example for the parallel read-out of bioaffinity reactions that we briefly describe concerns a simple 4  1 array composed of four Au stripes (cf. Figure 9.23). With this array being oriented relative to the exciting laser beam at an angle near the SPR reflectivity minimum, corresponding to the highest surface plasmon field enhancement and thus the highest fluorescence intensities, the various target solutions were injected into the flow cell. Upon hybridization of the targets to the oligonucleotide catcher probe on the Au electrodes, the chromophore tags were excited by the evanescent tail of the propagating surface plasmon waves. The fluorescence photons emitted from the electrode array were imaged by a color CCD camera as shown in Figure 9.22.

307

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies 1.0 Reflectivity 0.8

1.6x105

Reflectivity/R

1.2x105 0.6

8.0x104

0.4 SPR

Fluorescence/cps

Fluorescence LR SP

4.0x104

0.2

0.0

0.0 46

48

50

52

54

56

58

Incident Angle/deg

Figure 9.21

Comparison of the angular reflectivity scans recorded with normal SPR (open circles) and LRSP (open triangles), together with the simultaneously measured angular fluorescence intensity curves (full circles for SPFS and full triangles for the LRSP fluorescence). The setup consisted of a prism, 500 nm thick cladding layer (in LRSP excitation), Au layer (40 nm in both setups, yellow slice) and 500 nm PTFE coating on top of the Au (both setups) and the chromophore-labeled protein layer (red slice) adsorbed from solution.

goniometer beam expander

laser

polarizers objective lens CCD - camera PC

Figure 9.22

Experimental setup in the Kretschmann configuration for surface plasmon field-enhanced fluorescence microscopy.

308

Chapter 9

T1’ - QD655 : 3′-CGT GGA CTG AGG ACA-QD655-5′ T2’ - QD565 : 3′- ACA TGT AGT GTT GAT -QD565-5′

PEO 12000

PL Intensity/cps

10000

P1 + P2

8000

6000

4000

P2

2000

0 500

550

600

650

Wavelength/nm

Figure 9.23

700

750

P1

SPFM images showing the hybridization following sequential introduction of T1 0 and T2 0 .

The functionality of the four stripes in Figure 9.23 was chosen as follows: the top one was covered by a poly(ethylene glycol) (PEG)–thiol passivation SAM, working as a negative control. The other three Au stripes were, from the bottom, functionalized with P1, P2 and a mixture of P1 and P2, respectively (cf. the nucleotide sequences given in Figure 9.11d). The target sequence T1 conjugated with quantum dots emitting at l ¼ 655 nm (T1 0 -QD655 with a red emission color) and the targets T2, coupled to quantum dots with an emission wavelength of l ¼ 565 nm (T2 0 -QD565 emitting in the green) could both be excited with a green He–Ne laser at l ¼ 543 nm as the light source, using Cr/Ag/Au (2/30/7 nm) as a multi-layer metal film thermally evaporated via a mask on to the glass slide for the preparation of the 4  1 array [40]. The DNA hybridizing experiment presented here was initiated by injecting first a c0 ¼ 200 nM solution of the red-emitting T1 0 -QD655 target sample in PBS. Next, the T2 0 -QD565 solution was applied to the system and allowed to hybridize to the surface for some time. Now the green fluorescence of the QD565 could be observed from the electrode that was exclusively P2 functionalized. The electrode containing a mixture of P1 and P2, however, changed its color from red to yellow due to the red/green/blue (RGB) color addition of the green fluorescence originating from T2 0 -QD565 with the red fluorescence from the targets T1 0 -QD655, as has been observed in a similar way for the case of inkjet prepared sensor spots [39] (cf. the fluorescence spectrum given in Figure 9.23). Only the electrode protected with the PEG–thiol shows no fluorescence signal, confirming the excellent passivation properties of a PEG SAM.

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

309

9.6 Conclusions This chapter has demonstrated that the combination of surface plasmon field enhancement mechanisms with fluorescence detection schemes allows for a number of bioaffinity studies that go far beyond the classical SPR concept. A critical point that is sometimes raised concerns the fact that ‘‘in this way the attraction of SPR as being a label-free detection scheme is sacrificed’’. This statement needs a few comments. First, whenever the mere presence of the analyte induces a change in the optical interfacial architecture at the transducer surface which is sufficient to be monitored by the (slight change in the dispersion relation of the) surface plasmon wave, there is, of course, no need to attach a chromophore, thus risking an unwanted change in the interaction between the affinity partners of interest. However, even in the case of label-free detection by SPR, one of the partners has to be surface immobilized. This is a constraint at least as severe as the attachment of a chromophore. Second, applying fluorescence detection schemes does not necessarily mean that the analyte has to be dye-modified. The stretching of a chromophore-labeled probe strand on the SPFS transducer surface upon hybridization of an unlabeled target strand illustrated in Figure 9.10 is an example of this. Although not discussed in this chapter explicitly, the SPFS equivalent of the widespread ELISA assay with a catcher antibody immobilized on the sensor surface, followed by the binding of the unlabeled analyte from solution and the final recording of the decoration (binding) of a second, chromophore-conjugated antibody in this sandwich assay, is another prominent example of this concept. Third, the major advantage of using fluorescence detection is the significant extension of the range of analyte concentrations that can be monitored. For example, we demonstrate that binding of proteins can be quantified by SPFS from solutions at sub-femtomolar concentrations, which is several orders of magnitude better than can be seen by SPR. So what does one sacrifice? One only gains considerably in the LOD of an analyte. Finally, the detection of fluorescence can be seen as a second information channel that allows for the simultaneous but independent observation of two analytes, e.g. a protein by SPR via its optical mass and the other one, e.g. a (relatively) small oligonucleotide via its fluorescence signal. On top of this, color multiplexing in fluorescence detection allows for multiple simultaneous recordings of analyte binding to the sensor surface, a feature that does not exist in SPR.

9.7 Questions 1. If you look at Figure 9.3a, the minimum of the reflectivity and the maximum of the fluorescence intensity measured in the angular scans do not occur at the same angle of incidence. Why? 2. Why is the fluorescence of a chromophore near a metal surface reduced (‘‘quenched’’). What is a typical dye–surface distance for which this happens?

310

Chapter 9

3. Assume that the Langmuir adsorption behaviour is correct; what type of measurements would give you the affinity constant for a binding reaction between a surface-attached receptor and a ligand from solution? 4. Imagine that you study the hybridization reactions of two different target strands (T1 and T2, differing in their sequence by one nucleotide) from solution to surface-attached catcher probe DNA strands, with KA(T1)¼10–9 M–1 and KA(T2)¼10–7 M–1. Which concentration of the targets would you chose to maximize mismatch discrimination? 5. What is the analogy between long-range surface plasmons and coupled pendulums?

9.8 Acknowledgements Some of the results presented in this chapter were collected during the PhD projects of Thorsten Liebermann, who was the first in our group to document surface plasmon field-enhanced fluorescence detection of hybridization reactions. We are grateful to Eva-Kathrin Sinner and Danfeng Yao for their help with PCR, to Peter E. Nielsen, Andrea Germini, Stefano Sforza, Roberto Corradini and Rosangela Marchelli for their support with PNAs and to Jakub Dostalek for helpful discussions concerning long-range surface plasmons. Financial support came from various projects funded by the EU (QLK12000-31658, ‘‘DNA-Track’’) and by the Deutsche Forschungsgemeinschaft (KN 224/13-1, 2).

References 1. T. Moses and Y.R. Shen, Phys. Rev. Lett., 1991, 67, 2003–2037. 2. J.D. Swalen, J. Phys. Chem., 1979, 83, 1438–1445. 3. E. Aust, S. Ito, M. Sawodny and W. Knoll, Trends Polym. Sci., 1994, 2, 313–323. 4. E. Burstein, W.P. Chen, Y.J. Chen and A. Hartstein, J. Vac. Sci Technol., 1972, 11, 1004–1019. 5. H. Raether, Surface Plasmons, Springer Tracts in Modern Physics, Springer, Berlin, 1988. 6. W. Knoll, Annu. Rev. Phys. Chem., 1998, 49, 569–638. 7. W. Knoll, M. Zizlsperger, T. Liebermann, S. Arnold, A. Badia, M. Liley, D. Piscevic, F.-J. Schmitt and J. Spinke, Colloids Surf. A, 2000, 161, 115–137. 8. A.W. Wark, H.J. Lee and R.M. Corn, Anal. Chem., 2005, 77, 3904. 9. J. Homola, Anal. Bioanal. Chem., 2003, 377, 528. 10. B. Liedberg, C. Nylander and I. Lundstro¨m, Sens. Actuators, 1983, 4, 299–304.

Surface Plasmon Fluorescence Techniques for Bioaffinity Studies

311

11. S. Lo¨fa˚s and B. Johnsson, J. Chem. Soc., Chem. Commun., 1990, 1526–1528. 12. C.E. Jordan, A.G. Frutos, A.J. Thiel and R.M. Corn, Anal. Chem., 1997, 69, 4939–4947. 13. J. Spinke, M. Liley, H.J. Guder, L. Angermaier and W. Knoll, Langmuir, 1993, 9, 1821–1825. 14. N. Huang, J. Vo¨ro¨s, S.M. De Pani, M. Textor and N.D. Spencer, Langmuir, 2002, 18, 220. 15. L. Ha¨ussling, H. Ringsdorf, F.J. Schmitt and W. Knoll, Langmuir, 1991, 7, 1837–1840. 16. F. Yu, D. Yao and W. Knoll, Anal. Chem., 2003, 75, 2610–2617. 17. K.A. Peterlinz, R.M. Georgiadis, T.M. Herne and M. Tarlov, J. Am. Chem. Soc., 1997, 119, 3401–3402. 18. J.P. Attridge, P.B. Daniels, J.K. Deakon, G.A. Robinson and G.P. Davidson, Biosens. Bioelectron., 1991, 6, 201–214. 19. T. Liebermann and W. Knoll, Colloids Surf. A, 2000, 171, 115–130. 20. T. Liebermann, W. Knoll, P. Sluka and R. Herrmann, Colloids Surf. A, 2000, 169, 337–350. 21. F. Yu, B. Persson, S. Lofas and W. Knoll, J. Am. Chem. Soc., 2004, 126, 8902–8903. 22. S. Ushioda and Y. Sasaki, Phys. Rev. B, 1983, 27, 1401–1404. 23. W. Knoll, H. Park, E. Sinner, D. Yao and F. Yu, Surf. Sci., 2004, 570, 30–42. 24. P. Schuck, Rev. Biophys. Biomol. Struct., 1997, 26, 541–566. 25. P.W. Atkins and J. de Paula, Physical Chemistry, 7th edn., Oxford University Press, 2006. 26. D.G. Myszka, T.A. Morton, M.L. Doyle and I.M. Chaiken, Biophys. Chem., 1997, 6, 127–137. 27. T.A. Morton, D.G. Myszka and I.M. Chaiken, Anal. Biochem., 1995, 227, 176–185. 28. C.R. MacKenzie, T. Hirama, S.J. Deng, D.R. Bundle, S.A. Narang and N.M. Young, J. Biol. Chem., 1996, 271, 1527–1533. 29. T. Neumann, M.L. Johansson, D. Kambhampati and W. Knoll, Adv. Funct. Mater., 2002, 12, 575–586. 30. H. Park, A. Germini, S. Sforza, R. Corradini, R. Marchelli and W. Knoll, Biointerphases, 2006, 1, 113–122. 31. D. Yao, J. Kim, P. Nielsen, E. Sinner and W. Knoll, Biophys. J., 2005, 88, 2745–2751. 32. D. Yao, F. Yu, J. Kim, J. Scholz, P. Nielsen, E.K. Sinner and W. Knoll, Nucleic Acids Res., 2004, 32, 177–192. 33. E.K. Sinner, K. Kobayashi, T. Lehmann, T. Neumann, B. Prein, J. Ru¨he, F. Yu and W. Knoll, in Biopolymers at Interfaces, M. Malmsten, Ed., 2nd edn., Marcel Dekker, New York, 2003, pp. 583–607. 34. F. Yu, B. Persson, S. Lofas and W. Knoll, Anal. Chem., 2004, 76, 6765–6770. 35. D. Sarid, Phys. Rev. Lett., 1981, 47, 1927–1930.

312

Chapter 9

36. D. Sarid, R.T. Deck, A.E. Craig, R.K. Hickernell, R.S. Jameson and J. Fasano, Appl. Opt., 1982, 21, 3993–3995. 37. G.G. Nenninger, P. Tobiesoeka, J. Homola and S. Yee, Sens. Actuators B, 2001, 74, 145–151. 38. A. Kasry and W. Knoll, Appl. Phys. Lett., 2006, 89, 101–106. 39. R. Robelek, L. Niu, E. Schmid and W. Knoll, Anal. Chem., 2004, 15, 6160–6165. 40. L. Niu and W. Knoll, Anal. Chem., 2007, 79, 2695–2707.

CHAPTER 10

SPR Imaging for Clinical Diagnostics ELAIN FU,a TIMOTHY CHINOWSKY,b KJELL NELSONa AND PAUL YAGERa a

Department of Bioengineering, University of Washington, Seattle, WA 98195, USA; b Verathon Corporation, Bothell, WA 98021, USA

10.1 Introduction Surface plasmon resonance (SPR)-based detection for clinical applications has attracted significant interest in recent years. Studies have demonstrated the detection of clinically relevant analytes such as prostate-specific antigen [1], C-reactive protein [2] and human ferritin [3] at physiologically relevant concentrations using both commercial and custom SPR-based systems. The commercial market for SPR imagers is competitive and includes instruments from multiple companies including Biacore, IBIS, GWC Technologies, Biosensing Instruments, K-Mac and Lumera (for more companies and websites, see Chapter 3). However, these instruments have generally been designed to target the research market. There is currently no commercial SPR imaging-based instrument on the market that qualifies as a true point-of-care (POC) clinical diagnostic system, i.e. a portable SPR imaging-based detection platform for the rapid, simultaneous measurement of multiple analytes in human samples that could be used in a clinical setting by minimally trained personnel. SPR imaging is well suited for use in a system for detecting clinical analytes for several reasons: 1. Detection is rapid. Because detection is based on changes in refractive index (RI) near the surface, specific binding events are quantified as they occur, without time-consuming rinses of unbound reagents; this reduces time-to-result compared with other methods such as fluorescence detection. 313

314

Chapter 10

2. Detection does not depend on labeling of the target substances. Labeling of a substance may affect its binding kinetics and affinity, in addition to increasing the complexity of the assay and the cost of reagents. 3. Immunoassays based on SPR imaging are capable of direct detection of many clinically relevant analytes. In cases in which the sensitivity or RI resolution (i.e. limit of detection) is not adequate, signal amplification methods can be used to improve detection sensitivity or RI resolution. 4. SPR imaging only requires simple optics that can be miniaturized to a size that is suitable for POC diagnostics in clinical applications. 5. SPR imaging can detect multiple types of binding events simultaneously using spatially addressed capture arrays. The number of independent measurements possible is limited by the lateral resolution of the instrument and the density of the components in the capture array. 6. The non-specific binding to the SPR surface by interferents present in complex samples can be mitigated in imaging systems by the ability to have multiple reference surfaces that may be used to correct for background signals. 7. Recent advances in microfluidics allow preconditioning of small-volume samples upstream of the SPR surface to minimize the impact of nonspecific surface fouling. In addition, preconditioning of a complex sample may significantly reduce the range of expected sample RI, thus simplifying sample analysis. In this chapter, we discuss the main requirements of a POC SPR imagingbased clinical diagnostic system: (1) mechanical and optical simplicity for miniaturization and low cost, (2) adequate imager performance for the detection of the analytes of interest, (3) robust operation despite temperature and bulk RI variations and (4) formatting of bioassays that allow for multiplexed, rapid, quantitative measurement of analytes from a complex sample.

10.2 Achieving Miniaturization and Low Cost SPR-based detection is inherently simple, with the potential for dramatic miniaturization and low cost. To date there have been multiple reports of miniature non-imaging SPR detection systems, many focusing on the use of fiber optics [4,5]. Alternatively, Kawazumi et al. [6] have described a very compact (16
9
6 cm3) instrument for the multi-channel detection of small analytes using a laser diode line source to create two-dimensional angle scans of the surface over a limited angular range of 51. The smallest reported device is the Texas Instruments (TI) multi-channel Spreeta sensor (B$50 per sensor), which uses a design based on angular interrogation without moving parts [7]. The Spreeta chip has been used as the basis for highly integrated multi-channel sensor systems [8–12]. There has been substantially less progress in the development of miniature SPR imaging systems. Imagers described in the literature are generally research

SPR Imaging for Clinical Diagnostics

315

instruments that are mechanically complex (e.g. include bulky rotation arms [13]) or specialized, costly instrumentation [14,15]. In this section, we describe application-driven SPR imager design principles that stress mechanical and optical simplicity.

10.2.1

Tuning in SPR Imager Design

SPR imagers require adjustability in incident wavelength or angle in order to achieve maximum sensitivity and dynamic range for the target range of sample RI. Several tuning methods that minimize mechanical complexity and thus increase compactness and ruggedness are described below. A simple and compact imager design that uses wavelength tuning has been reported by Fu and co-workers [16,17] The center imaging wavelength was varied over 65 nm by tilting a low-cost interference filter with respect to the collimated source beam. This wavelength range is more than adequate to detect the small changes in RI produced in the binding of sub-monolayers of biomolecules. An additional advantage of this design is the ability to extend the range of operating wavelengths by replacement of the interference filter [13]. Compact SPR imager designs that operate by angle tuning have also been reported [18,19]. Angle interrogation systems typically use bulky rotation stages to rotate the source and detector about the prism. One option to avoid this is to rely on the intrinsic field of view of the optics. Imaging optics accept rays emitted from an object point at a range of angles and focus them to a single image point. Sufficiently optimized optics will have a field of view adequate to intercept a range of angles large enough to permit adjustment-free operation over a useful range of incident angles. Figure 10.1a shows a prototype portable SPR imager developed for the rapid detection of small molecules in saliva that is based on this design consideration [18]. In this device, the angle of incidence is adjusted by mechanically translating a single LED across the optical axis of its collimator. Mechanical motion in this device may be eliminated entirely if, rather than translating the source, multiple light sources (e.g. an array of LEDs), positioned at various locations perpendicular to the optical axis, are switched on in turn [19].

10.2.2

Compact Design of Additional SPR Imager Optical Elements

In addition to wavelength or angle tuning, all SPR imagers require polarization control and sufficient optical path length to permit image formation. Strategies for meeting these two requirements in a compact package are discussed below. Many designs use the simple geometry shown in the top drawing of Figure 10.1b and require mounting of the light source and detector at the corresponding angle to the prism surface (e.g. for an equilateral prism, 601 from the normal to the sensing surface). For compact packaging it may be advantageous to fold the optical path. In the middle drawing of Figure 10.1b, two

316

Chapter 10 (a) Tablet PC for instrument control

Microfluidic card

Card mated with optics and external fluidics

(b)

(c)

Beam steering prisms Imaging lenses Detector

Light source

Figure 10.1

LCD polarizer Collimating lenses

(a) Portable SPR imager developed for the detection of small molecules in saliva. Folding of the optical path is a key factor in the compact size. Reproduced with permission from reference 18. (b) Schematic of the optical path for various prism configurations. (c) Schematic of the compact optical module that occupies approximately 25
10
5 cm3.

317

SPR Imaging for Clinical Diagnostics

additional equilateral prisms are added to direct light downwards, perpendicular to the sensing surface (the angle of incidence for the reflecting surfaces is above the critical angle, so no aluminization is needed). This configuration allows mounting of optical components along a plane parallel to the sensing surface. In the bottom drawing of Figure 10.1b, two further 451 prisms are added, creating a configuration in which the optical axes are parallel to the sensing surface. Polarization control, i.e. switching between transverse magnetic (TM) polarization (to excite surface plasmons) and transverse electric (TE) polarization (to normalize the SPR images for non-uniformity of measured intensity), is often implemented using a rotation stage, adding bulk and cost to the instrument. An alternative is the use of a liquid crystal-based device that incorporates a polarizer and a vertically-aligned nematic liquid crystal phase shifter. Application of an appropriate AC voltage rotates the polarization 901 from TM to TE. The necessary voltage source can be compactly built into the light source controller. Many of the design elements described in this section have been implemented in a portable SPR imager (see Figure 10.1a). The mechanism of tuning is angle interrogation via small range translation of a single near-infrared LED source and the use of stationary wide-field imaging optics. The polarization of the light source is electronically switchable from TM to TE through the use of a liquid crystal switchable polarizer. Folding of the optical path is a key factor in producing the compact size of the instrument. The result is an optical module, shown schematically in Figure 10.1c, that occupies approximately 25
10
5 cm3 and resides within the imager shown in Figure 10.1a.

10.3 Optimizing Imager Performance For any target application, the performance requirements for an SPR imager can be broken down into two categories: RI resolution and lateral resolution over the field of view. Optimization of these performance parameters in SPR imagers is discussed below.

10.3.1

Refractive Index Resolution

An SPR imager designed for clinical diagnostics requires adequate RI resolution for the detection of the lowest expected levels of the target analyte(s). In SPR imaging, the signal-to-noise ratio (SNR), the ratio of the SPR signal (the change in reflectivity) to noise (the uncertainty in the reflectivity measurement), can be expressed as SNR ¼

pffiffiffi dR Dn
pffiffiffiffi I
dn R

ð10:1Þ

where I is the illumination intensity, R is the reflectivity, Dn is the change in RI and the sensitivity, dR dn , is the derivative of reflectivity with respect to RI [19]. The

318

Chapter 10

reflectivity will change only slightly during measurement of a typical binding event, so for measurement of the RI resolution, DnLOD, eq. (10.1) can be expressed as 1 DnLOD / pffiffiffi dR I
dn

ð10:2Þ

Optimization of the RI resolution translates into maximizing the quantity pffiffiffi dR I
dn . The main parameters available to the user for maximizing dR dn are the interrogation wavelength and angle. Greater dR occur at longer wavelengths dn [20,21]. Choice of angle is slightly more involved because the incident angle which maximizes dR dn varies depending on the sample RI, but for a chosen illumination wavelength such an angle can be found. Increasing I should be accompanied by changes in the detector that enable an increased number of photoelectrons to be accumulated without saturating the detector. One way to increase I is to increase the range of imaging wavelengths and incident angles (Dl and Dy) for a given measurement. However, adjustment of either Dl or Dy will increase I but reduce dR dn . Determination of the optimal Dl and Dy is done by balancing these effects. The contour plot in Figure 10.2a  pffiffiffi shows the dependence of 1 ð I
dR=dnÞ on Dl and Dy, indicating that the positive effects of the increase I may outweigh the effects of a pffiffiffiin dR pffiffinegative ffi dR decrease in dR on the product I
[19]. The largest I
is predicted for dn dn dn Dl on the order of 30 nm and Dy on the order of 31. Nelson et al. reported a threefold increase in the SNR of their system (leading to improved RI resolution) through the use of a source filter with a 10 times wider bandpass (10 nm compared with a 1 nm wide bandpass filter) [13]. Image averaging and normalization can also be used to improve the RI resolution. An example of this is shown in Figure 10.2b [18]. The ‘‘Raw’’ curve represents the root-mean-square RI noise values over eight sample regions for each of four RI solutions. The ‘‘Normalized’’ curve represents root mean square RI noise values that were calculated by first normalizing the brightness of each sample region to the corresponding region of the ‘‘intensity reference’’ sector of the image and then averaging in the same manner. Because the reference sector is not in resonance, any changes in its brightness are expected to be due to light source variations, camera fluctuations and other systematic effects. Image averaging and normalization successfully reduce measured noise to a level that is inversely proportional to the square root of the number of pixels averaged [17,18,22]. In contrast, the noise level of the un-normalized data decreases at a slower rate for high levels of averaging, indicating the presence of noise components that are correlated across pixels and thus cannot be reduced with averaging. Uniform SPR response is required across the instrument field of view in order to make quantitative comparisons between multiple sites. A comparison of the normalized reflectivity of multiple regions across the image provides a useful measurement of the response uniformity. An example of this is shown in Figure 10.2c [18]. During sensing operations, the incident angle is typically adjusted such that the reflected intensity is approximately one-third up the left

319

SPR Imaging for Clinical Diagnostics log10(dn/dR /√ I) (a)

--3.5 -3.5

1.5

-3

-3

1

-2

0.5

-1

-0.5 0 0.5 Angular width, log10 degrees

1

RI measurement noise, σ

-4

2

-2.5

Spectral width, log10 nm

(b) 10-4 Raw Normalized Dark

10-5

10-6

10-7 100

101 102 103 104 Number of pixels averaged, N

(c)

Reflectivity

0.8

0.6

0.4

62

Figure 10.2

63 64 65 Incident angle, degrees

 pffiffiffi (a) Predicted dependence of 1 ð I
dR=dnÞ on light source properties. The calculation assumes a source at 670 nm and a three-layer system  pffiffiffi composed of BK-7 glass, gold and water. The largest 1 ð I
dR=dnÞ is predicted for Dl on the order of 30 nm and Dy on the order of 31. Reproduced with permission from reference 19. (b) Noise as a function of pixels averaged with and without normalization for a portable SPR imager (described in Section 10.2). Error bars indicate the standard deviation across eight sample regions and four solutions of different RI. Shown for comparison are the equivalent RI noise values measured with no illumination and a line showing the expected inverse square root relationship between the number of pixels averaged and RI noise. Reproduced with permission from reference 18. (c) Normalized reflectivity (reflected intensity of a gold surface in water divided by the reflected intensity measured using a bare glass surface) of 16 regions across the image for a portable SPR imager (described in Section 10.2). Reproduced with permission from reference 18. (d) Improved focus over the field of view by tilting the image plane according to the Scheimpflug condition. (e) Image of approximately 100 mm features taken with a portable SPR imager (described in Section 10.2). The test pattern is composed of PDMS and water. Reproduced with permission from reference 18.

