Nuclear Receptors: A Practical Approach

Nuclear Receptors

The Practical Approach Series SERIES EDITOR B. D. HAMES Department of Biochemistry and Molecular Biology University of Leeds, Leeds LS2 9JT, UK

See also the Practical Approach web site at http://www.oup.co.uk/PAS * indicates new and forthcoming titles

Affinity Chromatography Affinity Separations Anaerobic Microbiology Animal Cell Culture (2nd edition) Animal Virus Pathogenesis Antibodies I and II Antibody Engineering if Antisense Technology Applied Microbial Physiology Basic Cell Culture Behavioural Neuroscience Bioenergetics Biological Data Analysis Biomechanics - Materials Biomechanics - Structures and Systems Biosensors Carbohydrate Analysis (2nd edition) Cell-Cell Interactions The Cell Cycle Cell Growth and Apoptosis

if Cell Separation Cellular Calcium Cellular Interactions in Development Cellular Neurobiology * Chromatin if Chromosome Structural Analysis Clinical Immunology Complement if Crystallization of Nucleic Acids and Proteins (2nd edition) Cytokines (2nd edition) The Cytpskeleton Diagnostic Molecular Pathology I and II DNA and Protein Sequence Analysis DNA Cloning 1: Core Techniques (2nd edition) DNA Cloning 2: Expression Systems (2nd edition) DNA Cloning 3: Complex Genomes (2nd edition)

DNA Cloning 4: Mammalian Systems (2nd edition) * DNA Microarrays * Drosophila (2nd edition) Electron Microscopy in Biology Electron Microscopy in Molecular Biology Electrophysiology Enzyme Assays Epithelial Cell Culture Essential Developmental Biology Essential Molecular Biology I and I * Eukaryotic DNA Replication Experimental Neuroanatomy Extracellular Matrix Flow Cytometry (2nd edition) Free Radicals Gas Chromatography Gel Electrophoresis of Nucleic Acids (2nd edition) if Gel Electrophoresis of Proteins (3rd edition) Gene Probes 1 and 2 Gene Targeting Gene Transcription * Genome Mapping Glycobiology if Growth Factors and Receptors Haemopoiesis * High Resolution Chromotography Histocompatibility Testing HIV Volumes 1 and 2

* HPLC of Macromolecules (2nd edition) Human Cytogenetics I and II (2nd edition) Human Genetic Disease Analysis * Immobilized Biomolecules in Analysis Immunochemistry 1 Immunochemistry 2 Immunocytochemistry if In Situ Hybridization (2nd edition) lodinated Density Gradient Media Ion Channels if Light Microscopy (2nd edition) Lipid Modification of Proteins Lipoprotein Analysis Liposomes Mammalian Cell Biotechnology Medical Parasitology Medical Virology * MHC Volumes 1 and 2 * Molecular Genetic Analysis of Populations (2nd edition) Molecular Genetics of Yeast Molecular Imaging in Neuroscience Molecular Neurobiology Molecular Plant Pathology I and II Molecular Virology Monitoring Neuronal Activity Mutagenicity Testing * Mutation Detection

Neural Cell Culture Neural Transplantation Neurochemistry (2nd edition) Neuronal Cell Lines NMR of Biological Macromolecules Non-isotopic Methods in Molecular Biology Nucleic Acid Hybridisation Oligonucleotides and Analogues Oligonucleotide Synthesis PCR 1 PCR 2 Hybridization Peptide Antigens Photosynthesis: Energy Transduction Plant Cell Biology Plant Cell Culture (2nd edition) Plant Molecular Biology Plasmids (2nd edition) * Platelets Postimplantation Mammalian Embryos * Post-translational Modification

ir

* *

*

Preparative Centrifugation Protein Blotting Protein Expression Protein Engineering Protein Function (2nd edition) Protein Phosphorylation Protein Purification Applications Protein Purification Methods Protein Sequencing Protein Structure (2nd edition) Protein Structure Prediction Protein Targeting Proteolytic Enzymes Pulsed Field Gel Electrophoresis RNA Processing I and II RNA-Protein Interactions Signalling by Inositides Steroid/Nuclear Receptor Superfamily Subcellular Fractionation Signal Transduction Transcription Factors (2nd edition) Tumour Immunobiology

Nuclear Receptors A Practical Approach Edited by

DIDIER PICARD Department of Cell Biology, University of Geneva, Geneva

OXPORD UNIVERSITY PRESS

OXPORD UNIVERSITY PRESS

Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford and furthers the University's aim of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Athens Auckland Bangkok Bogota Buenos Aires Calcutta Cape Town Chennai Dar es Salaam Delhi Florence Hong Kong Istanbul Karachi Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Paris Sao Paulo Singapore Taipei Tokyo Toronto Warsaw and associated companies in Berlin Ibadan Oxford is a registered trade mark of Oxford University Press Published in the United States by Oxford University Press Inc., New York © Oxford University Press 1999 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press. Within the UK, exceptions are allowed in respect of any fair dealing for the purpose of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act, 1988, or in the case of reprographic reproduction in accordance with the terms of licenses issued by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms and in other countries should be sent to the Rights Department, Oxford University Press, at the address above. This book is sold subject to the condition that it shall not, by way of trade or otherwise, be lent, re-sold, hired out, or otherwise circulated without the publisher's prior consent in any form of binding or cover other than that in which it is published and without a similar condition including this condition being imposed on the subsequent purchaser Users of books in the Practical Approach Series are advised that prudent laboratory safety procedures should be followed at all times. Oxford University Press makes no representation, express or implied, in respect of the accuracy of the material set forth in books in this series and cannot accept any legal responsibility or liability for any errors or omissions that may be made. A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data nuclear receptor : a practical approach / edited by Didier Picard. (The practical approach series ; 207) Includes bibliographical references and index. 1. Steroid hormones - Receptors. 2. Nuclear receptors (Biochemistry) 3. Transcription factors. I. Picard, Didier. II. Series. QP572.S7S757 1999 573.4'48845 - dc21 99-24782 ISBN 0 19 963742 3 (Pbk) ISBN 0 19 963743 1 (Hbk) Typeset by Footnote Graphics, Warminster, Wilts Printed in Great Britain by Information Press, Ltd, Eynsham, Oxon.

Preface This book has been written about a big family for a big family. The former is the ever-growing superfamily of transcription factors referred to here as the steroid/nuclear receptors. The latter is the large number and variety of both basic and clinical researchers who are interested in a member of this superfamily. The first contact with this superfamily may be made for many different reasons. One may want to investigate: (a) The molecular, developmental, or physiological aspects of signalling by ligands such as steroids, thyroid hormone, and retinoic acid. (b) The basic molecular mechanisms of transcription or signal transduction. (c) The transcriptional regulation of a target gene of a member of the superfamily. (d) The molecular basis of human diseases. Thus, as in other families, scientists join this family for love or reason, by chance or on purpose. When they do, they need many of the tools of the trade. And they need new tools even if they have been 'married' for a long time. We are a fairly homogeneous group when it comes to agreeing on a common set of tools, which made it a pleasurable and straightforward task to assemble the book that you hold in your hands. Old and new members of this scientific community should therefore find in here many, indeed almost all, tools that are specific to this trade. While we agree on the approaches (the protocols!) and on which factors belong to the superfamily, we disagree on the family name. Simply put, this is because all of the names that have been used so far are inaccurate. They either implicitly exclude certain members or include members of other superfamilies. The early term 'steroid/thyroid receptor superfamily' was fine until receptors for other types of ligands were discovered. The use of 'nuclear receptors' tends to ignore the fact that several steroid receptors can be cytoplasmic proteins in the absence of ligand. The term 'intracellular receptors' would take care of this problem. However, both 'nuclear' and 'intracellular' are also applicable to other receptors such as the dioxin receptor that are not members of this superfamily. 'Ligand-regulated transcription factors' suffers from the same problem. In addition, the terms 'ligand-regulated' and 'receptors' erroneously suggest that all members are receptors (for known or yet to be discovered ligands) and that they can only be regulated by ligands. For those of us already in the field, this confusion is not really a problem, but for someone outside of it, it probably is. In trying to make clear what this book is about, we decided on a compromise. Perhaps, by the time this book will be re-edited, the issue will have been solved.

Preface Beyond picking a title, any book is a lot of work and only possible with the help of many people. I wish to thank Oxford University Press for gently pushing me to accept this task and the contributing authors for generously giving time and doing a great job. I am very grateful to my laboratory and my family for understanding the passion, the worries, and the time constraints of an editor. D. P.

Geneva October 1998

viii

Contents List of Contributors Abbreviations 1. Evolutionary biology of the nuclear receptor superfamily

xv xvii 1

H. Escriva, M. Robinson, and V. Laudet 1. Introduction

1

2. Molecular phylogeny of nuclear receptors General overview Sequence alignments Tree reconstruction methods Interpretation of sequence trees

3 3 4 7 13

3. Cloning of new receptor genes from divergent animal species Overview of the strategy Designing PCR primers PCR isolation of nuclear receptor gene fragments RT-PCR with degenerate oligonucleotides

19 19 20 22 27

References

28

2. Initial characterization of new orphan receptors

29

R. Sladek and V. Giguere 1. Introduction

29

2. Sources of receptor protein for DNA binding studies Protein expression in bacteria Protein expression in vitro Protein expression in eukaryotic cells

31 32 35 36

3. Identifying the DNA binding mode: monomers and dimers GST-pulldown Mammalian two-hybrid system

40 40 43

4. Identifying hormone-response elements Characterizing DNA binding proteins using the electrophoretic mobility shift assay Identifying response elements using consensus binding sites Using EMSA to detect RXR heterodimers and receptor homodimers

45 45 52 53

Contents Determining the polarity of RXR heterodimers using chimeric receptors Identifying response elements using SAAB selection Identifying DNA-protein contacts using the methylation interference assay

54 57 60

5. Transcriptional activation assays: transient and stable transfection

63

References

68

3. Adopting orphans: finding ligands

71

5. Kersten and W. Wahli 1. Introduction

71

2. First step: guessing ligand classes

71

3. Tissue fractionation

72

4. Transactivation assays Standard transactivation assay Gal4-LBD fusion transactivation system Quantitation of transactivation assays

74 74 76 77

5. Ligand binding assays Competition assay

77 77

6. Monitoring changes in receptor conformation Protease sensitivity Non-denaturing gel electrophoresis

80 80 82

7. Co-activator-dependent receptor ligand assay (CARLA)

83

8. Fluorescence spectroscopy Principles and utility of fluorescence spectroscopy Emission scans Fluorescence titrations

86 87 89 90

9. Conclusions

91

References

92

4. Kinetic analysis of nuclear receptor interactions

95

B. J. Cheskis and L. P. Freedman 1. Introduction

95

2. Analysis of nuclear receptor functions using surface plasmon resonance (SPR) technology The basic principle of interaction analysis with BIAcore

98 98

x

Contents Overexpression and purification of nuclear receptors for kinetic and equilibrium analyses Analysis of protein-protein interactions Analysis of protein-DNA interactions Kinetic analysis

3. Conclusion References

99 103 106 114

116 116

5. Functional characterization of co-activators using mammalian cell microinjection

119

D. W. Rose, T.-M. Mullen, M. G. Rosenfeld, and C. K. Glass 1. Introduction

119

2. Equipment and materials Equipment Antibodies Choice of cells Preparation of cells Preparation of samples to be injected Choice of DNA to inject

120 120 120 124 125 126 126

3. Microinjection technique Basic microinjection technique Monitoring expression Analysis of results

129 129 131 134

References

135

Why microinjection?

6. Analysis of steroid/nuclear receptor phosphorylation

119

137

B. G. Rowan and N. L. Weigel 1. Introduction

137

2. Phosphorylation of the nuclear receptor super family

137

3. General guidelines for studying nuclear receptor phosphorylation

138

4. Identification of phosphorylation sites in nuclear receptors Approaches to identify phosphorylation sites in nuclear receptors and other proteins Conventional method Approach for use with low protein levels

xi

140 140 141 153

Contents Additional experiments to provide more sequence information Indirect method

157 158

5. Identifying candidate kinases Introduction In vitro phosphorylation of nuclear receptors

158 158 159

6. Assaying the functional significance of selected phosphorylation sites Introduction Site-directed mutagenesis of nuclear receptors

162 162 163

References

163

7. Ligand- and cofactor-regulated transcription with chromatin templates 167 W. L. Kraus and J. T. Kadonaga 1. Introduction

167

2. Synthesis and purification of recombinant nuclear hormone receptors and cofactors Choice of expression systems Purification of recombinant receptors and cofactors Functional analyses of purified receptors and cofactors

167 167 168 173

3. In vitro chromatin assembly Chromatin assembly systems Assembly of plasmid DNA templates into chromatin Analysis of chromatin templates assembled in vitro Mock chromatin assembly of plasmid DNA templates

173 173 173 176 177

4. In vitro transcription of chromatin templates In vitro transcription systems Ligand- and cofactor-regulated in vitro transcription with chromatin templates Chromatin versus non-chromatin transcription experiments Single versus multiple round transcription experiments

179 179 179 185 186

5. Exploiting ligand- and cofactor-regulated in vitro transcription systems

188

Acknowledgements

188

References

188

8. Hormone-resistance syndromes

191

M. Gurnell and V. K. K. Chatterjee

1. Introduction

191

xii

Contents 2. Receptor analysis Gene sequencing cDNA sequencing Genotyping by restriction analysis Haplotype analysis of kindreds with identical mutations

192 192 194 197 198

3. Ligand binding and DNA binding studies Site-directed mutagenesis to generate mutated receptor cDNA Synthesis of mutant receptor protein and ligand binding assays DNA binding properties of mutant receptors

201 201 201 204

4. Assaying mutant receptor function by transient transfection 205 13-dependent transactivation 205 Dominant negative inhibition by mutant receptors 206 5. Interaction of mutant receptors with cofactors

209

References

210

9. In vitro assembly of steroid receptor-Hsp90 complexes 213 D. F. Smith 1. Introduction Background Proteins associating with unactivated steroid receptors Basic pathway for assembly of steroid receptor complexes

213 213 214 214

2. Receptor isolation Immunoaffinity purification In vitro expression to produce radiolabelled receptor

217 217 219

3. Assembly reactions in RL Lysate sources and considerations Standard assembly conditions and further considerations

219 219 220

4. Analysis of assembly reactions Identification of assembly components Analysis of assembly dynamics Inhibition of specific assembly steps and chaperone components

222 222 225 225

5. Assembly of receptor complexes in a purified system

227

References

229

10. Yeast as a system for the study of nuclear receptor function

233

M. J. Garabedian, J. A. Ihiguez-Lluhi, and A. Kralli 1. Introduction

233 xiii

Contents 2. Why study nuclear receptors in yeast? 3. Expression of nuclear receptors in yeast 4. Reporter genes The colorimetric reporter, LACZ Reporter genes for positive and negative selection 5. Yeast host strains 6. Mutagenesis and screens for receptor mutations High density random mutagenesis using error-prone PCR 7. Strategies for identifying cellular factors that affect nuclear receptor function Identification of gene products affecting NR function using yeast knockout strains Identifying components in NR signal transduction through random mutagenesis of the yeast genome Identifying components in NR signal transduction through dosage suppression analysis 8. Rescuing plasmids from yeast 9. Preparation of protein extracts from yeast for Western blotting Acknowledgements References

11. Regulation of heterologous proteins by fusion to a hormone binding domain

233 235 236 237 242 243 244 245 247 247 249 252 254 255 256 256

261

D. Picard 1. Introduction 2. Key features Mechanism Practical implications of the proposed mechanism 3. Problems and troubleshooting Potential complications Unexpected behaviour Troubleshooting 4. Application to yeast Acknowledgements References

Appendix I Appendix II Index

261 262 266 266 267 267 268 268 268 271 271

275 277 283 xiv

Contributors V. K. K. CHATTERJEE

Department of Medicine, University of Cambridge, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK. B. J. CHESKIS

Department of Nuclear Receptors, Women's Health Research Institute, Wyeth-Ayerst Research, 145 King of Prussia Road, Radnor, PA 19087, USA. H. ESCRIVA

UMR 49 du CNRS, Ecole Normale Superieure de Lyon, 46 Allee d'Italie, 69364 Lyon Cedex 07, France. L. P. FREEDMAN

Program in Cell Biology and Genetics, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA. M. J. GARABEDIAN

Department of Microbiology, Room MSB 237, NYU Medical Center, 550 First Avenue, New York, NY 10016, USA. V. GIGUERE

Molecular Oncology Group, McGill University Health Centre, 687 Pine Avenue W, Montreal H3A 1A1, Canada. C. K. GLASS

University of California at San Diego, Department of Medicine, Cellular and Molecular Medicine, 9500 Gilman Drive, La Jolla, CA 92093-0658, USA. M. GURNELL

Department of Medicine, University of Cambridge, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK. J. A. INIGUEZ-LLUHI

Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, CA 91143-0450, USA. J. T. KADONAGA

Department of Biology, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0347, USA. S. KERSTEN

Institut de Biologie animale, Batiment de Biologie, Universite de Lausanne, 1015 Lausanne, Switzerland.

Contributors A. KRALLI

Department of Biochemistry, Biozentrum, University of Basel, 4056 Basel, Switzerland. W. L. KRAUS

Department of Molecular Biology and Genetics, Cornell University, 465 Biotechnology Building, Ithaca, NY 14853, USA. V. LAUDET

UMR 49 du CNRS, Ecole Normale Superieure de Lyon, 46 Allee d'ltalie, 69364 Lyon Cedex 07, France. T.-M. MULLEN

University of California at San Diego, Department of Medicine, Molecular Pathology Graduate Program, 9500 Gilman Drive, La Jolla, CA 92093-0673, USA. D. PICARD Departement de Biologie Cellulaire, Universite de Geneve, Sciences 3, 1211 Geneve 4, Switzerland. M. ROBINSON

UMR 49 du CNRS, Ecole Normale Superieure de Lyon, 46 Allee d'Italie, 69364 Lyon Cedex 07, France. D. W. ROSE University of California at San Diego, Department of Medicine, Whittier Diabetes Program, 9500 Gilman Drive, La Jolla, CA 92093-0673, USA. M. G. ROSENFELD

University of California at San Diego, Department of Medicine, Howard Hughes Medical Institute, 9500 Gilman Drive, La Jolla, CA 92093-0658, USA. B. G. ROWAN Department of Cell Biology, Baylor College of Medicine, 1 Baylor Plaza, Houston, TX 77030, USA. R. SLADEK

Molecular Oncology Group, McGill University Health Centre, 687 Pine Avenue W, Montreal H3A 1A1, Canada. D. F. SMITH

Department of Biochemistry and Molecular Biology, Mayo Clinic Scottsdale, 13400 East Shea Blvd., Scottsdale, AZ 85239, USA. W. WAHLI

Institut de Biologie animale, Batiment de Biologie, University de Lausanne, 1015 Lausanne, Switzerland. N. L. WEIGEL Department of Cell Biology, Baylor College of Medicine, 1 Baylor Plaza, Houston, TX 77030, USA.

Abbreviations AA AAV ADH AF-2 AR 3-AT BSA CARLA CAT cER CMV cPR DBD Dex DHT DMEM DMF DMS dNTPs DTT E2 EMS EMSA ER ERE ERR FKBP 5-FOA GA GPD GR GRE GST hAR hER hPR HBD HIV HRE HspVO

amino acids adeno-associated virus alcohol dehydrogenase activation function 2 androgen receptor 3-aminotriazol bovine serum albumin co-activator-dependent receptor ligand assay chloramphenicol acetyltransferase chicken ER cytomegalovirus chicken PR DNA binding domain dexamethasone 5a-dihydrotestosterone Dulbecco's modified Eagle's medium dimethylformamide dimethylsulfate deoxyribonucleoside 5'-triphosphates dithiothreitol 17p-oestradiol ethylmethane sulfonate electrophoretic mobility shift assay oestrogen receptor oestrogen-response element oestrogen receptor-related receptor FK506 binding protein 5-fluoro-orotic acid geldanamycin glyceraldehyde-3-phosphate dehydrogenase glucocorticoid receptor glucocorticoid-responsive element glutathione-5-transferase human AR human ER human PR hormone binding domain human immunodeficiency virus hormone-response element heat-shock protein 70

Abbreviations Hsp90 heat-shock protein 90 HTLV-I human T cell leukaemia virus I ICI ICI-182 780 IPTG isopropyl-p-D-thiogalactopyranoside LED ligand binding domain LUC luciferase LXR liver X receptor MBP maltose binding protein MED modified manual Edman degradation m.o.i. multiplicity of infection MR mineralocorticoid receptor NP-40 Nonidet P-40 NR nuclear receptor OD optical density OHT 4-hydroxytamoxifen ONPG o-nitrophenyl-B-D-galactopyranoside ORF open reading frame PAGE polyacrylamide gel electrophoresis PBS phosphate-buffered saline PCIAA phenol:chloroform:isoamyl alcohol PCR polymerase chain reaction PGK phosphoglycerate kinase PMSF phenylmethylsulfonyl fluoride PPAR peroxisome proliferator activated receptor PR progesterone receptor PvOH polyvinyl alcohol RAR retinoic acid receptor RFLP restriction fragment length polymorphism RL reticulocyte lysate rNTPs ribonucleoside 5'-triphosphates RTH resistance to thyroid hormone RT-PCR reverse transcription PCR RXR retinoid X receptor SAAB selected and amplified binding sequence TFA trifluoroacetic acid TPCK n-tosyl-L-phenylalanine chloromethyl ketone TPR tetratricopeptide repeat TR thyroid receptor tRA all-trans retinoic acid TRE thyroid hormone-response element UAS upstream activating sequence VDR vitamin D receptor X-Gal 5-bromo-4-chloro-3-indolyl-B-D-galactopyranoside XVlll

1 Evolutionary biology of the nuclear receptor superfamily H. ESCRIVA, M. ROBINSON, and V. LAUDET

1. Introduction The study of the interplay between gene regulation, embryonic development, and the evolution of animal forms is receiving a growing interest. Numerous authors have studied the Hox gene complexes which control, in all metazoan phyla, the establishment of antero-posterior polarity of organisms. By deciphering the phylogenetic relationships between individual genes of the Hox complex, as well as through isolation of homologues of Hox genes in a wide variety of organisms, from Hydra to various insects and vertebrates, the major role played by these genes in the evolutionary process is under intense scrutiny (reviewed in ref. 1). But Hox genes are not the only players in this game, and other gene families are implicated in the evolution of the animal body plan. In fact one of the major results of these studies is the emergence of the concept of evolutionarily conserved gene families, which play comparable roles in very different animal phyla. Among gene families, the nuclear hormone receptors are particularly useful and the study of their role in evolution will undoubtedly be recognized in the future. Indeed, nuclear receptors (NRs) appear as a fruitful model for an evolutionary approach for several reasons: (a) In many cases NRs are, as indicated by their names, ligand-activated transcription factors. This means that the study of a given receptor across evolution not only gives information about the function of this protein but also about the hormonal system that it controls. For example, the study of steroid receptors in sharks may provide information on the early steps of the steroid signalling in vertebrates, and may lead to a better understanding of how the five major classes of steroid hormones in human (namely oestrogens, glucocorticoids, progesterone, mineralocorticoids, and androgens) may have originated. Thus, as for the Hox genes, the study of a nuclear receptor in evolution provides useful information on the evolution of the organisms themselves.

H. Escriva, M. Robinson, and V. Laudet (b) The impressive chemical diversity of nuclear receptor ligands. The fact that more than the half of known NRs (the so-called 'orphan' receptors) have no known ligand remains a very intriguing question. How could such a diversity arise during evolution? How could shifts between receptor types occur? The NRs provide a unique model, that allows the functional diversification of a gene family during evolution to be studied. (c) As for most other gene families, the nuclear receptor superfamily has evolved by means of a large number of gene duplication events. In particular, it has been shown that during the chordate-vertebrate transition the nuclear receptor superfamily, as other gene families, underwent several major events of gene duplication (2, 3). Using the functional information known for these molecules in vertebrates such as Xenopus, chicken, or mouse, it is thus possible to study the role that these gene duplication events may have played in the appearance and diversification of vertebrates. (d) The NRs offer several unique methodological advantages that make them very efficient models for evolutionary studies. In contrast to most other transcription factors, they contain two conserved functional domains, namely the DNA binding domain (DBD), known as the C domain, and the ligand binding domain (LBD), known as the E domain (see also Chapter 2). This structure is useful for two reasons. First, this allows sequence alignments and phylogenetic analyses to be performed using a relatively high amount of information. In most cases, DNA binding domains are the only conserved part in transcription factors. These domains are usually short (less than 100 amino acids) and strongly conserved, and thus contain a relatively weak amount of phylogenetic information. In contrast, if the DNA binding domain of NRs is indeed short (c. 80 amino acids) and well conserved (50-90%), the ligand binding domain is long (180-200 amino acids) and much more variable (25-80%). The combined study of both domains generates a high amount of phylogenetic information, which results in robust phylogenetic trees. Secondly, the existence of two conserved domains allows specific PCR primers to be designed in order to perform PCR or RT-PCR experiments, and to isolate relatively large fragments that can be used directly for the determination of the expression pattern or for sequence analysis. The aim of this chapter is to describe the basic knowledge necessary to perform evolutionary studies on NRs. Our aim is to provide the reader with the minimal amount of information necessary to carry out these analyses, and also to discuss laboratory protocols as well as relevant practical aspects that may help in such an approach.

2

1: Evolutionary biology of the nuclear receptor superfamily

2. Molecular phylogeny of nuclear receptors 2.1 General overview The rigorous analysis of historical information contained in nucleotide or amino acid sequences requires a minimum of knowledge of molecular phylogeny. Throughout this section we have tried to reduce as much as possible these general concepts in order to focus on the application of these methods to the case of NRs. Nevertheless, we have to emphasize that, in too many papers, sequence analyses (and the important conclusions drawn from them) are performed without precautions and without knowledge of the basic assumptions linked to these kinds of analyses. We recommend reading at least one general textbook, such as Fundamentals of molecular evolution (4). The construction of any phylogenetic tree will always comprise the steps described in Protocol 1. Of course such a protocol has to be considered only as a basic framework in which several steps can be modified but are nevertheless necessary.

Protocol 1. Overview of the phylogenetic tree reconstruction procedure 1. Carefully select the sequences that will be studied. Avoid the inclusion of redundant sequences such as alternative splicing variants of a gene as well as the inclusion of incomplete or too short sequences. 2. Perform the alignment either manually, if the identity level is high enough, or using computer programs if manual alignment is not possible, due to low sequence identity (see Section 2.2.3). 3. Carefully select a tree reconstruction method. The main choice is between parsimony, maximum likelihood, and distance methods. The latter seems to us best adapted to the type of studies described here (see Section 2.3.5). 4. For a distance analysis compute the distance matrix between all pairs of sequences. It is important to choose an adapted correction for multiple substitution, as described in Section 2.3.1. 5. Run the program of tree reconstruction you have selected in step 3. For parsimony analysis the use of PAUP is standard, as is PHYLIP for maximum likelihood. For distance analysis the neighbour-joining (NJ) algorithm is implemented in many programs, and strongly recommended. Do not use the UPGMA method, which is easily misleading, especially in the deepest dichotomies in the tree. Most softwares that implement step 5 for NJ also implement step 4.

H. Escriva, M. Robinson, and V. Laudet Protocol 1. Continued 6. Test the robustness of your tree using the bootstrap resampling method (see Section 2.3.3). After obtaining the bootstrap value for each branch of the tree, collapse all the branches which are supported by less than 50% values. These branches are too weak to be discussed. 7. Print your tree. Pay attention to very long branches (not visible in parsimony), which may cause the classical long branch attraction artefact. Eventually exclude these sequences from the analysis. 8. Search for a possible root for your tree. If you have homologues of a given receptor in various species, check whether you obtained the known species phylogeny (mouse and rat sequences should be more closely related than mouse and chicken!). Look for gene duplication events. 9. Interpretation of the tree is now possible.

10 Homo RXRA Mus RXRA Gallus RXRA

Danio RXRA

20

30

TCA GGC AAG CAC TAT GGA GTG TAC AGC TGC GAG GGG A G ..A T . .T C T ..A ..T G C

G ..A ..T

T ..T

T ..A ..A

Figure 1. Example of a sequence alignment used in molecular phylogeny analyses. Each position of the sequence is a character.

2.2 Sequence alignments 2.2.1 Why do an alignment? In all cases a molecular phylogeny study will start by an alignment of the sequences to be studied. Indeed the basic principle of any phylogenetic analysis, based either on morphological or molecular characters, is to compare homologous structures in order to infer an evolutionary scenario from such a comparison. Homologous structures are generally defined as structures derived from a common ancestor. For molecular data, each character is a position in a sequence alignment, as shown in Figure 1. Both in molecular and in morphological data the delineation of homology is not always obvious. In molecular data, this delineation is done during alignment, by assuming insertion/deletion events of sites, which are thus deemed non-homologous. Those sites which are paired to other sites, not to gaps, are supposed to be homologous to them. In practice, this is done by minimizing the differences that may exist between these sequences. A frequent mistake is to speak about 'per cent homology' between sequences. For example, we may read that a given receptor exhibits '75%

1: Evolutionary biology of the nuclear receptor superfamily homology' with another one. This vocabulary is misleading since the concept of homology is defined as a common evolutionary origin, and thus homology cannot be fractionated. Two structures or two sequences are, or are not, derived from a common ancestor. It is impossible that they share only 75% of common evolutionary origin! This incorrect use of the term homology often renders difficult discussions about the relationships between two genes. Thus it is strongly recommended to use the term 'sequence identity' between two sequences, which is more accurate and precise. Two sequences exhibit 75% identity if, of 100 positions compared between these two sequences, 75 are identical. The term 'sequence similarity', also frequently used for amino acid sequences, can be misleading since it takes into account the conservative changes between amino acids. 2.2.2 Nucleotide or amino acid comparison? When dealing with protein coding genes, such as NRs, amino acid or nucleotide sequences can be used to reconstruct evolutionary history. There is no general answer to the question of which to use, since this depends of the type of problem to be solved. Nevertheless, in the case of NRs, it is preferable to use amino acid sequences for the following reasons: (a) When divergent receptors such as the TRs and the RXRs are compared, it is extremely difficult to perform a nucleotide alignment in the regions coding for the ligand binding domains. Thus, the study of the nucleotide sequence is restricted to closely related receptors such as TRs and RARs. (b) Even when homologous versions of the same receptor are used, in some cases it is difficult to construct a correct nucleotide alignment, since, due to differences in base composition of the genomes, there are often strong biases toward GC or AT richness that make alignment difficult. (c) Because of the degeneracy of the genetic code it is necessary to exclude from subsequent analysis the third position of each codon, which is by far too divergent. This process may be tedious, depending on software available. Even the first position of codons, which is variable for some amino acids, may prove to be difficult to study. Thus, except when precise information has to be obtained on the pattern and rate of nucleotide substitutions, and/or when closely related sequences are studied, we always perform our phylogenetic reconstructions using amino acid sequences. 2.2.3 Aligning The best possible alignment between two sequences is the one in which the number of mismatches and gaps is minimized. Unfortunately, reducing the

H. Escriva, M. Robinson, and V. Laudet number of mismatches usually results in an increase in the number of gaps, and vice versa. In many cases, when the level of sequence identity is high, finding the correct alignment is relatively easy. This is, for example, the case for the DNA binding domain, for which most receptors share at least 45% identity. In addition, this domain contains strictly conserved amino acids (the zinc-coordinating Cys residues), or frequently found sequences (such as the P-box), that help in doing the alignment. In other cases it is necessary to use computer programs in order to perform the alignment. Two types of programs exists: fully automatic ones, and others that provide a visual help for the alignment which is nevertheless done manually. We will not enter here into the details of relevant algorithms. The most common, and very satisfying, automatic alignment program is ClustalW (5), which is freely available and relatively easy to use. In most cases ClustalW gives accurate results. Nevertheless it can give erroneous alignments in some cases. One classical example is when two sequences contain a well conserved domain located in different positions inside the sequence. For example, if someone wants to align the glucocorticoid receptor (GR) with the vitamin D receptor (VDR) using ClustalW, the program may generate an alignment in which the two DBDs are not aligned. This is because the GR contains a very long A/B domain whereas the A/B domain of VDR contains only 23 amino acids. But the program found it more 'economical', in terms of penalty, to align artefactually the C domain of VDR with the A/B domain of GR. This creates a lot of short gaps, rather than to create a long gap in VDR, in order to locate correctly its DBD along the GR DBD. Other possible artefacts of that kind may involve the D domain, which is of variable size. Once again the program may generate incorrect alignments in which the various LBDs, which are also relatively divergent, are not correctly aligned. Thus, it is critical to carefully check the results obtained with ClustalW (or any other program), and in most cases to make manual corrections on the proposed alignment. Because there are some caveats associated with the use of purely automatic alignment programs such as ClustalW, we favour the use of computer programs that also provide a visual help for the alignment. In programs such as MUST (6), SEAVIEW (7), or ClustalX (graphical version of ClustalW), a colour code is associated which each type of amino acid (and may be modified by the user). Furthermore, it may be preferable to show only differences to the first sequence, as in Figure 1. Usually, the user can change easily the first sequence, and the general order of sequences, so that visualization of identities and differences is more obvious. Of course, using such programs is more timeconsuming than using a simple automatic program such as ClustalW, but the results are far better and more reliable. The human mind can make a reasoning on the evolutionary history that the computer cannot. The principle of such a reasoning may be to minimize the number of positions containing gaps, or to

1: Evolutionary biology of the nuclear receptor superfamily add biological knowledge, such as the definition of important domains, which often profitably completes a purely statistical scoring.

2.3 Tree reconstruction methods Given the enormous literature on the subject and the flurry of new methods of molecular phylogeny constantly published, an exhaustive review of all the methods available is way outside the scope of this chapter. We simply summarize the most generally used ones, with some of their interest and caveats, and their application to the analysis of nuclear receptor sequences. The interested reader can find a broad description of these methods in ref. 4, and a thorough, technical, description in ref. 8. 2.3.1 Distance methods The principle of distance matrix methods is very simple and is described in Figure 2. Briefly, the idea is to summarize as a number the amount of differences existing between the various sequences. The principle of these methods rests on the idea that the more closely related two sequences will be, the less differences we will see between them. The first observation that allowed this principle to be set up was done in 1962 by Zuckerkandl and Pauling, who observed a correlation between evolutionary distances between organisms and genetic distances in the amino acid sequences of their haemoglobins (9).

Figure 2. Principle of the distance analysis.

7

H. Escriva, M. Robinson, and V. Laudet Protocol 2. Construction of a distance tree from aligned sequences 1. Carefully select the region of the alignment that will be studied. Discard from the alignment all ambiguously aligned regions for which sequence divergence is too high. 2. Select the type of distance matrix that will be used: Boolean matrix, similarity, or genetic code matrices.a Select eventually a correction for multiple substitution. 3. Compute the distance matrix. 4. Select an algorithm for computing the tree from the distance matrix: UPGMA, NJ, or Pitch.a 5. Run your program of tree reconstruction with the algorithm you have selected at step 4. In many programs steps 3 and 5 are only one command. 6. Test the robustness of your tree using the bootstrap resampling method. * See text for further explanations.

The first step in constructing a distance tree is to construct a distance matrix, i.e. to compute all numbers of differences between sequences taken two by two (Protocol 2). There is a number of possibilities for calculating such a matrix. The first and simplest one is the 'Boolean' matrix: each difference is taken as one, identical amino acids are counted as zero. Other possible matrices may take into account the chemical similarity between amino acids (similarity matrix). In that case for example Lys and Arg will be considered as identical whereas Arg and Gly will be considered as differences. Other matrices take into account the number of nucleotide substitutions necessary to explain the difference between the two compared amino acids ('genetic code' matrix). In computation of the matrix, it is possible in many programs such as MUST or PHYLO_WIN to choose the regions that will be used to calculate the matrix. Classically regions for which the accuracy of the alignment (and thus real homology between the compared amino acid) is questionable should be excluded. For NRs most often do not use the highly divergent A/B, D, and F domains, although this may depend on the level of the comparison. If for example TRs from various species are compared, then the A/B or D domain can be used since the degree of identity is sufficient. The treatment of gaps is not correctly resolved by molecular phylogeny

1: Evolutionary biology of the nuclear receptor superfamily methods. If a gap of five amino acids was included between two sequences it is not clear whether it has to be counted as a unique insertion (or deletion) of five amino acids or if it was five independent events of gain (or loss) of one amino acid. In the first case there will be a count of only one difference and in the other of five. For that reason, most authors prefer to simply exclude gaps from subsequent computation. In our analysis, we have often treated long gaps as unique events of gain (or loss) of several amino acids. In any case it is essential to test the influence of the gaps on the final result. To do so we compare the tree obtained by excluding all the positions containing gaps, with a tree obtained where some or all the gaps are used. For NRs the exclusion of the gaps does not change the topology of the tree, but simply reduces a little the robustness value of some internal branches. At this level the researcher has to do subjective choices, the importance of which can be tested. In many cases the matrix may then be corrected for multiple substitutions. In fact, when two identical amino acids are found in the same position of two sequences, we can consider two possible scenarios: (a) The amino acid has remained unchanged. (b) The amino acid has changed but has subsequently reverted to the same amino acid. That latter case is of course generally less frequent than the first one, but its probability increases with the overall genetic distance between two sequences. Thus uncorrected distance matrices underestimate genetic distances, especially when the distance between the sequences is high. Many authors have developed formula to take into account these multiples hits. These are particularly developed for nucleotide sequences (Jukes and Cantor correction, or Kimura correction; reviewed in ref. 10), but are also available for amino acid sequences (Kimura correction). Again it is important to test the effect of these various corrections, in order to estimate their usefulness, and to correctly interpret the resulting tree. Numerous methods of tree reconstruction using distance matrices are known. Only three popular ones will be discussed: the UPGMA method, the Fitch least square analysis, and the neighbour-joining method. All are available in most programs, but also in servers that can be used through the Web (see Section 2.3.5 and Table 1). The UPGMA (unweighted pair group method using arithmetic average) is a widely used and popular method, which unfortunately relies on a strict 'molecular clock' assumption that is generally false. In fact, it can be used only if the rates of evolution are approximately constant among the various lineages to be compared. This is in general wrong, and in the particular case of nuclear receptors completely so (see Section 2.5)! Thus, in most cases UPGMA cannot be used and gives wrong results. We strongly

H. Escriva, M. Robinson, and V. Laudet suggest that the readers use the neighbour-joining (NJ) method, that does not have this assumption, and which is fast and accurate. The Fitch least square method can also be used, but is less rapid and congruent than NJ. NJ is implemented in many programs (see Section 2.3.5), notably ClustalW, which thus allows using the same program for alignment and phylogeny. 2.3.2 Parsimony analysis The principle of minimum parsimony methods involves the selection of a tree that requires the smallest number of evolutionary changes to explain the differences observed among the sequences studied. As depicted in Figure 3, in the first step, all possible trees relating the sequences to be analysed are calculated. Then, in a second step, the number of mutations that are necessary to explain each tree is computed. The parsimony principle is thus applied. This principle suggests that Nature is parsimonious and that the most parsimonious tree will be the real one. It is possible (and frequent) that several most parsimonious trees are found. In that case the researcher has to consider these alternative scenario as equally probable, and can compute the consensus tree between these equally parsimonious trees. The most popular program of parsimony analysis is PAUP 3.1, which is available for Macintosh computers. It is easy to use, and has become the standard in parsimony analysis. A new updated version will be soon available, including on other platforms than Macintosh. Unlike most programs used in molecular phylogeny, which are freeware, PAUP is commercial. One problem, which limits the use of the parsimony analysis, is that the number of possible trees increases extremely fast with the number of sequences. Between four sequences three unrooted trees are possible, between five sequences 15 tree exists, between ten sequences more than two million trees exist! This considerably limits the number of sequences that can be used since with more than 15 sequences it is extremely difficult even with a big computer to calculate all possible trees. Thus, in PAUP it is possible to do a heuristic search that explores portions of all possible trees but not really all the trees in order to reduce computational time. As for distance analyses, the user of parsimony program will have to make choices on the regions to be analysed as well as on each character (i.e. each position of the alignment) to be used. To be valid a parsimony analysis has to be limited to sites selected as informative ones. A site in an alignment is phylogenetically informative for parsimony only if it favours some trees over others. A position in which the amino acid is different in each sequence will not be informative since it cannot be used to group sequences together, to the exclusion of others. The same reasoning holds for constant positions. The informative sites may be easily identified using PA UP and should be used to calculate the tree. Their inclusion may only add 'noise' to the phylogenetic signal. 10

1: Evolutionary biology of the nuclear receptor superfamily

Figures. Principle of the parsimony analysis.

2.3.3 Maximum likelihood Like parsimony, maximum likelihood (ML) considers complete information from each site of the alignment, not reducing it to pairwise distances. But unlike parsimony and like distance methods, maximum likelihood allows the use of evolutionary models to correct for multiple substitutions. The principle of maximum likelihood methods is to compute the probability of each possible tree given the sequence data, and to choose the most likely tree. In practice, it is not possible to consider all trees, and heuristics is used to explore the most probable ones. Unlike for parsimony, usual software does not allow using a heuristic approach for computational reasons. Indeed, maximum likelihood is computationally the most extensive phylogeny method, and many molecular biologists (frequently including us) are not able to use it simply for this reason. This is regrettable since it seems to be the most efficient, robust, and widely accepted method. The most widely used program for maximum likelihood is the PHYLIP package, which runs on PC, Macintosh, and UNIX. 2.3.4 Robustness of the trees: the bootstrap test Even if evolution worked exactly in the way for which a particular method is appropriate, with finite data an incorrect tree may be inferred due to chance 11

H. Escriva, M. Robinson, and V. Laudet events. For example, convergent substitutions might occur in a way that lead us to conclude that two sequences are related when in fact they are not. Thus, several authors have developed methods, such as the bootstrap resampling method, that allow the reliability of a given tree to be tested. The aim of this method is to measure the amount of information in a sequence alignment that supports each node of the tree. A full discussion of the interest and drawbacks of the bootstrap method may be found in ref. 8. Interestingly, bootstrap can be used both on distance matrix and parsimony generated trees. A new data set is first generated by the random selection in the alignment of a number of sites equal to the length of this alignment. If we compared four aligned receptors, which encompass 800 amino acids, the bootstrap program will generate an alignment of 800 amino acids, in which some positions of the genuine alignment may be present several times, whereas others are not present since the selection of the sites is done randomly. Then using this new alignment a new tree will be computed with exactly the same method as the one used for the original tree. This tree will be conserved in the memory of the computer, and the whole process (random selection of sites and tree reconstruction) repeated again and again. 1000 resamplings is an advisable number, although for parsimony analysis, due to computation time, as few as 100 replicates may be done. Finally the 1000 (or 100) trees are compared and the number of times a given branch is obtained is computed, this for each branch appearing on the original tree. For each branch the number of occurrences will give an estimate of its robustness, i.e. of the amount of information in the genuine data set that supports this branch. The result of the bootstrap method appears as a per cent of robustness for each branch. Different authors may set different thresholds, but it is clear that nodes with a bootstrap value below 50% should at least be considered as not resolved by the data, and thus should not be used in an interpretation of the tree. In the particular case of NRs, since the amount of phylogenetic information appears high when the sequences of both LBD and DBD are studied, and since the bootstrap values of many nodes is equal to 100%, we generally considered only nodes supported by more than 90% of the replicates. This gives a large confidence in the reliability of the nuclear receptor phylogeny. 2.3.5 Availability and choice of methods Choice of the phylogeny method used involves statistical, philosophical, and publishing considerations, as well as availability of programs. As molecular biologists, we tend to give strong importance to the latter two points: easy publication and availability. For publication, many 'evolution' reviewers have a favourite method, and ask for it very strongly. We recommend presenting the results of one method as a figure (a tree), and agreements and contradictions with other methods in the text. The practical reason to prefer NJ for the presented tree is its computational speed, which allows easy boot12

1: Evolutionary biology of the nuclear receptor superfamily Table 1. World Wide Web sites for molecular phylogeny PHYLIP homepage: http: //evolution. genetics .Washington. edu/phylip. html PAUP homepage: http://www.webcom.com/~sinauer/new.shtml#paup Listing by Joe Felsenstein of these and many other phylogeny programs, what they do, and how to obtain them: http://evolution.genetics.Washington.edu/phylip/software.html On-line retrieval and analysis of sequences: http: //bioweb.pasteur. fr/introuk.html#log Lists of other relevant sites: http://www. ii.uib.no/-inge/list. html and http://www-biol.univ-mrs.fr/english/logligne.htmHtalign

strapping, as well as comparing different approaches, such as using DNA or protein sequence, using different portions of the alignment, or different models (= distance methods). From a statistical point of view, we also believe it to be better supported than parsimony. This being said, if a reviewer insists to have a figure of the parsimony tree, it is probably not worth a fight! We have already introduced the two most widespread phylogeny programs, PAUP and PHYLIP. Although in their most recent versions both include the three types of methods, PAUP is more oriented towards parsimony, and PHYLIP towards maximum likelihood. The advantage of using these two programs is that, since they are widespread, communication with other scientists, notably evolutionists, will be facilitated. Two other programs we found of interest are MUST, for its transparent integration of many options, and PHYLO_WIN, which allows easy mouse-driven choices of sites, sequences, and methods. These and many others can be found at the excellent Web site maintained by J. Felsenstein (see Table 1). Finally, a most interesting development for molecular biologists is that of Web pages which allow on-line analysis of sequences, with more and more options and user-friendliness. It is difficult to keep up with the many new sites, but we specifically recommend that of the Pasteur Institute (Table 1), because it integrates all steps, from sequence retrieval to printing of the tree, including alignment and bootstrapping.

2.4 Interpretation of sequence trees 2.4.1 The root of the tree The vast majority of tree building methods give unrooted trees. On an unrooted tree the position of the common ancestor of the compared sequences is not known and thus in most cases the direction of evolution cannot be inferred. The placement of the root is important since it will help to decide what will be the ancestral state of a given character (Figure 4). For example, if the root of the nuclear receptor tree is placed between receptors able to homodimerize this will allow us to suggest that the first receptors were homodimers and that the other behaviours (heterodimers, monomers) evolved 13

H. Escriva, M. Robinson, and V. Laudet

Figure 4. Rooting of a tree. From one unrooted tree clustering four sequences, five rooted trees can be drawn. The position of the root can thus be of importance in order to discuss the nature (ancestral or derived) of the characters used.

later. All types of characters (ligand binding versus orphan, activators versus repressers ...) are interesting to study in that respect. The only method to accurately root a tree is to include what is called an outgroup in the analysis. An outgroup is a sequence for which there is clear evidence that it branched off earlier than the taxa being studied. When studying the phylogeny of mammals a good outgroup will be a bird or a reptilian. For NRs, a good outgroup to the TRs will be an RAR. The root is then placed between the outgroup and the node connecting it to the other sequences. A correct outgroup should of course have diverged before all other sequences. Nevertheless, a good outgroup should not be taken too distant from the ingroup since otherwise this may result in serious topological errors. The use of more than one outgroup sequence generally improves the analysis by reducing the artefacts that may be generated by the use of a unique peculiar sequence. We classically use a set of four or five increasingly divergent sequences. For example when studying TRs we include as outgroups RAR, PPAR, NGFIB, and HNF4. When the whole nuclear receptor superfamily is considered we have unfortunately no possible outgroup that can be used to root the tree. In fact, there is clear evidence both from the sequence of whole genomes such as the one of the yeast Saccharomyces cerevisiae and from three-dimensional structure determination that no correct outgroup is available for NRs. The NRs found in the most early organisms such as COUP-TF in Hydra (11) can be well included in a given subfamily of receptors and may not be viewed as 14

1: Evolutionary biology of the nuclear receptor superfamily outgroups. This means that the major dichotomies that lead to the six described subfamilies of receptors were done before the split between Hydra and the metazoans. Except if a nuclear receptor related sequence is found outside metazoans it is highly probable that a correct outgroup will remain unavailable for NRs. In the absence of an outgroup, the root may be positioned by assuming that the rate of evolution has been approximately uniform over all the branches. With such a hypothesis it will thus be placed at the midpoint of the longest pathway between sequences. Another solution, that we have used in our phylogenetic analysis of the nuclear receptor superfamily is to avoid the placement of the root. In that case, since we have defined six subfamilies, we have reasoned that the root should be placed between these subfamilies. But since we had no rigorous argument to choose a precise placement we have not located the root in the tree. Even in that case, we have no definitive argument to assert that the root is located between the subfamilies and the precise position of the root in the nuclear receptor tree is still an open question. 2.4.2 Species tree versus gene tree In constructing a tree using nuclear receptor sequences the researcher will inevitably face a question of choice of the sequences to be used. In most cases one will have to include sequences of homologues of various receptors in various species. For example there will be both human, mouse, rat, chicken, and Xenopus RXRa as well as the Drosophila USP gene, the human and Xenopus RXRp, and the human, mouse, and rat RXR-y. A tree containing these sequences will both represent a species and a gene tree. The mixing of these different types of sequences is of interest for two main reasons: (a) It allows the accuracy of the constructed tree to be tested. It is clear that the species tree (for example the relationship between human, rat, chicken, and Xenopus) should be consistent with the known phylogeny of these species. Thus, mouse and rat sequences should be clustered together and with the human one. Then these mammalian sequences should be connected to the chicken and Xenopus ones. Any other order should be artefactual and indicative of either an abnormal pattern of sequence evolution that may suggest that adaptative events took place, or that the sequences are in fact not real orthologues but represent versions of two different genes in different species. The chicken Rev-erbfi gene was identified in our laboratory using this type of reasoning (12). (b) On a molecular phylogenetic tree, branch lengths are not proportional to time but to the divergence existing between the sequences. This means that, since the molecular clock hypothesis does not hold for NRs (3), one cannot compute a divergence time using branch length. Thus there is no way to date a given event on the tree, as for example the splits which gave rise to the RXRa, B, and -y receptors. The only way to date such an event 15

H. Escriva, M. Robinson, and V. Laudet is to include homologues of the same receptor in various species since paleontological data provides us with divergence dates for most of the species used in molecular biology laboratories. Thus, for example the split between the three RXR takes place before the split between amphibia and other vertebrates, i.e. more than about 360 Myr ago. By using the Drosophila USP sequence, which is unique (i.e. not duplicated), we can also conclude that the three RXRs originate after the arthropod/ vertebrate split, 570 Myr ago (Figure 5). Thus the three RXRs appear between 570 and 360 Myr. To avoid confusion between homologous genes, which appeared through gene duplication events, and those which appeared by speciation, precise terms have been introduced. Orthologous genes are two genes that diverged through a speciation event. They usually have the same function. For example human and mouse RXRa are orthologues. By contrast, paralogous genes diverged by gene duplication events, such as human RXRa and human RXRB. Finally, metalogous genes may be defined as two genes derived from a gene duplication but present in different species. Thus, human RXRa and mouse RXRB are metalogues. Some authors do not distinguish metalogues and paralogues, but in any case these must be distinguished from orthologues. 2.4.3 Gene duplications Figure 5 presents a tree constructed by the distance analysis of a data set containing one representative of each type of vertebrate and arthropod receptors. In such a tree, we can observe the six receptor subfamilies, as defined by searching the most internal branches of the tree, supported by bootstrap values above 90%. In fact, the general picture that emerges from the tree is the rapid appearance of subfamilies and groups of receptors (i.e. TRs, RARs, RXRs, and so on) followed by a later diversification inside each group (i.e. emergence of TRa and B, RARa, B, and y, RXRa, (B,and y, and so on). This observation allowed us to propose that the present diversity of the nuclear receptor superfamily arose by two different waves of gene duplications: (a) The first wave of gene duplication took place at the origin of the superfamily and led to the six subfamilies and to the various groups of receptors. Since we observed that the Hydra COUP-TF branched off before the COUP-TF/EAR2 split, and since we did not find any nuclear receptor related sequence outside metazoans we think that this first wave was contemporaneous to the early diversification of metazoans. Without further work on NRs from early metazoans it is not possible to date this event more precisely. (b) The second wave of gene duplication led to a diversification inside each group of receptors and corresponded in fact to the appearance of the various paralogous genes. This wave was observed for many other gene families and appeared to be contemporaneous to the diversification of 16

1: Evolutionary biology of the nuclear receptor superfamily

Figure 5. Phylogenetic tree obtained by the neighbour-joining method of 63 nuclear receptors. The position of the root of this tree is unknown. One homologue of each known locus was selected from our complete nuclear receptor database. The subfamilies are indicated by brackets on the right. 1000 bootstrap replicates were performed and the bootstrap values of the branches supporting the six defined subfamilies are boxed. The black dots indicate the vertebrate-specific emergence of the various paralogues in each relevant group of receptors.

17

H. Escriva, M. Robinson, and V. Laudet early vertebrates (discussed in refs 1 and 3). Several arguments suggest that this major event of gene duplication was directly implicated in the appearance of the unique developmental and morphological characteristics of vertebrates. 2.4.4 Ligand binding and evolution When the tree of NRs is considered one important observation may be made: the ligand binding ability of a given receptor as well as the identity of its ligand is not related to its position on the tree. For example TRs and RARs bind completely different ligands but are brother groups on the tree. In contrast, RARs and RXRs which both bind retinoids with high affinity are members of different subfamilies (namely I and II). Furthermore, the orphan receptors are scattered in all the subfamilies. This suggests an independence of the ligand binding ability of the receptors from their evolutionary origin. This observation and the fact that in early metazoans we only found homologues of orphan receptors led us to propose that ligand binding was acquired during the evolution of NRs (3, 11). Such an observation has important functional implications. This suggests that the first members of the superfamily were orphan receptors. Thus, the first NRs were classical transcription factors, the activity of which was regulated through a conformational change similar to the one driven by the ligand in 'liganded' receptors such as ER. In the primitive orphan receptors this conformational change may have been induced not by ligand binding but by all sort of events such as phosphorylation or protein-protein interaction. There is no reason to believe that this level of regulation of NR activity was lost during evolution, thus the model also suggests that the present day receptors may be switched on by events different from ligand binding as well. This model may be testable by studying NRs in early metazoans. 2.4.5 Rates of evolution The rate of evolution, i.e. the number of mutations that accumulate in a sequence in a given amount of time, is strongly variable from one sequence to another as well as when different organisms such as rodents and primates are considered. This observation led to the commonly accepted view that there is no universal molecular clock. There is a number of methods designed to precisely determine the rates of evolution in nucleotide or amino acid sequences and the description of these methods is outside the scope of this chapter. The interested reader can find a description of these methods in ref. 8. The rates of evolution are strongly variable between the different types of NRs. The COUP-TF orphan receptors are the most strongly conserved receptors whereas ERRs, steroid receptors, and the unusual receptors containing only an E domain, such as Dax-1 and SHP-1 evolve rapidly. Interestingly, up to now there is no correlation between the orphan status of a receptor and its evolutionary rate. Whether this can be an argument to suggest that all the 18

1: Evolutionary biology of the nuclear receptor superfamily orphans have in fact a ligand is still a matter of controversy. With the exception of our preliminary report of variable evolutionary speed of the various NRs (3) no systematic studies of this interesting aspect have yet been published.

3. Cloning of new receptor genes from divergent animal species The first step in the evolutionary study of a gene family is to acquire the sequences to be treated, either from databases or by cloning and sequencing them directly from the species of interest. In this section, we will present various protocols for the isolation of different members of a gene family from a wide range of phylogenetically distant species using the PCR technique. This method is particularly suitable in the case of NRs since these molecules contain several domains that exhibit a strong level of sequence identity. Such a structure is convenient for the design of PCR primers.

3.1 Overview of the strategy The first cycles in a PCR reaction constitute the screening cycles in which specific DNA fragments start to be amplified. Thus, screening occurs before amplification and not the opposite as in classical molecular cloning such as library screening. Although the amplification phase is well performed by a classical PCR, the screening phase is not. For this reason, for identifying distantly related members of known gene families, several protocols have been designed in order to perform low stringency PCR reactions. During the PCR reaction, the primers assume a double role, as a probe for screening and as a vector for amplification. During the first cycles, each primer acts independently as a single probe, screening all the possible targets. Only when a pair of primers hybridizes in the correct orientation is the target selected. After this selection, the primers assume the role of amplification vector to amplify the selected target during the last cycles. Optimizing these two processes in the PCR reaction requires different, sometimes antagonistic, conditions. Thus the first problem in optimizing the PCR reaction for cloning of new sequences is the choice of the PCR conditions. When cloning distantly related members of a protein superfamily the researcher has to face a second major issue: primer specificity. If the primers are too specific, he/she will obtain the same member of the superfamily in each PCR reaction, but if they are not specific enough he/she will fail to amplify the desired DNA fragment. In each case, it will be necessary to optimize the primer sequences as well as the PCR conditions in order to obtain different but specific members of the superfamily. The last major issue is the choice between amplifying directly DNA fragments or doing an RT-PCR experiment from a RNA target. If the expression 19

H. Escriva, M. Robinson, and V. Laudet of the desired gene in the studied biological material is well known an RTPCR reaction on a total RNA extract may be performed. In other situations, the experiment should be performed on total genomic DNA. In the latter situation the fragments that will be amplified will be in most cases very short since the amplification should be done inside the same exon to avoid introns, the size of which are not known. A comparison of the genomic organization of the gene of interest between closely related species can be useful in such a case. It is clear that in vertebrates the position of the intron inside the coding sequence of a given NR is extremely well conserved. Thus, the genomic organization may be known in only one species (e.g. human or mouse), but it may be extrapolated without risk to most other vertebrate species. To our knowledge, the only exception to this observation concerns the RXRs genes that exhibit a variable organization in vertebrates. One important reason to choose DNA rather than RNA to perform PCR amplification may be the availability of the material. DNA is very well conserved in ethanol-preserved tissues and it is thus relatively easy to find a large variety of samples. The collection and preservation of good RNA preparations may be much more difficult.

3.2 Designing PCR primers As we have seen, the choice of primers is one of the most important steps in a PCR strategy for the cloning of members of a gene superfamily. In our case, due to availability of the biological material, we performed all the PCR reactions on total genomic DNA. Degenerate primers were thus chosen from a unique exon encoding the most conserved region of NRs, the C domain. The alignment of known NR sequences was performed, using the ClustalV program (previous version of ClustalW) and respecting the triplets corresponding to the ORF (11). Importantly, we realized that the use of highly degenerated primers, theoretically able to amplify all types of NRs, gave rise to very poor results probably because of a very low amplification efficiency. Thus, we performed separate alignments of all known members of the various groups of receptors that were searched: TRs, RARs, RXRs, PPARs, COUP-TFs, and so on. We have always used all known sequences, even the incomplete ones, in order to have the largest possible diversity in a given group. Such a strategy allows group-specific primers to be designed that are able to amplify any member of a given group of receptors. 3.2.1 The nested strategy Due to the low quantity of target DNA (a single copy gene) in the total genomic DNA, and even more importantly to the low efficiency of the low stringency PCR method that we used, two rounds of amplification are usually necessary to obtain enough DNA for cloning and sequencing. In such a case, we never use the re-amplification of the PCR fragment using the same oligonucleotides, 20

1: Evolutionary biology of the nuclear receptor superfamily

Figure 6. Principle of the nested PCR strategy. (A) Location of the primers used for RT-PCR experiments. For PCR with genomic DNA all the primers were designed from the C domain sequences. (B) Classical nested strategy. For the first PCR reaction, three independent tubes are used, each with a different primer combination. For the second run (nested PCR), each possible semi-nested reaction is performed with 2 ul of PCR reaction from Protocol 3, step 1. The nested reaction is thus composed of five independent tubes.

but we rather perform nested or semi-nested PCR reactions. In our hands, reamplification with the same primers gives rise to smears or at least complicated patterns of multiple bands that render a subsequent analysis really difficult. In addition to increasing the sensitivity of the detection assay, such a strategy allows the specificity of the PCR reaction to be improved, since the correct amplification of a NR fragment is dependent not only on the correct hybridization of two primers as in classical PCR, but of three or four different ones. In general, we design four PCR primers that we use in all possible combinations as depicted in Figure 6. The test of all possible semi-nested combinations is worthwhile. If a primer is, for one reason or another, unable to work correctly, at least one primer couple will yield a positive result. In addition this increases the chance to find a primer couple that finds a compatible hybridization temperature, a problem which is never trivial with degenerated primers. 3.2.2 Degenerate primers As we have seen, an alignment of all known sequences of the desired NR is necessary in order to design degenerate PCR primers that are as specific and as efficient as possible. In such a multi-alignment, it is extremely important to conserve the triplet organization of the codons of the reading frame. We 21

H. Escriva, M. Robinson, and V. Laudet

Figure 7. Example of the rules governing primer design. In this example, the five 3' bases of the primer are fully conserved. Each position with more than two possible bases is replaced by an inosine.

never used amino acid alignment directly nor did we take into account all the possibilities that a given codon (as one coding for Leu for example) may have. We simply used the available nucleotide sequences without searching to increase the level of degeneracy of the primers since this would undoubtedly have decreased the efficiency of the amplification. We observed that in a given phylum (such as vertebrates) the codon usage is sufficiently well represented by the major model species (zebrafish, Xenopus, chicken, mouse, rat, and human) to propose most of the actual variation observed in more 'exotic' species. For organisms with genomes exhibiting highly divergent genome composition (for example, that of Hydra which is approximately 70% AT-rich) the wobble position in the primers may be skewed towards this extreme codon usage. In general, our degenerate primers are 20-25 nucleotides long. When only two possibilities appear in the alignment, a degenerate position is used in the primer. But if more than two (i.e. three or four) possibilities are observed in the alignment, even if one of such possibility is present in only one sequence, an inosine residue is used instead, as shown in Figure 7. Inosine residues at the wobble positions increase considerably the global specificity of the primer. Nevertheless, we try to avoid primers that contain more than three inosines since we observed that the more a primer is degenerate, the less efficient it is. Finally, the primer should end at its 3' position with at least two or three conserved nucleotides. These positions are critically important for the efficiency of the PCR amplification.