320

Chapter 10 (d)

The Scheimpflug condition

Normal incidence on detector

74° incidence on detector

(e)

Figure 10.2 (Continued) side of the curve (vertical line in Figure 10.2c). Binding to the sensor surface will cause a local increase in RI, a shift of the SPR curve to higher angles and an increase in reflected intensity. Changes in intensity will be approximately linear with respect to RI, provided that the operating point remains on the portion of the SPR curve which is close to linear [23].

10.3.2

Lateral Resolution Over the Field of View

The SPR imager should have adequate lateral resolution to distinguish the surface structures of interest for the target application. In the case of an instrument designed to make measurements on multiple analytes simultaneously, the lateral resolution will determine the minimum area of the independent capture regions and also place an upper limit on the number of possible parallel measurements. Specifications for the target application will influence the desired quality of focus over the field of view of the instrument and the choice of operating wavelength. One factor that affects the lateral resolution of an SPR imager is the quality of focus over the instrument field of view. A notable feature of SPR imager optics is that the object (i.e. the SPR sensing surface) is tilted relative to the optical axis of the imaging optics. If standard imaging optics are used, much of the object is either closer or further away than is required for best focus. SPR imagers in the literature often accept this limited depth of field. A potentially better performance alternative is to use a tilted image plane as dictated by the Scheimpflug condition [24], i.e. if the object and image are tilted such that the

SPR Imaging for Clinical Diagnostics

321

object plane, the image plane and the lens plane meet in a single line, the entire image will be in focus. The use of a tilted image plane becomes most important when a large depth of field is needed and the collimation of the input light has been relaxed in order to increase light throughput. For an SPR imager, the object tilt is often fairly large and meeting the Scheimpflug condition may present some difficulties (e.g. typical lens mounts for commercially packaged imagers will block light that has a sufficiently oblique angle of incidence and some image detectors have properties that make them unsuitable for use at oblique angles of incidence). Figure 10.2d demonstrates the quality of focus that can be gained by satisfying the Scheimpflug condition. As stated above, the change in the SPR reflectivity per unit of RI is greater at longer wavelengths. However, the use of longer wavelengths also produces surface plasmons with longer propagation lengths (e.g. approximately 25 mm for a gold layer at 830 nm [13] compared with 0.1 mm for a gold layer at 488 nm [25]). This surface plasmon propagation length presents an inherent limit to the lateral resolution of the imager, but only in the dimension parallel to surface plasmon propagation. de Bruijn et al. [26] have demonstrated the utility of rotating an object such that the critical dimension is perpendicular to the direction of surface plasmon propagation. The lateral resolution for the critical dimension is then determined only by the optics of the system. A balance between the lateral resolution and the RI resolution should be established given the requirements for the target application. For example, a portable SPR imager operating at 880 nm on gold (described in Section 10.2) can easily resolve 100 mm features (Figure 10.2e) with an RI noise level of approximately 7
106 RI units (RIU) [18]. Given a 10
7.4 mm field of view, the instrument has the potential to monitor simultaneously over 1000 independent samples. Additionally, averaging and normalization (described in Section 10.3) over a greater region can decrease the RI noise level at the trade-off of fewer total independent samples. An RI noise level of 2
106 RIU would limit the number of independent samples to approximately 400.

10.4 Robust Operation A requirement of any clinical diagnostic system is robust operation. The system should perform to specifications on complex human samples (such as blood, urine and saliva) under varying conditions. For SPR-based instrumentation, temperature fluctuations and variations in the sample bulk RI are of particular concern. In this section, we focus on these two critical issues.

10.4.1

Effects of Temperature Fluctuations

For an SPR-based system, temperature effects may occur in a variety of ways. First, surface plasmons, which sense the effective RI of the sample in a volume adjacent to the metal/sample interface, are affected by changes in the temperature-dependent RI of the sample. The solvent component of a typical sample

322

Chapter 10

will experience the largest RI temperature dependence. For example, a difference in temperature of approximately 0.11C in a sample of water corresponds to a 105 RIU change, near to or greater than the (limit of detection) LOD of many reported SPR-based instruments [20,27]. Second, temperature variations in the metal layer may also significantly affect the generation of surface plasmons. At higher temperatures, surface plasmons may be significantly damped due to increased phonon–electron scattering in the metal [28]. The result is a broadening of the SPR curve and a decrease in dR dn with increasing temperature [28]. Third, variations in temperature may also affect the operation of critical components of the system such as the light source output, detector response and sensor geometry. Finally, changes in ambient temperature may affect the rate and/or efficiency of the binding event(s) of interest. For example, Zeder-Lutz et al. [29] found a 10-fold change in the ratio of the association rate to disassociation rate for an antibody–antigen binding event over the temperature range 5–301C. Measurable effects of temperature variation for a given sample on SPRbased systems are exhibited as changes in the shape (e.g. width) of the SPR curve and the location of the minimum. For imaging systems, these changes in the SPR curve appear as a change in the measured reflectance at fixed wavelength and angle. The result is a temperature-induced signal that is indistinguishable from the signal due to a targeted binding event. Thus, minimizing temperature-induced SPR effects (or compensating for them) is critical to achieving robust instrument operation.

10.4.2

Strategies to Alleviate the Effects of Temperature Fluctuations

There are two general strategies to alleviate the effects of temperature variation on SPR-based instrumentation: temperature stabilization and compensation. The most basic form of temperature stabilization is the judicious choice of materials in the fabrication of imager components. For example, use of conducting materials in imager elements that contact the sample cells and upstream sample tubing can greatly aid in sample temperature stabilization [8,9]. Active heating or cooling of the system to achieve a predetermined optimal setpoint temperature is another temperature stabilization method. Thermoelectric devices, based on the Peltier–Seebeck effect,1 are often used in SPR-based instrumentation since the devices are compact, lightweight and inexpensive. They allow for both heating and cooling and are relatively easy to implement (e.g. they require no coolant). Optimal performance of these devices requires good thermal coupling between the heating/cooling elements and the sample and proper insulation of the instrument housing [8], so care should be taken in the choice of materials used in instrument fabrication. Temperature control to within 0.011C or 106 RIU in a laboratory setting has been reported with a Spreeta-based system using a thermoelectric device [11]. 1

The conversion of electric voltage to heat flux and vice versa.

SPR Imaging for Clinical Diagnostics

323

Temperature compensation can be achieved through the use of a reference channel to correct the sample signal for changes in temperature. In this method, two SPR signals are acquired simultaneously. One measurement is performed on an active channel that contains a sensor surface functionalized for the target application. A second measurement is performed on a reference channel that is resistant to the binding events of interest, but is otherwise similar to the active channel. The compensated signal is calculated by subtracting the reference channel signal from the active channel signal. Using this method, Kawazumi et al. reduced the effects of long-term temperature drift in their system to nearly the same level as that measured for short times [6]. However, the use of temperature compensation over a wide-range of temperatures has limits. An investigation of the efficacy of reference channel compensation over the temperature range 4–401C for a Spreeta-based system showed significant degradation in reference channel compensation performance at the temperature extremes (calibration was performed at intermediate temperatures) [8]. In the case of a POC instrument that requires operation over a wide range of ambient temperatures or has a small mass, a combination of these methods may be needed. Naimushin et al. [8] reported the use of both active temperature control and reference channel compensation in their portable SPR instrument. They used a thermoelectric module to control the temperature of their Spreetabased system over widely varying ambient temperatures, from 4 to 371C, to within 0.71C or a noise level of approximately 7
105 RIU. Additionally, use of a reference channel reduced this 0.71C change in temperature to below the noise limit of their instrument, 107 RIU.

10.4.3

Bulk RI Compensation

In addition to temperature change, a change in composition of the sample is another significant contributor to sample bulk RI variation. In the case of a clinical diagnostic system, natural variation in the bulk RI of complex human samples may be of particular concern. Changes in sample bulk RI may also occur intentionally in the course of an assay, such as during the binding and rinse stages. Reference channel compensation can effectively correct for variations in bulk RI [30–33] due to inherent differences in sample composition and also variations in bulk RI due to temperature. Several reports using Biacore instruments with active temperature control demonstrated compensation levels of 98–99% for bulk RI changes on the order of 103 RIU [34–36]. Reference channel compensation has the potential to correct for additional interfering influences such as non-specific binding [8] (see the discussion below) and drift due to instability in the functionalization of surface layers. Accurate compensation for these effects relies on the assumptions that (1) the reference channel can be treated to have identical non-specific binding and drift characteristics as the sample channel and (2) the reference channel is resistant to the binding event of interest. Implementation of a system that even approximately satisfies these assumptions may be challenging.

324

Chapter 10

A second effective bulk RI compensation method that is much less dependent on surface effects is internal reflection refractometry (IRR). In this method, a region of the substrate without the metal layer is used to perform measurements of reflectivity near the critical angle to determine the bulk RI. These IRR measurements can be set up to be acquired simultaneously with the SPR measurements. The compensated signal is then calculated by subtracting the IRR signal from the SPR signal. The accurate measurement of the bulk RI of solutions of varying temperature and composition has been demonstrated using this method [37]. Chinowsky and co-workers implemented this method in a Spreeta-based instrument [38,39] and achieved compensation of bulk RI variations of 5
103 RIU to within (2–4)
105 RIU. Effective compensation using this method requires calibration over the RI range of interest [39]. As mentioned above, the issue of variability in sample RI is of particular concern for an SPR-based clinical diagnostic system that processes raw human samples. For example, the RI of blood varies by 1.4
105 RIU per 10 mg dl1 of glucose [40] which, can range over 100 mg dl1 in uncontrolled diabetics. Urine samples can also vary greatly in RI, with a reported range of 6.4
103 RIU [41]. Based on the clinical application, bulk RI compensation after sample preconditioning (described below) and adjustment of imaging conditions to those optimal for a given sample may be needed.

10.5 SPR Imaging Assays SPR imaging assays can enable multiplexed, rapid, quantitative analysis of complex human samples for a multitude of clinical applications. The target application will determine the required sensitivity, LOD and speed of the assay, and also provide constraints on the sample volume, the stringency of sample preconditioning and the range of analyte concentrations. The optimal imaging assay for an application will contain design features that balance performance requirements with complexity and cost of implementation. In this section, we discuss key design considerations, sample handling and the complexity and cost of assay implementation.

10.5.1

Microfluidic Immunoassay Design for Small Molecule Analytes

A straightforward example of a microfluidic SPR imaging assay [18,42] (see also Chapter 7 for SPR immunoassays) is shown in Figure 10.3. In this assay, sample fluid containing an unknown concentration of analyte is mixed with a known amount of buffer containing antibody and then passed to the detection region. The binding of free or singly-bound antibody (antibody molecules with only one of their two binding sites occupied) to the surface-immobilized analyte within the detection region results in a change in the RI that is quantified by SPR imaging. The design of a multiplexed assay, the choice of assay quantification method and the choice of assay parameters such as flow rate, reagent

SPR Imaging for Clinical Diagnostics

Figure 10.3

325

Parallel immunoassays for the anti-epileptic drug phenytoin conducted on a portable SPR imager (described in Section 10.2). (a) Difference image shows the outcome after 3 min of anti-phenytoin binding to the BSA–phenytoin-modified surface. Solutions contained 75 nM antiphenytoin IgG in phosphate buffer premixed with 50, 25 or 0 nM phenytoin (left to right streams, respectively). Flow is from bottom to top. The dark region at the bottom of the difference image is a nonfouling poly(ethylene glycol) (PEG)-terminated thiol surface. The channel dimensions (60 mm deep, 3.8 mm wide) and flow rates (750 nl s1 total, 250 nl s1 per flow stream) used in this assay ensured laminar flow with negligible diffusion between the streams. The scale bars represent 1 mm in each dimension. Squares demarcate the pixels used to calculate the average intensity values plotted in (b). Contrast has been adjusted for display. (b) The average intensities demonstrate that both the rate of binding and the total coverage of the antibody anti-correlate with the amount of competitor in the sample. Solid lines are the average intensities in the PEG regions and show transient refractive index changes during the initial (‘‘wash in’’) phase of the experiment. After 1.5 min, the transients have stabilized and the remaining intensity changes can be solely attributed to antibody binding.

326

Chapter 10

concentration and sample volume are discussed below in the context of this example. Simultaneous analysis of multiple samples may be accomplished using controlled parallel flows. In the multiplexed SPR imaging assay shown in Figure 10.3, three samples of phenytoin (a small-molecule therapeutic drug) in buffer (premixed with antibody) are measured simultaneously. Laminar flow, with negligible diffusion between the streams (i.e. low Reynolds number and high Peclet number), ensures independent measurements of each sample. This single-channel design provides for the self-balancing of flow rates, eliminating the need for precise control of fluid resistances in different channels, as flow rate variations will affect the rate of antibody binding. The SPR signals demonstrate that both the rate of binding and the total coverage of antibody anti-correlate with the concentration of phenytoin. The detection of 25 nM phenytoin for an application in which the low end of the therapeutic concentration in saliva is 200 nM [43] provides a comfortable LOD margin for potentially necessary manipulations (including dilution) of the sample. Note the rapid assay time of less than 3 min. Further multiplexing can be achieved by patterning multiple capture zones within a given fluid stream for the simultaneous detection of multiple analytes [44]. On-board positive and negative controls may be integrated similarly. For example, parallel flow streams alongside the sample stream in a single channel may be used for a run-time two-point calibration. Proper functioning of the calibrators requires that the antibody concentrations and degree of non-specific binding or interference be identical in all streams – an important consideration when using solutions with known composition as a reference for complex samples. The selection of appropriate assay parameters is largely based on the method used for quantifying the assay; typically either an analysis of the end-point signal or binding rate. Both quantification methods require detailed knowledge of the concentration gradients at the leading and trailing edges of the sample due to Taylor dispersion [45,46]. Measurement of the end-point signal is less demanding on image processing capabilities but requires precisely timed delivery of the sample volume with respect to the end-point measurement. Binding rate analysis may reduce the time to result but is only valid at relatively low fractional surface coverage, complicating comparisons between high- and lowrate assay outcomes. This method of quantification also requires well-behaved sample introduction to the binding region, with regard to leading edge Taylor dispersion, such that an appropriate, well-defined time period can be used to evaluate the binding rate. Comparisons between streams will naturally require that dispersion effects be similar. The choice of flow rate, reagent concentration and sample volume is critical for optimal assay operation. Selection of flow rate is based on a balance between time to result and reagent consumption, and also the performance characteristics of the method used to drive the fluids (e.g. pressure-driven flow). Higher flow rates reduce the time to result by increasing the flux of molecules of interest (e.g. antibodies) with access to the sensing surface. Conversely, lower

SPR Imaging for Clinical Diagnostics

327

flow rates increase the time to accumulate adequate signal, but reduce the total volume of sample and other reagents consumed [47–49]. Low flow rates also increase the time available for diffusion transverse to convective flow, increasing potential cross-talk between parallel streams in the format illustrated in Figure 10.3. The choice of reagent concentration, e.g. the concentration of antibody mixed into the sample, will mainly affect the dynamic range of the assay and should be selected appropriately for the target range of analyte concentrations. Achieving adequate sensitivity for low analyte concentrations will require similarly low antibody concentrations. However, this will increase the time needed to obtain sufficient surface coverage for adequate signal and affect the optimal choice of other assay parameters.

10.5.2

Assay Compatibility with Complex Samples

Complex human samples contain substances that interfere with the detection of surface binding events in SPR-based assays. This interference may occur either through non-specific adsorption of substances in the sample to the detection surface (increasing the background signal) or through interactions between substances in the bulk sample with the analyte such that transport or binding of the analyte to the detection surface is impeded. These effects will create either an inflated or deflated signal and are problematic for the implementation of a quantitative SPR-based assay. Two strategies for dealing with these effects will be discussed in the next section. The non-specific adsorption signal from complex human samples in SPRbased assays can be significant. As discussed previously, reference channel compensation may be used to ‘‘subtract out’’ the signal due to non-specific adsorption if an appropriate reference channel can be created. For example, Navratilova and Skladal described a competition assay in which the sample channel contained urine samples with human serum albumin (HSA) and anti-HSA, while the reference channel contained urine samples with HSA only, i.e. without anti-HSA [50]. Implementation of a reference channel is not as straightforward in all assay designs. For SPR measurements of analyte binding directly to the detection surface (relevant for large analytes), the analogous reference channel would require selective removal of the analyte of interest from the sample – an impractical requirement. An alternative is to prepare a reference region in the assay channel with a surface chemistry that is similar to that in the sample channel but does not bind the analyte of interest (as described above). The practical utility of this method depends heavily on the specific surface chemistries available and the ease with which they can be incorporated into the device. For some samples, compensation for non-specific adsorption alone is not adequate and processing of the sample to remove interfering substances before arrival at the detection region is needed. Yager et al. [51] described a two-stage preconditioning process for human saliva samples. The first stage consisted of a mechanical filter and the second stage consisted of an H-filter [52,53]. They reported removal of 73% of the glycoprotein content (predominantly mucins)

328

Figure 10.4

Chapter 10

(a) Example of an integrated diagnostic card created using a combination of polymeric laminate, PDMS and gold-coated glass. Reproduced with permission from reference 51. (b) Schematic of the on-card and off-card processes. Not all of the challenges of integration have been accomplished in the current design. Most notably, the valves and pumps have yet to be integrated onto the card. Reproduced with permission from reference 51.

SPR Imaging for Clinical Diagnostics

329

and retention of 92% of the small molecule analyte using the mechanical filter. The combination of both stages removed 98% of the glycoprotein content while retaining 27% of the small molecule analyte. These preconditioning processes permitted quantitative operation of an immunoassay for the detection of a small molecule analyte in saliva. Preconditioning of a complex sample in this manner has the added benefit of narrowing the target range of sample RI, thus reducing the burden on tuning of the optical system to the optimal imaging conditions for a given sample. This is especially useful for samples as variable across populations and over time as saliva.

10.5.3

Assay Implementation

An attractive approach to POC diagnostics for high performance with low pertest cost is to combine a permanent SPR imager (and fluid workstation) with a disposable card incorporating all sample-contacting elements. This approach also eliminates the need for cleaning the instrument between samples to prevent carry-over. In the example shown in Figure 10.4, many of the steps necessary for analysis of a saliva sample (subsequent to mechanical filtration) have been combined on to a disposable diagnostic card [51]. In this card design, the sample and waste are retained, avoiding sample contact with the electrical and optical components of the permanent instrument. The selection of materials for use in a disposable should balance application requirements against the inherent cost of the material plus the processing costs [51]. Fortunately a 50 nm layer of gold over 1 cm2 does not contribute significantly to the cost of the disposable, provided that an inexpensive fabrication method can be found. Lamination of laser-cut polymer sheets is extremely cost-effective for rapid prototyping [54] but cannot be used to form sub-micrometer features. Another material, polydimethylsiloxane (PDMS), is excellent for the production of sub-micrometer features, but is permeable to many small and hydrophobic molecules and is not readily or economically formed in high-throughput production. The example card of Figure 10.4 was created using a combination of PDMS to form a herringbone mixer [55] with small features, gold-coated glass for the SPR imaging surface and polymeric laminate for the bulk of the disposable. Practical devices used in POC settings will likely require a variety of materials and manufacturing methods.

10.6 Conclusion SPR imaging is a detection technique that is well suited for use in a clinical detection platform. In this chapter, we have discussed the key requirements of an SPR imaging-based POC diagnostic system and described progress towards implementation of these requirements into an integrated system. Assuming that some of the remaining challenges to miniaturization of instruments can be overcome, we expect to see fully integrated POC SPR imaging-based diagnostic systems introduced into the marketplace in the next 5 years.

330

Chapter 10

10.7 Questions 1. What are the main advantages of SPR imaging as a detection method for POC diagnostics? 2. What are the main challenges for SPR-based detection with complex samples? 3. How can temperature fluctuations affect SPR-based measurements? 4. Describe temperature compensation using a reference channel. 5. Describe the competing effects of an increase in Dl or Dy on RI resolution.

References 1. L. Huang, G. Reekmans, D. Saerens, J. Freidt, F. Frederix, L. Francis, S. Muyldermans, A. Campitelli and C. Van Hoof, Biosens. Bioelectron., 2005, 21, 483. 2. W. Hu, H. Hsu, A. Chiou, K. Tseng, H. Lin, G. Chang and S. Chen, Biosens. Bioelectron., 2006, 21, 1631. 3. S. Chou, W. Hsu, J. Hwang and C. Chen, Biosens. Bioelectron., 2004, 19, 999. 4. R. Slavik, J. Homola and E. Brynda, Biosens. Bioelectron., 2002, 17, 591. 5. G. Garabedian, C. Gonzalez, J. Richards, A. Knoesen, R. Spenser, S. Collins and R. L. Smith, Sens. Actuators A, 1994, 43, 202. 6. H. Kawazumi, K. Vengatajalabathy Gobi, K. Ogino, H. Maeda and N. Miura, Sens. Actuators B, 2005, 108, 791. 7. J. Melendez, R. Carr, D. Bartholomew, K. Kukanskis, J. Elkind, S. Yee, C. Furlong and R. Woodbury, Sens. Actuators B, 1996, 35, 212. 8. A. Naimushin, S. Soelberg, D. Bartholomew, J. Elkind and C. Furlong, Sens. Actuators B, 2003, 96, 253. 9. G. Neuert, S. Kufer, M. Benoit and H. Gaub, Rev. Sci. Instrum., 2005, 76, 054303. 10. A. Sesay and D. Cullen, Environ. Monit. Assess., 2001, 70, 83. 11. S. Soelberg, T. Chinowsky, G. Geiss, C. Spinelli, R. Stevens, S. Near, P. Kauffman, S. Yee and C. Furlong, Environ. Technol., 2005, 32, 669. 12. M. Suzuki, F. Ozawa, W. Sugimoto and S. Aso, Anal. Bioanal. Chem., 2002, 372, 301. 13. B. Nelson, A. Frutos, J. Brockman and R. Corn, Anal. Chem., 1999, 71, 3928. 14. S. Wang, S. Boussaad, S. Wong and N. Tao, Anal. Chem., 2000, 72, 4003. 15. S. Boussaad, J. Pean and N. Tao, Anal. Chem., 2000, 72, 222. 16. E. Fu, J. Foley and P. Yager, Rev. Sci. Instrum., 2003, 74, 3182. 17. E. Fu, T. Chinowsky, J. Foley, J. Weinstein and P. Yager, Rev. Sci. Instrum., 2004, 75, 2300. 18. T. Chinowsky, K. Johnston, T. Edwards, K. Nelson, E. Fu and P. Yager, Biosens. Bioelectron., 2007, 22, 2208–2215.

SPR Imaging for Clinical Diagnostics

331

19. T. Chinowsky, A. Mactutis, E. Fu and P. Yager, Proc. SPIE, 2004, 5261, 173–182. 20. J. Homola, S. Yee and G. Gauglitz, Sens. Actuators B, 1999, 54, 3. 21. K. Johansen, H. Arwin, I. Lundstrom and B. Liedberg, Rev. Sci. Instrum., 2000, 71, 3530. 22. T. Chinowsky, J. Quinn, D. Batholomew, R. Kaiser and J. Elkind, Sens. Actuators B, 2003, 91, 266. 23. J. Shumaker-Parry and C. Campbell, Anal. Chem., 2004, 76, 907. 24. W. Smith, Modern Optical Engineering, McGraw-Hill, New York, 2000. 25. C. Berger, R. Kooyman and J. Greve, Rev. Sci. Instrum., 1994, 65, 2829. 26. H. Debruijn, R. Kooyman and J. Greve, Appl. Opt., 1993, 32, 2426. 27. J. Homola, H. Vaisocherova, J. Dostalek and M. Piliarik, Methods, 2005, 37, 26. 28. H. Chiang, Y. Wang, P. Leung and W. Tse, Opt. Commun., 2001, 188, 283. 29. G. Zeder-Lutz, E. Zuber, J. Witz and M. Van Regenmortel, Anal. Biochem., 1997, 246, 123. 30. R. Karlsson and R. Stahlberg, Anal. Biochem., 1995, 228, 274. 31. G. Nenninger, J. Clendenning, C. Furlong and S. Yee, Sens. Actuators B, 1998, 51, 38. 32. W. Peng, S. Banerji, Y. Kim and K. Booksh, Opt. Lett., 2005, 30, 2988. 33. H. Roos, K. Magnusson, R. Karlsson, in Proceedings of the 8th International Conference on Solid-State Sensors and Actuators and Eurosensors IX, Stockholm, Sweden, volume 2, p. 517–520. General Chairman: Ingemar Lundstran Foundation for Sensor and Actuator Technology Stockholm, Sweden, 1995. 34. R. Ober and E. Ward, Anal. Biochem., 1999, 271, 70. 35. A. Frostell-Karlsson, A. Remaeus, H. Roos, K. Andersson, P. Borg, M. Hamalainen and R. Karlsson, J. Med. Chem., 2000, 43, 1986. 36. R. Karlsson, M. Kullman-Magnusson, M. Hamalainen, A. Remaeus, K. Andersson, P. Borg, E. Gyzander and J. Deinum, Anal. Biochem., 2000, 278, 1. 37. J. Grassi and R. Georgiadis, Anal. Chem., 1999, 71, 4392. 38. T. Chinowsky, A. Strong, D. Bartholomew, S. Jorgensen-Soelberg, T. Notides, C. Furlong and S. Yee, Proc. SPIE, 1999, 3857, 104–113. 39. T. Chinowsky and S. Yee, Proc. SPIE, 2002, 4578, 442–453. 40. K. Zirk and H. Poetzschke, Med. Eng. Phys., 2004, 26, 473. 41. A. Wolf and V. Pillay, Am. J. Med., 1969, 46, 837. 42. K. Nelson, N. Geisler, K. Tandon and P. Yager, presented at Micro Total Analysis Systems, Tokyo, 2006. 43. H. Liu and M. R. Delgado, Clin. Pharmacokinet., 1999, 36, 453. 44. K. Nelson, J. Foley, A. Mashadi-Hossein, P. Yager, presented at Micro Total Analysis Systems, Boston, MA, 2005. 45. A. Bancaud, G. Wagner, K. D. Dorfman and J. Viovy, Anal. Chem., 2005, 77, 833. 46. J. Ruzicka and E.H. Hansen, Flow Injection Analysis, Wiley, New York, 1988.