3.3 PCR isolation of nuclear receptor gene fragments A usual PCR reaction is composed of a cycle encompassing a succession of three temperatures. First, a high temperature for denaturing DNA, second an annealing temperature that is variable from one primer set to another depending on their Tm, and finally an extension temperature that is usually 72 °C, the optimal temperature for Taq DNA polymerase. When using degenerate primers it is impossible to know the precise annealing temperature since there is a mix of primers. If too high annealing 22

1: Evolutionary biology of the nuclear receptor superfamily temperature is used, this will prevent amplification and negative results will be obtained. In contrast, too low a temperature results in a great number of artefacts, even using a semi-nested second round of amplification. An optimized PCR amplification technique called 'touch-down PCR', based on a successive series of cycles with decreasing annealing temperatures has been shown to be specific and efficient enough to allow successful cloning of different members of the NR superfamily in a wide variety of species (see Section 3.3.2). 3.3.1 Preparing the DNA The preparation of the DNA template appears to most people as a trivial step in the PCR amplification. However, we observed that this step may be critically important. Good quality DNA with very low amounts of PCR inhibitors may be extremely useful. Strikingly, some organisms (e.g. tunicates) or organs may be refractory to DNA extraction and this may explain amplification failure. For marine organisms such as sea urchins or tunicates we strongly favour the use of gonads, eggs, or embryos for DNA extraction. In addition to their high DNA content, these samples allow the contamination of the DNA preparation with DNA from other organisms to be avoided. Since many marine invertebrates are filtering or necrophages, their tissues (especially the digestive tract) may be easily contaminated by DNA from their preys and this is a major source of DNA contamination. Even after three days of fasting we have found preparation of Hydra DNA that were contaminated by sequences from Artemia, a small shrimp used to feed them! 3.3.2 The touch-down PCR method The use of degenerate primers and the fact that, in most cases, the sequences of the target will slightly differ from the sequences of the oligonucleotides render it impossible to calculate a precise annealing temperature for the PCR reaction. For this reason it is desirable to test different annealing temperatures during the PCR process. Two major strategies may be used for such a purpose: the 'touch-up' and the 'touch-down' methods (13). The principles of these methods are identical: the annealing temperature changes every five cycles. In the 'touch-up' strategy the annealing temperature increases during the PCR whereas in the 'touch-down' the annealing reaction is higher in the beginning of the reaction and decreases step by step until the end of the reaction (Figure 8). We never obtained good results with the 'touch-up' reaction, which in our hands, yielded many PCR artefacts, whereas the 'touch-down' method gave very good results. This may be due to the fact that in the 'touch-down' method, since the annealing temperature is high at the beginning, only specific primers can hybridize. Artefactual mis-priming may take place only after the annealing temperature has decreased. This is in sharp contrast with the 'touch-up' method which allows first the amplification of non-specific bands and only afterwards that of the specific ones. 23

H. Escriva, M. Robinson, and V. Laudet Protocol 3.

The 'touch-down' PCR cycle

1. Prepare the PCR reaction mix adding in the following order: water up to the total reaction volume (50 ul or 100 ul), 330-350 ng of each degenerate primer, 1.5 mM MgCI2, 1.25 mM of each dNTP, 200 ng of genomic template DNA, 2.5 U of Taq DMA polymerase in the PCR buffer supplied with the enzyme. The total reaction volume is covered with two drops of mineral oil to avoid evaporation during the PCR reaction. Hot start PCR can also be performed. 2. Touch-down cycle: (a) Cycle 1:96°C for 2 min. (b) Cycle 2: 94°C for 1 min, 55°C for 1 min, 72°C for 1 min 30 sec. Repeat five times. (c) Cycle 3: 94°C for 1 min, 50°C for 1 min, 72°C for 1 min 30 sec. Repeat five times. (d) Cycle 4: 94°C for 1 min, 45°C for 1 min, 72°C for 1 min 30 sec. Repeat five times. (e) Cycle 5: 94°C for 1 min, 40°C for 1 min, 72°C for 1 min 30 sec. Repeat five times. (f) Cycle 6: 94°C for 1 min, 37°C for 1 min, 72°C for 1 min 30 sec. Repeat 25 times. (g) Cycle 7: 72 °C for 7 min. Soak at 4°C. 3. The annealing temperatures and different times of denaturation, annealing, and elongation may be optimized for each set of primers and for each PCR machine.

As explained above (see Section 3.2.1), we always performed a second round of semi-nested PCR with the same conditions using 2 ul of the first PCR as template. As an alternative to the cycle described in Protocol 3, it can also be useful to use a classical cycle with low annealing temperature (94°C/1 min, 45 °C or 50°C/1 min, and 72°C/1 min). Such a cycle can give efficient results when the target is closely related to the sequence of the primers, i.e. when very closely related sequences are expected. 3.3.3 Cloning and sequencing After testing the result of the PCR by agarose gel electrophoresis, the cloning of the obtained DNA fragments is essential, because in most cases a 24

1: Evolutionary biology of the nuclear receptor superfamily

Figure 8. Principle of the 'touch-down' PCR cycle.

mixture of different fragments of the same size, corresponding to different members of the superfamily is obtained. Thus, the visualized PCR band contains several PCR products. Since most Taq polymerases add an A at the 3' end of the PCR product, cloning in commercially available vectors with a protruding T is necessary. This has the advantage of increasing the cloning efficiency. After cloning, several clones should be sequenced to assess the presence of either one or several members of the gene superfamily among the observed PCR bands. In most applications at least ten individual clones (and frequently up to 25) are sequenced in order to be sure to isolate all types of amplified PCR products. When we searched for several paralogues of a given NR group (TRa and B for example), we often observed that it is more efficient to perform two independent identical reactions and to sequence five clones from each rather than to sequence ten clones from a single tube. This is due to the fact that, if the two sequences have an identical chance to be amplified, the one that will be recognized by the primers during the first cycles will be preferentially amplified in subsequent cycles. Since this choice is random the use of independent amplification tubes may increase the chance to get all possible sequences. 3.3.4 Assessing the authenticity of the results PCR, especially with degenerate primers, is very prone to amplification or contamination artefacts. Amplification artefacts are mainly due to the 25

H. Escriva, M. Robinson, and V. Laudet relaxed conditions used and can be decreased by optimization of the PCR cycle or the MgCl2 concentration. Such artefacts may hide a positive result and this is one additional reason to sequence numerous individual clones after an amplification. In one example, we obtained, after a 'touch-down' PCR cycle with genomic DNA, five different artefacts and the two expected PCR products! Contaminations are the nightmare of PCR performed with degenerate primers, especially when genomic DNA is used as a target since the DNA of the researcher itself can be a source of contamination. Extreme care in laboratory management and reagent handling should be taken to minimize risk as much as possible. The most common measures are: (a) Perform DNA extractions in a dedicated laboratory in which no PCR products or NR cDNA clones are used. PCR mix should be done in a second dedicated laboratory. The PCR machine, the visualization of the bands, as well as the cloning and sequencing steps can be done in the regular laboratory. In our case, the DNA (or RNA, see Section 3.4) extractions are done in a fully equipped specific laboratory, which is under positive air pressure like the PCR mixing room. All the material and reagents are strictly specific to each laboratory and absolutely nothing is allowed to enter in these dedicated areas that may be as far as possible from the regular laboratory. (b) Use aerosealed tips in all steps with pipetting devices specific to each room. (c) Aliquot all reagents including Tag DNA polymerase and oligonucleotides. The reagents and stock solution should never be stocked in the regular laboratory. Every aliquot tube once opened should be discarded after the experiment. The meticulous application of these control measures allows most of the contaminations to be avoided. Of course it is always difficult to avoid all sources of contamination, especially the most dangerous one, coming from the DNA or RNA preparation. When a contamination occurs we never try to understand what is contaminated or to decontaminate the products but we simply discard all reagents. A good solution is then to wait one or two weeks before performing the same PCR and then to start again with entirely new reagents. Since some contaminations may be particularly insidious, the authenticity of the cloned DNA fragments always has to be verified by different methods. If possible a Southern blot analysis with total genomic DNA from the studied organism, using the PCR amplified fragment as a probe, should be performed. Other indirect methods can help when a Southern blot is not possible. For example, the A/T content of some organisms is particularly high in coding sequences, and the same A/T content should then be found in the PCR frag26

1: Evolutionary biology of the nuclear receptor superfamily ment. The reproducibility of the amplification with other reagents and DNA (or RNA preparations) can be a criteria but this may also be due to a spurious contamination. Ideally these reproduction tests of the initial experiment should be done in a different laboratory. The isolation of genomic or cDNA clones from different libraries using the PCR fragment as a probe is also a good proof, as is obtaining in different PCR reactions the same DNA fragment. All of this taken together, along with the relative phylogenetic position of the fragment according to its biological origin, can help to check the authenticity of the PCR fragment.

3.4 RT-PCR with degenerate oligonucleotides When the expression of the desired gene has been confirmed in a biological sample, an RT-PCR on total RNA can be more convenient than a genomic PCR. Because of the lack of introns in mRNA the amplified region can be larger. In NRs both the C and E domains are highly conserved and it is thus possible to design primers able to amplify most of the C, D, and E domains. Differences between genomic PCR and RT-PCR lie only in the nucleic acid extraction and reverse transcription reactions. The design of primers and the PCR reaction itself are identical in both cases, and the same care should be taken in choosing the annealing temperature and the extension time of the reaction. Moreover, the same kind of semi-nested PCR has to be performed when RNA is used as template. 3.4.1 RNA preparation The RNA extraction can be performed by many different methods. We routinely use a method derived from ref. 14. The purification of poly(A)+ RNA is not necessary. As for DNA preparations, the quality of the RNA preparation is important to obtain efficient and reproducible results. 3.4.2 Reverse transcription The reverse transcription step is not fundamentally different from a regular RT-PCR step. Three types of reverse transcription (RT) reactions can be carried out depending of the type of primer used: (a) RT with oligo(dT). (b) RT with random primers. (c) RT with specific oligonucleotides. In our hands the first gives poor results. In fact, the efficiency of this reaction depends on the size of the transcript and, in particular, on its 3' untranslated region, which is large in many NR transcripts (15). Note that for RT with specific oligonucleotides the primers can be degenerate with the same principle as for regular PCR primers. 27

H. Escriva, M. Robinson, and V. Laudet

References 1. Sharman, A. C. and Holland, P. W. H. (1996). Neth. J. Zool, 46, 47. 2. Laudet, V., Hanni, C., Coll, J., Catzeflis, F., and Stehelin, D. (1992). EMBO J., 11, 1003. 3. Laudet, V. (1997). J. Mol. Endocrinol, 19, 207. 4. Li, W. H. and Graur, D. (1991). Fundamentals of molecular evolution. Sinauer, Sunderland. 5. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994). Nucleic Acids Res., 22, 4673. 6. Philippe, H. (1993). Nucleic Acids Res., 21, 5264. 7. Galtier, N., Gouy, M., and Gautier, C. (1996). Comp. Appl. Biosci., 12, 543. 8. Swofford, D. L., Olsen, G. J., Waddel, P. J., and Hillis, D. M. (1996). In Phylogenetic inference in molecular systematics, 2nd edn (ed. D. M. Hillis, C. Moritz, and B. K. Mable), p. 407. Sinauer, Sunderland. 9. Zuckerkandl, E. and Pauling, L. (1962). In Horizons in biochemistry (ed. M. Kasha and B. Pullman), p. 189. Academic Press, New York. 10. Zharkikh, S. (1994). J. Mol. Evol., 39, 315. 11. Escriva, H., Safi, R., Hanni, C., Langlois, M. C., Saumitou-Laprade, P., Stehelin, D., etal. (1997). Proc. Natl. Acad. Sci. USA, 94, 6803. 12. Bonnelye, E., Vanacker, J. M., Desbiens, X., Begue, A., Stehelin, D., and Laudet, V. (1994). Cell Growth Differentiation, 5, 1357. 13. Don, R. H., Cox, P. T., Waintwright, B. J., Baker, K., and Mattick, J. S. (1991). Nucleic Acids Res., 19, 4008. 14. Chomczynski, P. and Sacchi, N. (1987). Anal. Biochem., 162, 156. 15. Gronemeyer, H. and Laudet, V. (1995). Protein Profile, 2, 1173.

28

2 Initial characterization of new orphan receptors R. SLADEK and V. GIGUERE

1. Introduction Nuclear receptors are transcription factors with the intrinsic ability to be regulated by changes in the intracellular levels of small lipophilic compounds. Since the molecular cloning of the glucocorticoid receptor in 1985, over 50 members of the nuclear receptor superfamily have been identified (1). Most of these gene products were discovered as a result of their sequence homology with other family members; however, their sequence conservation often reveals little about their individual function as transcription regulatory proteins. The ultimate goal for many investigators is to identify ligands for these orphan receptors: basic characterization of receptor functions including their DNA binding mode and transactivation properties provides an important first step towards realizing this goal. Nuclear receptors have a highly conserved protein structure that localizes specific functions to well defined regions of the receptor protein (Figure 1). The receptor's DNA binding domain (DBD), which displays the highest protein sequence conservation among family members, consists of two zinc finger modules occupying between 66 and 70 amino acids in the central region of the receptor together with residues located in the adjacent carboxy terminal extension (CTE) of the core DBD. The ligand binding domain (LBD), which is less well conserved among family members, is located in the carboxy terminal region of the receptor. The LBD contains a ligand-regulated transactivation function (AF-2), as well as regions that regulate receptor dimerization. The amino terminal region of the receptor, which displays the poorest sequence conservation among family members, contains a ligand-independent transactivation function (AF-1), and in some cases modulates the receptor's DNA binding specificity and affinity (2, 3). Nuclear receptors activate gene transcription by binding to hormoneresponse elements (HREs) contained in the enhancer and promoter regions of target genes. Contacts between key residues located in the first zinc finger module of the receptor DBD and nucleotides within the HRE core motif are

R. Sladek and V. Giguere

Figure 1. Nuclear receptor and DNA response element structure. Schematic of a typical nuclear receptor's functional domains (upper panel). The receptor has two transactivation domains which interact with co-regulatory proteins: a ligand-responsive activation function (AF-2) is located in the C terminus (E), while the N terminus (A/B) contains a ligand-independent activation function (AF-1). Region C contains the conserved 66-70 amino acid DNA binding domain (DBD). It is linked to the receptor ligand binding domain (LBD) by the hinge region (D), which also provides surfaces for interaction with certain co-regulatory proteins. Receptor dimerization is regulated by protein surfaces contained in the DNA binding domain and the C terminus. Sequences of consensus nuclear receptor hormone-response elements (lower panel). Most steroid hormone receptors bind to response elements containing an inverted repeat of the consensus half-site AGAACA separated by three nucleotides. Non-steroidal hormone receptors and most orphan nuclear receptors bind to direct repeat response elements (as hetero- or homodimers) or to single extended half-sites (as monomers).

primarily responsible for determining the specific DNA sequence recognized by each receptor (reviewed in ref. 4). The oestrogen receptors, non-steroidal hormone receptors, and orphan nuclear receptors bind to response elements containing variants of the consensus core motif AGGTCA. In contrast, most steroid hormone receptors bind to elements containing palindromic repeats of the core motif AGAACA (Figure 1). Non-steroidal receptors bind either as monomers to a single core motif, or as homodimers or as heterodimers with the retinoid X receptor (RXR) to HREs containing direct repeats of the core motif. High affinity binding by monomeric and certain homodimeric receptors relies on contacts between the DNA binding domain and the response element as well as on stabilizing interactions between residues within the CTE and nucleotides upstream of the core motif. As a result of these contacts, 30

2: Initial characterization of new orphan receptors individual receptors usually recognize specific nucleotide sequences contained in the response element's upstream flank (2, 5, 6). Most receptors which bind as dimers discriminate among HREs on the basis of their half-site orientation (inverted, everted, or direct repeats), as well as on their half-site spacing. The preferred distance between half-sites is determined by dimerization surfaces located within the DBD (7). Direct repeat response elements are inherently asymmetric and are recognized by receptor heterodimers whose polarity depends on the receptors forming the heterodimer as well as the half-site spacing within the HRE. The common heterodimeric partner RXR will bind preferentially to either the up- or downstream half-site core motif in order to achieve specific and co-operative recognition of the HRE (8-11). The identification of synthetic and natural response elements is a critical first step towards identifying potential in vivo roles for newly discovered members of the nuclear receptor family. While this approach relies heavily on in vitro studies, its success has been well demonstrated by promoter characterization experiments, which have identified a wide spectrum of genes which may be regulated by classical steroid and thyroid hormones as well as by retinoids and synthetic agents such as peroxisome proliferators. The following chapter provides experimental protocols used to define the DNA binding, dimerization, and transcriptional properties of a novel orphan nuclear receptor.

2. Sources of receptor protein for DNA binding studies The following sections discuss the relative advantages and disadvantages of three sources that have been commonly used to provide protein for DNA binding and dimerization studies. When choosing among these sources, it is important to consider the amount and purity of protein, as well as the fraction of correctly folded or active protein that will be required for a particular experiment. It is also important to determine whether post-translational modification of the receptor protein may alter the properties that are being studied: for example, phosphorylation the DNA binding domains of the orphan nuclear receptors NGFI-B and HNF-4 in response to specific physiological events alters their DNA binding affinity (12, 13). If the factors which regulate expression of a particular receptor are known—such as growth factors, specific intracellular signals, or specific forms of physiological stress—then receptor proteins obtained from appropriately treated cell lines as well as unmodified bacterial protein may be studied. Functional differences between receptor protein isolated from these sources may provide a clue that the in vivo extracts contain modified forms of the receptor or additional proteins that change its behaviour (14). Proteins produced at high levels in either bacterial or mammalian expression systems may often be misfolded and may display absent or inappropriate post-translational modification. If an experimental treatment seems to increase the strength of a protein-DNA or a 31

R. Sladek and V. Giguere protein-protein interaction, make sure that it is due to changes in the affinity of the interaction, rather than to changes in the fraction of protein that is functionally active.

2.1 Protein expression in bacteria Bacterial expression systems are a rapid and inexpensive means of preparing large amounts of receptor protein. The major advantages of these systems include low cost, high yield, and the possibility of making highly purified protein by using a wide range of affinity purification protocols. Potential disadvantages of using bacterially synthesized proteins include the time required to optimize protein induction, the lack of appropriate post-translational modification, and problems with protein misfolding. In particular, the hydrophobic nuclear receptor carboxy terminal domain may form insoluble inclusion bodies when expressed in bacteria. Although inclusion bodies often contain large amounts of highly pure and undegraded receptor protein that can be easily purified, the protein is usually misfolded and may be difficult to renature. Incorrectly folded proteins as well as the use of inappropriately high concentrations of receptor protein are frequent causes of artefacts in DNA binding and protein interaction studies. The first step in making bacterial protein is to clone the receptor cDNA into an expression vector containing a fusion partner that can be used in subsequent affinity purification steps (15). Commercial vectors, such as the pGEX series (Pharmacia) use isopropyl-B-D-thiogalactopyranoside (IPTG) to induce protein expression under control of the tac promoter. The pGEX vectors can be used to create chimeric proteins that contain glutathione-5transferase (GST) as the fusion partner (16). GST fusion proteins bind to glutathione-Sepharose, which can be used to affinity purify the fusion protein in a batch reaction. The fusion protein can be recovered by incubating the protein-Sepharose mix in a buffer containing glutathione and retaining the supernatant. Alternatively, many commercial GST expression vectors contain proteolytic sites that can be used to cleave the GST moiety from the cloned receptor protein. Most commonly, the washed beads are used directly as a substrate for detecting protein-protein interactions by affinity chromatography. To create the fusion construct, a fragment of the receptor cDNA, which includes all, or part of the coding region is obtained by restriction enzyme digestion or PCR amplification. As the nuclear receptor carboxy terminal domain contains many regions important for regulating receptor dimerization and other protein-protein interactions, the fusion partner is usually inserted at the receptor's amino terminus. The fragment used to express the receptor protein must be positioned in the expression vector to ensure that the reading frame is preserved and that a translation termination codon is present. Proteins expressed using prokaryotic expression vectors containing GST, maltose binding protein (MBP), or His-tag epitopes can be easily purified by performing affinity chromatography under conditions that preserve the fusion protein's 32

2: Initial characterization of new orphan receptors structure and activity. It is important to avoid large fusion partners, as proteins larger than 100 kDa are expressed inefficiently in bacteria. In addition, different epitope tags may influence the chance that a nuclear receptor will form inclusion bodies when overexpressed in bacteria. If protein made using a short epitope tag is poorly soluble, fusion partners such as MBP or GST may improve protein yields by promoting proper folding of the receptor protein. Regardless of the expression system used, it is important to try a variety of culture and extraction conditions to maximize protein yield. Pilot experiments should be performed using small scale (10-25 ml) cultures and the protein yield monitored by SDS-PAGE. The yield of soluble receptor protein is frequently affected by changes in the bacterial culture and induction conditions. This can be studied by inducing protein expression when the bacterial cultures have reached different cell densities(OD600between 0.4-1 units) and by growing the induced cultures at temperatures between 15-37 °C. Proteins manufactured using the GST expression system can be induced by treating the bacterial culture for short periods of time (1-3 h) using high concentrations of IPTG (0.5-1 mM) as well as for longer time periods (8-16 h) using less IPTG (final concentration 0.01-0.1 mM). In addition, the detergent used in the extraction and wash buffers may be changed to help preserve protein solubility and disrupt any interactions with contaminating bacterial proteins that may occur during protein purification. Finally, different bacterial hosts may produce significantly different levels of receptor protein. Host strains such as XL1Blue (Stratagene) and BL21(DE3)LysS (Novagen) contain low levels of endogenous proteases and have been used successfully to express a wide range of receptor proteins. The BL21 host strain is useful for expressing toxic and insoluble proteins, particularly when used with the low-leakage pET family of bacterial expression vectors (Novagen). A general method for the preparation of GST fusion proteins is given in Protocol 1 (see also Chapters 3 and 4): once the optimal induction and culture conditions have been established, the method may be scaled up to prepare larger quantities of the GST fusion protein. Protocol 1.

Preparation of GST fusion proteins

Equipment and reagents • Glutathione-Sepharose 4B (Pharmacia) • Refrigerated microcentrifuge and preparative centrifuge • Equipment for protein gel electrophoresis (SDS-PAGE)

• Competent Escherichia coli bacteria suitable for plasmid transformation • pGEX vector engineered to express receptor fusion protein (see text for details) • Facilities for large scale liquid bacterial culture

A. Selection of GST fusion protein expressing clones 1. Transform competent E coli with the receptor cloned into an appropriate pGEX expression vector and incubate on LB ampicillin agar plates 33

R. Sladek and V. Giguere Protocol 1. Continued at 37°C for 16-24 h. As a positive control for protein expression studies, also prepare a bacterial culture that expresses the unmodified GST protein. 2. Pick transformed colonies and grow in 5 ml LB ampicillin until the optical density at 600 nm (OD600) reaches 0.4-1 units. Remove 100 uL of culture, pellet the bacteria, resuspend in 50-100 ul of 1 x SDS-PAGE buffer, and boil for 2 min. This sample provides a baseline control for studying protein expression. 3. Induce fusion protein expression by placing 20 ul of the overnight culture in 2 ml of fresh LB ampicillin and adding IPTG to a final concentration of 0.5 mM. Place the induced culture in a shaking incubator at 37°C for 3 h. To study the time course of protein expression, remove 100 (ul aliquots of the culture every 30 min. Once the incubation is complete, remove a 100 (ul aliquot of the culture and prepare the sample for SDS-PAGE analysis as described in step 2. 4. Study protein expression by running 20 ul of the bacterial whole cell lysate from the uninduced and IPTG induced cultures using a SDSPAGE gel. Following electrophoresis, stain the gel with Coomassie Blue to determine which clones express a fusion protein of the correct size. Choose several of these for further expression studies to determine growth conditions which maximize the yield of soluble protein (see text). B. Small scale purification of GST fusion protein extracts 1. Inoculate 25 ml LB ampicillin with 0.25 ml of a saturated overnight bacterial culture and incubate with shaking until the OD600 reaches 0.4-1 units. A parallel experiment should be prepared using an E. coli strain transformed with the unmodified pGEX plasmid. Remove a 100 ul aliquot of the uninduced culture and prepare it for SDS-PAGE analysis. Add IPTG to a final concentration of 0.5 mM and incubate with shaking at 37°C for 3 h. If the cultures are grown at lower temperatures, the incubation time should be increased. 2. Remove a 100 ul aliquot of the induced culture and save it for SDSPAGE analysis. Pellet the bacteria by centrifuging the culture (2000 g, 4°C for 5 min), and resuspend the bacterial pellet in 1 ml of ice-cold NET-N lysis buffer (150 mM NaCI, 1 mM EDTA, 50 mM Tris-HCI pH 8.0, 1% Triton X-100, 1 uM leupeptin, 1 uM pepstatin, 0.1 mM PMSF). Lyse the cells using a dry ice/ethanol bath. If the lysate is extremely viscous, draw it through a fine-gauge needle or sonicate briefly to disrupt the bacterial DNA. Following lysis, centrifuge the extract to pellet the bacterial debris (12000 r.p.m., 4°C for 10 min in a microcentrifuge).

34

2: Initial characterization of new orphan receptors The crude bacterial extract may be purified immediately or aliquots may be stored at -80°C for future use. 3. Pellet 25 ul of 50% slurry of swollen glutathione-Sepharose beads by centrifuging briefly. Resuspend the beads in 250 ul NET-N and pellet again. Add the bacterial extract to the equilibrated beads and incubate with gentle agitation at 4°C for 30 min. 4. Pellet the slurry by centrifugation (5000 r.p.m., 4°C for 1 min in a microcentrifuge). Resuspend the beads in 250 ul NET-N and mix briefly. Save a 20 ul aliquot of the slurry for SDS-PAGE analysis. Pellet the beads, and repeat the wash and centrifugation steps two more times. Following the last wash, resuspend the beads in 250 ul 1x SDS-PAGE buffer. 5. Analyse the expression of the fusion protein using an SDS-PAGE gel. Load samples containing the uninduced and induced whole bacterial lysate (20 ul of extract obtained in part B, steps 1 and 2) as well as the GST affinity purified extract (20 ul each of the pre- and post-wash slurry obtained in part B, step 4). The GST expression control should produce a strong 28 kDa band in both the induced whole cell extract and in the affinity purified protein extract. The fusion protein should form a higher M, band in the crude lysate and the affinity purified extract. The latter band will not be detected if the fusion protein fails to bind the affinity matrix, which may occur if the fusion protein is insoluble, if the wash conditions disrupt GST binding, or if the fusion protein is damaged by proteases. 6. Elute the fusion protein by centrifuging the glutathione-Sepharose beads and resuspending them in 250 ul NET-N containing 5-20 mM reduced glutathione. Let the mixture stand for 10 min, then pellet the beads in a centrifuge. Add glycerol to the supernatant to a final concentration of 15% and store at -80°C.

2.2 Protein expression in vitro Reticulocyte lysates provide a convenient method of synthesizing small amounts of receptor protein in vitro. They are manufactured from rabbit reticulocytes using a proprietary process that usually includes a nuclease treatment step (to remove endogenous RNA transcripts) and may involve the addition of reducing agents (DTT), detergents, and stabilizers. As reticulocytes contain no nucleus, they contain low levels of transcription factors and other DNA binding proteins. Although reticulocyte lysates cannot be used to synthesize large amounts of protein, the protein yield is sufficient to perform most experiments described in this chapter. Additional purification steps are usually unnecessary if the protein is to be used in electrophoretic mobility shift assays (EMSA), ligand binding experiments, or studies of protein35

R. Sladek and V. Giguere protein interactions. Reticulocyte lysates have endogenous protein kinase and phosphatase activity, which may pose problems if accurate post-translational modification of the expressed protein is important for its function. In addition, the proportion of synthesized protein that is correctly folded may be quite low. As reticulocyte lysates can be used to synthesize a single protein, they provide a good source of radiolabelled receptor and co-activator proteins for in vitro protein interaction studies. Receptor proteins may be synthesized very rapidly using this technique, usually without recloning the cDNA into special expression vectors. The main disadvantage of this protein expression system is its relatively high cost. Two types of reticulocyte lysate expression systems are commonly available. The first uses a two-step process in which RNA is synthesized in an in vitro transcription reaction and then purified prior to being translated to make protein. The second uses a one-step process that performs RNA and protein synthesis in a single reaction mixture. The two-step system typically requires 4-6 h to manufacture protein while the coupled system produces synthetic protein within 1-2 h. In addition, the coupled synthesis systems often produce higher protein yields, which may offset their higher cost. Both systems use bacterial RNA polymerases (such as T3, T7, or SP6 RNA polymerase) to initiate RNA synthesis and can manufacture protein from most general purpose cloning vectors. For best results, clone the receptor cDNA so that the translation start site is surrounded by a good Kozak consensus sequence (e.g. GCCACCatgG) within 20-50 nucleotides of the bacterial promoter. If the cDNA contains multiple potential translation initiation sites, the reticulocyte lysate may synthesize both full-length and truncated receptor proteins. The size and yield of protein manufactured in a reticulocyte lysate reaction can be determined by including [35S]methionine or [35S]cystine in the reaction mix and analysing the reaction products by SDS-PAGE. This also provides a convenient method of manufacturing small amounts of radiolabelled protein for studying protein-protein interactions in GST pulldown experiments (Section 3.1).

2.3 Protein expression in eukaryotic cells Expression of receptor proteins in mammalian cells provides an efficient and inexpensive method of producing small amounts of active protein that is suitable for a wide variety of applications. Proteins can be expressed in eukaryotic cells using a variety of methods, including transient transfection and viral infection of cell lines. Depending on the expression system, the receptor protein may need to be purified later. This can easily be accomplished by expressing tagged proteins containing a His-tag (17), Flag-tag (18), or other fusion epitope (19). Purification protocols may be designed to separate different subcellular fractions, allowing 'active' nuclear receptors to be isolated separately from 'sequestered' cytosolic proteins. Characterization of these different receptor populations may provide evidence that posttranslational modification of the receptor protein regulates its subcellular 36

2: Initial characterization of new orphan receptors distribution or its DNA binding properties. Several cell lines are well suited for the synthesis of exogenous receptor proteins. For example, COS cells can synthesize large amounts of exogenous receptor following transient transfection with expression vectors containing the SV40 origin (which are amplified in this cell line). Exogenous receptor proteins may be expressed in COS cells at levels many fold higher than endogenous DNA binding proteins, eliminating the need for subsequent purification. Receptor proteins synthesized in COS cells are usually functionally intact and stable. While transient transfection provides a good source of small amounts of receptor protein, large scale protein synthesis usually requires expression systems using virally-infected cells that overexpress wild-type or epitopetagged receptor. The receptor may be expressed in a cell line derived from a tissue in which the receptor is expressed in vivo: pharmacological agents or growth factors may be used to treat these lines, inducing post-translational modifications resembling those seen in vivo. Experiments may be performed using whole cell extracts prepared from transiently transfected or infected cells (Protocol 2), or crude nuclear extracts (Protocol 5) prepared from cell lines or animal tissues. While whole cell extracts are much simpler to prepare, nuclear extracts will contain much higher concentrations of other nuclear proteins, some of which may interact with a particular receptor protein to stabilize its interaction with DNA or to modify its DNA binding specificity. Protocol 2.

Preparation of whole cell extracts

Equipment and reagents • COS-7 or other cell line suitable for receptor protein expression (see text for details) • Facilities for tissue culture and transient transfection of cell lines

• Refrigerated microcentrifuge • Dry ice/ethanol bath and 37 °C water-bath • Bradford assay reagent (Bio-Rad) and visible light spectrophotometer

Method 1. Grow COS cells on 10 cm tissue culture plates until they have reached 50-70% confluence. Transfect the cells with the receptor expression vector and control plasmids using the calcium phosphate coprecipitation technique. A typical transfection reaction contains 5-10 ug of receptor expression vector, together with 5-15 ug of carrier DNA per plate. The commonly used cytomegalovirus (CMV) and Rous sarcoma virus (RSV) promoters provide high level protein expression in COS cells. Transfection efficiency can be monitored by using an internal control plasmid in the DNA mix (e.g. 0.5 ug of CMV-B-galactosidase). 2. Following transfection, allow the cells to grow for 24-48 h (until they reach 80-90% confluence). Wash the plates twice with ice-cold PBS and harvest the cells by scraping into 1 ml of ice-cold PBS. Pellet the cells by centrifuging (2000 g, 4°C for 5 min). 37

R. Sladek and V. Giguere Protocol 2.

Continued

3. Resuspend the cells in 200 nl of ice-cold lysis buffer (20 mM Hepes pH 7.9, 20% (v/v) glycerol, 400 mM KCI, 0.5 mM EDTA, 0.5 mM EGTA, 1 mM DTT, 1 uM leupeptin, and 0.5 mM PMSF). 4. Lyse the cells using three freeze-thaw cycles (dry ice/ethanol bath for 3 min followed by incubation at 37°C for 3 min). Centrifuge the lysed cells in a refrigerated microcentrifuge at full speed for 10 min to clear the cell extract. 5. Determine the protein concentration in the supernatant using the Bradford assay (Bio-Rad) according to the manufacturer's instructions. Freeze aliquots of the supernatant in a dry ice/ethanol bath and store at -80 °C.

As an alternative to mammalian expression systems, receptor proteins can be expressed at high levels in virally-infected insect cells. In the baculovirus expression system, insect cells are infected by baculovirus in which homologous recombination has been used to replace the polyhedron protein gene— which is not essential for viral infection of cultured cells—with cDNA sequences coding for the expression of a specific nuclear receptor (20). The recombinant virus is purified by plaque dilution and can be used to infect cultured insect cells. Baculovirus expression systems may be used to provide pure receptor protein for in vitro transcription assays, ligand binding studies, or structure determination. As insect cells and other eukaryotic cells process and fold proteins in a similar manner, baculovirus systems rarely manufacture non-functional protein. While baculovirus expression systems allow high level of protein expression protein yields created using different recombinant virus strains expressing the same protein may vary significantly. Insect cells and mammalian cells differ in their ability to cleave proteins at internal sites as well as to modify proteins post-translationally. When proteins are overexpressed in insect cells, post-translational glycosylation usually differs from that of the native mammalian protein, while correct post-translational phosphorylation may or may not occur. The main disadvantage of baculovirus expression systems is the time required to create the purified recombinant virus and to set up the insect cell culture system. Receptor proteins may also be studied by obtaining nuclear or whole cell extracts from animal tissues. This approach provides small amounts of high quality, appropriately modified receptor protein. While the receptor protein can be isolated by fractionation and affinity purification protocols, it is usually easier to use crude whole cell or nuclear extracts directly in DNA binding or protein interaction experiments. An immunological detection step, such as Western blotting or EMSA antibody supershift, can demonstrate that a specific nuclear receptor participates in the interaction being studied. Tissue 38

2: Initial characterization of new orphan receptors extracts provide a good source of protein to confirm in vitro studies of the regulation of receptor function by post-translational modification. In addition, they may be used to show that DNA-receptor interactions may occur in the context of other cellular proteins. This is particularly important to determine when studying natural response elements: as many receptors share overlapping DNA binding specificity, other cellular proteins may block DNA binding by the receptor of interest. By altering the initial homogenization steps, the technique used to obtain nuclear extracts from cultured cells, described in Protocol 3, may be modified to prepare nuclear extracts from animal tissues (21). Tissue nuclear extracts frequently contain several transcription factors that bind to a common response element: immunodepleted cell extracts or immunodetection techniques can be used to confirm that a specific receptor participates in the DNA-protein interaction being studied. Protocol 3. Preparation of nuclear miniextracts from cultured cells Equipment and reagents • Facilities for tissue culture . Refrigerated microcentrifuge • Dry ice/ethanol bath and 37 °C water-bath

• Bradford assay reagent (Bio-Rad) and visible light spectre-photometer

Method 1. Grow monolayer cultures on 10 cm tissue culture dishes until they reach 80-90% confluence. Wash twice with 3 ml of ice-cold PBS and harvest by scraping into 1 ml of ice-cold PBS. 2. Collect cells by centrifugation (2000 g, 4°C for 5 min) and resuspend in 400 ul of lysis buffer (10 mM Hepes pH 7.9, 1.5 mM MgCI2, 10 mM KCI, 0.5 mM DTT, and 1 mM PMSF). 3. Lyse cells by three freeze-thaw cycles in a dry ice/ethanol bath. 4. Centrifuge the cell extract at full speed in a cold microcentrifuge for 20 sec to obtain a crude nuclear pellet. 5. Remove the supernatant and add glycerol to a final concentration of 15%. This fraction contains a crude cytosolic extract and may be frozen in a dry ice/ethanol bath and stored at-80°C. 6. Resuspend the pellet in two-thirds volume of nuclear extraction buffer (20 mM Hepes pH 7.9, 25% (v/v) glycerol, 1.5 mM MgCI2, 420 mM KCI, 0.2 mM EDTA, 0.5 mM DTT, 1 uM leupeptin, and 0.5 mM PMSF). 7. Incubate on ice for 30 min with occasional inversion. 8. Centrifuge the extract at full speed in a cold microcentrifuge for 20 sec to pellet any nuclear and membrane debris. 39

R. Sladek and V. Giguere Protocol 3.

Continued

9. Remove the supernatant and determine its protein concentration using the Bradford assay (Bio-Rad). Freeze aliquots of the supernatant in a dry ice/ethanol bath and store at -80°C.

3. Identifying the DNA binding mode: monomers and dimers The structure of the nuclear receptor LED is strongly conserved among family members. The LBD mediates a number of receptor functions, including ligand binding as well as interaction with transcriptional co-activator proteins. The LBD also contains dimerization interfaces that allow nuclear receptors to interact with each other: these interactions, together with DNA-stabilized protein-protein contacts mediated by the DNA binding domain, will determine the receptor's DNA binding mode. Heterodimer formation, which most often involves the nuclear receptor RXR as the heterodimer partner, will significantly alter the receptor's DNA binding affinity and specificity. Therefore, a receptor's homodimerization and heterodimerization potential should be determined before attempting to characterize its cognate response element. The following sections present two methods to determine whether a nuclear receptor interacts with RXR or with itself: both techniques study DNAindependent dimerization properties of the receptor and do not require any knowledge of the receptor's putative HRE. A receptor that fails to interact with itself or with RXR will most likely bind DNA as a monomer, although further studies should be performed to ensure that it does not form DNAstabilized dimers. The receptor's DNA binding mode should be confirmed by characterization of the receptor's HRE as described in Section 4.

3.1 GST-pulldown The GST-pulldown assay provides a rapid and efficient method for identifying protein-protein interactions. In a pulldown experiment, a receptor protein is cloned in-frame to create a fusion protein with GST that is expressed in bacteria and affinity purified using glutathione-Sepharose beads (see Protocol 1). The fusion protein bound to glutathione-Sepharose is mixed with an extract containing one or more potentially interacting proteins. This extract may be prepared in vitro, in which case a single target protein or a mixture of proteins can be radiolabelled to high specific activity: interacting proteins can be detected by SDS-PAGE analysis followed by autoradiography. GST-pulldown experiments may also be performed using whole cell or nuclear extracts: in this case, interacting proteins can be detected by Western or Far-Western blotting. Experiments performed using nuclear extracts may better detect weak protein interactions as well as interactions which are stabilized by a mixture of 40

2: Initial characterization of new orphan receptors nuclear proteins. Whole cell or nuclear extracts may also be used to study the role of post-translational effects in regulating protein-receptor interactions. Protocol 4 describes how to perform GST-pulldown experiments using a batch purification scheme (see also Chapter 3). GST fusion proteins can also be used to purify cellular extracts using microaffinity columns: this approach is particularly useful for detecting protein interactions with targets that are expressed at low levels in nuclear extracts (22). Protocol 4. Co-precipitation using GST fusion proteins Equipment and reagents • Bacterial lysate containing overexpressed GST fusion protein (Protocol 7) tive • Radiolabelled protein obtained from cultured cells or reticulocyte lysate • Glutathione-Sepharose 4B (Pharmacia)

• Refrigerated microcentrifuge and preparacentrifuge • Equipment for protein gel electrophoresis (SDS-PAGE)

Method 1. Pellet 25 ul of 50% slurry of swollen glutathione-sepharose beads in a microcentrifuge tube by centrifuging briefly in a microcentrifuge. Resuspend the beads in 100 (ul NET-N buffer (Protocol 1) and pellet again. 2. Add 250 uJ of the bacterial extract (see Protocol 7) to the equilibrated beads and incubate with gentle agitation at 4°C for 30 min. If different bacterial protein extracts are being used in the same experiment, make sure that equal amounts of the fusion proteins are being added to each tube. 3. Wash the beads for 5 min with 1 ml NET-N or GST binding buffer (20 mM Hepes pH 7.9, 150 mM KCI, 0.1% CHAPS, 5 mM MgCI2, 0.02 mg/ml bovine serum albumin, 1 uM leupeptin, 1 uM pepstatin, 0.1 mM PMSF). Precipitate the beads by centrifuging briefly in a microcentrifuge. Repeat the wash and precipitation steps three more times. 4. Resuspend the pelleted beads in 150 ul of GST binding buffer. Add the radiolabelled target protein and incubate with gentle agitation at 4°C for 90 min. The protein extract may be prepared in reticulocyte lysates or cell extracts. Use 5 uJ of radiolabelled lysate or 5-10 ug of COS whole cell extract (Protocol 2) to detect interactions with exogenous (overexpressed) proteins, or 20-50 ug of nuclear extracts (Protocol 3} to detect interactions with endogenous proteins. Save an aliquot of the protein extract for SDS-PAGE analysis (approx. 10% of the amount used in the binding reaction should be loaded). 5. Wash the beads four times for 30 sec with 1 ml NET-N or GST binding buffer. If GST binding buffer is used, remove the bovine serum albumin for the final two wash steps.

41

R. Sladek and V. Giguere Protocol 4.

Continued

6. Add 50 ul of SDS-PAGE sample buffer and boil samples for 2 min. Analyse 20 xl of the eluate by SDS-PAGE. Detect radiolabelled interacting proteins by autoradiography or Western blotting. To control for protein loading, analyse a sample aliquot by SDS-PAGE followed by Coomassie staining.

To test whether a particular receptor forms heterodimers with RXR, radiolabelled receptor protein may be synthesized in reticulocyte lysates. The labelled proteins are mixed with purified bacterial GST-RXR fusion protein bound to glutathione-Sepharose and precipitated by centrifugation: this also precipitates the fusion protein and any other proteins that are bound to it. A control reaction containing GST bound to glutathione-Sepharose is used to detect interactions that occur between the radiolabelled protein and either GST or the glutathione-Sepharose beads. The reaction mixture is washed and reprecipitated several times to remove loosely interacting protein. The wash buffer components, including detergents and salts, will determine the wash stringency and may be altered to disrupt non-specific protein interactions. Typical strategies to increase the wash stringency include using high and low salt buffers in sequential washing steps, as well as using buffers containing Triton X-100 or NP-40 in place of CHAPS. After the last wash is complete, SDS-PAGE sample buffer is added to the Sepharose-protein mix and the reaction products are analysed by SDS-PAGE. A parallel SDS-PAGE gel should be stained with Coomassie Blue to verify that each binding reaction contains equal amounts of the intact fusion protein. GST-pulldown experiments can be used to determine if chemical ligands influence the receptor's dimerization properties. In these experiments, the receptor ligand is included in the binding buffer together with the GST fusion protein and the target protein extract. When studying the effects of ligands, be careful not to use detergents which inhibit ligand binding: for example, NP-40 may inhibit oestradiol binding to the oestrogen receptor. In addition, the receptor ligand may stabilize the conformation of the receptor carboxy terminus, increasing the fraction of correctly folded protein contained in the binding reaction. Ligand stabilization of a partially folded receptor protein may be difficult to distinguish from true changes in the affinity of a receptorprotein interaction. While GST-pulldown experiments are easy to perform, they frequently detect false protein-protein interactions. These may result from the relatively high concentrations of proteins used in the binding reactions, from the specific binding and wash conditions used in the experiment, and from the potential presence of partially denatured proteins in the binding reactions. In addition, interactions detected by GST-pulldown experiments in vitro may be subject to competitive inhibition by other nuclear receptors or cellular proteins in vivo. 42

2: Initial characterization of new orphan receptors GST-pulldown experiments may fail to detect significant protein interactions due to the absence of other cellular proteins which are required for stabilizing the protein-protein interaction, the specific buffer conditions used in the experiment, or due to the exclusion of critical receptor domains from the fusion construct. As a result of these limitations, it is worthwhile to study potential protein interactions using additional techniques.

3.2 Mammalian two-hybrid system The two-hybrid system exploits the modular structure of transcription activators in order to provide a convenient assay for characterizing protein-protein interactions in yeast or mammalian cells (23). The assay was initially designed to provide an in vivo technique that determined whether two known proteins could physically interact (24) and was later adapted for screening expression libraries in yeast to identify novel interacting proteins (25). The two-hybrid assay was first developed using the yeast Gal4 transcriptional activator. This protein contains a compact DBD which is unable to activate transcription by itself as well as a larger transcriptional activation domain which has no DNA binding activity. The assay studies whether two chimeric proteins—one containing a 'bait' receptor protein linked to the Gal4-DBD, the other containing a 'prey' protein linked to a transcriptional activation domain—bind to each other. Interaction of the bait and prey proteins tethers the hybrid transcriptional activation domain, increasing the activity of a reporter gene that is controlled by a Gal4-responsive promoter. As the receptor 'bait' protein contains a heterologous DBD, the two-hybrid system cannot be used to detect DNA-stabilized interactions between nuclear receptors; however, the technique will detect protein interactions whose stability depends on endogenous cellular proteins. The mammalian two-hybrid system can be used to determine whether a nuclear receptor forms homodimers or heterodimers with RXR (Figure 2). The bait protein is constructed from the LBD of the receptor under study linked to the Gal4-DBD, while the prey protein contains the VP-16 activation domain linked to the LBD of RXR (26). A reporter plasmid driven by multimerized Gal4-response elements is transfected together with the fusion proteins: interaction between the bait and prey chimeras results in a marked increase in reporter activity. Control experiments should be designed to determine whether either fusion protein transfected alone induces the reporter activity. In addition, a parallel experiment that monitors the interaction between known heterodimeric partners (such as RAR and RXR) should be performed as a positive control. If the receptor has significant constitutive activity, it can be used to construct the 'prey' rather than the 'bait' hybrid protein. In general, the mammalian two-hybrid system provides a rapid and reliable method to screen for protein-protein interactions. It can be used to detect 43

R. Sladek and V. Giguere

Figure 2. Detection of RXR heterodimerization using the two-hybrid assay. Heterodimerization was studied using bait proteins expressing the Gal4 DNA binding domain (Gal4-DBD) (amino acids 1-147) fused to the RARa or ERRa C terminus, and a prey protein containing the VP16 activation domain (VP16-AD) (78 amino acids in length) fused to the RXRa C terminus. In a typical experiment, the CMV-driven expression vectors (250 ng of bait protein and 500 ng of prey protein) were transfected into COS cells which had been plated into 12-well culture dishes. Interaction was detected using a luciferase (LUC) reporter plasmid (1 ug per well) containing two copies of a Gal4 binding site (UAS) upstream of thethymidine kinase (TK) minimal promoter. Reporter luciferase activity was corrected for transfection efficiency using 500 ng of co-transfected internal control plasmid (CMV-p-galactosidase). Co-transfection of Gal4-RAR and VP16-RXR markedly induces the reporter activity as a result of interaction between the C termini of the two receptors: reporter activity is not increased when VP16-RXR is added to transfection reactions containing Gal4-ERR, indicating that the two fusion proteins do not interact.

both homodimer and heterodimer formation as well as interactions between receptor proteins and transcription factors or other cellular proteins (discussed in Chapter 5). Protein-protein interactions detected by the two-hybrid assay should be confirmed by GST-pulldown experiments, or in the case of receptor interactions with RXR, by EMSA. If the two-hybrid experiment fails to detect 44

2: Initial characterization of new orphan receptors interaction between receptor proteins, make sure that both the bait and prey proteins are stably expressed in the target cell. If the intact fusion proteins are expressed, try performing the assay using smaller amounts of the prey expression vector: VP16-containing proteins expressed at high levels in transfected cells may sequester general transcription factors so that they cannot interact with the reporter gene promoter. Interaction between two nuclear receptors may not be detected if the hybrid proteins do not fold properly or if they are missing determinants which stabilize dimer formation. In this case, hybrid proteins may be constructed to include regions of the receptor protein lying outside the LBD.

4. Identifying hormone-response elements There are several theoretical issues that should be considered before attempting to characterize the interaction between nuclear receptors and DNA. First, it is important to determine which source of receptor protein is best suited for DNA binding studies: this is particularly relevant for receptors that are known to be post-translationally modified, as this may alter their DNA binding affinity as well as their ability to recognize specific HREs. Secondly, HREs that are bound with high affinity in vitro may not always accurately reflect the classes of response elements that are most important for receptor function in vivo. Nuclear receptors frequently activate genes containing response elements that do not resemble the high affinity consensus binding sites defined by in vitro experiments. While some non-consensus binding sites may be recognized with the aid of bridging or adapter proteins, many are recognized directly without any stabilizing protein-protein interactions. A biological screen to identify potential receptor target genes and their regulatory regions may provide a useful adjunct to the in vitro characterization of receptor binding sites. Finally, the promoter context and chromatin structure may play important roles in determining whether a response element will function in vivo. Although in vitro techniques are not guaranteed to identify response elements that function in vivo, they have none the less been very successful for identifying potential target genes whose regulation may be characterized further in receptor 'knockout' mice and other biological systems.

4.1 Characterizing DNA binding proteins using the electrophoretic mobility shift assay The electrophoretic mobility shift assay (EMSA) is a powerful technique that can be used to characterize protein-DNA interactions (27-29). The technique is based on the observation that the electrophoretic mobility of DNA can be decreased when a protein is bound to it. The assay is easy to perform and its results are highly reproducible. Modifications of the basic EMSA protocol can be used to measure the receptor's DNA binding affinity as well as kinetic parameters related to its interaction with DNA (30). EMSA can also be used 45

R. Sladek and V. Giguere to identify a receptor's consensus binding site as well as the bases within the receptor's response element that are required for high affinity binding. Protocols 5 and 6 describe how to prepare DNA fragments for use in EMSA experiments, while Protocol 7 provides a method for performing EMSA (see also Chapter 4) using receptor protein prepared in reticulocyte lysates. DNA probes prepared using Klenow enzyme (Protocol 5) or T4 polynucleotide kinase (Protocol 6) may be used interchangeably in most experiments; however, methylation interference studies performed using kinased probes often produce clearer results, while EMSA studies using receptor proteins manufactured in reticulocyte lysates should be performed using end-filled probes to minimize artefacts caused by endogenous single-stranded DNA binding proteins. Protocol 5. Preparation of end-filled oligonucleotide probes using Klenow enzyme Equipment and reagents • Radiolabelled nucleotide ([a-32P]dCTP, specific activity 3000 Ci/mmol) • Complementary oligonucleotide strands with putative DNA response element

• Klenow enzyme . Sephadex G50 NICK (Pharmacia)

spin

column

Method 1. Mix 10 ug of each probe strand in 200 ul of Klenow buffer (50 mM NaCI, 50 mM Tris-HCI pH 7.5, 10 mM MgCI2). Heat the reaction mixture to 80°C for 10 min in a half-full 1 litre beaker and allow the reaction to cool slowly to room temperature. The annealed probe remains stable for at least three months when stored at -20°C. 2. Mix 100 ng of the annealed DNA together with 25 uCi [a-32P]dCTP and 200 uM each of dATP, dTTP, and dGTP in 50 ul of Klenow buffer. Add 1 U Klenow enzyme and incubate at 37°C for 1 h. 3. Separate the free nucleotides from the labelled probe by adding 150 ul of STE buffer (100 mM NaCI, 10 mM Tris-HCI pH 8.0, 1 mM EDTA) and centrifuging through a 1 ml Sephadex G50 spin column.

Protocol 6. Preparation of end-labelled oligonucleotide probes using T4 polynucleotide kinase Equipment and reagents • Radiolabelled nucleotide ([y-32P]dCTP, specific activity 3000 Ci/mmol) • Complementary oligonucleotide strands with putative DNA response element

46

• T4 polynucleotide kinase . Sephadex G50 NICK (Pharmacia)

spin

column

2: Initial characterization of new orphan receptors Method 1. Mix 100 ng of oligonucleotide DNA together with 25 uCi [y-32P]dCTP in 50 nl of polynucleotide kinase buffer. Add 10 U T4 polynucleotide kinase enzyme and incubate at 37°C for 1 h. 2. Add cold ATP to the reaction tube to a final concentration of 200 uM. Separate the free nucleotides from the labelled probe using a 1 ml Sephadex G50 spin column. 3. Mix 100 ng of labelled DNA with 100 ng of the complementary probe strand in 200 ul of Klenow buffer (see Protocol 5). Anneal the oligonucleotides as described in Protocol 5, step 1.

Protocol 7.

Electrophoretic mobility shift assay (EMSA)

Equipment and reagents • Radiolabelled oligonucleotides with putative DNA response element (Protocol 5 or 6) • Receptor protein obtained from cultured cells or reticulocyte lysate • Vacuum gel dryer

• Vertical gel electrophoresis system (Life Technologies, Model V16) • Film cassette and film for autoradiography of dried gel

Method 1. Protein extracts suitable for EMSA may be synthesized in vitro, using reticulocyte lysates according to the manufacturer's protocol, or obtained from transfected COS cells (Protocol 2) or crude nuclear extracts (Protocol 3). DNA probes may be labelled with Klenow enzyme (Protocol 5) or T4 polynucleotide kinase (Protocol 6). 2. Set up a non-denaturing 4-6% acrylamide gel3 using 0.5 x TBE running buffer (1 x TBE is 90 mM Tris-borate pH 8.5, 1 mM EDTA). Pre-run the gel at 7.5 V/cm for at least 30 min prior to loading. 3. Incubate 1 ug of whole cell lysateb or 1-50 ug of nuclear extract for 20 min on ice in a 20 ul reaction in EMSA binding buffer (10 mM Hepes pH 7.9, 100 mM KCI, 1 mM DTT, 0.05% NP-40, 2 ug poly(dl-dC), 100 ng single-stranded oligonucleotide NS (see step 4), and 10% glycerol). 4. Add 0.25 ng of labelled probe with or without 100-fold molar excess of cold competitor oligonucleotide and incubate for 10 min at room temperature. The oligonucleotide NS (5'-AGCTTGCGAAAATTGTCACTTCCTGTGTACACCGA-3') annealed to its complementary strand may be used as a non-specific competitor for control experiments. 5. Perform supershift experiments or ligand binding studies by adding antibodies or receptor ligands to the binding reaction, and incubate at room temperature for an additional 10 min. 6. Load the reaction mix onto the gel. Run the electrophoresis at 7.5-12.5 47

R. Sladek and V. Giguere Protocol 7.

Continued

V/cm for 90-150 min.a The progress of the electrophoresis may be monitored by loading a blank binding reaction containing 0.01% (w/v) xylene cyanol and 0.01% (w/v) methylene blue. In a 5% non-denaturing gel, the methylene blue will co-migrate with a 30 nucleotide probe. 7. Transfer the gel to moistened filter paper and dry by heating to 70°C under vacuum. Perform autoradiography with an enhancing screen at -80°C for 6-24 h. • See text for further details. b With reticulocyte lysate use an alternative binding buffer (see Protocol 88); see text for additional considerations when using nuclear extracts.

In the EMSA assay, a radiolabelled DNA probe is mixed together with one or more DNA binding proteins (Figure 3). The binding reaction will contain free protein, free DNA probe, and DNA-protein complexes: these components can be separated by non-denaturing acrylamide gel electrophoresis. The rate of migration of each component will be determined by its net charge, its mass, and its shape. The free DNA probe will migrate more rapidly through the gel than will DNA that is bound to protein. The mobility of the DNAprotein complex is primarily determined by the size and shape of the protein component, with larger proteins usually migrating more slowly than smaller proteins. Interventions that modify the conformation of the receptor protein, such as its interaction with ligand, can alter the mobility of the DNA-protein complex. The complex mobility can also be altered by changes in the DNA conformation, including receptor-induced bending of the DNA probe. EMSA is usually performed using a non-denaturing acrylamide gel. Probes ranging from 30-300 nucleotides in length and protein complexes of up to 500 kDa in mass can be studied using a 4-6% poly acrylamide gel (with an acrylamide:bisacrylamide ratio of 29:1). While the acrylamide concentration may be adjusted to improve the separation of multiple DNA-protein complexes, the optimal acrylamide concentration depends mainly on the size of the DNA probe: short DNA probes are best studied using higher concentration gels. For best results, use a gel that is at least 15 cm X 15 cm in size and 1.5 mm thick. The sample wells should be at least 0.8 cm wide to provide good complex separation and to prevent 'smiling'. Make sure that the well bottoms are flat and do not contain any loose debris. During electrophoresis, the protein-DNA complexes are focused by the buffer-acrylamide interface at the bottom of the well: any irregularity will distort the retarded probe band in the gel. A low ionic strength buffer such as 0.5 X TBE is used to prepare the gel and the running buffer. Low ionic strength buffers increase the DNAprotein complex mobility and may improve band separation. They may also increase the affinity of the DNA-protein interaction. Good buffering capacity is required to avoid significant pH shifts resulting from buffer electrolysis. 48

2: Initial characterization of new orphan receptors

Figure 3. Electromobility shift assay and antibody supershift. COS-7 cells were transfected with a vector expressing the orphan receptor ERRo. Whole cell extracts (1 ug per lane) were incubated with 0,25 ng of labelled oligonucleotide probe and analysed on a 5% acrylamide 0.5 x TBE gel. A single DNA-protein complex is detected in ERRa transfected cells (lower arrow, lanes 2 and 6), but not in sham transfected cells (lane 1 and 5). The complex is efficiently competed by a specific competitor oligonucleotide (NRRE-1) (lane 3) but not by a non-specific competitor oligonucleotide (NS) at 200-fold molar excess [lane 4). Addition of polyclonal ERRa antiserum produces a new low mobility 'supershifted' complex (upper arrow, lane 7), which is not produced by preimmune serum (PI, lane 8). Further details are provided in ref. 41.