332

Chapter 10

47. A. Lionello, J. Josserand, H. Jensen and H.H. Girault, Lab Chip, 2005, 5, 1096. 48. A. Lionello, J. Josserand, H. Jensen and H.H. Girault, Lab Chip, 2005, 5, 254. 49. M. Zimmermann, E. Delamarche, M. Wolf and P. Hunziker, Biomed. Microdevices, 2005, 7, 99. 50. I. Navratilova and P. Skladal, Supramol. Chem., 2003, 15, 109. 51. P. Yager, T. Edwards, E. Fu, K. Helton, K. Nelson, M. Tam and B. Weigl, Nature, 2006, 442, 412. 52. J. Brody and P. Yager, Sens. Actuators A, 1997, 58, 13. 53. J. Brody, P. Yager, R. Goldstein and R. Austin, Biophys. J., 1996, 71, 3430. 54. P. Yager, D. Bell, J.P. Brody, D. Qin, C. Cabrera, A. Kamholz, B.H. Weigl, presented at Micro Total Analysis Systems, Banff, Canada, 1998. 55. A. Stroock, S. Dertinger, A. Ajdari, I. Mezic, H. Stone and G. Whitesides, Science, 2002, 295, 647.

CHAPTER 11

The Benefits and Scope of Surface Plasmon Resonancebased Biosensors in Food Analysis ALAN McWHIRTER AND LENNART WAHLSTRO¨M General Electric Healthcare, Biacore AB, Rapsgatan 7, SE-754 50 Uppsala, Sweden

11.1 Introduction Although surface plasmon resonance (SPR)-based biosensors are most commonly associated with protein interaction analysis in the life sciences and drug discovery, the technology is also well established in the food industry. SPR biosensors are used commercially to screen for drug residues and to quantify vitamins in food matrices. The main advantages of SPR-based detection over alternative analytical techniques such as microbiological assays (MBA) include ease of use, simpler and faster sample preparation and reduced assay time from days to minutes in some cases. SPR biosensors offer the clearest advantages in speed over alternative techniques that rely on biological readouts such as inhibition of microbial growth for detecting antibiotics. In addition to a review of the practical aspects of SPR-based food analysis, quality control and safety aspects will be discussed in this chapter. These issues are often driven by regulations set by national authorities in addition to widespread consumer demand that the information on product packaging is a true indicator of the content. Several case studies will be covered such as the detection of growth-promoting hormone residues in meat, multi-analyte screening of veterinary drug residues in body fluids, antibiotic screening in honey and the detection of adulterants in dairy products. Together, these application areas illustrate the tremendous scope of SPR-based biosensors in 333

334

Chapter 11

food safety and quality control, fields with ever-increasing performance demands on throughput and data quality.

11.1.1

Why Analyze Food?

The packaging of most food products provides information about its contents, including additives such as preservatives or vitamins. In addition, a label may assure the consumer that the product has been tested and shown to be free of harmful or undesired substances, that the raw ingredients are from approved suppliers or that the crops or animals from which the product is derived were reared according to ethically accepted standards and that the production process was performed in compliance with safety and hygiene standards. Many of these criteria are strictly regulated and demand rigorous adherence to internationally agreed standards. Failure to comply with these standards can have serious economic and even legal implications for manufacturers and it is clearly in their interest to be as confident as possible that their products are compliant or, if not, to have the opportunity to rectify the problem as early in the production process as possible. Whatever the aim of the analysis, such as confirming the presence of beneficial vitamin supplements in baby food or detecting potentially harmful residues of growth-promoting hormones in meat, the technology used must provide that information rapidly, reliably and economically. Same-day results allow closer control of product quality and minimize the risk of costly product recalls. This chapter seeks to show how label-free analysis of food products and raw materials using the Biacore Q SPR-based biosensor satisfies these criteria.

11.1.2

Food Analysis Steps and SPR Assay Formats

Minimal sample preparation is one of the most striking advantages of SPR-based assays (Figure 11.1). For example, using the Biacore Qflex Kit Sulfonamides simply requires homogenization and centrifugation of meat samples or dilution of honey or milk samples, whereas lengthy and potentially hazardous organic solvent or solid-phase extraction is required by alternative technologies such as ELISA, TLC or HPLC. The savings in cost and time for sample extraction, together with direct analysis, are a highly attractive option for busy routine screening laboratories. Usually, only minimal sample preparation such as homogenization, extraction, centrifugation and filtration is required and accurate measurements may be made even in complex matrices. The risk of matrix effects – interference due to components other than the analyte itself in the sample or sample extract – giving false positives is minimized, since the contact time of a few minutes between sample and specific binding protein within the continuous flow system of the instrument favors high-affinity interactions over low-affinity matrix interactions. The longer contact times between proteins and sample in a typical

SPR-based Biosensors in Food Analysis

Figure 11.1

335

1. Prepare the sample – often a simple dilution or homogenization/ centrifugation step. 2. Prepare the assay using a Qflex Kit. A growing range of Qflex Kits is available for the analysis of many commercially important food additives and contaminants, including vitamins and veterinary drug residues. 3. Perform an assay on Biacore Q. Wizards guide the user through all steps from assay setup to analysis. 4. Analyze the results by interpolation results on a calibration curve. Biacore Q software automatically evaluates results and generates a report.

336

Chapter 11

ELISA (approximately 1 h) significantly increase matrix effects, causing loss of sensitivity and an increased risk of false positives. Typically, an active derivative of the analyte of interest is immobilized on the gold sensor surface of a chip. A fixed concentration of specific binding protein (often an antibody) is then mixed with the sample and injected over the sensor surface. The SPR response is inversely proportional to the concentration of the sample (Figure 11.2), which may be estimated by interpolation of the response on a calibration curve. The sensor surface needs to be regenerated after each sample injection, ready for the next sample to be tested (Figure 11.3). Although the sensitivity, specificity and working range of assays are determined to a large extent by the affinity of the interacting partners, high immobilization levels are also desirable in concentration assays. This ensures that binding is highly mass transport limited and less dependent on affinity. Although SPR is an optical detection technique, measurements may be made on colored, opaque or even turbid solutions. When a sample or calibrant solution containing an interacting partner is injected over the prepared sensor surface, the response is recorded in real time as a sensorgram and is directly related to the concentration of the interacting partner in solution. In the analysis of food samples, the binding partner of interest may be a relatively small molecule such as a vitamin or an antibiotic. Since it is difficult to interpret accurately the small changes in SPR response elicited by interactions involving

Figure 11.2

Inhibition assays are used for measuring the concentration of low molecular weight target molecules in solution. Note that the ligand which is immobilized on the sensor surface is similar to the target molecule (analyte). An excess of the detecting molecule is added to the samples prior to injection. The remaining amount of free detecting molecule binds to the sensor and is measured. Inhibition assays are based on solution competition, in which the high molecular weight detecting molecules that remain in solution are free to bind to analyte or analyte analogue on the sensor surface. The SPR response is inversely related to the concentration of the target molecule in solution. More assay formats are described in Chapter 7 (see Figure 7.2).

SPR-based Biosensors in Food Analysis

Figure 11.3

337

The entire course of the interaction may be followed in a real-time trace (sensorgram), allowing the user to visualize association, final binding level and dissociation. Analysis is fully automated and the sensor surface may be regenerated after each sample injection, ready for the next sample to be tested.

small molecules, the assays performed on Biacore Q using Qflex Kits are based on inhibition or surface competition to alleviate the problem. Interacting partners in solution with a mass below 5 kDa can be difficult to measure directly with high precision in SPR-based assays because they elicit only small changes in mass at the sensor surface. Inhibition assays are useful in such cases, in which a purified preparation of the interacting partner in the sample is immobilized on the sensor surface (Figure 11.2). A high molecular weight detecting molecule that can bind to the interacting partner under analysis is then added in excess to the sample. At equilibrium, a proportion of detecting molecules remains in solution and is free to bind to the sensor surface. In inhibition assays, the binding response is inversely related to the analyte concentration. Difficulties may be encountered in designing an inhibition assay if, for example, the interacting partner in the sample cannot easily be immobilized on the sensor surface. An alternative format in such cases is surface competition assay, in which the analyte is covalently conjugated to a larger molecule and then added to the free analyte solution prior to injection. This high molecular weight complex then competes with free analyte for binding sites on the interacting partner immobilized on the sensor surface (Figure 11.4). The competing complex should be significantly larger than the analyte so that the SPR response is attributable almost entirely to the high molecular weight analogue. The shelf-life of enzyme conjugates used in ELISA-based methods is typically 6–12 months and the production of consistent batches is difficult and time consuming. In contrast, prepared sensor surfaces are highly stable and the immobilization procedure is reproducible. The shelf-life of the surface, in fact,

338

Figure 11.4

Chapter 11

Surface competition assays are indirect assays, in which a high molecular weight complex competes with free analyte for binding sites on the ligand. The response is inversely related to analyte concentration (see also Figure 7.2).

is limited primarily by the chemical stability of the immobilized binding partner itself. Regardless of assay format, regeneration of the sensor surface with a regeneration solution is normally required between sample cycles to remove interacting partners bound to the surface. The ideal regeneration solution will sufficiently remove all traces of bound material, but at the same time it must be sufficiently mild not to perturb irreversibly the biological activity of the immobilized partner.

11.2 Biacore Q and Qflex Kits – the Workhorse of Food Analysis Biacore Q (Figure 11.5) is a fully automated SPR-based instrument from Biacore AB (Uppsala, Sweden), designed for routine analysis. Samples and reagents are injected automatically to ensure reproducibility and reliability of results by elimination of the variability inherent in manual procedures. Biacore Q Control Software is wizard-based for immobilization and analysis and is optimized to allow flexibility in assay design in addition to routine testing. Results from individual tests are clearly and automatically presented within a few minutes of the start of analysis. Methods under revision or routinely used may be protected by a password while changes are registered to comply with GLP/GMP for routine measurements. To make the analysis process as simple and consistent as possible, an extensive range of Qflex Kits have been created for use with the Biacore Q system. An overview of the currently available kits, designed for quantifying or

SPR-based Biosensors in Food Analysis

Figure 11.5

339

Biacore Q, specially adapted to the demands of routine screening and detection in the food industry. The system can be used for different assays while wizard-based software controls all processes from immobilization to data interpretation.

screening specific, commercially important food additives and contaminants, is given in Table 11.1. Analyte levels in the low picomolar range can typically be measured, but the dynamic range must be empirically determined and is dependent on the experimental conditions and the molecular weights of the interaction partners. Although Biacore Q includes a surface preparation unit that allows easy immobilization of small molecules on the sensor surface, several Qflex Kits are delivered with pre-immobilized sensor chips, further reducing manual preparation. It is easy to alternate between different Qflex Kits on the same system. As indicated in Table 11.1, Qflex Kits, containing stability-tested reagents, are available for the quantification of several vitamins and for screening veterinary drug residues such as growth promoters and antibiotics in animal products. In addition to the Qflex Kit range, the procedures leading to AOAC certification of Qflex Kit Folic Acid (a B group vitamin especially recommended during pregnancy) are described here.

11.2.1

Qflex Kits for Screening Veterinary Drug Residues

Sulfonamides are a family of broad-spectrum synthetic bacteriostatic antibiotics, which can be used against most Gram-positive and many Gram-negative organisms and protozoa. The resistance of animal pathogens to sulfonamides is widespread as a result of more than 50 years of therapeutic use, but despite this, they are still widely used in combination with other medication. Strict withdrawal periods must be observed before slaughter to prevent the entry of residues into the human food chain. The assays used to ensure compliance with EU and US regulations must be sufficiently sensitive to detect the maximum permitted residue level of 100 ppb in edible tissue. Immunoassays are frequently

Porcine muscle and kidney, honey and avian muscle, serum and plasma Bile and tissues of slaughtered pigs Bile and tissues of slaughtered pigs Bovine milk, porcine muscle and kidney, honey

Chicken muscle, honey, bovine milk, shellfish and other matrices

Bovine, porcine and sheep liver

Bovine urine Urine and liver

Honey Cereals, milk powder, milk- or soya-based infant formula, fortified beverages, vitamin premixes and dietary supplements Various health and nutritional foodstuffs, cereal products Cereals, milk powder, milk- or soya-based infant formula, fortified beverages, vitamin premixes and dietary supplements Cereals, milk powder, milk- or soya-based infant formula, fortified beverages, vitamin premixes and dietary supplements Cereals, milk powder, milk- or soya-based infant formula, fortified beverages, vitamin premixes and dietary supplements

Sulfonamides (20 sulfonamides) Sulfadiazine Sulfamethazine Streptomycin

Chloramphenicol

b-Agonists

Clenbuterol Ractopamine

Tylosin (A and B) Biotin

Pantothenic acid (vitamin B5)

Vitamin B12

Vitamin B2

Folic acid

Intended matrix

40 samples in 9 h 20 samples in 15 h (urine), 20 samples in 17 h (liver) 40 samples in 12 h 20 samples in 6 h

20 samples in 6 h 20 samples in 6 h

20 samples in 12 h

20 samples in 12 h

1 ng ml
1 17.1 ng ml
1

0.06 ng ml–1

4.4 ng ml–1

40 samples in 8 h 40 samples in 8 h 40 samples in 8 h (milk), 20 samples in 8 h (muscle/kidney), 40 samples in 8.5 h (honey) 80 samples in 24 h (chicken), 20 samples in 10 h (honey/shellfish), 40 samples in 10 h (milk) 16 samples in 36 h

40 samples in 5 h

Throughput

2.7 ppb (equivalent to 16.9 ppb in tissue) 0.6 ng ml
1 0.3 ng ml
1 28 mg l
1 (milk), 50 mg kg
1 (kidney), 69 mg kg
1 (muscle), 15 mg l
1 (honey) 0.02 ppb (chicken), 0.07 ppb (honey), 0.025 ppb (milk), 0.073 ppb (shellfish) 0.02 ng g–1 (mabuterol)
1.46 ng g
1 (pirbuterol) 0.27 ng ml–1 0.11 ng ml–1 (urine), 0.17 ng g–1 (liver) 5.7 mg kg
1 1 ng ml
1

Limit of detection

Qflex Kits for the detection of drug residues and vitamins: specifications.

Kit (target components)

Table 11.1

340 Chapter 11

SPR-based Biosensors in Food Analysis

341

used as screening tests but they tend to detect only one or two compounds from the sulfonamide family and often give a positive result for the acetylated metabolites. In contrast, assays based on Biacore Q used together with Qflex Kit Sulfonamides are capable of detecting at least 20 compounds within the sulfonamide family and have the added advantage of not recognizing the acetylated metabolites. The sulfonamide-based antimicrobial compounds sulfadiazine and sulfamethazine are commonly incorporated into porcine feedstuffs for therapeutic and prophylactic reasons. As tissue extraction is a time-consuming process, a common strategy for the detection of residues of these antibiotics in slaughtered animals is the random screening of body fluids (typically bile, urine or serum) by ELISA as an indicator of their presence in edible tissue. An indication of violative tissue concentrations of sulfamethazine or sulfadiazine from the bile screen then requires confirmation by HPLC analysis of the suspected tissue. Qflex Kit Sulfamethazine and Qflex Kit Sulfadiazine, however, can be used to quantify these specific sulfonamide residues in both bile and tissues of slaughtered pigs. Biacore Q in combination with Qflex Kit Sulfamethazine and Qflex Kit Sulfadiazine has been tested on-site at the kill line in a slaughterhouse to determine how the screens performed in the cold and humid conditions of a modern pig processing factory [1]. No adverse effects on any mechanical or electrical components were encountered either during the testing period or after the instrument was returned to the laboratory. Bile samples were first screened as a predictor of sulfonamides in the corresponding animal tissue and positive samples were confirmed using HPLC. Samples were compared with values from a calibration series containing 5% guaranteed sulfonamide-free pig bile. It is important to construct calibration curves in the presence of bile, as this medium appears to be responsible for an appreciable matrix effect. False positive prediction rates were calculated and compared with a standard enzyme immunoassay (EIA). For sulfamethazine, false-positive rates were 0.14% for the Biacore Q assay and 1.54% for EIA and for sulfadiazine the rates were 0.34% and 1.44%, respectively. No falsenegative results were obtained using the Biacore method. EIA analysis, in contrast, delivered false-negative rates of 0.14 and 0.24% for sulfamethazine and sulfadiazine, respectively. Despite the complexity of bile as a matrix, direct analysis is possible using the Biacore Q assay without clean-up or extraction. The results are available within 3–4 min, permitting analysis in real time in the slaughterhouse and, importantly, before the meat can enter the human food chain. Qflex Kit Sulfamethazine and Qflex Kit Sulfadiazine can be adapted to muscle extracts using a simple and low-cost mechanical extraction procedure involving homogenization and centrifugation. It is therefore possible to use these assays for both on-site screening and confirmation, thereby avoiding the need for HPLC. It has been demonstrated, however, that a small number of false positives were found using Biacore Q in a screening method [2]. Analysis of samples containing N4-acetyl metabolites always gives higher results than

342

Chapter 11

HPLC, due to the cross-reactivity of the antibodies, whereas HPLC only detects parent sulfonamides. This highlights the importance of using ‘‘real’’ incurred samples in evaluating assays as opposed to the simpler and more widely used technique of spiking blank samples. The validation data reported in this work were gathered from pigs that had been fed with sulfonamidesupplemented feed and then withdrawn for a specified time before slaughter and analysis. Streptomycin, in combination with penicillin, is the most commonly used antibiotic for the treatment of mastitis. It is also used for the treatment of bacterial honeybee diseases, such as European foulbrood, and to control fire blight, a devastating bacterial disease affecting fruit trees during blossom. Public health concerns persist, however, based on unlicensed use of the drugs and on lack of compliance with the withdrawal periods. Residues of antibiotics present a potential hazard to the consumer in terms of toxicity, allergic reaction and the development of bacterial resistance. Regulatory authorities have established permitted residue limits for streptomycin and dihydrostreptomycin in milk, porcine kidney and muscle. None have been set for honey. The permitted levels are based on extensive toxicological, pharmacological and microbiological data from clinical trials and specify a concentration of a drug residue that is considered safe (200 mg l–1 in milk, 1 mg kg
1 in kidney and 0.5 mg kg
1 in muscle). Biacore Q and Qflex Kit Streptomycin may be used to determine the concentration of residues of streptomycin and dihydrostreptomycin in a wide range of foodstuffs including whole milk, honey, porcine kidney and muscle. Chloramphenicol is a broad-spectrum antibiotic with excellent pharmacokinetic properties. In humans, however, its use is often associated with the adverse development of aplastic anemia, a rare but serious blood disorder. For this reason, it is been banned from use in food-producing animals, including honeybees, in the EU, USA, Canada and many other countries. Chloramphenicol, however, is still widely available in developing countries and is in common use in animal production. Affected products include poultry, shrimps and honey. As a prohibited substance, zero tolerance applies. Methods of detection, therefore, require very low detection limits and with the introduction of increased testing, high sample throughput is also an important factor. Conventional methods of detection include MBA, immunoassays, chromatography and mass spectrometry. Some of these methods lack the necessary sensitivity whereas others require long sample preparation. A provisional limit of detection of 0.3 ppb has been proposed as a level that any applied method should be able to achieve, but many food companies have introduced their own limits, forcing the development of ultra-sensitive assays able to detect o0.1 ppb. Qflex Kit Chloramphenicol can be used to screen for residues of chloramphenicol and chloramphenicol glucuronide in poultry muscle, honey, whole milk [3] and shellfish. b-Agonists are used in human and veterinary medicine as bronchodilators and also as agents to induce relaxation of the uterus. At high doses

SPR-based Biosensors in Food Analysis

343

(approximately 10 times the therapeutic dose), they are effective growth promoters in farm animals, increasing protein deposition and decreasing the fat mass. Although the economic benefits of this practice can be substantial, toxic residues of the drugs can remain in the meat. Several cases of acute food poisoning from the ingestion of contaminated meat containing the b-agonist clenbuterol have been reported. Consequently, b-agonists are banned from use in livestock production in many countries. Biacore Q and Qflex Kit b-Agonists can be used to screen for a wide range of b-agonist residues in liver samples from cattle, sheep and pigs. The assay has the advantage of using minimal organic solvent. In the EU, although not in the USA, all b-agonists, including ractopamine, are prohibited from use in livestock production. For monitoring purposes, however, the US Food and Drug Administration (FDA) has established specific tolerance levels for residues of ractopamine hydrochloride in edible swine tissues of 0.05 ppm in muscle and 0.15 ppm in liver. In the EU, there is zero tolerance to the presence of ractopamine in food of animal origin. Immunoassays have been developed as screening tests for ractopamine. HPLC/mass spectrometry is used for confirmatory analysis, analyses which require significant sample preparation. In comparison, Qflex Kit Ractopamine exploits the specificity of a high-affinity ractopamine-binding protein, providing a rapid screening method where results are obtained in hours rather than days. Some countries permit the use of certain antibiotics to treat diseased bee colonies, raising the possibility that antibiotic residues will contaminate the honey. Tylosin is a macrolide antibiotic that inhibits peptide growth in susceptible microorganisms. Its use in apiculture is permitted in the USA, but banned in the EU. Tylosin A is the predominant form but, at the low pH associated with honey, tylosin A is converted to tylosin B (desmycosin), which is also biologically active. The detection of both compounds in honey is essential to avoid the risk of underestimation. Traditional methods of analyzing tylosin include screening methods such as ELISA and confirmatory methods such as LC/MS/MS.

11.2.2

Qflex Kits for Quantifying Vitamin Content

Fortified foods with vitamins are subject to regulations from government bodies and consumer organizations to justify the labeling claims. Quality control procedures for the confirmation of vitamin levels are often rate limiting in the throughput of an entire processing facility. Assays that combine rapid analysis time with accurate results are therefore in demand. To reduce the overall assay time and increase throughput, in addition to simplifying assay preparation and reducing errors inherent in manual processes, Qflex Kits containing pre-immobilized sensor surfaces are available for quantifying the vitamin content of biotin, folic acid, vitamin B2, vitamin B12 and pantothenic acid. Many of these kits are certified as Performance Tested Methods by the AOAC. Consistency between the quality-controlled chip surfaces ensures that data can be reliably compared, even in multi-center studies.

344

Chapter 11

Biotin plays an important role in the metabolism of both humans and animals. As such, it is commonly added as a supplement to various health and nutritional foodstuffs such as cereals. Traditional methods of analyzing biotin include MBA, which require lengthy sample preparation and can take up to 2 days to obtain results. In SPR assays, food samples are prepared using simple extraction procedures, minimizing the pretreatment time. This indirect SPR inhibition assay exploits the high affinity of a biotin antibody to measure the biotin content in samples. Independent AOAC International (Association of Analytical Communities)-approved studies have demonstrated that analysis with Biacore Q and Qflex Kit Biotin correlates well with the ratified MBAs. Collaborative studies performed on a broad range of sample matrices have shown that the Biacore method also correlates well with other standard MBAs across a range of concentration values [4]. Although MBA has long been the method of choice for the analysis of folic acid, a widespread food supplement and an especially important dietary requirement during pregnancy, it is now possible to quantify folic acid content in hours rather than days, using SPR. Samples are simply mixed with a fixed amount of folic acid antibodies and injected into the system. Vitamin B2 (riboflavin) is routinely added to foods as a nutrient during processing and as a colorant in some cases. Traditional methods for the quantification of vitamin B2 are MBA, HPLC and fluorimetry. Qflex Kit Vitamin B2 PI assay (supplied with pre-immobilized chips) is designed as an indirect inhibition assay. Independent studies at the Nestle´ Research Centre, Lausanne, Switzerland, have shown an excellent correlation between Qflex Kit Vitamin B2 PI and a validated HPLC method across a range of matrices, detecting both fortified and non-fortified levels [5]. Vitamin B12 is routinely added as a supplement to a variety of processed foods such as infant formulas, breakfast cereals and nutritional products. Reliable quantification can be complicated by the low concentrations (nanogram quantities) normally present in typical samples and the presence of interfering compounds in sample matrices. Traditional methods of analyzing vitamin B12 include MBA, which results in a typical sample turnover of 2–3 days. Pantothenic acid (vitamin B5) has multiple functions and is essential for normal growth and development in humans. It is routinely added to many types of fortified foods, including dairy and non-dairy infant formulas, breakfast cereals and pet food. Any analytical method must therefore be sufficiently robust to perform well in a wide diversity of matrices. The most widely used method to determine pantothenic acid levels is an MBA employing Lactobacillus plantarum. Qflex Kit Pantothenic Acid is an inhibition assay designed for use with Biacore Q.