Electrophoresis is performed at room temperature using a voltage of 7.5-12,5 V/cm. The gel should be equilibrated by applying power for 30 min prior to loading the binding reactions. If the receptor-DNA interaction is weak, better visualization of the DNA-protein complexes may be obtained by performing the electrophoresis at 4oC or by using 0,25 X TBE to prepare the gel and running buffer. Following electrophoresis, the gel is transferred directly without fixing to a sheet of filter paper and dried. Autoradiography is performed overnight using an intensifying screen, which should permit detection of a retarded band containing a few hundred c.p.m. of labelled probe. In EMSA experiments, electrophoresis is performed for time periods that often exceed the DNA-prolein complex half-life: the amount of DNA contained in a retarded complex represents a dynamic balance between DNA49

R. Sladek and V. Giguere protein complex formation and dissociation. The acrylamide matrix promotes complex reassociation by limiting the rate of migration of probe fragments that have recently dissociated from the receptor protein. In most cases, the retarded band will be sharply focused within the gel; however, low affinity interactions, which result from slow complex formation or rapid complex dissociation may often produce faint and diffuse bands. Formal characterization of the kinetics of DNA-protein complex formation may often reveal that receptors binding a common response element do so with significantly different rates of complex formation or disassociation (31). Kinetic analysis of receptor-DNA interactions should be performed whenever two or more receptors display overlapping DNA binding specificity (for further discussion of this topic, see ref. 30). The DNA probes used in EMSA can be made from synthetic oligonucleotides or from restriction fragments obtained from cloned genomic DNA. The DNA probe should be between 25-250 nucleotides in length: longer probes may produce DNA-protein complexes that cannot be well separated from the free probe, while shorter fragments may not form stable receptor-protein complexes or may denature during electrophoresis. DNA probes should be designed so that they have 5' overhanging ends which match the consensus sequence of two different restriction enzymes: if the first base of each overhang is different, then the two strands may be independently labelled by Klenow enzyme (see Protocol 5), simplifying the preparation of probes for methylation interference studies (Section 4.6). To study the interaction of a specific receptor protein with DNA, it is usually easiest to study protein binding to oligonucleotide probes. Once the receptor's basic response element has been characterized, the element's function can be studied within the context of a natural promoter. If the response element forms part of a complex regulatory unit, co-operative interactions with other DNA binding proteins contained in nuclear extracts may stabilize binding of a particular receptor. Alternatively, a complex response element may contain overlapping binding sites for multiple transcription factors that may prevent a particular receptor from binding to its response element by competing for its binding site. Once detected by EMSA, these interactions can be confirmed by functional promoter studies. The DNA sequences required for specific protein binding can be determined by adding non-radiolabelled competitor DNA fragments to the binding reaction: these fragments will compete with the labelled probe for receptor binding sites. The relative concentrations and relative affinities of the labelled and unlabelled DNA fragments in the reaction mix will determine the intensity of the retarded band. The specificity of the receptor-probe interaction can be demonstrated by showing that a DNA fragment that is completely unrelated to the receptor binding site, called a 'non-specific competitor', can prevent the receptor from binding to its response element. If the non-specific competitor is able to displace equimolar concentrations of the bound probe, then the 50

2: Initial characterization of new orphan receptors DNA-protein interaction has low sequence specificity. More detailed information about the probe nucleotides that are involved in forming a specific protein-DNA complex can be obtained by using competitor oligonucleotides which differ at one or more positions within the response element. Oligonucleotides which contain substituted bases at residues important for DNAprotein complex formation may bind the receptor with lower affinity than the unmutated site: in competition studies, they will displace bound probe less efficiently than the wild-type oligonucleotide, resulting in a smaller decrease in the intensity of the retarded band. These results can be combined with methylation interference studies (described in Section 4.6) to determine which response element residues are important for receptor-DNA interaction. The EMSA experiment described in Protocol 7 can be used to study the DNA binding properties of nuclear receptors that have been synthesized in transfected COS cells (Protocol 2). Good results are usually obtained when 0.5-1 ug whole cell extract are used in the binding reactions. When reticulocyte lysates are used in EMSA experiments, an alternative binding buffer (Protocol 8) produces clearly defined receptor-protein complexes when between 2-5 ul reticulocyte lysate is included in the binding reactions. Reticulocyte lysates often contain significant single-stranded DNA binding activity, which may produce interfering DNA-protein complexes. This can be prevented by increasing the amount of single-stranded oligonucleotide contained in the binding reaction. EMSA may also be performed using nuclear extracts: since this protein source may contain a complex mixture of transcription factors which bind the response element with widely varying affinities, the optimal binding reaction conditions must be determined experimentally. In particular, the total amount of nuclear extract as well as the amount of competitor polynucleotides must be carefully adjusted to obtain good results. This may be accomplished by performing two or three preliminary experiments. The first experiment determines the minimum amount of nuclear extract required to detect DNAprotein complexes: this is done by preparing several parallel binding reactions that contain between 1-50 |xg crude nuclear extract, together with 0.25 ng probe DNA, in a binding buffer that has no poly(dI-dC) or other competitor DNA. The amount of probe DNA may be increased for larger fragments. As the nuclear extract has been prepared in a high salt buffer (Protocol 3), the binding buffer salt concentration will need to be adjusted for each sample. In this titration experiment, increasing amounts of nuclear proteins will trap an increasing fraction of the probe in the gel well. The minimum amount of protein extract required to trap the labelled DNA completely should provide good EMSA results and should be used in all further studies. A second experiment is performed to determine the amount of poly(dI-dC) that provides optimal complex resolution. Parallel binding reactions containing between 1-20 ug poly (dI-dC) are prepared and analysed by electrophoresis: the reaction that provides the best complex morphology should be used in 51

R. Sladek and V. Giguere subsequent experiments. If these preliminary experiments produce faint or smeared complexes, consider incubating the binding reaction for a longer period of time, performing the electrophoresis at a lower ambient temperature, or increasing the concentration of the acrylamide gel. It may also be worthwhile to adjust the detergent, DTT, EDTA, and divalent cation (Mg2+ and Zn2+) concentrations in the binding buffer in order to increase the affinity of the interaction between the receptor and DNA. Other buffer conditions, such as the buffer, Na+ or K+ concentrations, and pH will less often affect EMSA results with most nuclear receptors. EMSA experiments performed using nuclear extracts commonly detect multiple DNA-protein complexes: as many receptors have similar electrophoretic properties, it is often impossible to identify which receptor forms a retarded complex based on its mobility alone. A receptor-specific antibody can be added to the binding reaction to help determine whether a specific receptor binds the HRE (Figure 3). The receptor-specific antibody may disrupt DNA-protein binding leading to the disappearance of the retarded complex, or may bind stably to the receptor-DNA complex, increasing the complex mass and producing a new low mobility 'supershifted' band. Receptor antibodies may also stabilize weak DNA-protein interactions: this effect is often non-specific and should not be used as evidence that a specific receptor forms part of the DNA-protein complex without careful control experiments. In addition, the pre-immune serum may affect the formation of DNA-protein complexes and may produce supershifted bands by binding non-specifically to DNA.

4.2 Identifying response elements using consensus binding sites A convenient initial approach towards discovering which binding sites are recognized by a particular receptor uses EMSA to determine the receptor's ability to bind to consensus response elements. All known nuclear receptors bind to at least one response element that contains one or more consensus half-sites, oriented as direct, inverted, and in some cases everted repeats which are separated by a spacer region containing between 0-5 nucleotides (Figure 7). The specific half-site sequence that the receptor recognizes will be determined by the amino acids contained in the receptor 'P Box', which consists of five amino acids at the base of the first zinc finger of the core DBD (see also below). Orphan receptors contain T box' residues which directly recognize nucleotides contained in the in the HRE core sequence (A/G)GGTCA. Consensus response sites are identified by performing EMSA using a panel of oligonucleotide probes, each of which contains a consensus HRE with half-sites separated by 0-5 base pairs in the direct and inverted repeat conformations. Parallel binding reactions are prepared using each of the consensus oligonucleotides: those sites which bind stably to the receptor may be 52

2: Initial characterization of new orphan receptors characterized further, for example, by determining whether changes in the composition of the spacer elements or of the 5' flanking region result in changes in their DNA binding affinity. Oligonucleotides for these subsequent studies can be designed based on natural response elements with the same site spacing as the target HRE or by saturation mutagenesis. Alternatively, binding site selection studies (Section 4.5) may be performed using Oligonucleotides in which random nucleotides have been placed in the area of interest (8). Although the consensus site method is conceptually simple, it has three significant limitations: (a) It requires synthesis of nearly 40 Oligonucleotides to determine whether the receptor binds to direct, inverted, and everted repeat response elements. (b) It cannot identify whether the receptor will bind to HREs that contain non-consensus half-sites. (c) It is not well suited for studying receptors which bind monomeric response elements, as the sequence of the 5' flanking motif appears to be specific to each receptor.

4.3 Using EMS A to detect RXR heterodimers and receptor homodimers EMSA experiments may be used to confirm whether receptor heterodimerization with RXR results in functional changes in DNA binding. Although protein interaction assays may not demonstrate direct interaction between a particular receptor and RXR, the two proteins may still form DNA-stabilized heterodimers: consequently, EMSA should be used to study the dimerization properties of all novel orphan receptors. To identify whether a receptor forms heterodimers with RXR, its DNA binding properties are compared to those of a protein mixture containing both receptor and RXR synthesized in vitro. As stable receptor dimers may form rapidly after the protein is synthesized, the experiment should be performed using three different reticulocyte lysate synthesis reactions: in two reactions, the receptor and RXR are manufactured individually; in the third reaction, the two proteins are synthesized simultaneously. The experiment may produce one of three outcomes. (a) The receptor may bind DNA independently of RXR, suggesting that the receptor binds DNA as a monomer or homodimer. (b) The receptor may bind DNA only when RXR is co-expressed, suggesting that the receptor binds DNA as a heterodimer with RXR. Participation of both receptors in the heterodimeric complex can be demonstrated by performing supershift experiments using antibodies that recognize the individual receptor proteins. 53

R. Sladek and V. Giguerc

Figure 4. Detection of receptor homodimerizaton by EMSA. The wild-type receptor forms a DNA-protein complex (R-R, lane 2) with lower mobility than a complex containing truncated receptor protein (AR-AR, lane 3). Combining the wild-type and mutant receptor proteins produces a DNA-protein complex with intermediate mobility (R-AR, lane 4), showing that the receptor binds DNA as a homodimer.

(c) The receptor's DNA binding specificity may be altered by RXR: for example, thyroid hormone receptor (TR) homodimers bind to HREs containing inverted repeats while TR-RXR heterodimers bind to HRHs containing direct repeats. Homodimer formation may be demonstrated by performing EMSA using a protein mixture that contains the wild-type receptor and an amino terminal truncated mutant. As the mobility of the mutant protein will usually be faster than that of the intact receptor, homodimer formation will result in the appearance of a new DNAprotein complex whose mobility lies between that of the wild-type and mutant receptors (Figure 4).

4.4 Determining the polarity of RXR heterodimers using chimeric receptors Nuclear receptors that bind to DNA as heterodimers with RXR will recognize response elements in which the two consensus half-sites are placed in a direct repeat configuration. These response elements differ from those containing inverted or everted repeats in that they have intrinsic polarity. To fully characterize the binding mode of RXR heterodimers, it is necessary to determine which half-site is occupied by RXR and which one by the new nuclear receptor. The receptor binding polarity, together with the response element spacing, plays a critical role in determining the potential interactions which may necur between the RXR-receptnr heterodimer and co-activator or co54

2: Initial characterization of new orphan receptors represser proteins which in turn regulates the ligand responsiveness of the receptor heterodimer (32). The binding polarity of RXR heterodimers can be determined using a minimal DNA binding domain switch assay (10). This assay studies the binding properties of chimeric receptors whose binding site specificity has been altered by mutating critical residues in the first zinc module T box' (Figure 5). In most orphan receptors, the 'P box' residues match the consensus 'EGCKX' which binds to the half-site (A/G)GGTCA: mutating three residues recreates the glucocorticoid receptor 'P box' 'GSCKV which will bind to the half-site AGAACA (called the glucocorticoid-responsive element (GRE) halfsite). EMSA experiments are performed using consensus response elements in which either the upstream or downstream half-site has been replaced by the GRE half-site. The ability of these synthetic HREs to form DNA-protein

Figure 5. Determination of heterodimer polarity using chimeric receptors. Orphan nuclear receptors recognize HREs containing the consensus half-site AGGTCA. Mutation of three residues in the 'P box' contained in the nuclear receptor DBD creates a chimeric receptor that recognizes response elements containing the consensus half-site AGAACA. Synthetic hormone-response elements containing two AGGTCA motifs will be recognized by heterodimers of the wild-type receptors (upper right). A mixture of the wild-type receptor and its 'P-box' mutated heterodimeric partner will recognize response elements in which one half-site has been switched to AGAACA: the specific site polarity recognized by the heterodimer identifies the half-site which is occupied by the mutant heterodimeric partner (middle and lower right). The method may be used to study receptor heterodimerization using EMSA or transient transfection assays.

55

R. Sladek and V. Giguere complexes is studied using mixtures of the wild-type and mutant receptor proteins, together with the mutant and wild-type forms of RXR (Figure 5). These results can be confirmed by performing transient transfection assays in which the ligand-dependent activation of reporter plasmids containing a single copy of a hybrid response element is studied using different combinations of expression vectors expressing the wild-type and mutant receptors. Heterodimer polarity may also be studied using oligonucleotides in which nucleotides in one half-site have been replaced with bromodeoxyuridine, a DNA-protein cross-linking agent that is activated by ultraviolet radiation (8). Following cross-linking, the size of the receptor that bound to the substituted

Figure 6. Identification of receptor DNA response elements using binding site selection (SAAB selection). Nuclear receptor protein synthesized in vitro is incubated with random oligonucleotides and the resulting DNA-protein complexes are resolved by EMSA. Programmed (lanes 2 and 4) and unprogrammed (lane 1 and 3) reticulocyte lysates are incubated together with labelled random oligonucleotide probe (lanes 3 and 4). The resulting protein complexes are resolved by EMSA. Parallel reactions containing a consensus oligonucleotide (lanes 1 and 2) are used to estimate the position of DNA-protein complexes containing the random oligonucleotide. The region of gel containing the bound random probe fragments is cut out (represented by the white square) and the eluted probe fragments are amplified by PCR. The PCR product is used as the input to the next round of selection. After several rounds of EMSA selection and PCR amplification, the bound oligonucleotides are cloned and sequenced.

56

2: Initial characterization of new orphan receptors half-site may be determined by SDS-PAGE analysis: Western blotting may be used to distinguish similarly sized receptors.

4.5 Identifying response elements using SAAB selection SAAB (selected and amplified binding sequence) selection is a PCR-based procedure used to identify receptor HREs (33). This method may be used to characterize monomeric HREs as well as to determine the precise half-site sequence recognized by a novel nuclear receptor (2). In the experiment, described in Protocol 8, a pool of random oligonucleotides is incubated with receptor protein expressed in reticulocyte lysates or transfected cells. Following EMSA, the bound nucleotide fragments are recovered, amplified by PCR, and used as probes for another round of EMSA (Figure 6). As the complexity of most receptor binding sites is high enough to prevent a distinct retarded complex from being visible following the first few rounds of selection, a parallel binding reaction containing a known receptor-DNA complex is used to determine the approximate position of the receptor-protein complex. If the receptor is known to form heterodimers with RXR, parallel binding reactions should be prepared which contain and omit the heterodimeric partner. If necessary, a truncated form of RXR may be used to better separate the retarded bands selected by RXR-RXR homodimers from those selected by receptor-RXR heterodimers. In subsequent selection rounds, the bound fragments usually form a clear retarded band. Following five or six rounds of selection, the fragments are subcloned and approximately two dozen fragments are sequenced. The fragments can be inspected visually to see if they contain a consensus arrangement of one or two half-sites, or they can be analysed by computer programs such as MatInd (34). If consensus sites are not readily recognized, further rounds of selection should be performed. Protocol 8. Identifying hormone-response elements using PCR amplification of random oligonucleotides (SAAB selection) Equipment and reagents • Receptor protein synthesized using reticulocyte lysate (TNT kit, Promega) • Vertical gel electrophoresis system (Life Technologies, Model V16)

• Vacuum gel dryer . Film cassette and film for autoradiography of dried EMSA gel • Apparatus for DNA amplification by PCR

A. Preparation of the random oligonucleotide probe 1. Label 100 ng of forward (5'-CGCGGATCCTGCAGCTCGAG-3') and reverse (5'-TGCTCTAGAAGCTTGTCGAC-3') primers using T4 polynucleotide kinase. Separate the labelled primers from the free oligonucleotides using a spin column (Protocol 6).

57

R. Sladek and V. Giguere Protocol 8.

Continued

2. Prepare a 50 ul PCR reaction containing 20 pmol of the random oligonucleotide 5'-CGCGGATCCTGCAGCTCGAG-N30-GTCGACAAGCTTCTAGAGCA-3' together with 400 pmol each of 32P-labelled forward and reverse primers. Amplify the random primer using three cycles of denaturation for 1 min at 95°C, annealing for 2 min at 55°C, and extension for 2 min at 72°C. B. EMSA selection of receptor response elements 1. Synthesize receptor protein in vitro using the TNT reticulocyte lysate kit (Promega). Prepare a binding reaction using 5 ul of programmed reticulocyte lysate together with 5 ul of the PCR product prepared in part A in a 20 ul reaction in RRL binding buffer (the final binding reaction buffer composition is: 10 mM Tris-HCI pH 8.0, 40 mM KCI, 1 mM DTT, 0.05% NP-40, 2 ug poly(dl-dC), 100 ng unlabelled forward and reverse primers, 0.1 ul boiled salmon sperm DNA, 10 p.g BSA, and 6% (v/v) glycerol). Prepare a control reaction containing 5 nl of unprogrammed reticulocyte lysate together with 5 uJ of the PCR amplified probes. To help identify the position of the bound PCR fragments set up parallel binding reactions containing 5 ul of programmed and unprogrammed reticulocyte lysate together with 0.25 ng of a consensus oligonucleotide that is bound by the receptor protein. 2. Perform electrophoresis using conditions that clearly separate the retarded DNA-protein complexes from the unbound DNA. 3. Transfer the gel to moistened filter paper and dry by heating to 70°C under vacuum. Expose to film at -80°C for 4-24 h using an enhancing screen. C. Recovery and PCR amplification of receptor-selected response elements 1. Align the autoradiograph and the EMSA gel, and then use a scalpel or single-edged razor blade to excise gel slices containing the bound DNA. 2. Elute the dried gel containing the bound fragments at 37°C for 2 h in 500 uJ of elution buffer (500 mM ammonium acetate, 1 mM EDTA). Extract the DNA with phenohchloroform and precipitate with ethanol. 3. Resuspend the selected fragments in 20 ul TE. Use 2 ul of the product in a 50 ul PCR reaction containing 100 pmol each of 32P-labelled forward and reverse primers. Amplify the fragments using 10-15 cycles of denaturation for 1 min at 95°C, annealing for 2 min at 55°C, and extension for 2 min at 72°C. Load 5 ul of reaction products on a 1.4% agarose gel. Perform only as many PCR cycles as required to barely see the amplified probe band. 4. Repeat part B, steps 1-3 and part C, steps 1-3 up to seven more times. 58

2: Initial characterization of new orphan receptors D. Cloning and sequence analysis of receptor-selected response elements 1. Once the random DNA probe produces a clearly visible retarded band, amplify the selected fragments and clone the reaction products in a reporter plasmid (e.g. TKLuc). 2. Sequence 20-25 of the cloned selected sites. If no consensus sequence can be identified, perform two further iterations of the selection process up to a maximum of eight iterations.

The SAAB selection procedure can be performed in the absence of other information about the receptor's response element; however, prior identification of a potential receptor response element simplifies recovery of the bound fragments during the initial selection round. SAAB selection is useful for identifying the optimal half-site sequence for monomeric and dimeric response elements as well as the preferred flanking residues in monomer binding sites. For dimeric sites, it has also been used to demonstrate 5' flank preferences (8). While EMSA experiments can be performed using an unbiased oligonucleotide pool, the study of the spacer and flanking regions may be performed more efficiently by using oligonucleotides which contain one or more consensus half-sites flanked by short regions of random nucleotides. SAAB selection can also be performed using genomic DNA in order to identify potential natural response elements. While this approach may identify natural response elements as well as the target genes that they regulate, it often selects functionally unimportant sites that are present in bulk chromatin. Selection of natural response elements may be facilitated by creating the DNA probe from fragments contained in 'active' chromatin. For example, DNA fragments can be isolated from the non-nucleosomal chromatin fraction using the CpGSAAB protocol (35). Binding site selection procedures can be performed using a bacterially expressed receptor fusion protein to affinity purifying the random fragments: this is a worthwhile approach if no binding site can be identified using the consensus site oligonucleotides (Section 4.2). One disadvantage of this approach is that it does not allow separation of complexes containing multimerized binding sites—in contrast to the EMSA-based protocol, which allows monomer and dimeric binding sites to be distinguished by their relative mobilities. The major limitation of SAAB selection is that it tends to discover monomer binding sites in preference to dimeric and more complex elements. This problem can be accentuated if the SAAB selected sites are overamplified. If a nuclear receptor can bind specifically to two or more distinct response elements, SAAB selection experiments will only identify the less complex response element— unless the receptor binds the more complex element with a high enough affinity to overcome the selection disadvantage resulting from the increase in sequence complexity. In addition, ligand binding may stabilize receptor heteroand homodimers: if the receptor ligand is not present in the EMSA binding 59

R. Sladek and V. Giguere reaction, then the dimeric site will not be selected and amplified. Finally, while the selected response elements may contain readily identifiable monomeric sites, these sites may be part of a dimeric response element containing a second cryptic half-site. Methylation interference studies can be performed using one or more of the selected sites to determine whether the receptor contacts nucleotides lying outside the consensus response element.

4.6 Identifying DNA-protein contacts using the methylation interference assay Methylation and ethylation interference assays can be used to identify individual nucleotides within the HRE that contact the receptor protein directly (36, 37). When SAAB selected fragments are used as the DNA substrate, the methylation interference assay may be used to determine the 5' extent of a monomeric binding site as well as to identify incompletely selected or cryptic dimeric response elements. In addition, if a receptor regulates known target genes, interference studies may be performed to identify natural HREs within the gene promoter. While methylation and ethylation interference assays both characterize the receptor binding of structurally modified DNA, they provide distinct information. Ethylation interference primarily identifies backbone phosphates that are in close proximity to a DNA-bound protein, while methylation interference detects specific contacts made between the DNA binding protein and purine residues. Methylation interference can be performed using DNA substrates between 30-250 nucleotides in length. Ideally, the receptor binding site should not be contained in the terminal 10% of the DNA probe. The methylation interference assay uses EMSA to study receptor binding of a chemically modified DNA as a probe (Protocol 9). Two DNA probes, each radiolabelled at one end, are treated with dimethylsulfate (DMS) so that a fraction of the nucleotides are chemically modified. DMS methylates the N-7 position of G residues—which is positioned in the DNA major groove—as well as the N-3 position of A residues—which is positioned in the minor groove (38). The modified DNA is incubated with receptor protein and bound and free DNA are separated by EMSA. Certain specific base modifications will interfere with receptor binding; as a result, the bound probe will not contain DNA that is modified at these residues. The bound and free DNA fractions are recovered from the gel and the modified DNA is cleaved chemically and analysed by denaturing PAGE. The accompanying protocol uses a cleavage method that cleaves probe fragments at methylated 'A' and 'G' residues; however, different cleavage protocols may be used to analyse the modified probe (39). Band intensities are compared between the cleaved 'bound' and 'free' probe fragments: probe fragments which are modified at nucleotide positions that block receptor-DNA interaction will not be contained in the bound fraction, resulting in absence of the corresponding cleavage product. 60

2: Initial characterization of new orphan receptors Successful methylation interference experiments depend on establishing electrophoresis conditions which separate bound complexes and free DNA: this is a particular problem in studies of promoter response elements, which often require isolation of multiple DNA-protein complexes. If electrophoresis cannot adequately separate the protein-DNA complexes, DNase I footprinting studies may be used to characterize receptor binding sites contained in longer DNA fragments. In comparison with methylation interference studies, DNase I footprinting provides a better view of the overall promoter structure; however, footprinting studies are often difficult to perform—particularly if the protein-DNA interaction is labile—and provide less specific information about the individual bases that are involved in protein-DNA contacts. Receptor-DNA interactions may also be studied by methylation protection experiments. In these studies, DNA fragments obtained from intact nuclei, cloned genomic DNA, or oligonucleotides are incubated with DNA binding proteins in vitro. When DMS is added to the reaction mix, the bound proteins will protect the DNA from modification. Following inactivation of the DMS, the purified methylated DNA is cleaved chemically and analysed by denaturing gel electrophoresis to detect the protected protein footprints. Methylation protection studies identify specific nucleotides that are masked by DNA binding proteins, but often do not provide a clear protein 'footprint', particularly if the protein complex has rapid association and dissociation rates. Protocols.

Methylation interference footprinting

Equipment and reagents • Receptor protein synthesized using reticulocyte lysate (TNT kit, Promega) • Dimethylsulfate (DMS), fume-hood, and facilities for DMS disposal • Vertical gel electrophoresis system (Life Technologies, Model V16)

• Film cassette and film for autoradiography of wet EMSA gel • DEAE membrane NA45 (Schleicher and Schuell) • Denaturing polyacrylamide gel electrophoresis system suitable for DNA sequencing

A. Preparation of hypomethylated probe DNA 1. Label oligonucleotide probes using T4 polynucleotide kinase as described in Protocol 6. Two probes are prepared, each containing a single radiolabelled strand annealed to an unlabelled strand. 2. Assemble the methylation reaction on ice by mixing 106 c.p.m. probe and 10 ug poly(dl-dC) in a final volume of 200 ul methylation buffer (50 mM sodium cacodylate pH 8.0, 100 mM MgCI2, and 1 mM EDTA). 3. Add 1 ul dimethylsulfate (DMS), mix the reaction by pipetting, and incubate for 10 min at room temperature. The incubation time will depend on the probe length. 10 min is usually adequate for probes of up to 50 nucleotides in length while 5 min is suitable for probes containing 50-100 nucleotides. The appropriate incubation time can be determined experimentally for each probe (see step 5). DMS is highly 61

H. Sladek and V. Giguere Protocol 9.

Continued

toxic: perform this step in a fume-hood and dispose of all reaction components in an appropriate biohazard waste container. 4. Terminate the reaction by adding 50 uJ of stop buffer (1.5 M sodium acetate pH 7.0, 1 M 2-mercaptoethanol). In order to remove the DMS completely from the methylated probe, precipitate the DMA twice using 750 ul of absolute ethanol and wash the pellet each time with 200 ul of cold 75% ethanol. Resuspend the DNA pellet in 20 ul TE buffer. 5. Test the methylated probe by adding 1-19 (uJ of cleavage buffer (10 mM sodium phosphate pH 7.0, 1 mM EDTA) and incubating at 90°C for 15 min. Add 2 uJ of 1 M NaOH and incubating at 90°C for 30 min. Precipitate DNA, resuspend in 5 ul of TE, and analyse using a 6-10% denaturing urea acrylamide gel. Dry the gel and perform autoradiography for 2 h at -70°C using an enhancing screen. A correctly prepared methylated probe should show a clear 'ladder' with approximately equal signal intensity at each cleaved 'G' residue. Bands corresponding to 'A' residues will be less intense as they are less efficiently methylated during probe preparation. B. EMSA isolation of receptor-bound probe fragments 1. Perform binding reactions as described in Protocol 7. Typically, the binding reaction is scaled up by a factor of six to ten and run in adjacent gel lanes. Use the entire probe reaction prepared in part A, step 4. Electrophoresis conditions should be chosen to provide good separation of the retarded DNA-protein complexes. 2. Expose the wet gel to X-ray film at 4°C for 4-24 h, depending on the intensity of the retarded band. C. Recovery and characterization of receptor-bound fragments 1. Align the autoradiograph and the EMSA gel, and then use a scalpel or single-edged razor blade to excise gel slices containing the free probe and the retarded DNA-protein complex. 2. Electroblot the DNA samples onto DEAE membrane (DEAE membrane NA45; Schleicher and Schuell). 3. Elute the DNA by submersing the DEAE membrane in 250 ul of high salt NET elution buffer (1 M NaCI, 0.1 mM EDTA, 20 mM Tris-HCI pH 8.0, 0.5 mg/ml proteinase K) and incubating the reaction at 60°C for 1 h. Remove the buffer and wash the membrane with an additional 50 uJ of high salt NET elution buffer. Pool the eluate and wash fractions. 4. Extract the eluate using 300 ul phenol:chloroform:isoamyl alcohol (25:24:1). Precipitate the DNA fragments by adding 10 ug of tRNA and 62

2: Initial characterization of new orphan receptors 1 ml of absolute ethanol and incubating at -70°C for 30 min. After centrifugation, wash the DNA pellet twice with cold 75% ethanol to remove excess salt. 5. Resuspend the DNA precipitate in 20 ul of cleavage buffer (10 mM sodium phosphate pH 7.0, 1 mM EDTA) and incubate at 90°C for 15 min. Add 2 ul of 1 M NaOH and incubate at 90°C for a further 30 min. 6. Precipitate the cleaved DNA by adding sodium acetate and ethanol, and wash twice with cold 75% ethanol. Resuspend the DNA in 5 nl TE buffer and count the probe. 7. Load equal numbers of probe counts in each lane of a 6-10% denaturing acrylamide urea gel. Dry the gel and perform autoradiography overnight at -70°C using an enhancing screen.

5. Transcriptional activation assays: transient and stable transfection Transcriptional activation experiments determine whether a receptor can activate or repress gene transcription when it binds to its HRE. These experiments are usually performed by transiently expressing the receptor in cells that contain a synthetic reporter gene whose expression is controlled by the putative HRE. Usually, the reporter gene is also transiently introduced into cells (transient co-transfection assay); however, the assay can also be performed using cell lines in which a synthetic target gene has been permanently introduced into a cellular chromosome (stable transfection assay). The transient co-transfection assay is usually the simplest method to demonstrate a functional interaction between a receptor and its HRE: stable transfection assays should be used whenever the chromatin conformation is likely to influence receptor activity significantly—for example, in studies of the transcription activation properties of many receptor co-regulatory proteins. Transient transfection techniques have been extensively reviewed in other volumes of the Practical Approach series (40). The following sections discuss how these techniques may be applied in characterizing the function of a novel orphan receptor (see also Chapters 3 and 8). Transient transfection assays may be performed using primary cell cultures or established cell lines. Cell lines are easier to maintain in culture and are usually simpler to transfect. Initial experiments should be performed using a cell line that is robust and that can be easily transfected: as steroids and lipids regulate the activity of many nuclear receptors, cell lines that do not grow well in media which have been depleted of these agents will not be suitable for experiments that study ligand-induced receptor functions. Good starting choices include HeLa, NIH-3T3, HEK293, CV-1, or COS cells. Subsequent experiments may be performed to study receptor function in lines that model 63

R. Sladek and V. Giguere tissues in which the receptor is expressed in vivo. For example, receptor activity can also be studied in cells as they differentiate using cell lines such as NIH-3T3-L1—which provides an easily transfectable model of adipocyte differentiation—or L6 and C3H10T1/2—which are commonly used to study muscle differentiation. Using transient transfection assays, receptor activity can be studied for up to 48 hours following transfection: differentiation studies can be performed over longer periods of time by using stably transfected lines. If the model cell expresses the receptor you wish to study, consider replicating the results using chimeric receptors (Section 4.4) to minimize the effects of endogenous proteins. In addition, it may be worthwhile to study receptor function in steroid secreting lines—such as the adrenocortical Yl line, the placental JEG-3 and JAR lines, and the testicular Tm3 and Tm4 lines. In these cell lines, steroid synthesis can be studied under basal conditions or following pharmacological stimulation. To study the transactivation properties of receptors that are highly expressed in these lines, the culture media can be removed from plates containing the steroid synthesizing cell and used to treat transfected flbroblasts: this conditioned medium contains high concentrations of intermediates and end-products of steroidogenesis. When performing ligand studies, make sure that the medium does not contain phenol red: this agent mimics the effects of oestrogens in increasing receptor activity. In transfection assays, the receptor's transcriptional activity is detected by a reporter plasmid which contains a cDNA expressing some easily detectable protein—such as firefly luciferase, chloramphenicol acetyltransferase, or Bgalactosidase—whose expression is controlled by a natural or synthetic promoter (Figure 7). The expression vector promoter should contain a HRE to which the receptor can bind in vitro, together with a basal promoter region to allow binding of general transcription factors. The basal promoter can be as simple as a TATA box, in which case efficient activation by receptor may require multimerized binding sites, or it may be more complex, in which case activation can often be detected by a single copy of the appropriate response element. Reporter plasmids may also be constructed using natural promoters. Confirmation of the receptor's interaction with the promoter can be provided by mutation of the putative HRE. As transient transfection assays measure the steady state reporter protein activity, the experimental results may not always reflect the rate of reporter gene transcription. For example, reporter genes whose transcript or protein have a prolonged intracellular half-life may not be suitable for measuring the activity of transcriptional repressers. In general, this is not a problem for most reporter systems, and changes in enzyme activity reflect changes in intracellular mRNA levels. However, rapid transcriptional responses are better studied using nuclear run-on experiments or in vitro transcription assays, both of which assess the receptor's transcriptional activity directly. The receptor DNA is cloned into an expression vector, which contains a strong promoter and splicing sequences that allow high level expression of the 64

2: Initial characterization of new orphan receptors

Figure 7. Transient transfection studies. The transactivation potential of the ERRa orphan receptor was studied using transient transfection of HeLa cells. Reporter constructs contain firefly luciferase (LUC), which is expressed under the control of nucleotides -105 to +51 of the thymidine kinase (TK) minimal promoter (TKLUC) together with a single copy of the ERRa extended half-site (ERRE, 5'-TCAAGGTCA-3') or the consensus oestrogen-responsive element (ERE, 5'-AGGTCACTGTGACCT-3'). An expression vector containing the cytomegalovirus immediate early promoter (CMV) was used to express ERRa and the VP16-ERRa fusion protein in the transfected cells. The cells were plated onto 35 mm culture dishes and co-transfected with 1 ug (per dish) of the indicated reporter construct, 200 ng of each expression vector, and 1 ug of the internal control plasmid (CMV-p-galactosidase). Co-transfection of VP16-ERRa markedly induces activity of the ERRE- and ERE-containing reporters: the wild-type receptor has no transcriptional activity in this study. The experiment demonstrates that a fusion receptor containing the ERRa DBD interacts with the target response elements: control experiments show that the fusion protein does not activate the TKLUC backbone and that VP16 alone does not activate any reporter construct. Further details are provided in ref. 41.

receptor protein. The promoter region is chosen to be active in a broad range of cell lines and is often obtained from promoter or enhancer regions that regulate the expression of a viral gene. Expression vectors usually contain constitutive promoters, such as the CMV immediate early promoter, the SV40 early or major late promoter, or the RSV-LTR enhancer; inducible pro65

R. Sladek and V. Giguere moters, such as the dexamethasone-regulated MMTV promoter, or synthetic tetracycline-responsive promoters, may be used in stable transfection experiments. The expression vector should contain RNA splice and polyadenylation sites to ensure that the transcribed message is stably expressed. The efficacy of a specific expression vector and transfection method can be tested by cloning an easily detected marker, such as (J-galactosidase, into the expression vector and studying its activity following transient transfection experiments. As many receptors are activated by ligands, they may not be transcriptionally active in cell culture systems that do not contain sufficient ligand concentrations. If the ligand has been identified, it can be added directly to the culture plate. If the ligand is unknown, receptor-HRE interactions may be demonstrated using fusion proteins in which a strong constitutively active transcriptional domain, such as the VP16 activation domain, is coupled to the amino terminus of the receptor. To ensure that the transfection efficiency is high and that the experimental results are reproducible, all plasmid DNA must be of high purity. DNA purified by caesium chloride gradient ultracentrifugation is the most appropriate source for plasmids that will be used in transfection experiments. Column purified DNA may also be used in transfection studies, provided that fine resin or hydroxyapatite particles were not eluted from the column together with the DNA. If the transfection efficiency is poor, consider reprecipitating the DNA to remove any impurities. For efficient transfection, the cell line should be healthy and never grown to full confluence. Passage the cells on the day before transfection and seed the culture plates at a high enough density so that they are 30-70% confluent on the day of transfection. DNA can be introduced into cells by a variety of techniques: the calcium phosphate coprecipitation technique works well for the fibroblast cell lines discussed above, and will introduce DNA into between 10-50% of the cells. If this method is not successful, consider transfecting the cells using lipofection, DEAE-dextran transfection, or electroporation. In transient transfection experiments, a mixture of receptor and reporter plasmids is introduced into mammalian cells using reagents that allow nucleic acids to cross the cell membrane. A third plasmid, called an internal control, is included to estimate the amount of exogenous DNA that has entered the cells: this plasmid commonly contains the p-galactosidase gene expressed under the control of the same promoter used in the expression vector. Using the calcium phosphate transfection protocol, the DNA is incubated with the tissue culture cells for between 6-24 hours after which the cells are washed twice with warm PBS. If there has been extensive cell lysis, additional washes may be required to remove all the precipitate and cell debris. After the cells have been washed, fresh complete media is added to the plate for an additional 12-36 hours. This medium may contain potential ligands or other pharmacological agents. On the final day of the experiment, the cells are washed twice with PBS, scraped from the plates, spun down, and lysed in 66

2: Initial characterization of new orphan receptors reporter assay buffer. The reporter activity can be detected using appropriate chemiluminescent or enzyme assays. Transfection experiments are usually performed using replicate plates to allow statistical comparison of the experimental results. Control experiments should be designed to detect interactions between endogenous DNA binding proteins and the reporter vector as well as interactions between the receptor and the reporter backbone. The amount of receptor and reporter plasmids used in the transient transfection experiment will vary with the specific cell line being studied and should be determined in a series of pilot experiments. The first pilot experiment determines the optimal amount of internal control plasmid: using the calcium phosphate co-precipitation technique, transfection mixes containing increasing amounts of the internal control plasmid are incubated together with the cells for 6-24 hours. For transfections that are performed on plates with 4 cm2 wells, the transfection should be performed using 0.1-1 ug of the internal control plasmid together with 1-4 ug carrier DNA per well. If the transfection causes significant cell death, reduce the amount of DNA transfected per culture plate or use a different transfection protocol. At least three replicates should be prepared for each combination of plasmid concentration and incubation time: the minimum amount of internal control plasmid and minimum transfection time that provides reproducible reporter activity lying within the linear range of the enzyme assay should be used in subsequent studies. The second pilot experiment establishes the correct amount of reporter plasmid. Using the optimal conditions established in the first experiment, 0.2-2 ug of the reporter plasmid should be transfected in place of equal amounts of carrier DNA to determine the amount of reporter plasmid that is required to increase the measured enzyme activity to ten times greater than background levels. The third pilot experiment determines the optimal amount of receptor expression vector: transfection experiments are performed using several different ratios of receptor and reporter plasmid to determine which combination causes maximal induction of the reporter. The amount of receptor plasmid used in this experiment will depend both on the promoter contained in the expression vector and on the complement of nuclear receptors which may compete for HRE binding sites in the reporter plasmid, as well as on other cell line-specific parameters. As many nuclear receptors bind similar response elements, a receptor's transcriptional properties should be confirmed using chimeric proteins that alter its DNA binding specificity. Two approaches can be used to produce chimeric receptors. The large domain swap approach replaces the entire DBD with a heterologous DBD. This approach has commonly replaced the nuclear receptor zinc cluster with the Cys2His2 zinc finger domain from the yeast Gal4 protein. The resulting chimera binds to a dimeric response element (UASgal) and activates reporter plasmids whose promoters contain this element. The receptor's DNA binding specificity can also be altered using more selective 'P box' mutations (Section 4.4). This strategy results in limited disruption of the 67

R. Sladek and V. Giguere receptor DBD and may provide a better substrate for transactivation and DNA binding studies than chimeric receptors that contain large domain changes. This approach is less successful when it is applied to complex response elements, where receptor interaction with flanking nucleotides and adjacent half-sites may compensate for small mutations in the core response element.

References 1. Willy, P. J. and Mangelsdorf, D. J. (1998). In Hormones and signaling (ed. B. W. O'Malley), Vol. 1, p. 307. Academic Press, San Diego. 2. Giguere, V., Tini, M., Flock, G., Ong, E. S., Evans, R. M., and Otulakowski, G. (1994). Genes Dev., 8, 538. 3. Wong, C.-W. and Privalsky, M. L. (1995). Mol. Endocrinol., 9, 551. 4. Glass, C. K. (1994). Endocr. Rev., 15, 391. 5. Wilson, T. E., Fahrner, T. J., and Milbrandt, J. (1993). Mol. Cell. Biol., 13, 5794. 6. Wilson, T. E., Fahrner, T. J., Johnson, M., and Milbrandt, J. (1991). Science, 252, 1296. 7. Rastinejad, F., Perlmann, T., Evans, R. M., and Sigler, P. B. (1995). Nature, 375, 203. 8. Kurokawa, R., Yu, V., Naar, A., Kyakumoto, S., Han, Z., Silverman, S., et al. (1993). Genes Dev., 7, 1423. 9. Perlmann, T., Rangarajan, P. N., Umesono, K., and Evans, R. M. (1993). Genes Dev., 7, 1411. 10. Predki, P. F., Zamble, D., Sarkar, B., and Giguere, V. (1994). Mol. Endocrinol, 8, 31. 11. Zechel, C., Shen, X.-Q., Chambon, P., and Gronemeyer, H. (1994). EMBO J., 13, 1414. 12. Viollet, B., Kahn, A., and Raymondjean, M. (1997). Mol. Cell. Biol, 17, 4208. 13. Hirata, Y., Kiuchi, K., Chen, H.-C., Milbrandt, J., and Guroff, G. (1993). J. Biol. Chem., 268, 24808. 14. Glass, C. K., Devary, O. V., and Rosenfeld, M. G. (1990). Cell, 63, 729. 15. Harper, S. and Speicher, D. W. (1997). In Current protocols in protein science (ed. J. E. Coligan, B. M. Dunn, H. Ploegh, D. W. Speicher, and P. T. Wingfield), p. 6.6.1. John Wiley & Sons, Inc., New York. 16. Smith, D. B. and Johnson, K. S. (1988). Gene, 67, 31. 17. Hochuli, E., Bannwarth, W., Dobeli, H., Gentz, R., and Stuber, D. (1988). Bio/Technology, 6, 1321. 18. Hopp, T. P., Prickett, K. S., Price, V. L., Libby, R. T., March, C. J., Cerretti, D. P., et al (1988). Bio/Technology, 6, 1204. 19. Kolodziej, P. A. and Young, R. A. (1991). In Methods in enzymology (ed. C. Guthrie and G. R. Fink), Vol. 194, p. 508. Academic Press, San Diego. 20. Bernard, A., Payton, M., and Radford, K. R. (1995). In Current protocols in protein science (ed. J. E. Coligan, B. M. Dunn, H. Ploegh, D. W. Speicher, and P. T. Wingfield), p. 5.4.1. John Wiley & Sons, Inc., New York. 21. Roy, R. J., Gosselin, P., and Guerin, S. L. (1991). BioTechniques, 11, 770. 22. Formosa, T., Barry, J., Alberts, B. M., and Greenblatt, J. (1991). In Methods in enzymology (ed. R. T. Sauer), Vol. 208, p. 24. Academic Press, San Diego. 68

2: Initial characterization of new orphan receptors 23. Phizicky, E. M. and Fields, S. (1995). Microbiol. Rev., 59, 94. 24. Fields, S. and Song, O. (1989). Nature, 340, 245. 25. Chien, C. T., Bartel, P. L., Sternglanz, R., and Fields, S. (1991). Proc. Natl. Acad. Sci. USA, 88, 9578. 26. Perlmann, T., Umesono, K., Rangarajan, P. N., Forman, B. M., and Evans, R. M. (1996). Mol. Endocrinol, 10, 958. 27. Revzin, A. (1989). BioTechniques, 7, 346. 28. Carey, J. (1991). In Methods in enzymology (ed. R. T. Sauer), Vol. 208, p. 103. Academic Press, San Diego. 29. Dent, C. L. and Latchman, D. S. (1993). In Transcription factors: a practical approach (ed. D. S. Latchman), p. 1. Oxford University Press, New York. 30. Stone, S. R., Hughes, M. J., and Jost, J. P. (1991). In A laboratory guide to in vitro studies of protein-DNA interactions (ed. J. P. Jost and H. P. Saluz), p. 163. Birkhauser Verlag, Basel. 31. Tini, M., Tsui, L.-C., and Giguere, V. (1994). Mol. Endocrinol., 8, 1494. 32. Kurokawa, R., Soderstrom, M., Horlein, A., Halachmi, S., Brown, M., Rosenfeld, M. G., et al. (1995). Nature, 377, 451. 33. Blackwell, T. K., Kretzner, L., Blackwood, E. M., Eisenman, R. N., and Weintraub, H. (1990). Science, 250, 1149. 34. Quandt, K., Freeh, K., Karas, H., Wingender, E., and Werner, T. (1995). Nucleic Acids Res., 23, 4878. 35. Shago, M. and Giguere, V. (1996). Mol. Cell. Biol, 16, 4337. 36. Jiricny, J. and Gorman, N. (1991). In A laboratory guide to in vitro studies of protein-DNA interactions (ed. J. P. Jost and H. P. Saluz), p. 121. Birkhauser Verlag, Basel. 37. Lakin, N. D. (1993). In Transcription factors: a practical approach (ed. D. S. Latchman), p. 27. Oxford University Press, New York. 38. Siebenlist, U. and Gilbert, W. (1980). Proc. Natl. Acad. Sci. USA, 77, 122. 39. Maxam, A. M. and Gilbert, W. (1980). In Methods in enzymology (ed. L. Grossman and K. Moldave), Vol. 65, p. 499. Academic Press, San Diego. 40. White, R. and Parker, M. (1993). In Transcription factors: a practical approach (ed. D. S. Latchman), p. 143. Oxford University Press, New York. 41. Sladek, R., Bader, J.-A., and Giguere, V. (1997). Mol. Cell. Biol., 17, 5400.

69

This page intentionally left blank

3

Adopting orphans: finding ligands S. KERSTEN and W. WAHLI

1. Introduction This chapter outlines the strategy for finding natural ligands for orphan nuclear receptors. The first part discusses the overall approach for identifying compounds or a class of compounds that can activate the orphan receptor. The second part focuses on the methods that can be employed to demonstrate that the previously identified activator is indeed a bona fide ligand, meaning that the compound is able to directly bind to the orphan receptor with high affinity. In the era of chemical-based libraries and high throughput screening to find novel synthetic ligands, the strategy for finding natural ligands may seem comparatively primitive and outdated. This, however, does not mean that the strategy is less effective. Also, it is the aim of this book to present methods that are accessible to an academic biomedical research laboratory. Therefore, high tech screening procedures as routinely employed by industrial laboratories for the identification of synthetic ligands fall beyond the scope of this chapter. Finding a ligand for an orphan nuclear receptor seems like a daunting task. It almost resembles finding a partner: you hit numerous dead ends until you finally run into the right one. The number of dead ends can be limited by following a sound and careful strategy. However, as often happens in science, walking off the main corridor may sometimes give you the important key that will lead you to the door of success.

2. First step: guessing ligand classes The first step in finding a ligand is purely intellectual. The search for the ligand can often be narrowed down based on some information that you already have about the orphan nuclear receptor. Various types of information can be used in this context. For instance, the expression pattern of the receptor in various tissues may give some insight into what tissue extract to use in the ligand screen, by assuming that the actual ligand is likely to be present in high concentrations in the tissue in which the orphan receptor is strongly expressed.

S. Kersten and W. Wahli Sometimes, a different tissue can turn out to be a better source for ligands. A perfect example in this regard was the use of breeding bull testis as a source for ligands of the liver X receptor a (LXRa), a receptor which, as its name implies, is most abundantly expressed in liver (1, 2). Ironically, transcripts of LXRa could not be detected in the testis. In some cases, the biological function of an orphan receptor has been deciphered in sufficient detail to allow for a more precise appraisal of the chemical class of the ligand. To illustrate this point, the peroxisome proliferator activated receptors (PPARs) had been known to play a critical role in the maintenance of lipid homeostasis. By employing the knowledge of the biological function of PPAR, fatty acids could be identified as activators (and thus potentially ligands) for this receptor (3, 4). Subsequent studies showed that fatty acids directly bind to PPARs (5-7). Another type of information that may prove useful is the extent of homology of the orphan receptor to other nuclear receptors with known ligands. If the homology is extremely high, the ligands may very well be identical, as was demonstrated for the nuclear receptors LXRa and OR1 (also named LXRP) (2).

3. Tissue fractionation If very little is known about the biology of the orphan receptor, one is forced to follow a long and often arduous road in the search for the ligand. The overall goal of tissue fractionation is to selectively purify an activator by following a series of partitioning steps that enrich for the activator (see Figure 1). Since most of the nuclear receptor ligands classified to date fall within the lipid category, it is likely that any of the endogenous ligands for orphan nuclear receptors are similarly small lipophilic molecules. The first step therefore consists of homogenizing tissue in organic solvents such as hexane or a 50% mixture of chloroform:methanol. In the next step, one needs to find a series of solvent mixtures that separate into two clear phases and in which the activator selectively partitions into one but not the other phase. The behaviour of the activator can be monitored by testing each of the phases in a transactivation assay (see below). If at each step the activator has a strong preference for one of the two phases, after a few rounds of partitioning the final solution should have become enriched for the activator. After cleaning up the solution by filtering it, drying it, and reconstitution in the appropriate solvent, the sample can be further fractionated by reverse-phase HPLC. Each of the HPLC fractions is subsequently verified by transactivation assay (see below) for abundance of the activator. The HPLC fraction that contained the activator may have to be reloaded on the HPLC for further purification (using the same or a different solvent). The HPLC fraction(s) that displays strong activity in the transactivation assay will be the source for chemical analysis to determine the identity of the active component. It is important to realize that compounds that are very sensitive to oxidation 72

3: Adopting orphans: finding ligands

Figure 1. Tissue fractionation using organic solvents. The tissue of interest is homogenized in organic solvent and subsequently enriched for the activator by several rounds of partitioning. At each round the phases are checked for activity by a transactivation assay. The final phase is run on HPLC. Fractions coming off the HPLC column are again checked in a transactivation assay. The solvents mentioned are from ref. 8 and are given for illustrative reasons only.

73

S. Kersten and W. Wahli and/or UV light or are otherwise labile may not survive the above treatments (this includes many known nuclear receptor ligands). Care should thus be taken to minimize degradation by avoiding light exposure and keeping the samples under vacuum or argon.

4. Transactivation assays Transactivation assays are cell-based assays that comprise the cornerstone in the identification of ligands for orphan nuclear receptors. Their main advantage is that they allow for screening of a large number of compounds in a reproducible fashion. Another asset is that mixtures of compounds or even organic extracts can be tested. The latter property makes transactivation assays the method of choice in the early stages of ligand searching. One of the disadvantages of the transactivation assays is that they only allow for identification of receptor activators, of which only a subclass will actually be bona fide ligands. This is mainly because precursors in the biosynthetic pathway of an endogenous ligand, that may be unable to bind directly to the receptor themselves, may be converted into the actual ligand inside the cell. It is also possible that a compound, when added to cells, stimulates the production and/or release of the actual ligand, or works through a different signalling pathway. For these reasons, a relatively large number of compounds may show a positive response in the transactivation assay, but only a small percentage of these compounds will show binding in a direct binding assay. In this regard, an example that is worth mentioning is the case of the retinoid X receptor. It had been demonstrated that the retinoid all-trans retinoic acid (tRA) was able to activate the retinoid X receptor in a classical cell-based transactivation system (9). However, follow up studies showed that activation is due to binding of a metabolite of tRA, 9-cis retinoic acid, to RXR and that tRA itself binds to RXR only with very low affinity (10, 11). Transactivation assays rely on the potential of nuclear receptors to activate transcription in response to binding of ligand. This is achieved by transfecting a cell line with an expression vector for the receptor(s) studied, and a reporter vector that has a binding site for the receptor. Binding of the liganded receptor to this binding site leads to activation of transcription of a so-called reporter gene. The reporter gene encodes a protein that, when incubated with the appropriate substrate, yields a product that can be measured. Two transactivation systems are generally used within the nuclear receptor field, which differ mainly in the way the receptor is positioned onto the DNA, and accordingly how the event of ligand binding is translated into transcriptional activation (Figure 2 and see below).

4.1 Standard transactivation assay The first system is illustrated in Figure 2A. In this system the expression vector codes for the full-length orphan receptor, that is, with its DNA binding 74

3: Adopting orphans: finding ligands

Figure 2. Schematic depiction of the transactivation assay. (A) Standard transactivation assay. Expression vectors coding for the orphan receptor and possibly RXR are transfected into cells together with a reporter vector. The reporter contains a high affinity binding site for the orphan receptor within a strong promoter linked to a reporter gene (LUC or CAT). (B) Gal4-LBD transactivation system. An expression vector coding for a fusion between the Gal4 DNA binding domain and the orphan receptor ligand binding domain are transfected into cells together with a reporter vector. The reporter contains multiple high affinity binding sites for Gal4 within a strong promoter linked to a reporter gene.

domain (DBD) and ligand binding domain (LED). For orphans which efficiently bind to DNA only as a heterodimer with the retinoid X receptor, it may be necessary, depending on the endogenous levels of RXR, to cotransfect an expression vector for RXR. The reporter vector contains a high affinity binding site (response element) for the orphan receptor (or orphanRXR complex) in front of a potent promoter linked to a reporter gene. Application of this reporter system is therefore dependent on knowledge of what constitutes a high affinity binding site (see Chapter 2). For some orphan receptors, this information is not necessarily available and in those cases one is forced to use the Gal4 fusion system, which is discussed below. It is common practise to use a reporter vector in which several copies of the response element are lined up together. This is done in order to augment the overall signal and to boost the ratio of the signal between the presence and absence of ligand (fold-activation). It should be realized, however, that most response elements for nuclear receptors found in natural promoters contain only one copy of a direct or inverted repeats of a consensus half-site sequence. Besides expression and reporter vectors, cells are transfected with a control vector that 75

S. Kersten and W. Wahli encodes the (B-galactosidase enzyme to correct for differences in transfection efficiency.

4.2 Gal4-LBD fusion transactivation system The Gal4-LBD fusion system is illustrated in Figure 2B. In this system the expression vector encodes a fusion protein between the DNA binding domain of the Gal4 transcription factor and the ligand binding domain of the orphan nuclear receptor of interest (12). The reporter vector contains a number of high affinity binding sites for Gal4 (called UAS, upstream activating sequence) in front of a potent promoter linked to a reporter gene. Note that in this system knowledge of the binding site preference of the orphan receptor is not required, as binding to DNA is mediated by the Gal4 DBD. Yet even in the case when the preferred response element for the orphan receptor is known, the Gal4 fusion system may offer some advantages, the most important one of which is the absence of interference by endogenous receptors. A more complete description of transactivation assays can be found in Chapter 2. Because the main use of the Gal4-LBD fusion system is the screening of potential ligands, a protocol for this system has been included here. Protocol 1. Transactivation via Gal4-LBD fusion proteins Equipment and reagents • 24-well plates for cell culture • Lipofectamine (Life Technologies) • Dulbecco's modified Eagle's medium (DMEM) • Salmon sperm DNA • Fetal calf serum, delipidated

Method 1. Plate CV-1 cells in 24-well plates in DMEM medium supplemented with 10% delipidated fetal calf serum. 2. Transfect the following using lipofectamine:3a reporter vector (50-100 ng), expression vector (5-10 ng), (3-galactosidase expression vector (10-100 ng), and carrier DNA (salmon sperm DNA, up to a total of 500 ng).b 3. 8 h after transfection, add compounds to the concentration of interest. Make sure the stock solution (in DMSO or ethanol) is diluted at least 1000 times in the culture medium. 4. Harvest cells after another 36 h and determine reporter (CAT or LUC) and p-galactosidase activities. Normalize data for transfection efficiency using the p-galactosidase activity as internal control. aA cheaper, but perhaps less efficient, way to transfect is by calcium phosphate coprecipitation. "Amounts of vectors may need to be optimized.

76

3: Adopting orphans: finding ligands

4.3 Quantitation of transactivation assays To compare the relative potency of compounds to activate transcription via an orphan receptor one can perform the transactivation assay at an identical concentration for all compounds and compare the signal. A preferred way to test the relative potency is by determining the dose-response relationship for each compound (range 10-10 M to 10-4 M) and from that the concentration at which half-maximal activation is achieved. The lower this value, the higher the potency and thus the higher the chance the compound is a true ligand. Overall, the transactivation assay should lead you into the direction of what class of compounds may constitute a ligand or ligands for the orphan nuclear receptor that you are studying. The actual identification of a certain compound or group of compounds as direct ligands for the orphan receptor rests on in vitro binding assays. Because of their time-consuming nature, in an academic laboratory most binding assays are unsuitable as a screening tool to test a large number of compounds. Instead, the binding assays are more appropriate to fish out ligands from a relatively narrow pool of compounds that showed activity in the transactivation assay. The strength of the binding assays is that, unlike the transactivation assays, they can provide direct evidence that a certain compound binds to the orphan receptor. However, each of these assays carries along a number of pitfalls that should be acknowledged in the interpretation of the data. Below, the most commonly used binding assays will be explained and described in detail.

5. Ligand binding assays Ligand binding assays can be divided into several classes, based on the main principle of the method. Some measure the competition for binding to the receptor with a known ligand, others visualize the change in conformation of the receptor in response to binding of ligand. Each of these assays has a particular requirement for materials, such as radioactive ligand or purified receptor protein. These requirements may impose limitations on the applicability of certain methods to the orphan receptor that is being studied.

5.1 Competition assay The competition assay is based on the principle that two ligands for a receptor will compete for occupancy of the ligand binding site (Figure 3). An underlying assumption is that the receptor being studied has only a single binding site for a ligand, a condition which, according to the prevailing paradigm, is met for all nuclear receptors. An important condition that may limit the applicability of the competition assay is the availability of a known ligand (either natural or synthetic) that has a reasonably high affinity for the receptor and is available in radioactive form. For most orphan nuclear receptors, for 77

S. Kersten and W. Wahli

Figure 3. Competition binding assay with radiolabelled ligand. The orphan receptor is incubated with a known ligand that is radiolabelled. The candidate ligand (not radiolabelled) is added in excess after which free from receptor-bound figand is separated on a column. The eluate is collected into tubes and counted.

which by definition the ligand is unknown, this condition presents an insurmountable obstacle. In some cases, people have been very successful in creating synthetic ligands for a certain receptor, and these synthetic ligands can subsequently be used in the search for natural ligands. For instance. Kliewer et al. (6) used a radiolabelled fibrate to demonstrate by competition assay that fatty acids directly interact with PPARs. Competition assays are most optimally carried out using a relatively pure preparation of the receptor protein. Nowadays, recombinant proteins can be expressed in bacteria or insect cells as fusion proteins and can be easily purified to an acceptable level of purity (see Chapters 2, 4, and 7). It is not necessary to have a construct of the full-length receptor as long as it contains 78

3: Adopting orphans: finding ligands the entire LBD. Alternatively, one can use nuclear extracts from cells in which the receptor of interest has been overexpressed. Reactions should be carried out in a buffer that is optimized for the receptor of interest. When setting up the reactions, a few things should be taken into account. One should choose the concentrations of protein and radiolabelled ligand in such a way that all the radiolabelled ligand added should be bound to the protein in the absence of competitor. This requires an excess of protein over radiolabelled ligand. The magnitude of the excess depends on the absolute affinity of the radiolabelled ligand for the receptor. Preferably, one would like to limit the amount of non-labelled ligand added to get noticeable competition, as solubility problems of the ligand may arise at higher concentrations (this is a major problem for the hydrophobic nuclear receptor ligands and there is little one can do about it). The amounts given below are for relatively low affinity ligands (Kd in hundreds of nM or low uM range) and may have to be adjusted for high affinity ligands. Protocol 2. Competition binding assay with radiolabelled ligand Equipment and reagents • Radiolabelled ligand . PD-10 desalting columns (Pharmacia) • Bradford reagent (Bio-Rad)

• Buffer: 10 mM Hepes pH 7.5, 1 mM EDTA, 50-100 mM NaCI, 5% glycerol

Method 1. Set up eight microcentrifuge tubes on ice. In each tube, in a 50 uJ volume, mix 10 ug of (semi) pure receptor proteina with radiolabelled ligand (final concentration of radiolabelled ligand should be 50 nM). 2. Add experimental ligand to a concentration of 0, 0.1, 0.3, 1, 3, 10, 30, and 100 uM. Incubate for 1-4 h. 3. Equilibrate columns with the appropriate buffer.b Layer the 50 ul sample on top of the column. Let sample run into columns and add buffer.c 4. Collect 3-4 ml of eluate.d 5. Count the total radioactivity of the eluate with a liquid scintillation counter. Express the radioactivity as a percentage of the radioactivity observed for the first sample (no experimental ligand added). • Assuming that the protein preparation is at least 40% pure, this will yield a final concentration of receptor between 1-10 uM. bVarious types of columns can be used that allow for separation of protein from ligand. We recommend PD-10 desalting columns from Pharmacia. c As a pilot experiment, it is recommended to determine after what volume of buffer the protein will elute. This can be done by collecting small fractions (300 ul) that are coming off the column and adding 100 ul of Bio-Rad Bradford reagent to each. Most likely, the protein will have eluted after approx. 4 ml of buffer. d Based on the elution profile of the protein, one can decide to discard the first 1-2 ml that come off the column. Alternatively, one can collect the full volume (3-4 ml).

79

S. Kersten and W. Wahli

6. Monitoring changes in receptor conformation X-ray crystallographic studies have revealed that binding of ligand to a nuclear receptor is associated with a specific change in conformation of the receptor. Changes in receptor conformation can thus be used as a read-out for binding of ligand. Four types of methods can be employed, each with their specific advantages and disadvantages.

6.1 Protease sensitivity This technique is based on the principle that proteolytic enzymes can recognize and cleave proteins at specific sequences. Due to the folding of the receptor protein, not all of these sequences are accessible to the enzyme and some may therefore not be cleaved by the enzyme. Incubation of a nuclear receptor with a certain protease (at fixed concentrations) thus yields a specific pattern of degradation products that are visible as a specific profile of bands on an SDS-PAGE gel (Figure 4). The essence of the method is that the conformational change concomitant with ligand binding may result in an enhancement or reduction of the accessibility of potential proteolytic cleavage sites, and therefore may alter the profile of bands observed by SDS-PAGE. It

Figure 4. Protease sensitivity assay. The orphan receptor is incubated with increasing concentrations of a certain protease (chymotrypsin, trypsin) and resolved by SDS-PAGE. The whole procedure is repeated in the presence of the candidate ligand. The change in protein conformation following binding of ligand may give rise to a slightly different pattern of degradation products, indicated in this cartoon as the emergence of a specific extra band at high protease concentrations.

80

3: Adopting orphans: finding ligands should be emphasized that large changes between presence and absence of ligand are rarely observed. Most often one can discern the emergence of a minor band when ligand has been added. This technique has been successful in the elucidation or affirmation of ligands for the nuclear receptors RXR, RAR, VDR, PPARa, and LXRa (2, 13-16). Enzymes that people have used include trypsin, chymotrypsin, and subtilisin. This does not mean that either or each of these enzymes will work for your particular orphan receptor but it makes perfect sense to try these enzymes first. Proteases can be purchased from various companies; we recommend Boehringer Mannheim. The receptor protein should be relatively pure (80-100%) and can either be synthesized by in vitro transcription/ translation (IVTT) or by bacterial or insect cell culture, and subsequently purified by column chromatography. The amount of protein required naturally depends on the methods of synthesis. For IVTT protein (using [35S]methionine) 10 fmol of protein/reaction will suffice. For protein produced in bacterial or insect cell culture, the amount required is manifold higher, up to 100 pmol if staining is by Coomassie Blue. Due to the high concentrations required, chymotrypsin and trypsin are less suitable when staining is done with Coomassie Blue. They may actually give a visible band themselves. Protocol 3 describes in detail the protease sensitivity assay for an orphan receptor produced by IVTT and digested by chymotrypsin. Protocol 3.