11.2.3

AOAC Certification of Qflex Kits

AOAC International is the foremost independent body providing internationally recognized standards of testing, with over 120 years of experience in method

SPR-based Biosensors in Food Analysis

345

testing and validation and the expertise of more than 2700 members worldwide. Certification of an analysis requires documented evidence of the quality of several assay parameters and provides assurance that an independent third party has tested the assay and found that the product fulfills all performance claims. Qflex Kit Folic Acid, for example, received AOAC approval after rigorous performance trials carried out in independent laboratories [4]. A summary of the procedures leading to certification and the results is presented below. In the certification procedures, various assay parameters were determined: specificity, accuracy, repeatability, reproducibility, precision, recovery, limit of detection (LOD) and limit of quantification (LOQ). To ascertain the specificity of the assay, cross-reactivity of the monoclonal antibody to folic acid provided in the kits was tested with a target analyte, 5-methyltetrahydrofolic acid (5MTHF) and a related vitamer, tetrahydrofolic acid (THF). The crossreactivities, calculated from inhibition curves, were 100% for 5MTHF and 4% for THF. To determine the accuracy of the assay, four accredited laboratories measured folic acid concentration in five types of sample (milk powder, milk-based infant formula, soya-based infant formula, cereal and premix) using a microbiological technique and Qflex Kit Folic Acid. The results from the two assays did not differ significantly. Repeatability and reproducibility were tested by 10 operators in four laboratories in a total of 22 samples of cereals, milk powder, premix, milk-based infant formula and soya-based infant formula. Repeatability (the variation in the results within a laboratory, obtained by the same operator) was between 0.6 and 3.6% and reproducibility (the variation in results from different laboratories and operators) was between 5.0 and 9.9%. The intra- and inter-kit precisions of folic acid concentrations were measured in soya-based infant formula. The results showed not only that variation between kit batches was insignificant but also that variation was low even when calibration solutions were prepared on different occasions. Three operators analyzed milk powder, milk-based infant formula, soya-based infant formula and cereal samples on different occasions and recorded recovery efficiencies of between 87 and 95%. The LOD for folic acid was below 1 ng ml–1 and the LOQ was between 2.0 and 70 ng ml
1. Analysis of a sample spiked with folic acid at 2.0 ng ml
1 gave a concentration within 4% of the expected concentration (CV 2.7%, n ¼ 4) and analysis of a 70 ng ml
1 sample gave a concentration within 4.5% of the expected concentration (CV 3.3%, n ¼ 4).

11.3 Examples of Applications for SPR-based Biosensors in Food Analysis 11.3.1

Quantifying Antibiotics in Honey

Although antibiotics or their residues are generally not permitted in honey, some may be used in the control of bee diseases, provided that they are employed at the right time, using the right method and correct dosage.

346

Chapter 11

Early in 2002, all honey imports from China to the EU were temporarily stopped after the discovery in some samples of unacceptably high levels of the antibiotic chloramphenicol. Chloramphenicol has been banned from use in apiculture in the EU, North America and many other countries. A Minimum Required Performance Limit of 0.3 mg kg–1 has been established for chloramphenicol assays but many food companies have introduced their own limits, forcing the development of ultrasensitive assays able to detect levels below 0.1 mg kg–1. Conventional methods of chloramphenicol detection include MBA, immunoassays, chromatography and mass spectrometry. Sensitivity is not the only important factor in determining the suitability of a method; in addition, sample preparation must be quick and easy with a rapid sample turnover to cope with the ever-pressing testing schedule. The Qflex Kit portfolio (Table 11.1) includes tests for the presence of several antibiotics used in the honey industry. Although optimized for honey, the assays may also be applicable to other bee products such as royal jelly in addition to other food matrices such as muscle and kidney. Streptomycin residues are not permitted in European apiculture, but the antibiotic is permitted in the USA for the treatment of American foulbrood, a serious bacterial disease of honeybees. Withdrawal of its use must take place at least 4 weeks before the main honey flow. HPLC with post-column derivatization and fluorescence detection is one possible analytical method for screening but requires elaborate sample preparation. Enzyme immunoassays are also available although these tests have a tendency to generate large amounts of false-positive results due to cross-reactivity. LC/MS/MS with a sample preparation that involves an SPE step and concentration before being applied to LC is used for confirmation. Qflex Kit Streptomycin is a rapid and simple screening alternative for streptomycin, requiring little more than pH adjustment of the sample. Sulfathiazole is the only sulfonamide permitted for the treatment of European foulbrood in the USA, but no sulfonamides are permitted in the EU. Again, LC with different detection modes may be used for confirmation. Sample preparation for LC/MS/MS and LC/fluorescence-based detection can be tedious, involving hydrolysis (to counter the effects of the formation of glucose adducts, giving erroneously low results), pH adjustment, liquid–liquid extraction, evaporation, SPE and concentration before injection. Further, the majority of biological assays for sulfonamides in honey tend to have very limited cross-reactivity within the sulfonamide family, do not reach the required sensitivity for honey analysis or tend to cross-react with N-acetyl metabolites. Qflex Kit Sulfonamides is a rapid method for the detection of more than 20 sulfonamides in honey. There is good sample recovery (approximately 95%) from both real and spiked samples and analysis is rapid, with a sample injection time of 90 s.

11.3.2

Screening for Veterinary Drug Residues

In animal production, the outbreak of disease or sub-optimal growth of livestock can lead to considerable economic losses. Veterinary drugs, therefore,

SPR-based Biosensors in Food Analysis

347

are often administered under prescription to food-producing animals for the prevention, control and treatment of disease. Product licenses for all veterinary drugs state a withdrawal time that must be observed by producers before treated animals can be sent for slaughter. In recent years, a number of highly publicized instances drew attention to violations of the stated withdrawal periods, unlicensed use or deliberate abuse of the drugs, all leading to a fall in consumer confidence. Although the use of hormone preparations such as the b-agonist clenbuterol to enhance animal growth is forbidden in the EU, isolated cases of malpractice have come to the attention of regulatory authorities. Johansson and Hellena¨s together with the Swedish National Food Administration (NFA) have developed rapid SPR-based assays for detecting clenbuterol in urine and hair samples [6,7]. As a complement to controls in slaughterhouses, controls are also made during animal husbandry so that violations can be detected as early as possible. Although convenient to sample, trace quantities of clenbuterol in urine can only be detected up to 1 week after exposure to the drug. On the other hand, residues can be detected in hair samples several months after exposure. The test for clenbuterol and related compounds is based on an inhibition assay in which clenbuterol is first immobilized on a sensor surface over which specific antibodies are injected after incubation with test material. In order to cope with the demand for high throughput testing, industry is increasingly calling for tests that can screen for families of veterinary drugs rather than testing for each individual member of these groups. Qflex Kits for screening families of veterinary drug residues in foodstuffs have been developed to address this demand for generic testing. Qflex Kit Sulfonamides, for example, contains the reagents necessary to detect at least 19 members of the sulfonamide family of drugs in porcine muscle, including sulfamethazine, sulfadiazine, sulfathiazole and sulfaquinoxaline, all in one analysis cycle. A major advantage of this assay is that there is no cross-reactivity with inactive N-acetyl metabolites, thus reducing the probability of false-positive results. Samples giving values above the maximum residue level (MRL) can then be further analyzed by confirmatory methods where the specific sulfonamide present can be identified and quantified.

11.3.3

Milk Testing

Penicillins and cephalosporins, members of the b-lactam family of antibiotics, are the most frequently used group of antibiotics for the treatment of bacterial infections, e.g. mastitis (inflammation of the udder) in dairy cows. Consequently, they are also the most frequently occurring type of drug residues in milk. The most commonly used methods for the detection of b-lactams in milk are based on inhibition of microbial growth. These tests are inexpensive, but they are very time consuming, requiring days to complete an analysis. In recent decades there has been an increase in the number of rapid receptor- and antibody-based tests, for example, the enzyme-based Penzym test from UCB

348

Chapter 11

Bioproducts and the Charm II test from Charm Sciences, based on the use of whole bacterial cells in combination with labeled tracer. Simpler receptorbinding assays have also been developed, such as the SNAP test from IDEXX Laboratories, the b-STAR test from UCB Bioproducts and the Charm Safe Level test from Charm Sciences. Rapid immunoassays are also available, such as the Parallux test from IDEXX Laboratories. These tests offer high sensitivity and specificity, but they are difficult to automate and are therefore limited in their capacity for increased sample throughput. An assay for the detection of b-lactam residues in milk has been developed in Biacore Q. Most Biacore assays for the detection of drug residues in various foods use antibodies as the detection molecule but in work by Gustavsson et al., a broad spectrum b-lactam receptor protein was used instead [8]. The advantage of using such a receptor protein instead of antibodies is that a generic assay can be designed, i.e. the whole group of b-lactams can be detected. This type of assay makes it possible to detect the active form of the b-lactam ring structure, an important consideration, as the legislated residue limits only cover the active form. Three assays were developed, all based on the same receptor protein, a carboxypeptidase. The b-lactam ring structure is easily hydrolyzed, resulting in inactivation of the substance. This, together with the very strong interaction between b-lactams and the receptor protein, made it necessary to use an assay different in design from most previous assays in which antibodies are mixed with the sample and injected over the sensor surface containing immobilized antibiotics. The formats of the b-lactam assays based on the enzymatic activity were similar, but differed in some important aspects. The b-lactam receptor protein carboxypeptidase hydrolyzes a tripeptide into a dipeptide, but this reaction is inhibited in the presence of b-lactams. Milk sample is mixed with the receptor protein and the tripeptide and the enzymatic reaction is allowed to proceed. Antibodies directed against the dipeptide (two-peptide assay) or tripeptide (three-peptide assay) are then added and the sample is injected over the sensor surface to which the respective di- or tripeptide is immobilized. The excess unbound antibody is free to bind to the surface. SPR results of milk samples from different producers were in good agreement with those from a number of commercially available screening tests commonly used for milk analysis and also with an HPLC method. The SPR assays based on the enzymatic activity of the b-lactam receptor are highly precise and are sufficiently sensitive to detect several b-lactams at or near their respective MRLs set in European legislation. These assays are thus viable alternatives for automated screening of b-lactam antibiotics in milk. The compositional standards for most milk products require that they contain no other proteins than milk proteins, unless declared. The low price of some non-milk proteins makes them attractive as potential adulterants in dairy products. Soya protein is probably the most commonly used non-milk protein in milk substitutes such as simulated yogurts, coffee whiteners and frozen desserts and it is likely to be a potential adulterant. A multiplex biosensor immunoassay was developed by Haasnoot et al. [9] for the simultaneous detection of soya, pea and soluble wheat proteins (SWP) in

SPR-based Biosensors in Food Analysis

349

milk powder by using polyclonal antibodies raised against these plant proteins as the immobilized binding partner [9]. The specificity of the assay was confirmed by analyzing proteins extracted from other food products. Both anti-soya and anti-pea antibodies partially cross-reacted with extracts from sources other than those to which the polyclonal antibodies were originally raised, showing that the assay could be used for tracing a broad range of non-milk proteins. Supplementing bovine milk with cheaper milk from other species is financially attractive. Within the EU, however, it is mandatory for producers to state the type of milk used for manufacturing dairy products as this may be important information for allergic individuals. Methods used to detect adulteration must be rapid and have an LOD of o1%. In an assay developed for bovine milk proteins, two monoclonal antibodies (Mabs) to bovine b-casein were used as immobilized binding partners [10]. After confirming the specificity of the assay by testing bovine milk protein solutions of k-casein, b-casein, g-casein, a-casein and whey proteins such as a-lactalbumin and b-lactoglobulin, high responses were obtained for bovine milk whereas responses for ewes’ and goats’ milk were much lower. From the milk species tested, therefore, the two Mabs were sufficiently specific for cows’ milk to enable adulterated milk to be detected. An inhibition assay was also developed in which k-casein is immobilized on the sensor surface and the binding response is measured after incubation of milk samples with Mabs to bovine k-casein. Different combinations were tested, aiming for a fast assay with 50% inhibition at around 0.5% cows’ milk in the milk of ewes and goats. Both assay formats performed well, giving o0.1% LODs and run times of around 5 min. Although the direct assay is characterized by its simplicity (single reagent format), the use of small amounts of antibodies and a broad calibration curve (0.1–10% cows’ milk), the inhibition assay has other advantages due to the possible application of non-purified Mabs, the higher responses, the higher sensitivity at relevant low percentages of cows’ milk and superior robustness (4800 cycles per chip). ‘‘Bioactive minor proteins’’ such as IgG, folic acid binding protein and lactoferrin are becoming valued for their beneficial effects as additives to foods such as infant formula. Direct biosensor-based immunoassays have been developed by Indyk et al., in which approximately 60 samples can be analyzed within 12–18 h, with each cycle completed in 8–15 min [11]. Briefly, sensor surfaces are prepared with anti-bovine IgG, folic acid derivative or anti-bovine lactoferrin and analytes are prepared for analysis by simple dilution. The specificity of immobilized ligands is evaluated by measuring cross-reactivity against individual casein and whey proteins at levels expected in milk. In comparison with alternative techniques (LC, ELISA, RID) for the analysis of milk products, the mean determined protein levels are consistent with ranges reported in the literature from assays employing alternative techniques. The minimal requirement for sample manipulation reduces the risk of recovery losses commonly associated with procedures such as filtration and centrifugation.

350

11.3.4

Chapter 11

Detecting Antibodies to Salmonella in Meat

Salmonella is a major cause of food-borne bacterial infections with a significant proportion of these cases associated with the consumption of pork. There are several stages during which the meat can become contaminated, e.g. infections may arise at the farm, during transport and in the slaughterhouse. It is important, therefore, to be able to monitor the whole production process. Serological assays are currently the most commonly used methods and, although time consuming, they provide useful information indicating the prevalence of Salmonella on individual farms. Evidence indicates that these ubiquitous monitoring steps can help reduce the frequency of human Salmonella infections. A more rapid SPR-based assay has been developed to detect antibodies against Salmonella in pig sera in a busy routine setting [12]. The assay is based on the immobilization of antigenic bacteria-derived lipopolysaccharides (LPS) on a sensor surface (the Salmonella serotypes are defined according to the sugars present on the LPS). The assay was compared with a commercially available ELISA-based Salmonella assay. The results of the assays were very similar. Large differences in the prevalence of Salmonella antibodies at different farms were noted, with some farms having positive sera in almost all deliveries while others had only negative sera in all deliveries. These results indicate that provided enough sera are taken, it is feasible to screen pig herds for Salmonella antibodies using an SPR-based biosensor assay. Where a quick result is needed, for instance when a pig herd is suspected of infection with a notifiable disease such as classical swine fever or foot and mouth disease, SPR-based testing can provide an answer within a few minutes and control measures can be implemented promptly.

11.3.5

Genetically Modified Organisms

In the wake of an investigation in 2003 into the content of food products, the Swedish National Food Administration (NFA) found gene-modified organisms (GMOs) in several foods that were not declared as such. GMOs were found in 14% of food products containing soya or corn. Although labeling demands are generally being observed, trace amounts of GMOs are finding their way into final products. In addition, both the source and means of introduction of GMOs into the manufacturing process remain to be conclusively identified. Gene-modified crops are known to be able to spread in nature and, in addition, raw ingredients must pass through many environments from cultivation to the outlets, including growers, mills, lorries, silos and deliverers, all of which present a potential point of entry for GMOs into the manufacturing process of an ostensibly GMO-free food product. Sensitive assays have been developed by Feriotto and colleagues to address this issue [13,14] based on DNA hybridization. In a protocol described by Feriotto et al., hybridization was monitored by immobilizing PCR products generated using DNA isolated from normal or

SPR-based Biosensors in Food Analysis

351

transgenic soyabeans on a sensor surface. Oligonucleotides or PCR-generated probes from suspected sources were then injected over the prepared surface. PCR-generated probes are far more efficient than oligonucleotides in detecting GMOs, allowing the detection of trace quantities. Their results indicate that the need for automated detection systems for GMO screening in food makes SPR-based biosensors strong candidates for routine procedures.

11.4 Conclusions Although the screening and quantification assays using Biacore Q and Qflex Kits or custom-built assays in this chapter cover a wide range of additives to and contaminants of food products, they are all characterized by a set of common advantages. SPR assays are highly automated, they are easily controlled and evaluated via software, no fluorescent or radioactive labels are required and the samples seldom require clean-up or solvent extraction. In addition, the availability of kits providing pre-immobilized sensor surfaces further reduces the manual input required by the user and provides a standardized sensor surface. Lastly, the possibility to easily regenerate and reuse sensor surfaces with no significant loss in assay performance is a major attraction in the quest for an assay that delivers consistent data over several hundred runs, permitting valid comparative studies between geographically disparate locations or at different times. Together with the inherent robustness of the technology and the proven resistance of Biacore Q to stressful industrial environments, these features add to the efficiency and productivity of screening and quantification. For example, the ability to identify the presence of veterinary drug residues in animal carcasses on the slaughterhouse floor means that steps can be taken to address the problem immediately, before the meat has left the factory. Import controls are a further area in which a short screen-to-result time can allow regulators to act on imported food containing banned substances before it has become widely distributed throughout the recipient country. There remains scope for the development and application of an instrument that can handle more samples simultaneously in order to control food safety hazards within the production chain. To this end, a multi-channel, high-throughput biosensor prototype instrument was recently developed at Biacore [15] and is designed to be used in conjunction with a commercially available automatic sample pipetting station in both laboratory and abattoir environments. A number of differences exist between this instrument and its predecessors. The most important of these changes is the prototype autosampler design which allows eight samples to be collected and analyzed simultaneously. The application wizard permits automatic result evaluation. The analysis time for one full 96-well microtitre plate for the sulfonamide assay was approximately 50 min as opposed to 9 h on a conventional instrument.

352

Chapter 11

Although SPR-based biosensors remain to make significant inroads into many areas of food safety testing and analysis, all comparative tests to date demonstrate that in addition to delivering data of a quality at least as high as its main competitors, it is safe and easy to use, with clear advantages of speed and throughput.

11.5 Questions 1. In SPR instruments the ligand is always coupled to the sensor surface. Explain why the terminology ligand and analyte in inhibition tests is sometimes confusing. 2. Why should we apply other assay formats for low molecular weight protein detection? 3. Explain the four main SPR assays and give reasons for choosing these assays in typical applications. 4. Try to address the complexity of the determination of kinetic rate constants in sandwich assays. 5. Why are SPR assays as shown in the Biacore kits attractive for the analysis of compounds in the food industry?

References 1. G.A. Baxter, M.C. OConnor, S.A. Haughey, S.R.H. Crooks and C.T. Elliott, The Analyst, 1999, 124, 1315–1318. 2. P. Bjurling, M. Caselunghe, C. Jonson, B. Persson, G.A. Baxter, M. OConnor and C.T. Elliott, The Analyst, 2000, 125, 1771–1774. 3. V. Gaudin and P. Maris, Food Agric. Immunol., 2001, 13, 77–86. 4. L. Wahlstro¨m and G.A. Baxter, Biacore J., 2005, 5, 8–11. 5. Biacore Application Note 51, Product No. BR-9003-55 (available from www.biacore.com). 6. M.A. Johansson and K.-E. Hellena¨s, Food Agric. Immunol., 2003, 15, 197–205. 7. M.A. Johansson and K.-E. Hellena¨s, Int. J. Food Sci. Technol., 2004, 39, 891–898. 8. E. Gustavsson, P. Bjurling, J. Degelaen and A. Sternesjoe, Food Agric. Immunol., 2002, 14, 121–131. 9. W. Haasnoot, K. Olieman, G. Cazemier and R. Verheijen, J. Agric. Food Chem., 2001, 49, 5201–5206. 10. W. Haasnoot, N.G.E. Smits, A.E.M. Kemmers-Voncken and M.G.E.G. Bremer, J. Dairy Res., 2004, 71, 322–329. 11. H.E. Indyk, E.L. Filonzi and L.W. Gapper, J. AOAC Int., 2006, 89, 898–902. 12. R. Achterberg, J. Maneschijn-Bonsing, R. Bloemraad, M. Swanenburg and K. Maassen, Biacore J., 2005, 5, 16–18.

SPR-based Biosensors in Food Analysis

353

13. G. Feriotto, M. Borgatti, C. Mischiati, N. Bianchi and R. Gambari, J. Agric. Food Chem., 2002, 50, 955–962. 14. G. Feriotto, S. Gardenghi, N. Bianchi and R. Gambari, J. Agric. Food Chem., 2003, 51, 4640–4646. 15. C. Situ, S.R.H. Crooks, G.A. Baxter, J. Ferguson and C.T. Elliott, Anal. Chim. Acta., 2002, 473, 143–149.

CHAPTER 12

Future Trends in SPR Technology RICHARD B.M. SCHASFOORTa AND PETER SCHUCKb a

Biochip Group, MESA+ Institute for Nanotechnology, Biomedical Technology Institute (BMTI), Faculty of Science and Engineering, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands; b Protein Biophysics Resource, DBEPS, National Institute of Biomedical Imaging and Bioengineering, National Institutes of Health, Bethesda, MD 20892, USA

12.1 Introduction We anticipate that surface plasmon resonance (SPR) technology will remain the gold standard in direct biomolecular interaction sensing during the next decade. Although in the past only one company (Biacore AB) mainly dominated the market (B90%) of high-quality SPR systems including optics, liquid handling and sensor chips, new players and new trends can be identified. As highlighted in Chapter 3, with 25 SPR-related companies the market is now more open than ever before and competition between companies takes place on several aspects of the SPR system. The customers for instruments will profit further from this competition, offering more flexibility, innovation and costeffectiveness. The field of clinical analysis including proteomics in all its facets is expected to undergo a revolutionary change with the introduction of new multianalyte diagnostic SPR systems. Mainly the combination of SPR imaging and dedicated microfluidic ‘‘lab-on-a-chip’’ may drive the technology to another level of commercialization where point of care (POC) devices for specific applications may be the ultimate objective. For applications in the field of kinetic and thermodynamic characterization of molecular binding parameters, higher computational power readily available with desktop PCs brings more sophisticated approaches within reach. In this Chapter, anticipated trends regarding the SPR systems are highlighted. Necessarily, this may sometimes have a speculative character and we 354

Future Trends in SPR Technology

355

recognize that, unavoidably, such an attempt may be regarded as biased to some extent by the interests and unique perspectives of the authors. However, we hope that this Chapter will, at the minimum, provide a description of some exciting prospects for the coming decade in the main areas of SPR development. It is structured along the three essential units that make up SPR instruments: (1) the detection instrumentation, (2) the fluidics and (3) sensor surfaces/chips. These units are complementary to each other and all critical for the quality and performance of the total SPR system and they are complemented by the data analysis. In the following, we will discuss our perception of the trends in each of these fields.

12.2 Trends in SPR Instrumentation Although the improvements and functionality of the Biacore line of instruments during the almost two decades of SPR technology are impressive, new trends in segmented parts in the market are appearing for more instruments. The proteomics area demands screening of a large number of analytes in complex samples, exceeding by far the currently readily available number of 4–20 and multi-analyte parallel diagnostics of kinetic parameters are desired. The FLEXChip instrument of Biacore is the answer for screening many biomolecular interactions simultaneously. For the next 5 years it is predicted that hundreds to thousands of simultaneous biomolecular interactions will need to be measured for screening the quality of binders. SPR technology should follow the protein microarray technology and kinetic parameters should be determined in a highly parallel manner. Instruments with SPR imaging will meet these requirements. Another new development is the combination of SPR with a complementary technology, so-called hyphenation SPR. Some trends are given in this Chapter and we expect that the combination with mass spectrometry (SPR–MS) will increase in importance. The implementation of lab-on-a-chip technology with SPR is the second important trend for a huge number of new applications. Right from the beginning of the introduction of SPR technology, one goal was the use of SPR for point of care diagnostic devices, but several attempts were unsuccessful. Why is this development of point of care SPR devices so problematic? Is it the sensitivity which should compete with other immunochemical tests (e.g. dipsticks) using labels? SPR has a resolution in the order of 10
5 refractive index change. If new instruments are developed based on the plasmonic effect in nanoparticles as described in Chapter 2, a new generation of instruments with unmatched high sensitivity is foreseen. However, the limiting factor will not be the sensitivity of refractive index changes, but the ratio between specific and non-specific binding. This determines the limit of detection for concentration measurements and therefore label-free SPR cannot compete with the best labeling technology. In the latter, any non-specifically interacting unlabeled species will not contribute to the signal. The nanoparticle SPR approach is unequivocally highly relevant; however, it will take many years

356

Chapter 12

to implement other important aspects for the study of biomolecular interactions, such as fluidics, nanoparticle surface functionalization and the investigation of this technology for the determination of reliable kinetic parameters. The first Biacore instrument in 1990 contained a so-called optogel for use as the medium between the prism in the optical unit of the instrument and the sensor chip, which turned out to be a crucial factor in the early development of the instruments. The optogel, which is considered part of the success of Biacore, ensures optical contact, simplifying exchange of the sensor chip. Some commercial SPR players in the market still apply refractive index matching oil, although alternatives have been published. For example, Masadome et al. [1] developed a refractive index matching polymer film for a portable SPR system. It is expected that successful SPR instruments will have a solution for the optical matching problem and will no longer need RI matching oil. Several manufacturers of commercial SPR instruments already use alternative optical arrangements with (expensive) disposable prisms. An alternative optical arrangement using polymeric diffractive optical coupling elements (DOCEs) was published by Thirstrup et al. [2], which is an attractive solution to prevent the optical matching problem.