Protease sensitivity assay

Equipment and reagents • Chymotrypsin (Boehringer Mannheim)

• Equipment and reagents for SDS-PAGE

Method 1. Set up four microcentrifuge tubes per ligand tested (plus four tubes for 'no ligand'). Mix approx. 10 fmol of 35S-labelled receptor with ligand (concentrations may vary from 1 nM to 100 uM) in a 20 ul volume. Incubate for 15 min at room temperature. 2. Add chymotrypsin to final concentrations of 0 uM (tube 1), 3uM (tube 2), 15 uM (tube 3), and 50 uM (tube 4).a 3. Incubate for 20 min at room temperature. It is critical that the time of digestion is identical for all samples. 4. Stop the reaction by adding 5 x SDS sample buffer. Boil for 5 min and load on a 12-15% SDS-PAGE gel. aConcentrations for trypsin digestion are similar; for subtilisin tenfold lower concentrations are used.

81

S. Kersten and W. Wahli

6.2 Non-denaturing gel electrophoresis 6.2.1 Electrophoretic mobility shift assay (EMSA) The changes in protein conformation that often accompany binding of ligand can sometimes be visualized as an alteration in the mobility of the protein on a native gel. The most powerful application is the electrophoretic mobility shift assay (EMSA, also referred to as DNA gel shift), in which the receptor is run on a non-denaturing gel together with a small oligonucleotide that tightly binds the receptor. The principles for EMSA and the protocols for running it can be found in Chapter 2. In brief, samples of receptor plus oligonucleotide are incubated in the presence and absence of potential ligands and run in parallel on a native gel. The receptor can be synthesized by various means, such as bacterial or insect cell expression, IVTT, or obtained from nuclear extracts. Purification of the receptor is possible but this is not always necessary. Although this technique seems very straightforward and easy to use, it has only been employed to a limited extent in the nuclear receptor field. This is mostly due to the fact that the changes in mobility are often hard to discern. Prolonging the running time can sometimes be helpful. Matters are much facilitated if a known ligand is already available and can be tested. If the known ligand does not cause any change in mobility, it is not very likely that any other ligand will, although there may always be exceptions. Another possible limitation is that knowledge about the preferred nucleotide binding sequence is a prerequisite for running an EMSA. If the orphan receptor has only been superficially characterized, this information is often not available. 6.2.2 Ligand-induced complex formation (LIC) As its name already implies, the LIC assay reports on an enhancement of the formation of a complex induced by ligand (5). Most nuclear receptors bind to DNA as either homodimers or heterodimers, and, except for a few possible cases, each of the subunits of the dimeric complex is able to bind its ligand. The binding of ligand to any of the two subunits can have a marked effect on the propensity of the subunits to associate, and, as a result, on the binding of the dimeric complex to DNA. This effect can either be stimulatory or inhibitory. An example of this phenomena is the reported enhancement of DNA binding of RXR homodimers by a ligand for RXR, 9-cis retinoic acid (17). In the LIC assay, the increase in complex formation is most easily visualized by running an EMSA with the orphan and its partner receptor in the presence and absence of putative ligands. Protocols for running an EMSA can be found in Chapter 2. It should be emphasized that finding the proper conditions under which enhancement or reduction of complex formation is clearly observed can be a tough experiment (mainly finding the proper relative concentrations of the receptor(s) and of DNA). For this reason, it is of great benefit if a known ligand is already available with which one can optimize the conditions. 82

3: Adopting orphans: finding ligands 6.2.3 Non-denaturing gel electrophoresis in the absence of DNA One can also carry out non-denaturing gel electrophoresis in the absence of a DNA oligonucleotide. In this case one attempts to visualize the conformational change that occurs in response to binding of ligand to the receptor when it is not bound to DNA. The applicability of this technique depends on the availability of a receptor protein preparation of extremely high purity (> 90%). The best chance of achieving this is by overexpressing the protein as a fusion protein in bacterial cells (see Chapters 2 and 4). The protein should be purified to homogeneity by one or more columns (affinity columns and gel filtration). Non-denaturing gel electrophoresis is identical to SDS-PAGE except that SDS is omitted from all buffers (this includes the loading buffer). A parameter that can be altered to improve resolution is the pH of the running gel (7.5-9.5). Depending on the amount of protein loaded, the gel can either be stained with Coomassie Blue (5-50 ug) or by silver staining (100 ng to 1 ug). It is recommended to keep the voltage relatively low, as high voltages may result in smearing of the band(s). It needs to be said that the success of this method depends on a significant amount of luck, as only a small number of proteins resolved under non-denaturing conditions give a band tight enough for easy interpretation of the data. Playing with the conditions could help a little bit but success is not guaranteed.

7. Co-activator-dependent receptor ligand assay (CARLA) CARLA is based on the principle that binding of ligand to a nuclear receptor stimulates its interaction with a nuclear receptor co-activator protein (7). This ligand-dependent association is part of the normal pathway for transcriptional activation in which binding of ligand to a nuclear receptor leads to enhanced DNA transcription via several intervening proteins. CARLA can thus be viewed as a more functional binding assay that reports on the molecular consequences of ligand binding. Several co-activators for nuclear hormone receptors are known and, in theory, each of these co-activator proteins may be used in the CARLA assay (this includes SRC-1, CBP/p300, Tif2, Rac3, GRIP-1, and RIP140). Nuclear receptors have shown a moderate preference for a particular co-activator protein and it is recommended to start CARLA with the co-activator that shows the strongest ligand-dependent association with the receptor of interest. Unfortunately, knowledge about the preferred co-activator is most likely not available for newly cloned orphan receptors. In that case it is recommended to test one of the 'broad spectrum' co-activators such as SRC-1. If this proves unsuccessful, one can test the next co-activator in line. The principle of CARLA is fairly simple. In essence, CARLA is nothing 83

5. Kersten and W. Wahli else than a GST-pulldown assay using a GST fusion of the orphan receptor (see also Chapters 2 and 4) and in vitro transcribed/translated co-activator protein (35S-labelled) (Figure 5). The GST fusion protein is immobilized on glutathione-Sepharosc beads and incubated with the co-activator protein in

Figure 5. Co-activator-dependent ligand assay (CARLA). A GST fusion protein of the orphan receptor is bound to glutathione beads. In vitro transcribed/translated co-activator protein PS-labelled, in this cartoon SRC-1) is added together with the test compound. Ordinary GST-pulldown procedures are followed. Beads are loaded on an SDS-PAGE gel. Staining is first by Coomassie Blue to check whether equal amounts of beads with the GST fusion protein have been loaded. The gel is dried and autoradiography is performed to reveal the pulled clown co-activator. Cartoon has been adapted from ref. 18. 84

3: Adopting orphans: finding ligands the presence and absence of potential ligands. An actual ligand for the nuclear receptor will enhance the interaction of the receptor with the coactivator and thereby increase the amount of co-activator that is pulled down in this assay. Because some pulldown of the co-activator may occur even in the absence of any ligand, it is of benefit if one already has a known ligand to include as a positive control. The protocol outlined below has been tested and validated for the PPAR nuclear receptors. Some modification of this protocol is likely to be necessary for other nuclear receptors. Protocol 4. Co-activator-dependent receptor ligand assay 25000 g for 45 min). Save the supernatant for CARLA at -70°C (make multiple aliquots of 3 ml). 3. Use 2.5 ml of supernatant per 50-200 ul of beads for ten CARLA reactions.c Equilibrate 50-200 ul of beads in NETN. Spin down beads in microcentrifuge at low speed (3000 r.p.m.) and wash twice. Mix beads with 2.5 ml of supernatant and incubate at 4°C for 30-60 min on rotating wheel.

85

S. Kersten and W. Wahli Protocol 4.

Continued

4. Spin down the beads at low speed (1000 r.p.m. in a big rotor for 2 min). Wash beads three times with NETN and resuspend in 5 ml NETN with 0.5% (w/v) dried skimmed milk. 5. In microcentrifuge tubes incubate 500 ul of beads with 2 ul of 35Slabelled SRC-1 lysate. Add NETN buffer (with added milk) to total reaction volume of 1 ml. Add ligand to the concentration of interest. 6. Incubate at 4°C for 2-4 h on rotating wheel. 7. Spin down beads in microcentrifuge at low speed (3000 r.p.m.). Wash three times with NETN buffer (without milk). Make sure not to disturb the pellet while removing the supernatant. 8. Dry the pellet in vacuum dryer. Resuspend in 15 ul H2O. Prepare samples as normal for SDS-PAGE. Load on 7% acrylamide gel (including beads). Run gel, dry, and visualize by autoradiography. a These are guidelines specific to the PPAR receptors. Consult Chapters 2 and 4 on how to prepare GST fusion proteins from bacterial cultures. b Be aware that some of these treatments can denature the protein so be careful and make sure the lysate is kept on ice. c lf your protein expresses very well and a very large amount of it is soluble (visible as a nice fat band when 5 ul of the bacterial supernatant is run by SDS-PAGE) you may downscale accordingly (e.g. 1 ml of supernatant per 50-100 ul of beads for ten CARLA reactions).

In the initial screen for ligands, it is recommended to use a relatively high concentration of ligand, such as 100 uM. Compounds that test positively in the CARLA assay can then be further analysed by determining the doseresponse relationship. The relative intensity of the bands can be compared by scanning the autoradiograph and quantifying the bands with a suitable analysis program. After plotting the concentration of ligand against band intensity, the EC50 (concentration of ligand at which half the total enhancement is observed) can be calculated. Because CARLA is not an equilibrium method (see below), values obtained will underestimate the true ligand binding affinity.

8. Fluorescence spectroscopy Fluorescence spectroscopy, when properly applied, can prove very useful in the identification of new ligands, mainly in the later stages when the actual binding affinity has to be determined. The main obstacle in applying fluorescence spectroscopy is the high cost of equipment. A fluorometer is usually not part of the standard equipment of an ordinary molecular biology or even biochemical laboratory. But almost every campus harbours at least a few fluorometers with plenty of expertise surrounding them. It is often worth it to 86

3: Adopting orphans: finding ligands spend some time and effort to familiarize yourself with the broad utility that fluorescence spectroscopy provides. The main asset that fluorescence spectroscopy contributes to the spectrum of ligand binding techniques that have been discussed so far is that it is an equilibrium method. What this means is that separation of free from bound (as in radioactive binding assays, assay based on running of gels, and in CARLA) is not necessary. The whole binding reaction takes place in a 0.5-5 ml cuvette in solution. As a result, fluorescence spectroscopy allows for an accurate measurement of the dissociation constant of binding (Kd). For instance, new data now suggest that Kd values for binding of certain fatty acids and of leukotriene B4 to PPARa are approximately 100-fold lower than those obtained with non-equilibrium methods such as competition assays, CARLA, or LIC (19).

8.1 Principles and utility of fluorescence spectroscopy Fluorescence spectroscopy is based on the principle that certain molecules are capable of absorbing light of a certain wavelength (called excitation) and reemitting light of a different (longer) wavelength. Proteins are inherently fluorescent due to the presence of fluorescent amino acids such as tryptophan and tyrosine. This results in the ability of proteins to absorb light at 280 nm and to re-emit it at 340 nm. Proteins can also be labelled with fluorescent probes such as fluorescein or rhodamine. These probes often react specifically with certain parts of the protein molecule. An interesting property of fluorescent molecules and an important principle on which many protein fluorescence methods are based is the sensitivity of the fluorescence to changes in the chemical environment. For instance, movement of a fluorescent group to a more hydrophobic environment results in a shift of the emission maximum to a shorter wavelength (blue shift). It is thus easy to understand how changes in protein conformation (such as those following binding of ligand) can have measurable effects on the fluorescent signal at a certain wavelength. Alternatively, it is possible that the protein conformation changes without significant changes in the environment around the fluorescent amino acids or the attached probe. In that case, ligand binding is unlikely to cause measurable changes in fluorescence. Some compounds, such as the retinoid ligands for the RAR and RXR nuclear receptor, have an absorption spectrum that directly overlaps with the emission spectrum of proteins. This allows for a very convenient and highly quantitative way to measure ligand binding (20, 21). Unfortunately, few compounds display this type of behaviour and therefore the applicability of this particular method to the field of nuclear receptors is limited. Besides the availability of the proper equipment, another restriction on the use of fluorescence spectroscopy is the requirement for reasonable quantities of purified receptor protein (> 80%). Guidelines for synthesis of recombinant proteins can be found in Chapters 2, 4, and 7. 87

S. Kersten and W. Wahli As was alluded to before, proteins can either be used without modification (utilizing the intrinsic fluorescence of proteins) or after labelling with a fluorescent probe. Finding the proper condition for fluorescent labelling of a protein can be a time-consuming business. Numerous probes can be tried and it is often a matter of trial and error to find out which probe gives high and reproducible labelling (see the Molecular Probes catalogue for a complete list). The various probes differ by their fluorescent group and also in the reactive group that attaches to the protein (e.g. amine reactive, carboxyl reactive). The most important variable that is controllable and that determines labelling efficiency is the pH of the solution in which labelling is carried out. If labelling is not very successful, it may be worthwhile to repeat the labelling at higher or lower pH, taking into account the stability limits of the protein. Protocol 5.

Labelling of receptor protein with a fluorescent probe

Equipment and reagents Fluorescent probe: e.g. fluorescein, rhodamine, dansyl (see the Molecular Probes catalogue on detailed properties of fluorescent probes) Dialysis bags (Spectrum)

Buffer: 10 mM Hepes pH 8.0, 0.5 mM EDTA, 0.5 mM DTT, 100-400 mM KCI or NaCI, 5% glycerol Bradford protein assay reagent (Bio-Rad)

Method 1. Dilute protein to 10-20 uM in 0.5-2 ml buffer. Add five-to tenfold molar excess of the probe (probe is stored as a 10 mM stock in DMF or DMSO). 2. Incubate at 4°C or at room temperature for 1 h to overnight. If labelling efficiency is high, 2 h at 4°C may suffice. 3. Remove free probe by dialysis at 4°C. Change buffer twice (after 5 h each). 4. If necessary, spin down aggregates. Save protein at 4°C. 5. To check labelling, perform an absorption scan of the (diluted) protein on a spectrophotometer. If labelling was successful, a peak should be observed at a wavelength corresponding to the absorption maximum of the probe (remember, the peak at 280 nm is the protein, not the probe!). Read off absorption and convert to concentration using the molar extinction coefficient of the probe (see catalogue). 6. Determine the concentration of the protein by standard protein assays such as Bradford (Bio-Rad) or Lowry.a One can now determine the labelling efficiency. a Theoretically, the protein concentration can also be obtained by reading the absorption at 280 nm. However, the attached probe may distort the protein absorption spectrum resulting in an inaccurate estimate of the protein concentration.

88

3: Adopting orphans: finding ligands

8.2 Emission scans The strength of fluorescence spectroscopy is its ability to determine actual binding affinities of a small group of compounds that were demonstrated to be ligands by some other method. Nevertheless, to some extent, fluorescence spectroscopy can also be applied as a non-quantitative method at an earlier stage in the ligand screening. For this particular application, emission scans (fixed excitation wavelength, variable emission wavelength) are performed with the purified receptor protein in the absence and presence of putative ligands. For instance, if one were to look at the intrinsic protein fluorescence, the excitation wavelength would be 280 nm and the range of the emission wavelength would be from 300-400 nm (wavelengths are correspondingly different for proteins labelled with a fluorescent probe). Binding as a ligand is visualized as a change in the spectrum of the protein in the presence of an excess of a certain compound. An approximate tenfold excess of compound over protein is recommended (for instance: 1 uM protein, 10 uM compound). This will guarantee that binding will be close to saturation, at least for ligands that bind with moderate affinity. Shifts in emission maxima are most indicative of actual binding but increases or decreases of the peak fluorescence value without concomitant shift of the emission spectrum may be informative as well (Figure 6). Since the fluorescence signal is an absolute value, it is pertinent that any handling that may lead to a change in magnitude of the

Figure 6. Use of fluorescence spectroscopy to study ligand binding. An emission scan is performed with the purified receptor in the absence and presence of the test compound (left). In this example, the (apparent) binding of the compound is associated with a rightward shift of the emission maximum, resulting in a decrease of the absolute fluorescence value at the indicated wavelength. Note that changes in absolute fluorescence may occur without a concomitant shift in the emission maximum. To perform a fluorescence titration, the wavelength at which the difference in fluorescence between minus and plus ligand is maximal is selected and fixed. To the non-complexed protein (apo-receptor), ligand is added stepwise until saturation is reached. An appropriate analysis program is applied to calculate the dissociation constant and the number of binding sites.

89

S. Kersten and W. Wahli signal is avoided. It is therefore recommended to compare the signal in the absence and presence of a certain compound in the same exact solution (take scan without compound, add compound and re-take scan). Also, it is essential that the response to a candidate ligand is compared to that of a non-ligand and (if possible) to that of a known ligand. Once a clear effect has been established, the next step is to validate the observed response by performing a titration (see Figure 6).

8.3 Fluorescence titrations The aim of the titration is twofold. First, one would like to verify whether binding is stoichiometric, meaning that 1 mol of ligand is complexed with 1 mol of receptor. This is to rule out the possibility that the observed binding may be non-specific, in which case binding would likely be non-stoichiometric. Secondly, titrations have the advantage that they allow for accurate quantitation of the ligand binding affinity under equilibrium conditions. The principle of how to set up a titration based on the pattern observed in the emission scan is illustrated in Figure 6. Protocol 6. Fluorescence titrations Equipment and reagents • Fluorescence quartz cuvette (four walls transparent) • Syringe that allows for accurate pipetting of 1 ul volumes of solvents such as ethanol

. Buffer: 10 mM Hepes pH 8.0, 0.5 mM EDTA, 0.5 mM DTT, 100-400 mM KCI or NaCI, 5% glycerol . Fluorometer

Method 1. Determine the emission wavelength where the difference in fluorescence between minus and plus ligand is maximal by using the emission scans of the receptor protein in the absence and presence of the test compound (Figure 6). 2. In a fluorescence quartz cuvette, prepare a 1 uM solution of the protein in buffer and measure the fluorescence at fixed excitation and emission wavelengths. Stepwise add ligand from a concentrated solution (depending on the binding affinity, take 100-500 nM increments), mix by inversion, and measure fluorescence after each step. 3. Keep adding ligand until the drop in fluorescence levels off and continues as a more shallow, linear decrease.a 4. Correct the curve for inner filtering. Apply the appropriate analysis program to compute the dissociation constant. a This latter phase is due to the quenching effect of free ligand in the solution on the fluorescence signal (called inner filtering).

90

3: Adopting orphans: finding ligands

9. Conclusions If anything should be clear from this chapter it is that the road towards the identification of ligands for orphan receptors is likely to be rough and bumpy. Even worse, it lacks road signs! This chapter has attempted to make things a little bit easier by providing an overview and detailed protocols for each of the steps along the way. Because no nuclear receptor and their putative ligands are alike, it is impossible to provide a perfect roadmap that will guarantee an easy ride. Each track requires a thorough analysis of the situation and can sometimes only be passed by trying out several routes. In the process of finding a ligand or ligands for orphan receptors it is important to realize that the concept of ligand is not an all or nothing principle. Naturally, a myriad of compounds are clearly NOT ligands for a certain receptor and a certain compound may show all the properties of a true bona fide ligand, but often one is working in the grey zone where the classification of whether a certain compound is a ligand or not is not easy to make. What this means is that numerous factors should be taken into account in the evaluation whether a compound is a true ligand in vivo. A very important factor that has been repeatedly addressed in this chapter is the absolute binding affinity, expressed as Kd (dissociation constant). The lower the Kd, the higher the affinity, and thus the higher the chance that a compound is a true ligand. Most of the assays discussed in this chapter allow for at least a semiquantitative way to measure the binding affinity. The second piece of information that should be taken into consideration is the concentration of the putative ligand at the site where it is binding with the orphan nuclear receptor, i.e. the cell nucleus. Estimates for the actual concentration of a ligand in the nucleus are impossible to come by and instead one relies on the plasma concentration, which may or may not be very accurate. To illustrate this point, lets say we are dealing with the following problem. Compound A may have a relatively high absolute binding affinity but cannot be detected in plasma or cell extract, while compound B has a 100-fold lower binding affinity but is present at high concentrations in the plasma. Compound A thus works beautifully in in vitro binding assay but its absence in cells and plasma would suggest that it cannot represent an in vivo ligand for the receptor of interest. On the other hand, compound B, because of its abundance, has a good chance of being an actual in vivo ligand. Of course, one should always take into account that the lack of detection of a certain compound does not mean a compound is not present but rather that its concentration is below the detection limit. The confusion in the classification of whether a certain compound is a ligand or not is further complicated by the concept that perhaps several nuclear hormone receptors do not conform to the traditional paradigm of one receptor binding one high affinity ligand. Instead, some nuclear receptors may bind a whole array of related compounds with moderate or low affinity. In this 91

S. Kersten and W. Wahli context, nuclear receptors such as the pregnane X receptor, benzoate X receptor, PPAR, and LXR come to mind (2, 5-8, 22). It is very difficult to judge whether low affinity binding of several ligands to a certain nuclear receptor reflects the situation in vivo or signifies that the true, high affinity, ligand just has not yet been identified. From the above discussion it seems that it is almost impossible to come up with all the evidence showing that a certain compound is a ligand for a certain orphan receptor without leaving some gaps. Inspection of past efforts in the nuclear receptor field tells us that this has been universally the case but one can find consolation in the fact that the gaps do eventually get filled.

References 1. Willy, P. J., Umesono, K., Ong, E. S., Evans, R. M., Heyman, R. A., and Mangelsdorf, D. J. (1995). Genes Dev., 9, 1033. 2. Janowski, B. A., Willy, P. J., Rama Devi, T., Falck, J. R., and Mangelsdorf, D. J. (1996). Nature, 383, 728. 3. Gottlicher, M., Wildmark, E., Li, Q., and Gustafsson, J.-A. (1992). Proc. Natl. Acad. Sci. USA, 89, 4653. 4. Keller, H., Dreyer, C., Medin, J., Mahfoudi., Ozato, K., and Wahli, W. (1993). Proc. Natl. Acad. Sci. USA, 90, 2160. 5. Forman, B. M., Chen, J., and Evans, R. M. (1997). Proc. Natl. Acad. Sci. USA, 94, 4312. 6. Kliewer, S. A., Sundseth, S. S., Jones, S. A., Brown, P. J., Wisely, G. B., Koble, C. S., et al. (1997). Proc. Natl. Acad. Sci. USA, 94, 4318. 7. Krey, G., Braissant, O., L'Horset, F., Kalkhoven, E., Perroud, M., Parker, M. G., et al. (1997). Mol. Endocrinol, 11, 779. 8. Blumberg, B., Kang, H., Bolado, J. Jr., Chen, H., Craig, A. G., Moreno, T. A., et al. (1998). Genes Dev., 12, 1269. 9. Mangelsdorf, D. J., Ong, E. S., Dyck, J. A., and Evans, R. M. (1990). Nature, 345, 224. 10. Levin, A. A., Sturzenbecker, L. J., Kazmer, S., Bosakowski, T., Huselton, C., Allenby, G., et al. (1992). Nature, 355, 359. 11. Heyman, R. A., Mangelsdorf, D. J., Dyck, J. A., Stein, J. B., Eichele, G., Evans, R. M., et al. (1992). Cell, 68, 397. 12. Luckow, B. and Schutz, G. (1987). Nucleic Acids Res., 15, 5490. 13. Leid, M. (1994). J. Biol. Chem., 269, 14175. 14. Dowell, P., Peterson, V. J., Zabriskie, T. M., and Leid, M. (1997). /. Biol. Chem., 272, 2013. 15. Leidel, S., LeMotte, P., and Apfel, C. (1994). Mol. Cell. Biol, 14, 287. 16. Nayeri, S., Kahlen, J. P., and Carlberg, C. (1996). Nucleic Acids Res., 24, 4513. 17. Zhang, X. K., Lehmann, J., Hoffmann, B., Dawson, M. I., Cameron, J., Graupner, G., et al. (1992). Nature, 358, 587. 18. Devchand, P.R., Ijpenberg, A., Desvergne, B., and Wahli, W. (1999). In Eicosanoids and other bioactive lipids in cancer, inflammation and related diseases (ed. K. V. Honn, S. Nigam, L. J. Marnett, and E. Dennis). Plenum Press. 92

3: Adopting orphans: finding ligands 19. Lin, Q., Ruuska, S. E., Shaw, N. S., Dong, D., and Noy, N. (1999). Biochemistry, 38, 185. 20. Cogan, U., Kopelman, M., Mokady, S., and Shinitzky, M. (1976). Eur. J. Biochem., 65, 71. 21. Kersten, S., Dawson, M. L, Lewis, B. A., and Noy, N. (1996). Biochemistry, 35, 3816. 22. Kliewer, S. A., Moore, J. T., Wade, L., Staudinger, J. L., Watson, M. A., Jones, S. A., et al. (1998). Cell, 92, 73.

93

This page intentionally left blank

4

Kinetic analysis of nuclear receptor interactions B. J. CHESKIS and L. P. FREEDMAN

1. Introduction Steroid/nuclear receptors modulate transcription by virtue of their binding as part of a multiprotein complex to specific DNA sequences in promoter regions of target genes. This process directs a cascade of protein-protein and protein-DNA interactions affecting gene expression. Careful biochemical analysis of protein-protein and protein-DNA interactions is critical to the understanding of how genes are regulated by this family of ligand-modulated transcription factors. It is also a key to comprehending how small molecules, the cognate ligands for these receptors, affect receptor activity by directing the mode and amplitude of gene transcription. Important progress in our understanding of steroid/nuclear receptors action has been generated from recent X-ray crystallographic and NMR analysis. These studies have provided critical structural details of the various interactions involving nuclear receptors and their co-activators, as well as specific receptor-DNA interactions. Three-dimensional structures, however, are inherently static and cannot reveal the nature and energies that contribute to the formation and stabilization of structural changes that prompt specific complex formation. A detailed analysis of macromolecular interactions, in addition to structural information, must include information about thermodynamics and kinetics. This analysis should involve stoichiometry, the existence and degree of co-operativity, and thermodynamic and kinetic rates of the interaction. A major difficulty in studying ligand-protein, protein-protein, and proteinDNA interactions is the high affinity, often with sub-nanomolar equilibrium dissociation constants, that generally characterize these interactions. Such a high affinity significantly complicates direct measurements since they need to be made at receptor concentrations comparable to the dissociation constants, and therefore, requires an extremely sensitive detection method. This limitation has driven the development of numerous indirect and in some cases nonequilibrium methods, such as filter binding and the electrophoretic mobility shift assay (EMSA).

B. J. Cheskis and L. P. Freedman EMSA is based on the observation that electrophoretic mobility of nucleic acids in a non-denaturing polyacrylamide gel matrix is altered when a protein is bound to it (see also Chapter 2). This method allows detection of specific protein-DNA complex formation even in a crude cell or nuclear extract when other binding proteins are present. EMSA is relatively simple to carry out, and specific protein-DNA complexes may be directly visualized. Importantly, even complexes that differ in their stoichiometries can be resolved, which yields important information about the nature of the protein-DNA complex (i.e. monomers, dimers, heterodimers). This approach, however, has a somewhat limited application to studying the affinity and kinetics of protein-DNA interactions, particularly when applied to unstable complex formation. As with any other method where free and bound reaction components are fractionated, the separation or running time is a critical parameter. A prolonged gel run may affect the equilibrium and a very significant dissociation may take place while loading and running the gel. A simple mathematical analysis helps to understand whether or not a separation procedure permits an equilibrium analysis. Assuming that the association rate constant (ka) for protein binding to DNA or ligand is 106 M-1 sec-1 and that the dissociation rate constant (kd) is monoexponential:

where R = receptor; D = DNA or ligand; RD = receptor-DNA complex; ka = association rate constant; kd = dissociation rate constant. Then: and it can be calculated that a separation procedure needs to be completed within 0.15 t1/2 to prevent receptor-DNA complex dissociation for more than 10% of the total complex. The relationship between KD values and acceptable separation time, based on these assumptions, is presented in Table 1 (taken from ref. 1). If, for example, the receptor binds DNA with a KD in the nanomolar range, separation of a protein-DNA complex from free DNA must take only several minutes. If the association rate constant is slower than 106 M-1 sec-1, and the affinity is the same, the dissociation rate constant will be slower also, allowing more time for the separation process. Thus, as with any other method where free and bound components need to be separated, the possibility exists that the separation procedure may disturb the equilibrium. Therefore, considerable care needs to be used in interpreting binding constants obtained by EMSA. A number of methods have been developed to study macromolecular interactions that do not require separation of bound from free complexes. Some of them are based on changes that occur in a spectroscopic property of either a ligand or macromolecule upon formation of a complex (2) (see also Chapter 3). 96

4: Kinetic analysis of nuclear receptor interactions Table 1. Relationship between KD and allowable separation time for DNA-receptor complexes a KD (M)

Allowable separation time b (0.15

10-12

1.2 days

10-11 10-10 10-9 10-8 10-7 10-6

2.9 h 17 min 1.7 min 10 sec 0.1 sec 0.01 sec

t1/2)c

a This table was taken from ref. 1. b Separation time to avoid loss of more than 10% of complexes. c Calculation oft, (half-life dissociation) assuming an association rate constant of 106 M~' sec'1 as shown in Equation 1.

These methods permit continuous monitoring of binding, which in turn allows for kinetic measurements. Spectroscopic methods, however, are limited to molecules with suitable absorbent or fluorescent groups which in most cases requires some kind of labelling. Of course, one must be careful to determine whether the presence of a labelled probe affects the interaction. Fluorescence spectroscopy is a technique that is most widely used to study macromolecular interactions. In some cases, intrinsic fluorescence of the proteins, arising from aromatic amino acids that are affected by ligand or DNA binding, is used. Some of the fluorescence probes, derived recently and used for protein and DNA labelling, have sufficiently high quantum yields to enable experiments to be performed at nanomolar concentration, which permit not only the detection of protein-DNA binding, but also of conformational alterations. This approach was used to study TBP binding to DNA (3), kinetics of DNA unwinding by helicase (4), the molecular mechanism of Rep helicase binding to DNA (5), and the kinetics of DNA synthesis by DNA polymerases (6). Fluorescence anisotropy (or polarization) can also be used if the molecular size of the complex is significantly different from the free protein or ligand. This approach has been used to study CBP binding to phosphorylated CREB-DNA complex (7), 9-cis retinoic acid (RA) and all-trans RA binding by RAR-RXR heterodimers (8), DNA binding by glucocorticoid receptor zinc finger mutants (9), and trp repressor binding to DNA (10). A relatively new approach to assess macromolecular interactions using optical phenomena, called surface plasmon resonance (SPR), was developed and utilized recently in an instrument called BIAcore. This approach is rapidly becoming the standard for studying biomolecular interactions in real time. This chapter will describe how this highly sensitive and effective method can be used to examine the kinetics and affinity of steroid and nuclear receptor-DNA and protein interactions. 97

B. J. Cheskis and L, P. Freedman

2. Analysis of nuclear receptor functions using surface plasmon resonance (SPR) technology 2.1 The basic principle of interaction analysis with BIAcore In the BIAcore one of the interacting molecules is immobilized onto a solid phase—the chip surface—while the other is introduced in flow over the surface. When a soluble macromolecule binds to the immobilized one, it leads to an increase in the macromolecule concentration at the sensor surface, with a corresponding increase in the refractive index. SPR measures changes in the refractive index close to the chip surface. A linear relationship exists between the mass (concentration) of molecules bound to the surface and the refractive index change observed (11). The chip surface carries a carboxymethylated dextran to which one of the reactants is covalently linked (12). Binding, measured in arbitrary response units (RU), is recorded in real time and the data obtained contain information about the kinetics of macromolecular interactions. A signal of 1000 RU corresponds to a surface concentration change of approximately 1 ng/mm2 (13). In a typical experiment, one of the interacting molecules is attached to the dextran matrix (Figure 1). The other is injected over the surface during the socalled 'wash-in' phase. In the 'wash-out' phase, reaction buffer is injected for a prolonged period of time. Results from the experiment are presented as a sensogram, which is a plot of refractive index changes recorded as a function of time. The major advantage of the BIAcore system is that it allows label-free detection and real time analysis of macromolecular interactions. It should be noted that many macromolecular interactions in vivo occur between solid and liquid phases in dynamic flow, as in the BIAcore system. The range of equilibrium and kinetic constants measurable with BIAcore is provided in Table 2. Interactions can be studied over a wide range of temperatures from 4-40°C, hence information about the thermodynamics of this process can be

Table 2. BIAcore system performance a 10-4 ka

Equilibrium constant, KD Association rate constant, Dissociation rate constant, kd Minimum molecular weight Temperature range Flow rate Time resolution

to 10 -11M 103 to 106 M'1 sec"1 10~5to KT1 sec'1 200 Da 4-40°C 1-100 ul min'1 5 Hz

' Data are taken from the BIAcore, 2000 manual (Biacore Inc.).