12.2.1

SPR Imaging

SPR imaging has the ability to obtain a microscopic view of the sensor surface and define certain regions of interest (ROIs) for measurement of many biomolecular interactions at the same time. Reference spots, positive and negative controls to determine the non-specific binding and/or cross-over interactions, as well as triplicates or higher replicates of identical interactions for checking the variability of the sensor surface are helpful to obtain reliable, accurate and valid data. Common mode effects caused by temperature changes, bulk refractive index shift or flow direction shifts can be compensated using these reference spots. However, as explained in Chapter 3, simple reflectivity measurements can only give qualitative and not quantitative data which are necessary for in-depth studies of the kinetics of the binding process on spotted microarrays with different ligands. For every spot the shift of the SPR dip should be followed to allow subtraction and referencing. The data should not include an instrumental artifact but should show the real kinetic data of the parallel interaction process. Nevertheless, for expert users it is still possible to obtain relevant data from reflectivity instruments in carefully conducted experiments. In Chapter 3, Section 3.4.6, the companies that have SPR imaging instruments (including FLEXChip) are briefly described.

12.2.2

Hyphenation SPR Technology

New developments are combinations of SPR with a complementary technology, also referred to as ‘‘hyphenation’’ SPR. Often, these technologies are offline and should be used serially, but both technologies share the SPR sensor

Future Trends in SPR Technology

357

chip, which can be installed in the SPR instrument as well as inspected with the other ‘‘hyphenation’’ technique. An exception to this strategy is the combination with SPR excited fluorescence, which is treated in-depth in Chapter 9.

12.2.2.1

SPR–MS

Krone et al. [3] first combined SPR and mass spectrometry (MS), which created a unique approach for protein investigations. This technique was adapted and modified in several laboratories [4–7] and some of the commercial instruments are now also equipped with basic analyte recovery capability for MS. The basic idea is to follow up the characterization of interactions between proteins and surface-immobilized ligands by SPR with the determination of the identity of the bound proteins or peptides using MS. This has applications in protein interaction discovery in proteomics, and also the characterization of protein modifications critical for the interaction. Considering the SPR sensor surface as a miniaturized chromatographic matrix, SPR–MS is reminiscent also of the traditional affinity chromatographic purification preceding MS in protein discovery, but obviously with added real-time quantification of the capturing and elution process, in addition to exhibiting different elution behavior due to the much smaller scale matrix. Several fundamentally different interface approaches have been developed by different groups, for example (1) direct use of the SPR chip surface for MS by mounting the preloaded SPR chip on a MALDI platform [3] or analyzing the surface with a SELDI protein chip reader [7]; (2) on-chip digestion followed by microfluidic elution, recovery in a reversed-phase microcapillary column and ESI-MS/MS sequencing of peptides from the digest [5]; and (3) analyte dissociation, microfluidic elution and collection followed by external digestion and MALDI-TOF sequencing [4,9]. Recent progress in SPR–MS includes improved methods and operations, increased limits of detection, multi-protein analysis and protein-complex delineation. With the subsequent design of SPR protein arrays, SPR–MS enters into the field of high-throughput protein interaction discovery and miniaturized diagnostics. Below we briefly highlight an example of SPR detection of an enrichment and elution process of specifically bound analyte as an application of SPR–MS by Visser et al. (Heck’s group) at the University of Utrecht (The Netherlands) [8]. In this work, an SPR-based chemical proteomics approach in combination with nano-liquid chromatography/electrospray tandem mass spectrometry (nanoLC/ESI-MS/MS) was used to obtain both semi-quantitative and qualitative data on enriched proteins obtained from SPR sensor surfaces. The aim of the research was the characterization and identification of autoimmune antibody–antigen complexes present in RA patient sera. The autoimmune antibodies can be enriched using a citrullinated peptide (i.e. a specific peptide where the arginine is replaced by citrulline) as a ligand which is immobilized on a geltype G-COOH sp IBIS sensor chip. The work is connected to the research described in Chapter 7, which should be consulted for details and references.

358 COOH COOH COOH COOH COOH COOH COOH COOH

Chapter 12

1. EDC/NHS 2. Peptide: pos.(citr)

TB34 serum

3. EtAm Sequential elution: 1. Pept.neg. 2. Pept.pos. 3. Gly

1. Trypsin,O/N, RT 2. LC-ESI-MS/MS

Figure 12.1

Gly

Pept. pos

Pept.neg

Schematic set-up of the autoimmune antibody–protein complex detection and enrichment procedure in three steps from a specific RA serum (TB34) using a citrullinated peptide as the ligand for binding of specific autoimmune antibodies (Pept.pos ¼ a specific peptide, where the arginine is replaced by citrulline). The first elution was with negative peptides (Pept.neg ¼ where the arginine was not replaced by citrulline), followed by a second elution with the ligand in solution (Pept.pos) to replace the bound fraction with free competing ligands in solution and finally a full regeneration step (Gly).

The analysis cycle can be followed in real time as shown in Figure 7.10. A threestep sequential elution procedure (see Figure 12.1) was conducted to increase the selectivity and to permit discrimination between non-specific and specific binding. It is necessary to dissociate non-specifically and less-specifically bound material with low affinity from the sensor chip in the first elution step. Then selectively bound proteins should be competitively dissociated in a second elution step, which could in this case be achieved using a competing ligand peptide (Pept.pos) to that immobilized ligand. This differs from other approaches where reversible chemical denaturation is applied for the analyte release. This present approach works for analyte molecules where the capture is kinetically aided by rebinding arising from mass transport limitation. If the protein dissociates from the surface in the presence of the competitor, it will bind to free ligand in solution and rebinding to the surface will not occur. (Note that for SPR instruments the ligand is considered to be immobilized to the surface. The specific elution with ligand can be confused with an inhibition assay format as described in Figures 7.2 and 11.2.) In the third elution step, the

Future Trends in SPR Technology

359

sensor was regenerated with a low-pH glycine–HCl (gly) buffer, removing the remaining bound material from the sensor disc. Each of the fractions was collected and digested. The protein content was characterized using a nano-LC/ ESI-LTQ set-up. Preliminary results showed a three-fold increase in dissociation of the second elution with the competing ligand with respect to the first elution. The remaining bound material can be eluted completely in the regeneration step with the low-pH (gly) buffer. As with other methods for SPR–MS interfacing, to obtain a fraction containing sufficient protein which was specifically bound to the citrullinated peptide and which has been successfully eluted, is not easy. The eluted specific protein should be as clean as possible without interference from highly abundant non-specific proteins or non-isolated competing free ligands for successful detection with the LC/ESI-MS/MS. The off-line combination of SPR and nano-LC/ESI-MS/MS needs further to be optimized and may lead to enrichment and identification of autoimmune antibody–antigen complexes using immobilized target peptides combined with the sequential elution approach. New results will be published by Carol-Visser and colleagues soon. Some of these difficulties regarding high selective loading and recovery may be addressed with improved microfluidic liquid handling. As illustrated by Gilligan et al. [9], the manipulation of small distinct liquid volumes in the microfluidics with reversible and oscillatory flow patterns [10] can provide significantly increased amounts of recovered material and controlled washing conditions. In this approach, small liquid plugs containing the loading sample, washing buffers and elution solution are separately transferred to the sensor surface such that multiple cycles of loading, washing and recovery can take place, enriching the eluent concentration in the same plug of recovery buffer [9]. Further, by virtue of the oscillatory flow pattern once the specific liquid volumes are covering the flow cells, the contact times can be extended (while maintaining high mass transfer to the capturing molecules at the surface) until the slow binding kinetics arising from the low antigen concentration has achieved a plateau. The latter sample handling technique has been applied, for example, to minimize sample consumption and optimize detection efficiency in an SPR based assay for the detection of anti-idiotypic antibodies in patient sera [11]. Another effective approach for improved recovery from the SPR surface is the use of a larger surface [12].

12.2.2.2

Other Hyphenated SPR Techniques

A way to enhance sensitivity and to push the limit of detection (LOD) to lower surface coverages is the use of fluorescent chromophores covalently attached to the analyte molecules. In this approach of surface plasmon fluorescence spectroscopy (SPFS), the resonantly excited surface plasmon waves excite the fluorophores [13] and their emitted photons can be monitored by a simple detection unit attached to a conventional SPR set-up. In Chapter 9 the features and benefits of such an SPR excited fluorescence instrument are described. The

360

Chapter 12

preferred mode of operation is to monitor the fluorescence emitted directly from chromophores sufficiently separated from the substrate surface. The dye molecules are still within the substantially enhanced optical field of the surface plasmon mode, but they are not quenched. This combination of field enhancement and fluorescence detection forms the basis for the largely enhanced sensitivity applied for a wide range of bioaffinity studies (for selected examples, see Chapter 9). As shown in the previous section, SPR excitation in the SPFS instrument can be combined with instrumentation assembled at the liquid side. SPR optics in the Kretschmann configuration require only the assembly of optical components at the prism side and therefore approaching the instrument at the wet side with flow cells in microfluidic cartridges, cuvettes or even lab-on-a-chip devices is in principle feasible. The implementation of lab-on-a-chip devices is discussed in Section 12.3.3. The lateral resolution of SPR, which is equal to the propagation length of the plasmon wave (e.g. for a wavelength of 680 nm it is of the order of 10 mm), is sometimes unacceptably large for the imaging of small features, such as ligand clusters, aggregates or fibrils with submicrometer dimensions. If a higher lateral resolution is needed, then other combinations of instruments or hyphenation SPR technology are required. A combination of atomic force microscopy (AFM) and SPR [14] seems in principle a very interesting approach. Combining the dynamic topographical data from AFM with the kinetic data from SPR may be applied in many fields of materials research, particularly in the design of biomaterials, where dynamic surface changes, such as protein aggregate adsorption/desorption processes, play a role. Another approach is the exploitation of SPR as a complementary surface analytical tool to scanning probe microscopy (SPM). The potential for a combined SPM–SPR approach for the analysis of biomaterial surfaces was demonstrated in 1994 [15]. It is expected, although speculative, that commercial SPR imaging combinations with AFM, STM [16], SPM, etc., will enter the market in the next 5 years.

12.2.3

Nanoparticle SPR

SPR phenomena are not restricted to planar multilayers as discussed in this book: for metal particles, usually gold, with dimensions much smaller than the wavelength of the interacting light, surface plasmon effects can be much more prominent (see Chapter 2, Section 2.5.2). Nanostructured surfaces, such as nanoholes, can also be applied to exploit surface plasmon/plasmonic effects for sensing biomolecular interactions [17]. The use of metal nanoparticles as surface plasmon-assisted field amplifiers is described in Section 8.4. However, these particles can also be exploited as intrinsic refractive index sensors, analogous to the more familiar planar SPR experiments (for a review, see [18]). The physical basis of this application is the light extinction (absorption and scattering) which is heavily dependent on the nanoparticle’s dielectric constant, size and geometry and also on the dielectric constant of the

Future Trends in SPR Technology

361

surrounding medium. Van Duyne’s group [19] developed a silver nanoparticlebased LSPR nanosensor which yields ultrasensitive biodetection with extremely simple, small, light, robust and low-cost instrumentation. They used this LSPR spectroscopy to detect less than 1 pM up to micromolar concentrations of biological molecules. However, the silver nanoparticles applied are intrinsically less inert as sensing elements than gold nanoparticles. Gold nanoparticles of defined dimensions can be coated on a substrate in order to enable the easy exchange of liquids similar to flat SPR instruments. Hong and Kao [20] developed such a gold nanoparticle-coated film to achieve highly spatially resolved biosensing that is based on localized SPR. It was reported that unlike the planar gold film employed for conventional SPR sensing, the gold nanoparticle film relies exclusively on shifting the peak extinction wavelength for the detection of biomolecular interactions and that it does not depend critically on the angle of incidence. Magnetic particles, which can be recruited to the sensor surface by magnetic fields, are difficult to use in Kretschmann-operated SPR sensors, where the magnetic field would need to be raised from the prism using bulky coils. However, for the phenomena used in giant magnetic resistors (GMRs) [21], an electrical current is flowing to attract magnetized particles to a sensing metal line. A simple electrical configuration can be made using gold lines both as an electric current actuator to attract the particles and as an SPR sensing device. Essentially, it should be possible to apply SPR imaging in a hyphenation approach of magnetism and SPR. However, to our knowledge, such a combination has not yet been realized.

12.3 Trends in Fluidics For the binding to the SPR surface to have maximum sensitivity and the signal to reflect the intrinsic kinetic and thermodynamic properties of the molecules under study, microfluidics appears to be the most powerful approach. Attractive features are excellent baseline stability, low sample consumption and the potential for relatively high mass transfer when high flow rates and thin channels are used. As already described in Section 3.3.1 and Figure 3.9, flow channels are formed, in principle, by pressing a grooved surface against the sensor chip. Biacore introduced a microfluidic cartridge for sample delivery in 1990, with incorporated pneumatic values that allow closing of specific channels, thereby providing control over the flow paths. This allowed the sample or buffer liquid volumes to be directed to different surface spots. In this microfluidic cartridge, also integrated was an (HPLC-like) injection loop that allowed a fixed sample volume to be chased by running buffer and flowed over the surface. Although this technology has been shown to be fairly powerful and far superior with regard to stability in comparison with simple cuvette-based systems, some improvements were subsequently addressed in different designs. For example, the constraints of the injection loop with finite fixed sample

362

Chapter 12

volumes allowing only limited sample contact times and flow rates were overcome with the oscillatory flow technique [10]. This permits the use of sample plugs of smaller volume, yet allowing simultaneously very long contact times while maintaining high flow rates and mass transfer. This can be very important when working with limited sample volumes, for better characterizing the thermodynamic binding parameters by permitting the binding progress to reach the steady state [10] and be utilized to improve the sensitivity for analyte detection at low concentrations [11] and in interfacing SPR with MS as outlined above [9]. To improve the usage of each sensor flow channel, Biacore has more recently introduced the option of hydrodynamically addressing different spots within each channel, thus multiplexing the use of each sensor flow channel as shown in the Section 3.5, Figures 3.35 and 3.36. A different microfluidic design was implemented in the commercial Prote-On system by Bio-Rad, where a crisscross pattern of microfluidic channels allows one to measure the binding of a series of several samples to several different sensor spots in parallel – a design particularly suited to SPR imaging (see Figure 3.29). The latter multi-analyte detection permits experimental designs that do not require surface regeneration. The trend for multiple ligands to be attached to the sensor surface will continue in the future. The body of literature on lab-on-a-chip is expanding at a fast pace, but the field is still far from mature. The combination of lab-on-a-chip devices with SPR sensing is even in its infancy. Advanced techniques for microarray spotting are reviewed in more detail in Section 12.3.1, and prospects for point of care use of SPR devices are discussed in Section 12.3.2. Limitations inherent in laminar flow-based microfluidics arising from diffusion and viscosity can be overcome using different flow principles for sample delivery. This includes the use of the electroosmotic flow principle to pump liquids in microchannels, which will be treated in Section 12.3.3. A combination of flow with a separation system is the microfluidic free-flow electrophoresis device for proteomics, which is reviewed in Section 12.3.4. Another highly promising development is digital microfluidics using electrowetting principles, as reviewed in Section 12.3.5.

12.3.1

Microarray Spotting on Gold

With the introduction of protein microarrays in the late 1990s, there was tremendous excitement about the potential of protein arrays to improve further our understanding of protein expression, function and structure on a scale approaching the proteome. However, there was also from the beginning hesitancy by many scientists to adopt a technology that is often still perceived as unstable and irreproducible. Protein microarrays were developed largely by extending technologies used for gene chips [22]. Most protein arrays as currently developed rely on detection technologies that apply fluorescent tags. Definitely, fluorescence detection methods are successful for gene chips, but

Future Trends in SPR Technology

363

much less convenient with protein chips due to the heterogeneity of proteins, difficulties in the synthesis of conjugates and the potential for non-specific binding. Also, signal-producing reactions in solution catalyzed by commonly used enzyme-linked antibodies are difficult to implement in an immunoassay microarray format, where the product will diffuse away from the surface, diminishing the potential for the discrimination of spots. However, in Section 8.3 an enzymatic amplification method is described for SPR sensing, which applies a localized non-soluble precipitate, which is detected by SPR. Protein microarrays can be generally divided into two main categories: capture arrays and interaction arrays. Capture arrays have immobilized molecules such as antibodies or chemically treated surfaces, which bind generally with high affinity to a specific, known ligand. Interaction arrays have ligands which are used to identify functions or determine directly interactions with other, frequently unknown, analytes. The detection is generally accomplished with a non-interfering labeled conjugate in a two-step sandwich. Is SPR microarray imaging the answer for measuring biomolecular interactions in a reliable way without the need for labeling? In Chapter 3 many instruments including FLEXChip, Lumera, K-MAC, Bio-Rad and IBIS are described which can implement microarray technologies. A main challenge is how to spot the ligands to the SPR sensor surface. Initially, pioneers in the field used laboratory-made equipment as designed and published by Brown and co-workers (see http://cmgm.stanford.edu/pbrown). Recently, several commercial array systems have been developed for printing target DNA on microscope slides which can be adapted to create protein microarrays. In short, every arrayer consists of an X–Y–Z robot of which the printhead travels between a microtiter plate, containing the ligand and the sensor surface, which is a standard 3  1 inch glass microscope slide. Approximately 16 spots per mm2 can be printed when spots have a pitch width of 250 mm. The differences between the available systems are mainly based on the structure of the printhead: there are several different ways of picking up a small amount of ligand solution and printing small amounts of this solution in an orderly and systematic way on the sensor surface. Examples of arrayer methodologies include piezo technology (inkjet dispensers), quills (split-pen) and ‘‘pen and ring’’ systems (see the webpage cited above). As a consequence, the systems differ considerably in reliability, accuracy, capacity and the required (starting) volume of sample. One of the main difficulties that distinguish DNA and protein arraying is that DNA spots do withstand drying of the surface while many protein ligands are prone to various conformation changes and denaturation after evaporation of the liquid, which frequently affects biological binding properties or may eliminate binding altogether. In the next section we present briefly a strategy for producing high-density protein microarrays using a DNA coding technique in a confined microfluidics set-up including a self-assembly process of proteins onto individual addressable microstructures. Spots of 10 mm are feasible, which correspond to the propagation length of the plasmon wave, allowing the creation of a microarray with 2500 individually addressable protein spots per

364

Chapter 12 2

mm . This is the physical maximum for Kretschmann-configured SPR instruments.

12.3.1.1

DNA Coding Technology

The so-called DNA coding technology is proposed in Figure 12.2. First a substrate is coated with a homogeneous coating (e.g. streptavidin). To define the array, a PDMS device is fabricated with channels and reservoirs (see Figure 12.3). The reservoirs in the PDMS device are filled with multiple solutions of biotinylated single-stranded DNA, each having a specific sequence. When these solutions flow through the channels the single-stranded DNA is coupled to the streptavidinated substrate surface. After immobilization of the first lane, the streptavidin coating is inactivated with an excess of biotin. Next, a second, equal PDMS device is placed in a perpendicular position and the reservoirs are filled with multiple solutions of different mixtures of complementary DNA conjugates. Each of the DNA conjugates hybridizes in a self-assembly process to its specific lane. At each individual crossing of lanes the self-assembled

Figure 12.2

Schematic presentation of the construction of a microarray using DNA coding technology. First, lanes of biotin-labeled reverse oligonucleotides are immobilized on the streptavidin surface. The spot is defined by hybridizing a mixture of complementary DNA–protein conjugate in the perpendicular direction. During the self-assembly process the forward complementary oligonucleotides are immobilized and an ultra-highdensity protein microarray can be created. The DNA coding technology can be extended to fabricate, for example, a 50  50 microarray.

Future Trends in SPR Technology

Figure 12.3

365

Left: PDMS line-spotter (50 mm lane) on top of an iSPR sensor chip. Right: iSPR image of the six lines produced by the line spotter and the ROIs in red. The first DNA coded lanes can be created and the PDMS device can be removed. In principle, drying of the DNA-lanes is allowed to occur. A second similar device can be placed perpendicular to the first creating a 6  6 array of different protein ligands using the DNA coding technique.

molecules will define a different array square spot. For a high-density protein array, a mixture should be made of a DNA sequence coupled to a protein (protein–ssDNA or DNA conjugate). To address each spot, a mixture of different protein–DNA complementary codons should be prepared, which in principle can be produced by molecular synthesis and can be automated using liquid handler robots. The physical dimension of the array chip is defined by the channel width of the PDMS device and by the number of spots in the array. Although the lateral resolution of SPR imaging is limited by the propagation length of the surface plasmon wave (e.g. B10 mm) in the clean room of the MESA+ Institute, University of Twente, even 1 mm wide lines can be processed. The novel method allows the fabrication of a high-density multi-ligand array, without exposing the sample to air, thus avoiding denaturation of proteins. It is important that non-specific binding should be suppressed to a high degree to have optimal benefit from the self-assembly process. Kanda et al. [23] used PDMS microfluidic devices as a surface to pattern the gold and adsorb antigens from the sample – this device had the potential to produce arrays of 64 or more spots. In the biochip laboratories of the University of Twente, Beusink and colleagues developed PDMS devices for spotting ligands for immobilization in small, spatially separated lanes on a sensor chip. In Figure 12.3 a six line-spotter is shown for the immobilization of six ligands (unpublished results). A critical advantage of the proposed coding system is that it can create ultrahigh-density microarrays without using droplet-based spotting procedures. Line spotter microfluidic devices allow us to fabricate confined microarrays

366

Chapter 12

in a two-step process, as shown here, for an immunochemical biomolecular interaction. Although the technology is still in its infancy, applications are foreseen in the areas of disease monitoring, proteomics and genomics studies.

12.3.2

Prospects for SPR-based Point of Care Devices

Clinicians are beginning to use point of care (POC) testing of compact (in terms of size and weight) and flexible clinical chemistry testing devices suitable for use close to the patient. These analytical devices are designed to move diagnostic testing out of central laboratories into sites closer to the patient. In the future, households might also be equipped with small, user-friendly devices to monitor the daily health status based on measurements of small samples [24]. Miniaturization of devices will offer advantages when rapid and selective monitoring is required, e.g. of cardiac markers for diagnosing acute myocardial infarction [25] or whole blood chemistry relevant to intensive care medicine [26,27]. Why are POC SPR devices not yet in the market? We can identify the following five reasons: 1. SPR-based concentration measurement is never as sensitive as labeled techniques because of the intrinsic and inherent drawback of SPR: the detection of non-specific binding interferes with the specific binding signal intended to be measured. In contrast, an immunoassay that detects only the label will not be susceptible to signal interference from non-specific binding of unlabeled proteins. 2. For POC analysis there is still not yet the absolute necessity to detect kinetic rate–equilibrium constants of biomolecular interactions, which are the unique features of SPR. 3. Current SPR instruments are still bulky and expensive and are not in a state for production for high-volume markets. The SPREETA chip of Texas Instruments was the ultimate configuration for POC devices but this application has not developed as successfully as expected. 4. The advantages of direct detection and speed of SPR appeared to be not (yet) the crucial factor making it suitable for POC devices and giving it a decisive advantage, e.g. over dipstick tests, which seem to be fast enough for current applications. 5. The cost aspect of expensive labeling is not (yet) the remaining argument to replace labeled tests for SPR–POC devices. If POC tests can be designed where the kinetics of the biomolecular interaction determine the outcome of the test, then SPR–POC devices are attractive for the market and can be developed with a great intrinsic advantage. However, at the present time, in the absence of this type of POC test, it is still question able whether SPR can compete with current label technology in POC devices.

Future Trends in SPR Technology

12.3.3

367

Implementation of Lab-on-a-Chip Devices for SPR Systems

In 1990, Biacore introduced a fully automated pneumatic valve-operated microfluidic cartridge for biospecific interaction analysis, which was at that time the most high-tech and advanced fluidic system developed in a commercial instrument. However since the introduction in 1990 by Manz et al. [28] of miniaturized total analysis systems (m-TAS), an enormous research effort [29,30] has taken place in the area of miniaturized devices or labs-on-a-chip with thousands of papers in 2006. It may once have seemed an utopian dream to create highly parallel and automated microfabricated devices for SPR systems. However, considering the new lab-on-a-chip trends for SPR systems in this section, we hope to convince the reader that the rapidly unfolding reality of lab-on-a-chip technologies for new SPR instruments will pave the way to achieving highly parallel and automated microanalyses of biological processes. Table 12.1 contains a compilation of some potential building blocks for labon-a-chip devices that are useful in combination with SPR detection. In this chapter, only a few of these building blocks will be described in detail. The most important trends regarding the implementation of potential lab-on-a-chip approaches into SPR systems are discussed.

12.3.3.1

Pumping Liquids Using Electroosmotic Flow in Microfluidic Devices with Gold Layers

When an electric field is applied in the longitudinal direction of a charged surface (substrate) or capillary, the cations close to the wall move towards the cathode (Figure 12.4). Wrapped in the layer of cations, the bulk solution is transported in the direction of the cathode. The only plane of friction is between the stationary layer at the capillary wall and the layer of cations which is in motion. As illustrated in Figure 12.4, unlike in pressure-driven systems with parabolic flow profile (Poiseuille flow), the velocity of the bulk is constant resulting in a flat flow profile (plug flow). The stagnant diffusive layer

Table 12.1 Lab-on-a-chip building block Channel Pump Mixer Separator Collector Spotting Detector

Potential lab-on-a-chip building blocks for SPR devices. Description Formed by dry/wet etching, molding or soft lithography (e.g. PDMS) Enables fluid transport which can be driven electro- or hydrodynamically Splitting or coiling of a laminar flow Based on capillary or free flow electrophoresis (CE, FFE) Using hydrodynamic addressing or address flow Line, criss-cross or dead end spotting Functionalized gold surface; patterned gold sensor patch

368

Chapter 12

EOF

E

Figure 12.4

The electroosmotic flow profile is plug flow, while hydrodynamic pumping gives a laminar flow profile. The electric field (E) will drag cations in the double layer of the wall (B10 nm) to the cathode (negative electrode). If a conducting gold layer is deposited on the glass surface, the electrical field will be short-circuited and affected by the gold through reduction and oxidation processes at the gold surface. The gold layer can be considered as a bipolar electrode.

of e.g. 2 mm (see Chapter 6, Figure 6.5) is reduced to the thickness of the double layer (B10 nm), which is dependent on the ionic strength of the buffer solution. This may result in improved mass transport of the analyte to the immobilized ligand. The velocity of the electroosmotic flow (EOF) is given by the following equation [31]: vEOF ¼ ðe0 e=4pZÞ zE

ð12:1Þ

where e0 and e denote the dielectric constants of vacuum and the buffer, respectively, and z denotes the zeta potential, the potential at the first moving layer at the capillary wall. In order to modify the magnitude and direction of the EOF, either the lateral electric field E or the zeta potential should be modified. The principle of modifying or even reversing the EOF inside a capillary has been presented before, including the control of flows in integrated devices [32,33]. As the EOF is generated at the channel wall, the viscosity at the wall is one of the determining factors for the flow velocity. If a conducting gold layer is deposited on the glass surface, the electric field will be influenced and changes in the EOF arise. Moreover, this electric field may have an effect on the surface plasmons, which, in turn, may affect detection of biomolecular interactions using the SPR phenomenon. The gold patches for SPR sensing are not directly connected to a power supply but are floating. A lateral field is over the gold layer and the equipotential of the gold is considered to be the average between the potentials in the electric field in the liquid at both sides of the gold electrode. Under conditions where the voltage difference is low, o1 V, no reactions occur, no electric current will flow and the double layer of the metal/electrolyte interface

Future Trends in SPR Technology

369

behaves as an insulator. In contrast, when the voltage difference over the electrodes becomes too large, an electric current will flow through the metal and at both ends electrochemical processes or even electrolysis of the liquid will occur, resulting in bubble generation. If, on the other hand, the gold electrode is connected in an electronic circuit, then a changed potential of the electrode will modify the charge distribution near the metal surface and hence alter the ionic distribution in the double layer. In principle, these ionic changes will be measured with SPR, because changes in ionic concentration in the electrical double layer will alter the refractive index in the evanescent field. This so-called electrochemical SPR or E-SPR phenomenon depends on the ionic composition of the electrolyte and on the electron concentration in the metal. As described in the literature [34], to a first approximation, the refractive index of the double layer can be considered to vary with the change in the charge of the double layer. Lioubimov et al. [35] described a combination of oscillating electric potential and SPR measurement. Tadjeddine [36] discussed a multilayer model of the electrochemical interface in combination with SPR phenomena. An example of these effects can be seen in Figure 12.5: when an external voltage is applied to the channel, one side of the bipolar gold strip turns darker, while the other side turns lighter. There is a gradient of reflected light along the gold chip. When the voltage polarity is switched, the dark and light ends of the strip exchange accordingly. A delayed switch-on effect was also observed, in agreement with experiments observed by Lioubimov et al. [35]: the gradient in the reflected light did not appear instantaneously, indicating a true electrochemical process changing local refractive indices, as opposed to a physical SPR effect based on electron charge concentration differences in the gold, which may induce a change in free electron plasma oscillation.

12.3.4

Lab-on-a-Chip Implementation Using Free Flow Electrophoresis and SPR Imaging for Proteomics-on-a-Chip

A so-called proteomics-on-a-chip device should permit the separation, detection and identification of new biomarkers in a micro-fabricated device. A common problem with miniaturized separation systems based on, e.g., capillary electrophoresis is the low sample loading capacity. This problem, together with the often low concentration of relevant bioactive compounds, puts severe demands on the detection system. If a separated sample is flowing in peaks over the SPR sensor area, the contact time of the sample with the surface is limited, making it impossible to detect proteins of low abundance. Therefore, a peak flowing over the sensor area should be trapped (stopped) in order to allow diffusion of analyte to the sensor area. Lammertink et al. [37] described a peak recirculation approach to allow coupling of CE with SPR with higher contact times with the sensor area. However, the absolute number of lowabundant biomolecules in volumes of less than 1 nl will put a severe constraint

370

Figure 12.5

Chapter 12

Three screenshots (A, B, C) (unpublished results obtained by M. van der Ploeg) of an SPR imaging experiment in a microfluidic channel acquired at the University of Twente laboratories. The electric field drops from left to right. Two containers as shown in Figure 12.4 are outside this image and contain the Pt electrodes At A, no external voltage was applied, at B an external voltage of less than +2 V over the gold electrode was applied (note that the gray values at the two ends of the gold strip differ), at C an external voltage of o
2 V was applied. The reflectivities at the two ends of the gold strip exchanged as a result of the electric field switch. A 10 mM HEPES buffer was used in the channel.

on the SPR biosensing system. Therefore, we shifted to another combination of separation technology and SPR detection. Recently, we have developed a microfluidic free flow electrophoresis (FFE) device for coupling the device with SPR imaging as a separation and detection system for biomarker discovery. FFE is a continuous separation method, providing continuous bands and thus virtually unlimited amounts of separated components. In FFE an electric field is applied perpendicular to the flow direction and charged molecules are

371

Future Trends in SPR Technology

deflected from the carrier flow direction, in a way that is controlled by the electrophoretic mobility, the flow velocity and the electrical field strength. Figure 12.6 depicts the layout of the FFE section of the device [38]. Microfluidic FFE (m-FFE) was introduced by Raymond et al. [39]. Although the peak capacity in FFE is limited, a continuous supply of separated components is beneficial for the successful integration with affinity detection. In our group, we patented a new approach to an FFE device in combination with detection [40]. In this system, the separated proteins at the outlet of the FFE chip enter a socalled SPR affinity area where biomolecular interactions are studied at a microarray. The binding can be followed with imaging surface plasmon resonance detection (iSPR). Although the concepts shown here are still in their infancy, a first operational device with a separation and detection section is expected to be available soon. An improved free-flow isoelectric focusing chip has been fabricated by Kohlheyer et al. [41] and tested with a set of fluorescent isoelectric focusing markers. For the first time, high-resolution results could be obtained in such a microfluidic device. As illustrated in Figure 12.7, the chip contains five inlets. These inlets are used to infuse the separation chamber with different ampholyte solutions to build up a pH gradient inside the separation chamber, perpendicular to the flow. Figure 12.8 illustrates how the pH gradient is developing with the increasing residence time of the ampholytes and the sample in presence of the electrical field. The pH gradient starts with a step gradient and finally results in a linear pH gradient, with all components focused at their isoelectric points. Due to the use of a step pH gradient, the focusing times and, more importantly, Joule

flow direction total internal reflection at the glass/gold interface circular prism

SPR angle monochromatic and p-polarized light to CCD camera

Figure 12.6

The FFE–SPR combination.

372

Chapter 12

Figure 12.7

IEF Chip layout.

Figure 12.8

pH gradient development in a FFIEF device.

Future Trends in SPR Technology

Figure 12.9

373

Experimental results with fluorescent IEF markers.

heating are reduced. During experiments eight different isoelectric focusing markers were used to visualize the separation efficiency and the linearity of the pH gradient (see Figures 12.8 and 12.9). A new FFE chip has been fabricated, which contains a gold surface for SPR measurements (Figure 12.10). This chip was placed inside an IBIS SPR instrument and first preliminary results could be obtained as shown in Figure 12.10 (left). The photograph shows the gold region inside the chip with a centered sample stream. The surrounding water is in resonance, while the centered sample (2-propanol) is out of resonance. It is definitely a trend that the integration of lab-on-a-chip devices and SPR (imaging) instruments will be developed further for new applications in the life sciences. Broadening the options of parallelization and assay implementation, including sample treatment on-a-chip as shown here by the FFE principle, would certainly contribute to an increase of the range of applications. A limitation of miniaturization is imposed by the lateral resolution of SPR imaging, which is B10 mm and can only be improved by using other plasmonic strategies, for instance using nanoparticle sensing (see Section 12.3 and the discussion above). However, current spotting technologies are still an order of magnitude (250 mm) away from these physical limitations.

12.3.5

Digital Microfluidics

A new trend has been observed in the development of a microfluidic chip, in which single cells or excreted compounds from these single cells can be diagnosed for (early) diseases in a flexible and versatile way by combining two existing platforms. Although it can be categorized among hyphenation

374

Figure 12.10

Chapter 12

Left: FFE+SPR chip (20  20 mm). Right: an image showing the SPR gold region inside the FFE chip.

techniques, we consider this combination as a new trend in the combination of microfluidics and SPR imaging. So-called digital microfluidics using electrowetting (EW) principles can actuate cell-containing droplets and SPR imaging detects the binding affinity of the cell or excretion products from this single cell to a variety of specific ligands. This is the topic of current research being carried out in the groups of van den Berg and Mugele at the University of Twente. EW is arguably the most versatile tool for the manipulation of individual droplets in digital microfluidic systems (DMS) [42]. In certain respects, EWbased DMS is similar to other lab-on-a-chip concepts: there are several macroscopic reservoirs on the chip, for the analyte fluid, for reagents and perhaps also products. However, in contrast to conventional lab-on-a-chip methods which use continuous flow, in DMS all substances are handled in discrete amounts of individual droplets, which can be detached from on-chip reservoirs and moved along certain paths to dedicated locations, where (bio)chemical reactions can be initiated or optical/electrical measurements can be performed. The basic experimental setup of EW is shown in Figure 12.11. An aqueous droplet rests on a solid substrate with an electrode on top. Typically the substrate is a glass slide and the electrode a transparent indium tin oxide (ITO) layer, covered by an additional insulating layer of either SiO2/Si3N4 or a Teflon-like polymer with thickness of the order of 1 mm. Under this condition, the contact angle y can be reduced by several tens of degrees by applying a voltage U across the electrodes. This follows from the electrowetting equation [42]: cosy ¼ cosyY þ

e0 ed 2 U ¼ cosyY þ Z 2dslv

ð12:2Þ

Future Trends in SPR Technology

Figure 12.11

375

Sketch of electrowetting setups. (a) Basic setup with homogeneous substrate. (b) Substrate with two independent electrodes. (c) System with parallel plate geometry with the top electrode replacing the immersed wire. Taken from Mugele and Baret [42].

where slv is the liquid–vapor interfacial tension, d the thickness of the insulator layer, ed its dielectric constant and Z the dimensionless electrowetting number. This principle allows actuation of drops via a different wetting by two adjacent electrodes on the same side of the droplet [43]. The basic actuation scheme is shown in Figure 12.11b. If only one electrode is activated, the contact angle reduction takes place only on one side and hence the drop experiences a net force. This is generally sufficient to let droplets make a unit step, within a 2D pattern of electrodes laid out by the chip designer. To avoid the necessity of immersing a wire, the drop is typically sandwiched between two parallel plates (Figure 12.11c). The top plate then provides the second electrode, which may or may not be patterned also. Apart from serving as a second electrode, the top surface also reduces water evaporation dramatically, which can otherwise be a substantial problem. Droplets can also be provided with an oil environment. This will eliminate evaporation altogether, while the oil will also form a film between the droplets and the substrate, which can reduce actuation problems related to adsorption and pinning. The use of EW to actuate droplets has been used successfully many times already for simple liquids, typically salt solutions with concentrations ranging from zero (deionized water) to saturation [44], with the most active groups being based in the USA (Kim’s group at UCLA and Fair’s group at Duke University).

12.3.5.1

Cell Diagnosis and Monoclonal Antibody Screening Using SPR Imaging and Digital Microfluidics

In principle, digital microfluidics allows for various kinds of diagnostic tests on cell-laden droplets: optical, electrical, chemical and even mechanical testing. This diagnosis can take place on a flexible and custom-designed detector array on chip. Here, cells can be exposed to a variety of ligands (e.g. antibody, receptor or binding epitope), with each type of ligand being grafted to an isolated patch of gold, patterned on the lower glass surface of the microfluidic chip. Screening hybridoma cell lines for the production of specific monoclonal antibodies is an example of the application of the EW–SPR imaging combination. In biotechnology, the development of monoclonal antibodies is a complex, time consuming and thus expensive procedure involving the generation,

376

Chapter 12

maintenance and screening of thousands of hybridoma clones. The confident early identification of hybridomas that produce the best candidate antibodies is a critical step in successful, cost-efficient development. During the exposure of the cell to a specific sensor patch, the cell will secrete monoclonal antibodies into the droplet. While the cell itself will not be bound to the surface, the secreted antibodies will diffuse within the droplet and be measured in real time with respect to their binding to specific ligands at the surface. The cell can then be further processed for cultivation of the hybridoma clone, also using the digital microfluidics principle for actuation.

12.4 Trends in Sensor Surfaces Although dimensionally extremely small, the quality of the sensor chip surface coating has a tremendous influence on the performance of an SPR biosensor. SPR imaging of an area of 1 cm2 of various high-quality sensor chips shows defects in more than 90% of the chips, leading to potential affected sensorgrams and artifacts. For instance, irreproducible drying effects, caused, for example, by adsorbed air bubbles, often show cauliflower images of the surface. Further, dust particles that are always present in the air of non-sterile environments can be irreversibly adsorbed on the surface. The imaging feature reveals the quality of the sensor surface and inhomogeneous coatings can be visualized. The operator of the imaging instrument is able to reject suspicious sensor areas. The homogeneity of the nanoarchitecture of the sensor surface can be checked with a reflectivity image where the SPR angle is set in the inflection point of the left-hand flank of the SPR curve. For instance, in the IBIS-iSPR system of IBIS Technologies either an SPR image can be measured with improved contrast or the SPR image can be transposed further to an artificial color image. In order to avoid contamination, it is extremely important to expose the uncovered sensor chip as briefly as possible to the open atmosphere. Biacore introduced in 1990 a cassette for the sensor chip that only is opened automatically inside the Biacore instrument. Hydrogels with a thickness above 1 mm are useful to keep particulate contaminants or air bubbles outside the evanescent field, resulting in a very robust surface. However, heavy diffusion limitation is observed in such structures with a mean diffusion time of several seconds for small molecules across the hydrogel.

12.4.1

Smart Polymer Brushes

In the past, an enormous variety of sensor coatings were used for SPR detection. It is clearly demonstrated that the gold surface needs to be shielded from the influence of complex samples. The modification of surfaces with thin polymer films is still used to tailor surface properties such as the adsorption behavior of ligands, wettability and biocompatibility. For example, a polystyrene microtiter plate for ELISA measurements was mimicked in an SPR

Future Trends in SPR Technology

377

set-up by spin coating or spraying of a polystyrene polymer dissolved in toluene solution [45]. A table-top spin coater for this purpose can be obtained from, for example, Eco Chemie (Utrecht, The Netherlands) to deposit polymers on the sensor surface for SPR measurements. However, hydrophobic surfaces are prone to poor wettability, ligand desorption due to physical attachment and the adsorption of air bubbles detrimental to signal reproducibility. A hydrophilic coating is much more reliable in SPR biosensors. The hydrogel-based carboxymethylated dextran layer is the most popular matrix for SPR biosensors due to its high coupling yields and reliability with regard to ligand immobilization. The reaction conditions to couple proteins, peptides and small molecules to carboxymethylated dextran surfaces are well characterized and extensive optimization studies have been performed [46]. Figure 6.12 a ‘‘smart’’ composite hydrogel layer is shown, which consists of an unreactive, relatively long polymer chain in low density on top of a thin ligand layer. Such a long polymer composite layer will exclude particles and cells which might be present in the sample. Particulate samples such as whole blood or crude fermentation broths are thus filtered from the surface and do not have access to the evanescent field of the sensor surface. The synthesis of hydrogel layers to surfaces can be performed by grafting polymers with reactive end-groups on to surfaces, resulting in so-called polymer brushes. The advantage of polymer brushes over other surface modification methods (e.g. selfassembled monolayers) is their mechanical and chemical robustness, coupled with a high degree of synthetic flexibility towards the introduction of a variety of functional groups. There is also increasing interest in the use of functional or diblock copolymer brushes for smart or responsive surfaces, which can change a physical property (hydrophilicity, biocompatibility, swelling) upon an external trigger, such as heat (in the case of materials with a lower critical solution temperature), pH or salt concentration [47]. Two methods of grafting polymers to surfaces are applied to create polymer brushes: either (1) via chemical bond formation between reactive groups on the surface and reactive end-groups or (2) by physisorption of block copolymers with ‘‘sticky’’ segments. This ‘‘grafting to’’ approach is experimentally simple, but has some limitations. It is very difficult to achieve high grafting densities because of steric crowding of reactive surface sites by already adsorbed polymers. Relying on non-covalent adsorption of polymers to surfaces makes the adsorption a reversible process and such brushes desorb from the sensor surface resulting in an apparent off-rate, which has nothing to do with the dissociation of the analyte from the ligand. ‘‘Surface-initiated polymerizations’’ [48] (also called ‘‘grafting from’’) from initiators bound to surfaces are a powerful alternative to control the functionality, density and thickness of polymer brushes with almost molecular precision. First, the gold sensor surface should be modified with an initiator-bearing self-assembled monolayer (for example, thiols on gold). The sensor surface is then exposed to solutions containing catalyst and monomer (plus solvent if necessary). Ideally, the polymerization is only surface initiated and no polymerization in solution takes place. In order to achieve maximum control over

378

Chapter 12

brush density, polydispersity and composition, a controlled polymerization is highly desirable. Over the last few years, this field has evolved rapidly and many polymerization strategies have been used to grow polymer brushes on gold [49]. The surface-initiated route to polymer brushes initiated by Fukuda’s group [50] has expanded tremendously over the last 9 years. It opens up new possibilities for creating smart or responsive surfaces and we have only seen the beginning of research in this direction. Polymer brushes have interesting physical properties that are primarily related to the fact that the polymers are covalently tethered to the surface while the other end of the chain is freely moving in solution. No doubt this will lead to new applications and to the improvement of SPR devices regarding the controlled swelling and shrinking of polymer brushes for the purpose of enhancing the sensitivity of specific biomolecular interactions and for improving the limit of detection of low molecular weight analytes, which cannot be detected reliably without the use of these smart polymer brushes.

12.4.2

Photoactivation of Surfaces for Immobilization

A trend is observed in combining lab-on-a-chip unit operations and SPR imaging as treated in Section 12.3.2. After a chip-based separation process integrated in the SPR instrument, ligands should be immobilized on the surface of the sensor chip. For example, the FFE–SPR combination requires a method to trap separated molecules on the surface. It is desired that an interfering method triggers the activation of the surface, in order to allow covalent coupling of the separated ligands on a desired spatially resolved area. Light of a specific wavelength and intensity leads to activation of specific groups in the hydrogel, permitting covalent coupling of ligands. Benzophenone-containing substrates with low non-specific binding properties are a first choice [51]. The use of benzophenone as photoactivator has important advantages: benzophenones are more stable than other photoactivators such as arylazides [52] and can be manipulated in ambient light; activation wavelengths are around 350 nm, a range not damaging towards most proteins; and benzophenone can be activated by light repeatedly without chemical degradation. Two general approaches were pursued in our laboratories at the University of Twente for the preparation of photoactivatable substrates (unpublished results): 1. Chemical coupling of 4-benzoylbenzoic NHS and benzophenone-4-isothiocyanate on to an aminated hydrogel. 2. Chemical coupling of neutravidin on to a carboxylated hydrogel followed by binding of biotin–dPEG3–benzophenone. A relatively simple method for immobilizing benzophenone moieties consists of the chemical coupling of 4-benzoylbenzoic NHS or benzophenone-4-isothiocyanate on to amino group-containing substrates. First substrates of choice are coatings of branched polyethylenimine or hydrogels containing simple primary amino groups. For example, the incubation of bare gold with a diluted (0.01%)

Future Trends in SPR Technology

379

aqueous solution of branched polyethylenimine (10 kDa) leads to good coverage of the sensor surface. The advantage is that the efficiency of immobilization by photoactivation can be followed in real time. Another approach is to prepare, illuminate and evaluate gold sensor modifications in an IBIS imaging SPR instrument during the SPR measurement, using an LED as a light source for photoactivation. Then the process of immobilization and the effect of the photoactivation can be followed in real time. Neutravidin has been covalently coupled on to carboxylated hydrogel (XanTec HC 200m) and binding of biotin–dPEG3–benzophenone to this matrix was shown to be successful (unpublished results).

Figure 12.12

Schematic overview of the concentration gradient immunoassay method. (A) Device geometry (not drawn to scale). Three fluid streams converge within a single channel: one containing the analyte of interest (sample), one containing an antibody against the analyte and (optionally) a reference or control stream. Following convergence, the fluids flow down a channel for some distance to permit diffusive mass transport among the fluid streams. This portion of the channel surface is functionalized with PEG to resist protein adsorption. After interdiffusion establishes a gradient of antibody–analyte complexes transverse to flow, the fluids then encounter the sensing surface that is functionalized with surface-bound analyte. (B) Schematic view of competition between solution-phase analyte and surface-immobilized analogue for antibody binding sites. Diffusion of solution-phase analytes into the antibody stream establishes a gradient of occupied antibodies. Only antibodies with at least one open binding site may bind to the surface, leading to an SPR signal change. The view is through the channel from the outlet. (C) Cartoon of SPR difference image obtained after antibody accumulation to the sensing surface. The ‘‘assay shift’’ is the difference between the width of the center flow stream and the region of the surface on which antibodies accumulate. Axes indicate the view direction: +x is through the channel in the direction of fluid flow. Reproduced from Analytical Chemistry with permission of the American Chemical Society.

380

12.4.3

Chapter 12

Gradient Chemistries

Gradient chemistry strategies are attractive in combination with SPR imaging. In SPR imaging instruments, the optimal degree of coupling can be spatially resolved. An example of the gradient chemistry strategy was recently published by Yager’s group (see also Chapter 10). Concentration gradient immunoassay (CGIA) [53] is capable of the direct measurement of low molecular weight analytes in less than 10 min with simultaneous controls on a single fluid sample (Figure 12.12). Question: how can we coat an SPR sensor chip device in a steady gradient? In the previous example, a diffusion regime between two different flow regions causes a gradient at the intersection of the two flows. Gradients can also be created by timely contact of a ligand with the sensor chip. A microfluidic device is necessary to create the gradients in the chip which is represented schematically in Figure 12.13. In the sensing lane area of Figure 12.13 ligands are immobilized on sensor patches. In the non-specific binding section the analyte will bind non-specifically to the surface, while the ligand was not in contact with this section, hence ligands are not present on the surface. In the common mode rejection (clean) section, the analyte did not contact the surface and the area can be used for common mode rejection compensation for bulk refractive index shifts of buffer or regeneration liquids or for temperature correction. It is a challenge to determine the effectiveness of this new strategy for a certain application, however; the implementation of a microfluidic device and an SPR imaging instrument is needed. Not only defined areas of sensing lane, non-specific binding section or common-mode rejection section, but also a gradient of contact times of ligands and analytes can easily be created in a microfluidic device. Another approach is described in Section 12.3.2. A pH gradient can be obtained which may be used either for gradient coupling or for gradient denaturation of bound ligands. The technique may also be used for finding

Figure 12.13

In a microfluidic device a sample can be injected slowly at point B and reversed after a certain exposure to the sensor spots. The diffusion rate of the compounds and the contact time determine the spatial resolved accumulation or gradient of ligands and analytes. Because the supply of ligand and analyte in the sample is only from one side (B) to the other compartment (C) there is a gradual contact time difference of the sample with the sensing areas on the left (red) and right side (yellow to blank). Here the ligand coating solution (green antibody) is reversed at the fourth spot. The analyte (blue stars) is reversed at the seventh spot. The final three spots as indicated here in this linear flow channel will never be exposed to the ligand or analyte containing solution.

Future Trends in SPR Technology

381

the optimal conditions for reversible regeneration or elution properties, which should be as mild but as complete as possible. The combination of smart chemistries, microfluidic chips and SPR imaging will gain great potential for the search for the best interface behavior of biomolecular interactions of ligands with their specific analyte.

12.5 Trends in Measuring Reliable Kinetic Parameters 12.5.1

Introduction

The aphorism ‘‘God made the solid state, he left the surface to the devil’’, attributed to Wolfgang Pauli [54] or Enrico Fermi [55], seems to apply equally to the measurement of protein interactions in solutions versus at surfaces. Surface binding measurements, as facilitated by SPR and illustrated in this book, provide unique opportunities, among them small sample volumes [56], high affinity measurement with signal-to-noise ratio independent of KD and micropurification for interaction discovery [57] and multi-protein binding studies [58]. However, this comes at the price of immobilizing one binding partner to the surface. For proteins, this in itself bears the possibility that its conformational ensemble may be skewed or even significantly altered, even when using chemically and/or structurally uniform attachment strategies. Further, while in free solution all molecules experience the same environment, the microenvironment at the surface may be strongly variable depending on the location on the surface (e.g. from surface roughness) or depending on the location within an inhomogeneous matrix (e.g. from the obligate density distribution perpendicular to the surface of grafted polymers [59,60] or from ligand gradients perpendicular or parallel to the surface created during immobilization [61]). In solution the conformational ensemble usually exchanges rapidly enough for the binding thermodynamics to be well described by single values but, as a result of the immobilization and localization to the surface, this is frequently no longer the case for the surface sites. As a consequence, it should be expected that the surface binding energetics can experience a dispersion and assume a continuous distribution of parameters. In this chapter new insights will be evaluated regarding this distribution analysis of rate and equilibrium constants. Methods have been introduced that allow the characterization of this functional distribution of binding parameters of the ensemble of surface sites from experimental surface binding data [62–64], and, in particular, we have recently shown that it can be applied to the global analysis of SPR kinetic traces [65,66]. The goal is two-fold. First, a more detailed characterization of the distribution of binding properties should provide a useful tool in the optimization of surface immobilization, to study immobilization processes and surface properties in more detail, towards the efficient functionalization of biosensor and protein chip surfaces with uniform high-affinity binding sites. Second, if one is able to identify a peak in the distribution reflecting ‘‘high-affinity’’ sites, distinct from a range of other ‘‘low-affinity’’ and ‘‘non-specific’’ sites impaired in their function, this should allow one to characterize the interaction

382

Chapter 12

properties more reliably as they may reflect solution conditions. To some extent, that may be possible with conventional tools by an ad hoc assumption of two classes of discrete sites, but this artificial a priori constraint to two classes of sites may be too crude and introduce bias in the results. It is preferable to have a more refined, assumption-free method from which the determination of number of ‘‘classes of sites’’ may be made, if appropriate, as a final interpretative step considering the full continuous distribution. This model invokes better insight in the biomolecular binding process than simple discrete binding models. That the experimental data carry enough information for the characterization of the functional distribution of binding parameters is ensured by the excellent signal-to-noise ratio and reproducibility of SPR binding traces. Further, it is a common observation that the errors encountered from fitting with simple discrete binding models are not randomly distributed, but frequently show highly reproducible systematic residuals [67]. This indicates that discrete binding models may not be sufficiently detailed to account for the observed processes. In some cases this may well be the result of more complicated binding schemes (see examples in Chapters 4 and 5 and models reviewed in [58]). However, in cases where there is no independent supporting evidence for higher-order binding modes, a natural extension of a simple bimolecular reaction scheme is to account for an ensemble of surface sites with continuous distribution of binding parameters. This can frequently provide excellent fits of data commensurate with the high signal-to-noise ratio and reproducibility of SPR binding traces. As described in Chapters 4 and 5, another potential difficulty in the quantitative interpretation of surface binding kinetics is mass transport limitation. As a penalty from the measurement taking place at the surface, for the surface binding kinetics to reflect only the chemical reactions of interest the soluble binding partner has to diffuse sufficiently rapidly from bulk to the surface, such that the probability of binding to a surface site remains a relatively rare event and no concentration gradients of the soluble binding partner occur. The latter will depend on the diffusivity of the soluble analyte, the time-scale of the binding reaction and the density of immobilized sites [65]. As introduced in previous chapters, if this condition is not fulfilled, the binding process will be governed by mass transport limitation, where the experimental data will only indirectly reflect the molecular binding properties, and instead be governed by macromolecular transport properties, such as diffusion coefficients and transient non-specific interaction with the surface, as well as the permeability of the immobilization matrix (if one is used) [66,67]. Recently, the analysis of the distribution of surface sites was extended to include compartment-like first-order corrections for mass transport-influenced surface binding. For the first time, this permits the difficulties of surface heterogeneity and mass transport, both of which frequently occur in experimental SPR data, to be simultaneously addressed. This extends the range where the distribution analysis can be applied and also permits a more detailed study of the origin and behavior of mass transport limitation.

383

Future Trends in SPR Technology

12.5.2

The Model for Distribution Analysis of Rate and Equilibrium Constants

With the terminology of Chapter 5, the traces for different analyte concentrations [A] starting to interact at t0 for a duration tc with a single class of surface sites of maximum signal Bmax, as described in eqs. (5.4) and (5.6), can be combined as kon ½A Rt ðkoff ; KD Þ ¼ Bmax ðkoff ; KD Þ kon ½A þ koff ( 1
e
ðkon ½Aþkoff Þðt
t0 Þ
 1
e
ðkon ½Aþkoff Þðtc
t0 Þ e
koff ðt
tc Þ

t0 ot  t0 þ tc

ð12:3Þ

t4t0 þ tc

The expression Rt(koff, KD) highlights the signal contribution arising from the sites with equilibrium dissociation constant KD and dissociation rate constant koff (or with equilibrium association constant KA ¼ 1/KD and association rate constant kon ¼ koff/KD, respectively). The binding kinetics expressed in eq. (12.3) combines the association and dissociation phase and reflects a family of curves as shown, for example, in Figure 4.2 in Chapter 4. Depending on the rate and equilibrium constants, the shape of the curves follows different patterns and the following analysis is concerned with the inverse problem of seeking the combination of such patterns that fits experimental data best. For the description of a continuous distribution of surface sites Bmax(koff, KD) with a range of chemical off-rate constants koff and equilibrium dissociation constants KD, we define the infinitesimal quantity Bmax(koff*, KD*)dkoffdKD as the population of surface sites (in signal units) with an offrate constant between koff* and koff*+dkoff and the equilibrium constant between KD* and KD*+dKD. (With the usual transformation kon ¼ koff/KD, the distribution can be transformed to an equivalent distribution of on-rate and equilibrium constants; see below and Figure 12.14.) The total measured signal, Rtot,t, can then be expressed as a Fredholm integral equation: Rtot;t ¼

KD;max Z koff;max Z

Rt ðkoff ; KD Þdkoff dKD

ð12:4Þ

KD;min koff;min

The computational problem consists in a global least-squares fit of the integral eq. (12.4) to experimental binding traces R(exp)t([A]) acquired at different analyte concentrations. This can be accomplished by discretization of the two-dimensional space of binding parameters with a mesh of (koff,i, KD,i) values: ðexpÞ

Rt

ð½AÞ ffi

N X

Rt;i ðkoff;i ; KD;i Þ

ð12:5Þ

i¼1

and numerical optimization of Bmax,i(koff,i, KD,i), which is the discrete representation of the distribution Bmax(koff, KD) [65].

384

Chapter 12

Figure 12.14

Binding of soluble microglobulin to a monoclonal IgG immobilized on a carboxymethylated dextran surface (F1 chip) [66]. (A) Association and dissociation traces at an analyte concentration of 1 nM. (B) Contour plot of the affinity and rate constant distribution calculated from the 1 nM binding data of (A), on a grid of KD and koff value as indicated by the small circles. In this presentation, lines of constant kon values are diagonals and logkon values can be read as logkoff – logKD. (C) Association and dissociation traces at a concentration of 100 nM. (D) Surface site distribution calculated from the 100 nM trace alone.

In case of mass transport-limited binding, the distribution analysis model can be extended [66]. Briefly, it is approximated on a discretized mesh of binding parameters Bmax,i(koff,i, KD,i). However, in this case the binding traces for a single class of sites do not follow eq. (12.3). As outlined in Chapter 5, mass transport-limited binding can be approximately described by a two-compartment model that introduces, as a first-order approximation of spatial inhomogeneity, the distinction of analyte at the surface [A] and in the bulk [A]0. Equation (5.7) can be generalized to the case of many sites with binding properties (koff,i, KD,i), which results in the following differential equation for the signal arising from each site:   dRt;i ¼ kon;i ½A Bmax;i
Rt;i
koff;i Rt;i dt N   X d½A dRt;j ¼ ktr ½A0
½A
dt dt j¼1

for all i ð12:6Þ

with the abbreviations Rt,i and Bmax,i denoting Rt(koff,i, KD,i) and Bmax,i (koff,i, KD,i), respectively, and kon,i ¼ koff,i/KD. After a fast initial transition period after application of [A]0, the surface concentration [A] assumes an

Future Trends in SPR Technology

385

approximately constant value (steady-state conditions), which leads to   N   kon;i Rmax;i
Rt;i X dRt;i dRt;j ¼ kon;i ½A0 Rmax;i
Rt;i
koff;i Rt;i
dt dt ktr j¼1

ð12:7Þ

This is a system of rate equations coupled through the third term, which is the one describing the mass transport influence. The coupling reflects the physical possibility that an analyte dissociating from one class of sites may rebind to sites from another class. Again, optimization of the Bmax,i(koff,i, KD,i) values in a least-squares fit of experimental data [eq. (12.5)] provides a discretized representation of the continuous distribution Bmax(koff, KD). In order to make this distribution analysis computationally feasible, it is combined with maximum entropy [68] or Tikhonov regularization. This is a technique for stabilizing ill-conditioned or underdetermined data inversion problems and well known in many areas of physical data analysis [69]. Importantly, it acts to suppress detail in the distribution that is not statistically warranted by the information content of the data. This will lead to the most parsimonious, or broadest, distribution consistent with the data. The software (termed EVILFIT in our laboratory), is implemented on the MATLAB platform and is freely available from the authors. A simplified graphical user interface is anticipated for the future.

12.5.3

Examples of the Distribution Analysis Method

When inspecting the results from the distribution analysis method, it is important to be aware of the effects of regularization providing the ‘‘simplest’’ distribution of all that may be consistent with the data, following Occam’s razor. For example, by virtue of the regularization, it is possible to solve underdetermined problems, such as to obtain full two-dimensional distributions of affinity constants and kinetic rate constants from association and dissociation curves at a single concentration. This is illustrated in Figure 12.14, which in panels B and D depict the distributions in units of logkoff and logKD [in this presentation, lines of constant kon are diagonals (since logkon ¼ logkoff – logKD) and the volume under the peaks gives the total binding capacity of sites within the given parameter range]. Panel A shows the association and dissociation traces of 1 nM antigen binding to an immobilized antibody and panel B shows the distribution obtained from this single trace. As can be expected on the basis of the limited information carried in these data, the distribution has only very broad features. It is conservative in a sense that it suggests order of magnitude of KD and koff values, rather than attempting to provide single values. This aspect is highlighted by comparison with the distribution obtained at a single analyte concentration of 100 nM (Figure 12.14C and D, respectively). Since higher concentrations provide more information, which can be discerned here from the biphasic shape of the association trace, the distribution has more detailed

386

Chapter 12

features: it displays a narrower main peak for the high-affinity sites, alongside some minor peaks for sites with lower affinity. This example demonstrates how the level of detail in the distributions is automatically adjusted according to the information carried in the experimental data, such as to provide the most conservative interpretation. Importantly, the width of the distribution should not be interpreted in a sense that it necessarily reflects the exact distribution of surface sites, but that it reflects the most detailed statement that can be made about the surface sites given the data. For example, the single 1 nM trace may well be modeled by a single discrete 1:1 binding model, providing apparently unambiguous unique values of binding and rate constants. However, whether or not this number would reflect the true binding parameters cannot be decided on the basis of the single curve. Indeed, as the additional data at the higher concentration show even from visual inspection of the biphasic shape, there is heterogeneity of the surface sites and the discrete single-site model to the 1 nM data would have been misleading. Similarly, there may still be finer structure in the true distribution of surface sites than displayed in Figure 12.14D, but the given finite signal-to-noise ratio does not permit closer characterization. Although the analysis of single traces can have important applications, for example, for the study of binding sites prior to the exposure to chemical regeneration, obviously the goal of the analysis and the mode in which the distribution analysis is usually applied is the global analysis of all traces at all concentrations or even global analysis of different flow rates. The relationship between information content of the data and resolution of the resulting distribution is the same for global analyses. For this example, such a global analysis including curves at more analyte concentrations has been carried out [66], showing similar features to Figure 12.14D. The question of whether the observed width of the distribution arises from true microheterogeneity of the surface sites or from the finite signal-to-noise ratio in the data is addressed in the next example (Figure 12.15) [58]. Although a single major peak is obtained in the distribution (Figure 12.15C), some degree of heterogeneity may be discerned. The polydispersity of the binding sites is supported by a comparison of the quality of fit of the distribution model (residuals in red in Figure 12.15B) with the best-fit single-site model (residuals in blue in Figure 12.16B and dotted line in Figure 12.15A). The opportunity to study the behavior of binding sites in their dependence on the surface employed for immobilization is highlighted in Figure 12.16. This shows the same antibody–antigen interaction as in Figure 12.14, but immobilized on a long-chain carboxymethyldextran surface (Biacore CM5). On this surface, the binding is mass transport limited and the global analysis of binding traces at 0.1, 1, 10 and 100 nM with the model eq. (12.7) results in a distribution that exhibits a tail of low-affinity sites that amounts to 425% of all surface sites. Such a broad population of low-affinity sites was not observed in the analogously conducted experiment using the short-chain carboxymethyldextran surface [66]. Interestingly, the possible role of non-specific binding causing flow rate-independent contributions to mass transport limitation is supported

Future Trends in SPR Technology

Figure 12.15

387

Surface site distributions obtained from an antigen interacting with the antibody immobilized on a long-chain carboxymethyldextran surface (Biacore CM5) [58]. (A) Experimental association and dissociation traces (black) at three different concentrations. The blue dotted line is the best-fit model based on a single discrete 1:1 interaction. (B) Residuals of the fit with a single discrete 1:1 interaction model (blue) and with the continuous distribution model (red). (C) KD–koff distribution from the data in (A). For details, see [58].

by theoretical expectation of transient matrix interactions slowing the effective diffusion time of analyte through the matrix [66,69]. Finally, the surface site distribution can be applied to the study of immobilized biomolecules that exhibit naturally multiple classes of sites. This is highlighted, for example, in the study by Vorup-Jensen et al. [70] on acidic residues of fibrinogen exposed during tissue decay and their role as a pattern

388

Chapter 12

Figure 12.16

Binding of soluble microglobulin to a monoclonal IgG immobilized to a long-chain carboxymethylated dextran surface (CM5 chip) [66]. Shown is a plot of the affinity and rate constant distribution calculated from the global analysis of traces obtained at analyte concentrations of 0.1, 1, 10 and 100 nM.

recognition motif in the recognition by integrin aXb2 (Figure 12.17). VorupJensen et al. concluded that the increased affinity for the recognition of ‘‘damaged’’, i.e. partially unfolded or proteolyzed, fibrinogen plays an important role in tissue repair as a danger signal enabling enhanced recognition by leukocyte cell-surface receptors [70].

12.5.4

Conclusions and Perspectives of the Distribution Analysis Model

The binding model for a continuous distribution of sites is a natural extension of the 1:1 discrete binding site model that takes into account the heterogeneity of binding site properties commonly arising from heterogeneity in the surface microenvironment (surface roughness, charge and density distribution from polymer matrices) and from surface immobilization (steric, spatial and/or chemical heterogeneity of the attachment). At the same time, the model can account for low to moderate degrees of mass transport limited binding. This addresses the two most commonly observed complications in the use of SPR to study interactions of biological macromolecules [67]. Interestingly, heterogeneity in the spatial dimension parallel to the surface (as opposed to the functional heterogeneity) was addressed in Chapter 7, where it was shown how SPR imaging enables one to obtain a microscopic inspection

Future Trends in SPR Technology

Figure 12.17

389

Interaction between fibrinogen and integrin aX I domain. The binding of soluble integrin aX I domain to native, proteolyzed or guanidinetreated immobilized fibrinogen was probed by applying 10 aX I concentrations from 0.28 to 10.6 mM (A). For comparison, sensorgrams in (B) show the injections of the I domains at the highest applied concentration of 10.6 mM over the surfaces with native, proteolyzed or guanidine-treated fibrinogen (the end of injection phase is indicated with arrows). Two-dimensional off-rate constant and affinity distributions show a relatively homogeneous ensemble for native (C), but more heterogeneous ensembles for denatured (D) and plasmin-treated (E) fibrinogen. Reproduced with permission from reference 70.

of the spatial heterogeneity of the surface before and after a biomolecular interaction process. It has often been observed that an initial homogeneous spot shows heterogeneity after the binding process, implying that a distribution of a kinetic process takes place over the surface caused by clustering of biomolecules or by an initial heterogeneous distribution of binding sites. In this way, the functional and spatial heterogeneity may be linked. In the case of

390

Chapter 12

such an observation, it might be advantageous to apply the distribution analysis method to describe and verify the biomolecular interaction process. With the goal of the SPR experiment to characterize the binding affinity and rate constants, the model is parsimonious in the assumptions and the regularization ensures conservative data interpretation, automatically adjusting to the information content of the experimental traces. In addition to a range of binding constants, signal-average single numbers for the affinity constant and for the chemical rate constants may be determined by integration of the peaks. The model naturally displays multiple classes of sites, without the need for postulation of their number a priori, if their detection is statistically warranted by the data. Even if some of the sites are not of interest, such as low-affinity or ‘‘non-specific’’ sites considered surface-related artifacts, identifying them and accounting for their signal contributions is particularly important in order not to bias the characterization of the high-affinity sites of interest. In recent years, we have routinely applied this approach, for example, to the study of antibody–antigen interactions [71,72] and have observed that the distribution model in most cases provides a fit of the data stringently within the noise of data acquisition (such as in the examples with systems drawn from the Biacore startup kit shown in the present chapter and in Chapter 6, where simple discrete 1:1 models would fail and the extension to more complex binding schemes seems inappropriate since biologically not supported). This supports the view that the surface site heterogeneity model captures a true feature of analyte binding to surface-immobilized sites, which SPR binding data commonly have a sufficient signal-to-noise ratio to display readily. A second area of application for the distribution model is the optimization of sensor surfaces and surface immobilization. This is a very important and active area of research (see, e.g., Chapter 6). Although the best sensor surface to permit uniformly active protein attachment and to exhibit low non-specific binding will certainly depend on the nature of the proteins and analytes involved, the model for distribution of kinetic and affinity parameters currently provides the most detailed tool for the functional characterization of the ensemble of surface sites. As has been shown [66], the combined mass transport/surface heterogeneity model can also provide more detailed insights into the physical nature of the transport process and help to understand how different sensor surface properties may impact the relationship between the measured surface binding kinetics and the intrinsic chemical kinetics of the interacting macromolecules. Similarly, SPR imaging of (protein) microarrays with multiple spots of the same ligand immobilized at different densities is a promising approach, where the global analysis will become important. In this case, also, the combination of multi-spot global analysis with the distribution method may lead to a more detailed description of the ligand functional binding properties. We expect that with the introduction of reliable SPR imaging instruments with hundreds of spots, kinetic off-rate screening of samples [73] will be extended to screening of the affinity constant of many analytes to ligands (multi-kinetics) in only a single injection.

Future Trends in SPR Technology

391

12.6 Final Comments In this chapter, trends in SPR technology regarding instrumentation, fluidics, sensor surfaces and kinetic analysis have been described. It is speculative to lay down the likely degree of impact in the next few years regarding these trends in SPR technology. However, considering that biomolecular recognition and the quantitation of molecular interactions have become central in the study of structure and function of proteins and biological pathways, in molecular medicine and in biotechnology and pharmaceutical development, and considering SPR as being one of the most mature, widespread and direct detection principle, it seems certain to continue to undergo rapid development and expansion of applications. We hope that this chapter and indeed the entire book will have persuaded the reader to share this enthusiasm for SPR and surface-related research and technologies. If the surface is the domain of the devil, as suggested by the well known aphorism cited above, it seems fitting that it should provide us with many highly exciting new research avenues and opportunities to shed light on biological processes.

12.7 Questions 1. The implementation of a lab-on-a-chip device with integrated gold surface in an SPR instrument is not easy. What will happen with the channel surface and the gold surface if we transport analyte directly from diluted serum to the gold surface using electroosmotic flow (EOF) as pumping mechanism? Describe the effects. 2. What happens when an analyte in serum with low isoelectric point is not pumped by pressure from a sample loop but transported using the EOF principle? 3. The main analytical problem of coupling SPR and MS is that the detection limits of both techniques should be matched. What are the discrepancies and explain why SPR has in principle a lower detection limit than MALDI-MS. Does this also apply to small molecules? 4. The application of PDMS devices is beneficial to improve mass transport to the surface. Calculate the increase in the flow rate if we replace a standard flow cell of 1 cm  1 cm  200 mm (L  W  D) with a PDMS device with channels of 5 mm  200 mm  25 mm (L  W  D). What is the factor of sample reduction if we apply an equal flow rate in these flow cells? What is the disadvantage of applying a PDMS device for a spotted sensor surface? 5. A new trend is given in this chapter in kinetic evaluation regarding the simple monophasic interaction model by applying distribution analysis of the rate constants. Microarrays with different ligands and ligand surface coverages show different binding kinetics, e.g. changed mass transport limitation conditions. Why is the distribution analysis model of interest for microarrays and can be regarded as a new trend?

392

Chapter 12

6. Spotting and immobilizing ligands on sensor chips can be carried out using microfluidic devices. Explain how via a self-assembly process thousands of protein ligands can be immobilized in an ultra-high-density microarray with squares of 25  25 mm.

References 1. T. Masadome, Y. Asano, T. Imato, S. Ohkubo, T. Tobita, H. Tabei, Y. Iwasaki, O. Niwa and Y. Fushinuki, Anal. Bioanal. Chem., 2002, 373, 222. 2. C. Thirstrup, W. Zong, M. Borre, H. Neff, H.C. Pedersen and G. Holzhueter, Sens. Actuators B, 2004, 100, 298–308. 3. J.R. Krone, R.W. Nelson, D. Dogruel and P. Williams, P., R. Granzow, Anal. Biochem., 1997, 244, 124–132. 4. C.P. Sonksen, E. Nordhoff, O. Jansson, M. Malmqvist and P. Roepstorff, Anal. Chem., 1998, 70, 2731. 5. T. Natsume, H. Nakayama, O. Jansson, T. Isobe, K. Takio and K. Mikoshiba, Anal. Chem., 2000, 72, 4193. 6. J. Grote, N. Dankbar, E. Gedig and S. Koenig, Anal. Chem., 2005, 77, 1157. 7. E. Bouffartigues, H. Leh, M. Anger-Leroy, S. Rimsky and M. Buckle, Nucleic Acids Res., 2007, 35, e39. 8. N.F.C. Visser, A. Scholten, R.H.H. van den Heuvel and A.J.R. Heck, ChemBioChem, 2007, 8, 298–305. 9. J.J. Gilligan, P. Schuck and A. Yergey, Anal. Chem., 2002, 74, 2041. 10. M. Abrantes, M.T. Magone, L.F. Boyd and P. Schuck, Anal. Chem., 2001, 73, 2828. 11. N.R. Gonzales, P. Schuck, J. Schlom and S.V. Kashmiri, J. Immunol. Methods, 2002, 268, 197–210. 12. E. Melles, T. Bergman, M. Sta˚hlberg, C. Thirstrup, J. Wahren, H. Jo¨rnvall and J. Shafqat, J. Biomol. Tech., 2005, 16, 392. 13. J.W. Attridge, P.B. Daniels, J.K. Deacon, G.A. Robinson and G.P. Davidson, Biosens. Bioelectron., 1991, 6, 201–214. 14. K.M. Shakesheff, X. Chen, M.C. Davies, A. Domb, C.J. Roberts, S.J.B. Tendler and P.M. Williams, Langmuir, 1995, 11, 3921–3927. 15. J. Davies, C.J. Roberts, A.C. Dawkes, J. Sefton, J.C. Edwards, T.O. Glasbey, A.G. Haymes and P.M. Williams, Langmuir, 1994, 10, 2654–2661. 16. A. Zangwill, Physics at Surfaces, Cambridge University Press, Cambridge, 1988, p. 186. 17. R.P. van Duyne, Science, 2004, 306, 985–986. 18. C.L. Haynes and R.P. Van Duyne, J. Phys. Chem. B, 2001, 105, 5599. 19. A.J. Haes and R.P. Van Duyne, J. Am. Chem. Soc., 2002, 124, 10596–10604. 20. X. Hong and F.J. Kao, Appl. Opt., 2004, 43, 2868–2873.

Future Trends in SPR Technology

393

21. J. Daughton, IEEE Trans. Magn., 1994, 30, 4608. 22. G. MacBeath and S.L. Schreiber, Science, 2000, 289, 1760–1763. 23. V. Kanda, J.K. Kariuki, J.D. Harrison and M.T. McDermott, Anal. Chem., 2004, 76, 7257–7262. 24. P. Connoly, Biosens. Bioelectron., 1995, 10, 1–6. 25. M.P. Hudson, R.H. Christenson, L.K. Newby, A.L. Kaplan and E.M. Ohman, Clin. Chim. Acta, 1999, 284, 223–237. 26. B.J. Tortella, R.F. Lavery, J.V. Doran and J.H. Siegel, Am. J. Clin. Pathol., 1996, 106, 124–127. 27. J.H. Nichols, Blood Gas News, 1999, 8, 4–14. 28. A. Manz, N. Graber and H.M. Widmer, Sens. Actuators B, 1990, 1, 244– 248. 29. P.S. Dittrich, K. Tachikawa and A. Manz, Anal. Chem., 2006, 78, 3887. 30. P.S. Dittrich and A. Manz, Nature, 2006, 5, 210–218. 31. J.W. Jorgenson and K.D. Lukacs, J. High Resolut. Chromatogr., 1981, 4, 230. 32. R.B.M. Schasfoort, S. Schlautmann, J. Hendrikse and A. van den Berg, Science, 1999, 286, 942–945. 33. M.A. Hayes, I. Khetarpal and A.G. Ewing, Anal. Chem., 1993, 65, 27. 34. M.J. Jory, G.W. Bradberry, P.S. Cann and J.R. Sambles, Sensors and Actuators B, 1996, 35, 197–201. 35. V. Lioubimov, A. Kolomenskii, A. Mershin, D.V. Nanopoulos and H.A. Schuessler, Applied Optics, 2004, 43, 3426. 36. A. Tadjeddine, Electrochim. Acta, 1989, 34, 29–33. 37. R.G.H. Lammertink, S. Schlautmann, G.A.J. Besselink and R.B.M. Schasfoort, Anal. Chem., 2004, 76, 11. 38. D. Kohlheyer, G. Besselink, S. Schlautmann and R.B.M. Schasfoort, Lab Chip, 2006, 6, 374–380. 39. D.E. Raymond, A. Manz and H.M. Widmer, Anal. Chem., 1994, 66, 2858– 2865. 40. D. Kohlheyer and R.B.M. Schasfoort, Eur. Pat. Appl. EP 05076569.2, 2005. 41. D. Kohlheyer, J.C.T. Eijkel, S. Schlautmann, A. van den Berg and R.B.M. Schasfoort, Anal. Chem., 2007, 79, 8190–8198. 42. F. Mugele and J.-C. Baret, J. Phys.: Condens. Matter, 2005, 17, 705. 43. M.K. Chaudhury and G.M. Whitesides, Science, 1992, 256, 1539. 44. M.G. Pollack, A.D. Shenderov and R.B. Fari, Lab Chip, 2002, 2, 96. 45. W. Qian, D. Yao, F. Yu, B. Xu, R. Zhou, X. Bao and Z. Lu, Clin. Chem., 2000, 46, 1456–1463. 46. B. Johnsson, S. Lo¨fas and G. Lindquist, Anal. Biochem., 1991, 198, 268– 277. 47. B. Zhao and W.J. Brittain, Prog. Polym. Sci., 2000, 25, 677–710. 48. S. Edmondson, V.L. Osborne and W.T.S. Huck, Chem. Soc. Rev., 2004, 33, 14–22. 49. D.M. Jones, J.R. Smith, W.T.S. Huck and C. Alexander, Adv. Mater., 2002, 14, 1130–1134.

394

Chapter 12

50. M. Ejaz, S. Yamamoto, K. Ohno, Y. Tsujii and T. Fukuda, Macromolecules, 1998, 31, 5934–5936. 51. G. Dorman and G.D. Prestwich, Trends Biotechnol., 2000, 18, 64–77. 52. S.A. McMahan and R.R. Burgess, Biochemistry, 1994, 33, 12092. 53. K.E. Nelson, J.O. Foley and P. Yager, Anal. Chem., 2007, 79, 3542–3548. 54. G. Binnig and H. Roehrer, Sci. Am., 1985, 253, 50–56. 55. F.S. Ligler and T. Cass, Immobilized Biomolecules in Analysis, a Practical Approach, Oxford University Press, Oxford, 1998. 56. M. Abrantes, M.T. Magone, L.F. Boyd and P. Schuck, Anal. Chem., 2001, 73, 2828–2835. 57. J.J. Gilligan, P. Schuck and A.L. Yergey, Anal. Chem., 2002, 74, 2041. 58. E.J. Sundberg, P.S. Andersen, I.I. Gorshkova, P. Schuck, in Protein Interactions, Biophysical Approaches for the Study of Complex Reversible Systems, P. Schuck (Ed.), Springer, New York, 2007, Protein Reviews, vol. 5, 97–142 (Review). 59. P.G. de Gennes, Macromolecules, 1980, 13, 1069–1075. 60. A. Chakrabarti and R. Toral, Phys. Rev. B, 1990, 42, 2445. 61. T. Zacher and E. Wischerhoff, Langmuir, 2002, 18, 1748–1759. 62. S. Haber-Pohlmeier and A. Pohlmeier, J. Colloid Interface Sci., 1997, 188, 377–386. 63. A.M. Puziy, Langmuir, 1999, 15, 6016–6025. 64. V.M. Gun’ko, R. Leboda, V.V. Turov, F. Villieras, J. Skubiszewska-Zieba, S. Chodorowski and M. Marciniak, J. Colloid Interface Sci., 2001, 238, 340–356. 65. J. Svitel, A. Balbo, R.A. Mariuzza, N.R. Gonzales and P. Schuck, Biophys. J., 2003, 84, 4062–4077. 66. J. Svitel, H. Boukari, D. Van Ryk, R.C. Willson and P. Schuck, Biophys. J., 2007, 92, 1742–1758. 67. P. Schuck, Annu. Rev. Biophys. Biomol. Struct., 1997, 26, 541–566. 68. U. Amato and W. Hughes, Inverse Problems, 1991, 7, 793–808. 69. P. Schuck, Curr. Opin. Biotechnol., 1997, 8, 498–502. 70. T. Vorup-Jensen, C.V. Carman, M. Shimaoka, P. Schuck, J. Svitel and T.A. Springer, Proc. Natl. Acad. Sci. USA, 2005, 102, 1614–1619. 71. Z. Chen, P. Earl, J. Americo, I. Damon, S.K. Smith, Y.H. Zhou, F. Yu, A. Sebrell, S. Emerson, G. Cohen, R.J. Eisenberg, J. Svitel, P. Schuck, W. Satterfield, B. Moss and R. Purcell, Proc. Natl. Acad. Sci. USA, 2006, 103, 1882. 72. Z. Chen, M. Moayeri, Y.H. Zhou, S. Leppla, S. Emerson, A. Sebrell, F. Yu, J. Svitel, P. Schuck, M. St. Claire and R. Purcell, J. Infect. Dis., 2006, 193, 625. 73. M. Steukers, J.M. Schaus, R. van Gool, A. Hoyoux, P. Richalet, D.J. Sexton, A.E. Nixon and M. Vanhove, J. Immunol. Methods, 2006, 310, 126–135.

Subject Index Note: Page numbers in italics refer to figures or tables. absorbance, enhancement of 28–9 adhesion linking layers 173, 174, 181–3 adsorbate 82 adsorption modes 102–5 multilayer 105, 113–15 non-specific 174–5, 183–5 role in life processes 82 adsorption kinetics 81–122, 282–6, 380–2 basics and terminology 81–6 distribution analysis model 383–90 mass transfer 90–8, 284–5 solute transport and interaction 87–90 mechanisms 98-9 effect of competing reactions 101–2 idealized partition process 99–101 surface functions 102–5, 105 multilayer growth 105, 113–15 saturation 104–12, 105 optical quantification 86–7 reaction rate 87–90 reaction- vs transfer-limited rate 87–90 affinity 8 amine coupling reductive amination 202–3 via reactive esters 199–202 aminopropyltriethoxysilane 183 analysis cycle 5, 221–4 analyte calibration curves 6–7 correction for depletion 128–9

defined 4 small molecules 7, 193, 277, 324–7 angle scanning instruments 40–1, 59–60 microarray analysis 230–2 miniaturization 315 angle shift 36, 37 units 38 angle tuning 315 antibiotics 340, 342, 346, 347 antibodies microarrays see immunoassays immobilization of 213 Salmonella 350 AOAC certification 344–5 APTES 183 assay 3–6, 226–8 inhibition 227–8, 336, 337 surface competition 337, 338 see also immunoassays association 5, 6 constants 7–8 and mass transport limitation 125–6, 284–5 atomic force microscopy 360 attenuated total reflection 20 autoimmune diseases see immunoassays baseline 4–5, 5, 6, 224–5 β-agonists 342–3, 347 β-lactams 347–8

396

Biacore instruments 11, 12, 53, 69–78 A100 70, 71, 73–5 basic systems reviewed 69–70 FLEXChip 61, 69, 70, 75–7 Q sensor 334, 338–9 T100 70–3 binding competing ligand 137–41 intermolecular bivalent 147–52 models 131–5, 143–52, 152–4 rebinding 137–41, 226 sites 82–4 thermodynamics 154–61 bioaffinity 8 bioinert/immobilization matrices 173, 174, 181 bioinert hydrogels 185–7 non-specific adsorption 183–5 biomolecular interactions 174–7 kinetics/thermodynamics see adsorption kinetics; ligand–receptor interactions biosensors definition 9 historical development 9–12 see also sensor chip biotin 344 biotinylation 212–13, 232–3 bivalent binding 147–52 bovine milk proteins 349 Brewster angle 18 buffer solution 4–5, 223–5 and electrostatic interaction 176–7 bulk effect 222 calibration curve 6–7 calorimetry 161–3 capture arrays 363 carbodiimides 199–200 see also EDC carboxymethylated dextran see under hydrogels cell diagnosis 375–6 cephalosporins 347 ceramic substrates 182–3 Charm tests 348

Subject Index

chip see sensor chip chloramphenicol 340, 342, 346 chromophores 278, 281–2, 359 CLAMP software 133–4, 147 clenbuterol 342, 347 clinical diagnostics 313–32, 355–6 advantages of SPR 313–14 bulk refractive index compensation 323–4 immunoassay complex samples 327–9 concentration gradient 279–80 disposable card 328, 329 small molecules 324–7 instrument miniaturization optical system 316 wavelength and angle tuning 315 optimizing imager performance 317–21 temperature fluctuations 321–3 see also lab-on-a-chip; microfluidic devices coherence length 23–4 color multiplexing 306–8 competing ligand 137–41 competition reactions 101–2 competitive assay 226–7 complex samples 327–9 concentration gradient 379–80 conformational change 144–7, 184 convection 178 counter-ion evaporation 185, 196 coupling reactions 203–12 cuvette-based instruments 11, 48–9, 61–2 correction for analyte depletion 128–9 mass transfer process 90–1, 92, 95, 143, 178 Cytop 251, 253, 254 desorption 84 dextran brush 277, 298 dextran hydrogels see under hydrogels diagnostic card 328 diffraction grating 20–1 instrumentation 41–2, 301 diffusion and diffusion layer 178–81, 192 mass transfer kinetics 94–5

Subject Index

digital microfluidics 373–6 dip 2–4, 22–3 fluorescence and absorbance enhancement 29 presentation in sensorgrams 36–8 silver vs gold layer 27 direct assay 226, 227 direct detection 4, 174 directed immobilization 212–13 dispersion relation 17–19, 20 disposable diagnostic card 328, 329 dissociation 5, 6 competing ligand 139–41 constants 7–8 and mass transfer limitation 125–6 rebinding model 137–9 distribution analysis 180, 181, 383–9 dithiothreitol 204 DNA amplified detection gold nanoparticles 260–2 microRNA detection 264–9 RNase H 257–8 coding technology 364–6 immobilization 211 microarrays see microarray imaging drug development 123–4, 165–7 dry immobilization 197–9 dynamic range of scans 230–2 EDC–NHS chemistry 125, 175, 192, 199–200 coupling procedure 201 hydrazide activation 207–8 EDC/(sulfo) coupling 200–1 electrochemical SPR 369 electroosmotic microfluidic devices 367–9 electrostatic immobilization 210, 211 electrostatic interactions 175–7, 185 electrostatic preconcentration 195–7, 196 electrowetting 374–6 entropic stabilization 185, 186 enzymatic enhancement 247, 254, 264–5 epoxy-mediated coupling 208–10 equilibrium constants 7–8, 84 model for distribution analysis 383–5

397 equilibrium signal 126–8, 286 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide 199-200, see also EDC–NHS evanescent field 3, 17, 86, 177–8, 191 and hydrogel thickness 186, 192 and optical requirements 41 evanescent wave 16–17 EVILFIT software 385 excitation 19–21 Eyring plots 163–5 fan-shaped beam instruments 39, 50–4, 55 field enhancement 21–2 field of view 320–1 filter layers 193, 194 fire blight 342 fixed angle instruments 41, 54–9 and microarray analysis 228–30 FLEXchip 61, 69, 70, 75–7, 355 flow cells 45–8 mass transfer process 90–8, 143 microfluidic see microfluidic devices flow rates 94–7, 178 fluidics see microfluidic devices fluorescence 234, 362 enhancement of 28–9 fluorescence spectroscopy 275–8, 309, 359–60 color multiplexing 306–8 fluorescence imaging 306–8 grating coupling 301–3 kinetic studies 286 hybridization of oligonucleotides 286–98 protein binding 298–300 Kretschmann configuration 278–81, 300–1, 307 long-range surface plasmons 303–6 protein binding 298–300 wavelength-resolved spectra 299–300 folic acid 344 food analysis 333–4 assay formats/steps 334–8 genetically modified organisms 350–1 honey 342, 343, 345–6

398

instrumentation 334, 335, 338–9 milk 347–50 Qflex kits 334, 340 AOAC certification 344–5 Salmonella antibodies 350 veterinary drug residues 333, 340 antibiotics 342, 346, 347 β-agonists 342–3 sulfonamides 339, 340, 341–2, 346 Tylosin 343 vitamins 340, 343–4 foulbrood 342, 346 Frank–van der Merwe growth 113, 114 free flow electrophoresis 369–73 Fresnel equation 27–8 gauche defects 182 genetically modified organisms 350–1 genotyping 247, 262–4 glass substrates 182–3 global kinetic analysis 152–4 bimolecular models 131–5 complex binding models 143–52 glycidylpropyltrimethoxysilane 183 glycoproteins 207–8 gold layer 26–7, 44 adhesion linking layers 181–2 and coherence length 23–4 and electroosmotic flow devices 367–9 sputtered surface 190, 191 gold nanoparticles see nanoparticles GPTMS 183 gradient chemistries 379–80 grating coupler 21 fluorescence spectroscopy 301 instrumentation 41–2 growth promoters 333, 343 His6-tagged ligands 211 honey 342, 343, 345–6 hybridization reactions 286–97 hydrazide activation/coupling 207–8 hydrodynamic addressing 47–8, 73, 362 hydrogels dextran 10, 12, 175, 187, 191–3 epoxy activation 208–11

Subject Index

for different sensor applications 188 film density 192–3 film thickness 186, 192 filter layers 193, 194 polymer brushes 277, 298, 376–8 protein-compatible polymers/ functionalities 186, 187 surface chemistry 185–7 synthetic polycarboxylates 192 three-dimensional nanoarchitecture 191–4, 277 two-dimensional surfaces 189–91, 276–7 hydrophilic interactions 181 protein-compatible polymers 187 hydrophobic interactions 175, 176, 177, 184 hyphenation technology 356–7, 359–60 mass spectrometry 357–9 image averaging 318 imaging instruments 43, 63–9, 356 IBIS iSPR for multiplex analysis 228–34 see also microarray imaging immobilization see ligand immobilization immobilization matrices see bioinert matrices immunoassays 222, 225–8 analysis cycle 222–4, 228–32 assay formats 225–8 buffer solutions 224–5 dynamic range/reliability 230–1 experimental setup 228–30 limit of detection 232–4 monoclonal antibody screening 375–6 non-milk proteins 348–9 serum antibodies of rheumatoid arthritis 235–42 SPR and mass spectrometry 357–9 see also clinical diagnostics; microfluidics inhibition assays 227–8, 336, 337 inorganic dielectrics 182–3

Subject Index

instrumentation 35–80 basic principles 8–9, 35–8 future trends 354–6 general optical requirements 44–5 history of 11–12 imaging instruments 43 liquid handling systems 45 cuvettes 48–9, 232 flow cells 45–8, 229, 232 miniaturization 314–17 optical systems 38–9 angle scanning 40–1, 315 fan-shaped beam 39 fixed angle 41 grating coupler 21, 41–2 interferometers 43 resonant mirror 42, 62 wavelength interrogation 42 instruments reviewed 50, 51, 52 angle-scanning ESPRIT (Eco Chemie) 40, 59–60 SPRINGLE (Eco Chemie) 40, 59–60 Biacore instruments see Biacore fan-shaped beam BI-SPR (Biosensing Instruments) 50, 54 DKK-TOA SPR-20 (Tacadanobaba) 52, 55 Plasmonic Biosensor (Wallenfels) 53, 55 SensiQ (Nomadics) 52, 54 SR7000 DC (Reichert) 52, 55 fixed-angle β-SPR (Sensia) 56, 57 BIOSUPLAR-321 (Sinzing) 56, 57 K-MAC SPRi/SPRLAB (Korea Materials) 57–8 Moritex (Myutron) 54, 56 Multiskcop of Optrel GBR (Kleinmachnow) 56 Nanofilm EP3 (Göttingen) 58–9 Resonant Probes SPTM (Goslar) 54 imaging instruments GenOptics 64 IBIS iSPR (IBIS Technologies) 67–8, 69, 228–30

399 LFIRE (Maven Biotechnologies) 64, 66 MultiSPRinter (Toyobo) 64, 66, 67 Plasmon Imager (Graffinity Pharmaceuticals) 67, 68 Proteomic Processor (Lumera) 65, 66 ProteOn XPR36 (BioRad Laboratories) 64, 66 SPRi-Lab+ (GenOptics) 64, 65 SPRi-Plex (GenOptics) 64, 65 SPRimager II (GWC Technologies) 63 other systems IAsys Neosensors (Sedgefield) 61–2 SPR 100 module (Thermo Electron Corp.) 60–1 integrated microfluidic cartridge 465 interaction arrays 363 interference filter 315 interferometers 43 intermolecular bivalent binding 147–52 ionic immobilisation 210, 211 kobs kinetic analysis 130–1 kinetics determination of parameters 5–6, 7–8 instrumentation for 43, 45 kinetic models see adsorption kinetics; ligand–receptor interactions surface enzyme 254–7 Kramers–Kronig relation 29 Kretschmann configuration 2, 9, 10, 20, 279 fluorescence spectroscopy 279, 304, 305, 307 lab-on-a-chip 354, 362, 367, 378 biomarker imaging 369–73 electroosmotic microfluidic devices 367–9 labeling 174–5, 225 Lactobacillus plantarum 344 Langmuir absorption model 282–4 see also adsorption kinetics; ligand–receptor interactions

400

lateral resolution 360 ligand immobilization 125–6, 173, 174, 175, 194–5, 221–2 adsorptive methods 195 covalent coupling 195 amine coupling through reductive amination 202–3 amine coupling via reactive esters 199–201 epoxy-mediated 208–11 hydrazide activated aldehydes 207–8 thiol coupling 203–7 directed (biotinylated) methods 212–13 electrostatic methods 210–11 preconcentration 195–7 and kinetic analysis 180–1 membrane protein immobilization 213–15 photoactivation of surfaces 378–9 preconcentration procedures 195 dry immobilization 197–9 electrostatic 195–7 summary of strategies 214–15, 216–17 ligand–receptor interactions 4–5, 123–71, 180–1 affinity constants from equilibrium signals 126–8 from kinetic analysis 129–34 in solution vs at the surface 154–8 complex binding models 143 conformational change 144–7 intermolecular bivalent bonding 147–52 drug research 123–4, 165–7 mass transport limitation see transport-limited interactions rate constants 129–31 thermodynamics binding constants in solution 154–8 Eyring transition states 163–5 SPR compared to calorimetry 161–3 van’t Hoff analysis 158–61 ligation chemistry 259–60 light harvesting complex 299–300 limit of detection 232–5 line spotter 365

Subject Index

lipid bilayers 214 liquid handling systems see under instrumentation long-range surface plasmons 251, 253, 254 and fluorescence spectroscopy 303–6 maleimide coupling 205–6 manufacturers see instruments reviewed mass spectrometry 357–9 mass transport/transfer 87–90, 178–80, 184 transfer in biosensors 90–8, 295 see also transport-limited interactions matrix sites 82 meat 350 membrane protein immobilization 213–15 mercaptoalkyls 190 metal substrates 182 micelles, mixed 213 microarray imaging 43, 67, 269, 362–3 fluorescence spectroscopy 306–8 instrumentation and surface chemistry 247–54 kinetic and thermodynamic parameters 251 long-range surface plasmons 251, 253, 254 microchannel flow cell 251, 252 surface probes/target molecules summarized 249 nanoparticle-amplified sensing 260–2 microRNA detection 264–9 SNP genotyping 262–4 optimizing clinical performance 317–23 spotting on gold 362–4 DNA coding technology 364–6 surface enzymatic enhancement 247, 254, 264–9 enzyme kinetics 254–7 RNA microarrays with ligation 259–60 RNase H-amplification of DNA 257 see also immunoassays

401

Subject Index

microfluidic devices 251, 252, 359, 361–2 concentration gradient immunoassay 379–80 digital microfluidics/electrowetting 373–6 electroosmotic flow 367–9 free flow electrophoresis 369–73 integrated cartridge 465 line spotter 366 microRNA detection 264–9 milk 347–9 miniaturization 314–17 see also lab-on-a-chip molecular interactions 174–7 monoclonal antibody screening 375–6 multi-layered systems, analysis of 27–8 multilayer adsorption 105, 113–15 multiplex analysis cycles 228–34 N-hydroxysuccinimide see EDC–NHS nanoarchitecture 174, 187–8 three-dimensional hydrogels 191–4 two-dimensional surfaces 189–91 nanoparticles 29–31, 360–1 amplified DNA detection 260–2 microRNA detection 264–9 NHS see EDC–NHS nitrilotriacetic acid 210, 211 non-specific binding 174, 175, 183–5, 327 NTA 210, 211 oligonucleotides 210, 286–98, 364 and gene-modified organisms 351 optical systems see under instrumentation optogels 10, 44, 356 oscillatory flow 362 p-polarized light 2, 17–18, 21, 44 pantothenic acid 344 partition processes 99 passivation layer see adhesion linking penicillin 340, 342, 347 pharmaceutical research 123–4, 165–7 phase jump 22–3 phenytoin 325, 326

photoactivation 378–9 physics of SPR 15–33, 86–7 planar flow cell 46–7 plasma deposition 183 plasmons see surface plasmons plastic substrates 183, 195 platinum 182 point-of-care diagnostics see clinical diagnostics polarization control 317 polycarboxylate hydrogel 192 poly(dimethylsiloxane) 329 flow cell 252, 252 line spotter 365 microarray fabrication 364 polymer brushes 277, 298, 376–8 portable SPR see clinical diagnostics preconcentration methods 195–9 protein A-modified surfaces 213 protein binding fluorescence spectroscopy 297, 298–301 see also immunoassays; ligand– receptor interactions protein microarrays 362–3 see also microarray imaging proteins compatible matrix polymers 186–7 coupling reactions 199–207 disulfide reduction 204 immobilization 213–15 milk assays 348–9 non-specific absorption 184–5 proteomics 357 lab-on-a-chip 369–73 Qflex kit 334, 335 quantum dots 306, 308 ractopamine 343 Raman spectroscopy 30, 31 rate constants 7–8, 84 model for distribution analysis 383–5 reaction rate 87–90 REBIND software 137–8 rebinding 137–41, 226

402

receptor–ligand interactions see ligand–receptor reductive amination 202–3 reflectance 18, 29 reflectivity change 36, 38, 45 refraction and evanescent wave 16–17, 86–7 SPR principles 2–3, 222–3 refractive index bulk compensation 323–4 optical matching 44, 356 resolution 317–20 temperature dependence 44–5, 321–2 regeneration 5, 6, 175, 176 solution 225, 338 resolution lateral 360 refractive index 317–21 resonance 2, 18, 279–80, 281 resonant mirror measurements 42, 62 rheumatoid arthritis 235–43 auto-immune antibody complexes 357–9 riboflavin 344 RNA microarrays 259–60 RNA–DNA heteroduplexes 254–60 RNase H amplified detection of DNA 257–8 and surface enzyme kinetics 254–7 Salmonella 350 SAMs 182, 190, 191, 277, 300 sandwich assay 227, 228 saturation, surface 104–12, 105 scanning probe microscopy 360 scanning SPR see microarray imaging Scheimpflug condition 319, 320–1 Schiff bases 202 selective capture 3 self assembled monolayers 182, 190, 191, 277, 300 sensitivity, optimization of 26–7 sensor chips applications and surface structure 188 development/history 1, 9–12 elements of 173–4, 181

Subject Index

nanoarchitecture 174, 187–9 three-dimensional hydrogels 191–4 two-dimensional surfaces 189–91 optimal surface selection 177–81 polymer brushes 277, 298, 376–8 surface interactions 174–7 see also adhesion linking layers; bioinert matrices; ligand immobilization sensorgrams 3, 5, 6, 223 types/presentation of dip 36–8 SERS 30 serum antibodies 235–42, 357–9 silanes 182–3 silver layer 26–7, 44 adhesion linking layer 182 and coherence length 23–4 single nucleotide polymorphisms 262–4 small molecules 7, 193, 324–7 smart polymer brushes 376–8 spotting, microfluidic 362–6 SPREETA chip 314, 366 SPRI see microarray imaging stagnant layer 95–7, 178 streptavidin-modified surfaces 212 streptomycin 340, 342, 346 substrate, chip 173, 182–3, 195 photoactivatable 378–9 sulfadiazine 340, 341–2 sulfamethazine 340, 341–2 sulfathiazole 340, 346 sulfonamides 340, 341–2, 346 surface chemistry see sensor chip surface competition assay 337, 338 surface conditioning 4 surface enzymatic enhancement 247, 254, 264–5 surface plasmon fluorescence spectroscopy see fluorescence spectroscopy surface plasmon resonance (overview) 1–13, 221–4, 275–6 assay 3–6 calibration curve 6–7 dip to real time measurement 3–4 experimental parameters 26–7

403

Subject Index

history and biosensor development 1, 9–12 instrumentation 8–9 kinetic parameters 7–8 multi-layered systems 27–8 surface plasmons 15–16 analysis of multi-layered systems 27–8 coherence length 23–4 dispersion relation 17–19, 20 enhanced fluorescence and absorbance 28–9 field enhancement 21–2 long-range 251, 253, 254 and nanoparticles 29–31 phase jump 22–3 resonance 18 surface-enhanced Raman spectroscopy 30 symbols 32–3, 118–19, 168 T4 DNA ligase 259–60 temperature stabilization/compensation 44–5, 322–3 thermodynamics affinity in solution vs at the surface 154–8 compared to calorimetry 161–3 Eyring plots 163–5 van’t Hoff analysis 158–61

thioethers 182 thiols 182 coupling reactions 203–7 tilted image plane 320–1 transition state analysis 163–5 transport see mass transport transport-limited interactions 382 adsorption kinetics 87–90, 284 detection and modeling 135 assay of high off-rates 141–2 association and viscosity 135–7 competing ligand 139–41 dissociation rebinding model 137–9 quantitative considerations 142–3 ligand–receptor affinity/kinetics 125–35 oligonucleotide hybridization 296–7 Tylosin 343 van’t Hoff analysis 158–61 veterinary drugs 333, 339–46 viscosity and association 135–7 vitamins 340, 343–4 Volmer–Weber growth 113, 114 Vroman sequence 184 wall-jet flow cell 47 wavelength interrogation 42 wavelength tuning 315 wavevectors/wave equations 16–17