98

4; Kinetic analysis of nuclear receptor interactions

Figure 1. Macromolecular interactions can be directly monitored with BIAcore.

obtained. The SPR detection system is sensitive to the molecular mass of the interacting molecules; the lowest reported limit of detected molecular mass is 200 Da (13).

2.2 Overexpression and purification of nuclear receptors for kinetic and equilibrium analyses To study the structure and function of nuclear receptors, full-length proteins or distinct functional domains have been overcxpressed and purified. Most often, bacterial overexpression systems have provided high quantities of these receptors, although some members of the superfamily have been notoriously difficult to express as full-length proteins; in these cases, insect cell expression 99

B. J. Cheskis and L. P. Freedman using baculovirus vectors has been more effective. Alternatively, isolated functional domains, such as the DNA or ligand binding domains (DBD and LBD, respectively) have been overexpressed to very high quantities. The expression and purification of these domains have led to detailed threedimensional structures, as solved by both 2D NMR and protein crystallographic analyses (14-17). We describe here two protocols for the purification of DBDs that are either soluble or insoluble following overexpression in E. coll (Protocols 2 and 3). For this, we prefer using expression systems driven by the phage T7 promoter (the pET system). For full-length receptors, we often fuse cDNAs to affinity tags, such as GST, poly-His, or FLAG. In Protocol 4, we describe one method for the purification of a GST fused receptor (see also Chapters 2 and 3). Protocol 1. Overexpression of receptor derivatives in E. coli 1. A culture of BL21(DE3)/pLysS carrying inducible versions of the glucocorticoid or vitamin D receptor DNA binding domains (GR-DBD and VDR-DBD, respectively) is grown overnight at 37°C in LB medium (10 g tryptone, 5 g yeast extract, 5 g NaCI, per litre) containing 50 ug/ ml ampicillin and 34 ug/ml chloramphenicol. BL21 host bacteria contain a stable lambda lysogen carrying the T7 gene 1, encoding T7 RNA polymerase that is under the control of the lac operator/promoter. The DBDs are expressed by a T7 promoter, which is in turn dependent on T7 RNA polymerase. 2. The saturated overnight culture is used to diluted 1:100 in fresh LB containing only ampicillin and grown at 37°C with shaking for aeration. 3. The culture is induced at mid-logarithmic growth (OD600 = 0.6) by addition of isopropyl-D-thiogalactoside to a final concentration of 0.5 mM. 4. After 3 h of induction, collect the cells by centrifugation (4000 r.p.m. in a Sorvall RC3C), and purify overexpressed proteins according to Protocols 3 or 4, depending on their relative solubilities in E. coli.

The GR-DBD is very soluble even at millimolar concentrations, while the VDR-DBD is extremely insoluble. Although these two nuclear receptor derivatives are highly homologous in amino acid sequence (approximately 50% identity) and presumably fold into similar three-dimensional structures based on the conservation of critical residues, it is difficult if not impossible to predict a priori whether or not a bacterially overexpressed protein or protein fragment will turn out to be stable, soluble, and easy to purify. We 100

4: Kinetic analysis of nuclear receptor interactions provide purification protocols for both proteins below as relatively simple strategies to isolate proteins depending on such parameters. Alternatively, cells can be grown and induced at room temperature; in many cases, the slower growth rate of the bacteria reduces the fraction of insoluble overexpressed protein. Protocol 2. Purification of an overexpressed soluble receptor DBD Equipment and reagents • Poly(ethylenimine) (Aldrich)

Method 1. Resuspend the cell pellet in 3 vol. of lysis buffer (50 mM TrisHCI pH 7.5, 1 mM EDTA, 10% glycerol, 500 mM NaCI, 4 mM CaCI2, 40 mM MgCI2, and added fresh: 5 mM DTT, 0.5 mM PMSF, lysozyme to 0.13 mg/ml, and DNase I to 20 ug/ml) and incubate on ice for 10 min. 2. Add sodium deoxycholate slowly over a period of 10 min to 0.05%, and mix the mixture for 15 min at 4°C. Centrifuge the lysate at 60000 g for 30 min. 3. Add poly(ethylenimine) (10% stock, pH 7.5) slowly over a period of 10 min to the supernatant to a final concentration of 0.2%, and mix the solution for an additional 10 min at 4°C. This suspension is centrifuged at 15000 r.p.m. for 30 min in an SS-34 rotor, and the pellet discarded. 4. Precipitate the supernatant with ammonium sulfate (50% saturation), stirred gently for 10 min at 4°C, and centrifuged at 10000 r.p.m. 5. Resuspend the pellet in TGED50 buffer (50 mM Tris-HCI pH 7.5, 0.5 mM EDTA, 10% glycerol, 5 mM DTT, 50 mM NaCI) and dialyse against TGED50. The dialysate is applied to a Bio-Rex 70 (Bio-Rad) column equilibrated with 5 vol. of TGEDZ50 buffer (TGED50 plus 0.05 mM ZnSO4). 6. Wash the columns with 3 vol. of TGEDZ50 and elute the protein with a 50-600 mM NaCI gradient, and collect 0.5 ml fractions. 7. The fractions are run on a 15% SDS-PAGE gel to determine which fractions contain the purified protein. The pooled eluent is typically 90-95% pure protein, as assessed by SDS-PAGE.

101

B. J. Cheskis and L. P. Freedman Protocol 3. Purification of overexpressed insoluble VDR-DBD 1. Resuspend the cell pellet in 3 vol. of buffer 1 (PBS containing 1 mM DTT, 1 mM PMSF, and 0.1 mM ZnCI2), and sonicate on ice for three 20 sec intervals. 2. Centrifuge the lysate at 14 000 r.p.m. in a SS-34 rotor for 20 min, and resuspend the pellet in 3 vol. of buffer 1 plus 1 M NaCI. The mixture is sonicated on ice three times for 20 sec, centrifuged at 14 000 r.p.m. for 20 min, resuspended in 3 vol. of buffer 1 plus 1 M NaCI, and sonicated and centrifuged as before. 3. Resuspend the protein pellet in 10 ml TGEDZ50 plus 5 M urea, and denature overnight by inversion at 4°C. 4. After centrifugation for 10 min at 14000 r.p.m. in a SS-34 rotor, the supernatant containing denatured, solubilized VDR-DBD is retained and subjected to stepwise renaturation by dialysis for 2.5 h each against TGEDZ50 plus 0.1% Triton X-100, and then against TGEDZ50. 5. Centrifuge the dialysate at 14000 r.p.m. for 10 min, and then apply to a CM Sepharose CL-6B (Pharmacia) column equilibrated with TGEDZ50, and elute with a 50-800 mM NaCI gradient.

While the preceding protocols describe methods to purify native receptor derivatives, the use of tagged fusion proteins has become widespread. The advantage of a tag is the ease of purification, since the tag is typically designed to bind to an affinity or antibody matrix with very high affinity. In addition, the tag can be used to selectively 'pulldown' a protein and whatever other proteins are associated. For this, GST fusions have been especially effective. Protocol 4. Purification of a full-length nuclear receptor fused to GST Equipment and reagents • Bio-Rad protein assay kit

• Glutathione agerose beads (Sigma)

Method 1. Grow and induce a bacterial culture exactly as described in Protocol 1. 2. Harvest cells by centrifuging at 4°C, 10000 r.p.m. for 10 min. Re-

102

4: Kinetic analysis of nuclear receptor interactions suspend cell pellet in 0.5 ml PBS plus 1 mM DTT and PMSF (Bfr 1), and sonicate on ice three times for 10 sec each time. 3. Spin again (14 000 r.p.m. for 10 min in a microcentrifuge), retain supernatant. Add 50 ul glutathione agarose beads which have been pre-swollen. To swell the beads combine 20 mg beads in 0.4 ml Bfr 1, and allow them to incubate at room temperature for 30 min; the beads are then washed three times in Bfr 1 before using. 4. Incubate at 4°C for 30-60 min with gentle rocking or inversion. 5. Wash the beads in 0.5 ml Bfr 1 three times by spinning for 2 min in a clinical centrifuge at 2000 r.p.m. and carefully aspirating and discarding the supernatant each time. 6. Elute GST-nuclear receptor fusion protein (GST-NR) from the glutathione agarose beads by resuspending the beads in 100 ul freshly prepared 10 mM reduced glutathione in 50 mM Tris-HCI pH 8.3. 7. Incubate at room temperature for 20 min. Spin beads as in step 5 and recover supernatant (Sup 1). Repeat, but retain this supernatant in a separate tube (Sup 2), and examine purity by SDS-PAGE. If elution was efficient, the majority of GST-NR should be in Sup 1. 8. Determine protein concentration by the Bradford method.

2.3 Analysis of protein-protein interactions Protein-protein interactions play an important role in nuclear receptormediated transcriptional regulation, for example, bringing to DNA enzymes that control chromatin structure, thereby adjusting the rate of gene expression to the spectrum and concentration of available transcription factors. The BIAcore system allows for the direct evaluation and kinetic/ equilibrium analysis of nuclear receptor homo/heterodimerization, receptor interactions with co-activators, and the role their cognate ligands play in regulating these processes. The manner in which one of the interacting molecules is attached to the sensor surface is critical for the generation of a uniform binding response. There are a number of immobilization strategies that can be used for molecular attachment. Several immobilization kits are available from Biacore Inc. They include reagents for amine coupling, using reactive esters, for coupling by thiol-disulfide exchange, and for aldehyde coupling to a hydrazineactivated surface. However, direct statistical covalent immobilization may introduce heterogeneity to the attached molecule. We found that immobilization of VDR or RXR by amine coupling completely abolished their activity. Among the possible reasons for this is that functionally important amino acids were modified. To create a homogeneous interaction surface and 103

B. J. Cheskis and L. P. Freedman to avoid random receptor modification, receptor molecules can be captured directly by a specific antibody, especially if epitope binding and interaction motifs are known and well separated; or indirectly, using antibodies to fusion tags. It is also important to remember that immobilization may affect the thermodynamics of the binding reaction by constraining a molecule's rotational and diffusional freedom. To examine VDR heterodimerization with RXR, in the absence of DNA, covalent immobilization of anti-glutathione-S-transferase (GST) antibodies followed by the capturing of GST-RXR fusion proteins through antibodyantigen interactions, can be carried out (18) (Figure 2).

Protocol 5. Preparation of the sensor chip Equipment and reagents • BIAcore 2000 system (Biacore Inc.) • Amine coupling kit containing N-hydroxysuccinimide, N-ethyl-N-(3-diethylaminopropyl) carbodiimide (EDO, and ethanolamine hydrochloride (Biacore Inc.)

• Sensor chips CM 5 (certified) (Biacore Inc.) « Anti-GST binding kit (Biacore Inc.) . HBS buffer: 10 mM Hepes pH 7.4, 150 mM NaCI, 3.4 mM EDTA, 0.05% Tween 20

A. Anti-GST antibody immobilization 1. Activate carboxylic groups on the sensor surface with a 7 min injection of a solution containing 0.2 M N-ethyl-N-(3-diethylaminopropyl) carbodiimide (EDO and 0.05 M N-hydroxysuccinimide (NHS). 2. Dilute anti-GST antibody to 10 ug/ml (800 ug/ml stock) with 10 mM Na acetate buffer pH 5.0. 3. Inject 30 ul of the anti-GST antibody at 10 ug/ml. 4. Block remaining ester groups by injection of 60 (J of 1 M ethanolamine.a B. GST-RXR immobilization 1. Inject 20 ul of purified GST-RXR (0.2-0.5 mg/ml) (see Protocol 4). A typical immobilization level for GST-RXR is 300-400 RU. 2. To remove immobilized GST-RXR from the surface of the sensor chip, inject 10 ul of 10 mM glycine pH 2.0. a A typical immobilization level at these conditions is 4-5000 RU.

To study the affinity and kinetics of protein-protein interactions, a range of concentrations of the interacting molecules needs to be used. Conveniently, the same interaction surface can be utilized for multiple injections. It is important that repetitive injections are run in an automated, computer controlled way. 104

4: Kinetic analysis of nuclear receptor interactions

Figure 2. Surface preparation for the analysis of VDR-RXR heterodimerization. (A) AntiGST antibody immobilization using an amine coupling protocol. Carboxylic groups on the sensor surface were activated with a 7 min injection (70 ul) of 0.2 M N-ethyl-N'-(3diethylaminopropyl) carbodiimide (EDO and 0.05 M N-hydroxysuccinimide (NHS). Then, 30 ul of the anti-GST antibody at 10 ug/ml in 10 mM Na acetate buffer pH 5.0 were injected. Remaining ester groups were blocked by injection of 60 ul of 1 M ethanolamine. (B) GST-RXR was immobilized through GST capture on the anti-GST sensor chip. After injections of 20 ul of GST-RXR at a protein concentration of 0.2 mg/ml, surface with approximately 300 RU of immobilized RXR was obtained.

105

B. J. Cheskis and L. P. Freedman Figure 3A presents overlaid results of injections of VDR at concentrations ranging from 13-730 nM over the surface with immobilized GST-RXR. After each experiment, the chip surface was regenerated by injection of 10 ul of 0.05% SDS. Protocol 6. Analysis of VDR-RXR heterodimerization Equipment and reagents • BIAcore 2000 system « HBS buffer (see Protocol 5)

• Sensor chips cM 5 (certified) with immobilized anti-GST antibody and GST-R+R (see Protocol 5)

Method 1. Maintain a continuous flow of the HBS buffer at 10 ul/min during the binding cycle. 2. Inject 30 ul of VDR solution in HBS buffer. 3. Wash the formed complex with the buffer for an additional 1000 sec. 4. After each injection regenerate the surface with injection of 10 ul of 0.03% SDS. 5. Continue with next injection of VDR.

To evaluate if a ligand has an effect on the interaction, injections of the liganded and unliganded VDR, at the same protein concentration, over the surface with immobilized RXR can be compared. Figure 3B presents overlaid sensograms of VDR injections in the presence and in the absence of 1,25dehydroxyvitamin D3 (taken from ref. 18). It is evident that ligand binding by VDR is augmenting the affinity of VDR-RXR heterodimerization. Kinetic analysis of such protein-protein interactions is presented in Section 2.5.

2.4 Analysis of protein-DNA interactions DNA binding by transcription factors is a complex process that brings multiple components of the transcriptional machinery to the DNA template. Some of these transcription factors can bind DNA directly, while others form multiprotein complexes. Nuclear receptors regulate the rate of gene transcription by binding to specific DNA sequences and delivering to the template other transcription factors that modulate chromatin structure and participate in the assembly of the pre-initiation complex. Depending on the nuclear receptor, these proteins bind DNA as monomers, homodimers, or heterodimers. The transcriptional activity of these species may be very different. Moreover, for RXR dimeric receptors, a ligand can have a dramatic influence on the receptor complex that predominates (18, 106

4: Kinetic analysis of nuclear receptor interactions

Figure 3. 1,25(OH)2D3 affects VDR-RXR heterodimerization. (A) Overlaid sensograms showing injections of VDR at increasing concentrations over the surface with 320 RU of GST-RXR immobilized on the surface of a sensor chip. (B) 1,25(OH)2D3 binding enhances the affinity of VDR-RXR interaction. Two overlaid sensograms of injections of 7.3 mM of liganded (VDR-D3) and unliganded VDR over the surface with GST-RXR immobilized are shown. Reprinted from ref. 18 with permission from The American Chemical Society'.

107

B. J. Cheskis and L. P. Freedman 19). Therefore, an accurate analysis of the thermodynamics and kinetics of the receptor-DNA interactions and how this process is influenced by ligands, may aid in the development of ligands that drive this process towards the desired receptor dimer complex. In approaching protein-DNA interactions by BIAcore, the way in which a specific oligonucleotide is attached to the chip surface is a very important parameter. Homogeneous immobilization and direct attachment is a strict requirement to generate accurate data. We found that all of these conditions could be fulfilled if DNA is synthesized as a self-annealing oligonucleotide that forms a hairpin duplex upon heating and quick cooling. The oligonucleotide is biotinylated with Klenow enzyme and attached through streptavidinbiotin interactions. Several approaches can be used to study the specificity of nuclear receptor binding to DNA. By competition, an excess of unbound DNA can be added to the receptor before it is injected. Alternatively, oligonucleotides with different sequences can be immobilized. The sensograms presented in Figure 4A demonstrate that VDR-RXR binding to a specific vitamin D-response element (VDRE) derived from the mouse osteopontin (Spp-1) gene promoter (20) can be almost completely abolished by addition of an excess of unbound specific competitor DNA. To evaluate specificity of ER/oestrogen-response element (ERE) interactions, three potential oestrogen receptor binding elements were used (21). One oligonucleotide contained an inverted repeat of the sequence AGGTCA spaced by 3 bp. This is a classic ERE derived from the vitellogenin A2 gene (Vit. A2) promoter of Xenopus laevis (22). A second oligo was designed using the same sequence but organized as two directly repeated half-sites spaced by 3 bp (DR3). This sequence represents a high affinity binding site for the vitamin D3 receptor (23). A third oligo (C3) was derived from the mouse complement component C3 gene promoter (C3), which is regulated by oestradiol (24). Like the Vit. A2 ERE, it contains two half-sites organized as inverted repeats spaced by 3 bp, except that one half-site is a perfect AGGTCA and the other—AGTCTA—deviates at three positions. Human ER (hER) bound with E2 was injected over the surfaces coated with these different oligonucleotides (Figure 4B). High affinity interaction was detected in the flow cell with the immobilized Vit. A2 ERE. At the same time, binding of hER to immobilized DR3 or C3 surfaces was very low. Thus, the mass increase on the surface of the sensor chip in response to the injection of a given nuclear receptor is due to specific receptor binding to DNA. Injection of ER or VDR/RXR over the surfaces with no DNA immobilized results in no detectable binding (data not shown). Using the BIAcore 2000, it is possible to simultaneously detect interaction events on four different parts of the sensor surface—flow cells (FC). The flow of the BIAcore allows the sample to be addressed to individual FCs or serially to predefined combinations of the FCs. Therefore, by immobilizing different 108

4: Kinetic analysis of nuclear receptor interactions

Figure 4. Two approaches to study the specificity of receptor-DNA interactions. (A) VDR-RXR heterodimer binding to specific DNA can be almost completely abolished by addition of specific competitor DNA. Two overlaid sensograms of 0.7 mM of VDR and GST-RXR injected over the surface with 817 RU of VDRE immobilized on a surface of sensor chip are shown. (B) Analysis of the specificity of the ER-ERE interaction. 30 ul hER at 150 nM liganded with E2 was injected over the surface with 1094 RU of DR3, 943 RU of Vit. A2, and 1063 RU of C3 oligos, see text for details. Panel A is reprinted from ref. 18 with permission from The American Chemical Society'. Panel B is reprinted from ref. 21 with permission from The American Society for Biochemistry and Molecular Biology'.

109

B. J. Cheskis and L, P. Freedman amounts of one of the biotinylated oligos in different FCs, it is possible to create a four step 'gradient' surface. Interaction analysis at the 'gradient' surface allows one to calculate stoichiometry and improves the quality of the collected data.

Protocol 7. Preparation of the sensor chip for protein-DNA interaction analysis Equipment and reagents • BIAcore system (Biacore Inc.) • Sensor chips CM 5 (certified) (Biacore Inc.) • Amine coupling kit containing N-hydroxysuccinimide, N-ethyl-N'-(3-diethylaminopropyl) carbodiimide (EDO, and ethanolamine hydrochloride (Biacore Inc.) • Streptavidin (Sigma) • Oligonucleotides. Several potential EREs were synthesized as self-annealing Oligonucleotides that form a hairpin duplex upon heating and rapid cooling. A 75 bp oligonucleotide (Vit. A2) containing a specific binding site for ER was derived from the vitellogenin A2 gene promoter of Xenopus laevis. Its sequence is 5'-AGCTCTTTGATCAGGTCACTGTGACCTGAACTTACTCCCC

CCGAGCAAGTTCAGGTCACAGTGACCTGA TCAAAG-3'. The second oligo (DR) was designed as two directly repeated half-sites. Its sequence is 5'-AGCTCTTTGATCTGACCTCTGTGACCTGAACTTACTCCCCCCGAGCA AGTTCAGGTCACAGAGGTCAGATCAAAG3'. The third oligo (C3) was derived from the mouse complement component C3 gene promoter Its sequence is 5'-AGCTCTTTGATCTGACCTCTGTGACCTGAACTTA CTCCCCCCGAGCAAGTTCAGGTCACAGAG GTCAGATCAAAG-3'. . HBS buffer: 10 mM Hepes pH 7.4, 150 mM NaCI, 3.4 mM EDTA, 0.05% Tween 20 . 10 x react 2 buffer: 500 mM Tris-HCI pH 8.0, 100 mM MgCI2, 500 mM NaCI

A. Biotinylation of oligonucleotide 1. Dilute large fragment of DNA polymerase I to 0.5 U/ul with 100 mM potassium phosphate buffer pH 7.0, 10 mM 2-mercaptoethanol, 50% (v/v) glycerol. 2. To a 1.5 ml microcentrifuge tube on ice add: • 10 x react 2 buffer

3 ul

• 0.4 mM biotin-14-dATP

3 ul

• 0.5 mM dCTP

1 ul

• 0.5 mM dGTP

1 ul

• 0.5 mM dTTP

1 ul

• annealed oligo (1

ug/ul)

• large fragment of DNA polymerase I • autoclaved distilled water

1 ul 1 ul 19 ul

3. Mix gently and incubate on ice for 15 min. 4. Load reaction mixture on Chromaspin 10 column (Clontech). 5. Centrifuge for 3 min at 3000 r.p.m. in a microcentrifuge to purify from unincorporated biotinylated nucleotides.

110

4: Kinetic analysis of nuclear receptor interactions B. Streptavidin immobilization 1. Activate carboxylic groups on the sensor surface with a 7 min injection of 0.2 M N-ethyl-N'-(3-diethylaminopropyl) carbodiimide (EDO and 0.05 M N-hydroxysuccinimide (NHS). 8 2. Dilute streptavidin to 10 ug/ml with 10 mM Na acetate buffer pH 5.0. 3. Inject the streptavidin solution for 1, 2, 4, and 6 min over the flow cells 1, 2, 3, and 4 correspondingly. 4. Block remaining ester groups by injection of 60 ul of 1 M ethanolamine over all flow cells. C. Oligonucleotide immobilization 1. Inject 10 ul of 0.1% SDS. 2. Inject 50 ul biotinylated Oligonucleotide solution at 33 ng/ml. * Flow rate for all steps was 10 ul/min.

Figures 5A and B (taken from ref. 21) illustrate immobilization of a gradient of specific DNA on a surface chip. First, a gradient of immobilized streptavidin in different flow cells was obtained by varying the streptavidin injection time as illustrated in Figure 5A. As a result, a surface with 2918 RU, 1922 RU, 1363 RU, and 908 RU of streptavidin in FCs 1, 2, 3, and 4 respectively, was obtained. Injection of 50 ul biotinylated Vit. A2 at 33 ng/ul resulted in a surface with 1817 RU, 1290 RU, 917 RU, and 605 RU of immobilized Vit. A2 in FCs 1-4, respectively (Figure 5B). This gradient surface was used to evaluate the stoichiometry of hER interaction with Vit. A2. Serial injections of unliganded ER and ER incubated overnight with 10-6 M of E2, 4(OH)tamoxifen, raloxifene, or ICI-182 780, at protein concentrations ranging from 35-270 nM, were run over the sensor chip with an immobilized gradient of the Vit. A2. Figure 6 (taken from ref. 21) demonstrates overlaid sensograms of the injections of hER, liganded with E2, at different protein concentrations, over the gradient of immobilized Vit. A2. A clear correlation between the change of the refractive index and the level of immobilized DNA was detected. We found that there is a linear relationship between the surface density (amount of immobilized DNA) and the saturation response (refractive index at the end of injection) obtained with 270 nM injection of hER in the interval of the DNA densities used for this experiment (data not shown). The calculated stoichiometry for the formed ER-ERE complex (21), was 1:1.92 ± 0.11. This result confirms that hER binds an ERE as a homodimer. The contribution of ligand in steroid receptor binding to DNA has been somewhat controversial for many years. Real time interaction analysis allows one to unequivocally address this question. Figure 7 shows overlaid senso111

Figure 5. Immobilization of a DNA gradient. (A) A gradient of immobilized streptavidin in different flow cells (FCs) was obtained by varying the streptavidin injection time. As a result a surface with 2918 RU, 1922 RU, 1363 RU, and 908 RU of streptavidin in FCs 1, 2, 3, and 4, respectively, was obtained. (B) Injection of 50 ul of biotinylated Vit. A2 at 5 ng/ul over flow cells 1-4 with different amounts of immobilized streptavidin resulted in a surface with 1817 RU, 1290 RU, 917 RU, and 605 RU of immobilized Vit. A2 ERE in FCs 1-4, respectively. Reprinted from ref. 21 with permission from The American Society for Biochemistry and Molecular Biology'.

112

4: Kinetic analysis of nuclear receptor interactions grams obtained by injections of unliganded hER and hER liganded with E2, 4(OH)tamoxifen, raloxifene, or ICI-182 780 over a surface with immobilized Vit. A2. Even visual analysis clearly demonstrates that ER ligands dramatically affect hER interaction with specific DNA. It can be seen that binding of oestradiol, a 'pure' agonist, induces rapid formation of an unstable ER-ERE

Figure 6. Titration of the immobilized ERE DNA with hER at different concentrations. Serial injections of hER were pre-incubated overnight with 10-6 M E2 at the following protein concentrations: 35, 87.5, 175, and 270 nM, and run over sensor chips with increasing amounts of immobilized Vit. A2 ERE: (A) 605 RU, (B) 917 RU, (C) 1290 RU, and (D) 1817 RU (FCs of Figure 5). Reprinted from ref. 21 with permission from The American Society for Biochemistry and Molecular Biology'.

113

B. J. Cheskis and L. P. Freedman complex, and furthermore, binding of a 'pure' antagonist such as ICI-182 780 leads to the slow formation of a very stable receptor-DNA complex.

2.5 Kinetic analysis The simplest model that can be used to describe a bimolecular interaction is:

where: A = immobilized molecule; B = soluble molecule; AB = formed complex; ka = association rate constant; kd = dissociation rate constant. According to the Langmuir theory, the rate of the complex formation can be described as: where: Rmax = total binding capacity of the surface, maximum amount of molecule B, A can bind; C = solution concentration of B; R = BIAcore response, directly proportional to the change in mass on the surface of sensor chip due to B binding to A. It is assumed, in this model, that free molecule B is homogeneously distributed. The interaction is exclusively controlled by intrinsic kinetic rate constants (ka and kd) and diffusion of the free molecule is not rate limiting. This model also defines that the rate of B dissociation from the surface in the wash-out phase as: assuming that there is no rebinding for B. Rearranging Equation 4: A plot of dR/dt versus R should give a straight line of slope -(kaC + kd) and y intercept kaCRmax. Furthermore, a plot of the slope, ks—derived from a plot of dR/dt versus R—versus C should give a straight line of slope ka and y intercept kd. The linear transformation of Equation 5 that describes the process of dissociation is: where R1 is the response at time t = 1 and Rn is the response at time t = n on the dissociation curve. A plot of In(R1/Rn) versus (tn - t1 should produce a straight line of slope kd. Therefore, according to this approach, linear transformation of the primary data can be used to obtain association (ka) and dissociation (kd) rate constants. Initially, kinetic information from the data collected with BIAcore was extracted by a linear kinetic curve-fitting software package provided with the instrument, fitting straight lines to portions of sensograms (25). Linearization, 114

4: Kinetic analysis of nuclear receptor interactions however, can only be used to estimate the rate constants for a simple bimolecular, or Langmuir, interaction. Another disadvantage of this method of data analysis is that it transforms the data and so transforms the error associated with the parameter estimates (26). An alternative approach was to use non-linear least squares analysis and to fit data directly. This can be done by integration of the rate equations that describe the interaction. The rate equation for a simple bimolecular interaction has been integrated analytically (26). It has been suggested that the dissociation rate constants be calculated first and then used to constrain values for the association rate constants (27). Linearization or analysis based on the integrated rate equation can be used to describe a very simple bimolecular interaction. In many cases, however, the interactions under study do not fit the simple bimolecular model. This may be a consequence of physical limitations of the BIAcore itself, such as, for example, mass transport limitations or steric hindrance (28, 29), or it may indicate that interactions between some molecules has a complex nature. For such complex models, analytical integration becomes very complicated. To understand the molecular mechanisms of a complex biological reaction and to describe them kinetically, simple analysis based on a simple Langmuir model is insufficient. Instead, a model of the reaction has to be created and verified (30). To make sure that the model is correct, it needs to be validated

Figure 7. ER ligands modulate receptor interaction with DNA. Overlaid sensograms of hER injections over the surface with 943 RU of immobilized Vit. A2 oligo. Unliganded hER and hER, pre-incubated overnight with 10 -6 M of E2, 4(OH)tamoxifen, raloxifene, and ICI182780, at a protein concentration of 100 nM, were used for this experiment. Reprinted from ref. 21 with permission from The American Society for Biochemistry and Molecular Biology'.

115

B. J. Cheskis and L. P. Freedman using an independent biochemical approach. Furthermore, the entire sensogram, not just parts of it, must be fitted with varying flow rates and reagents concentrations (31). Numerical integration can then be used to analyse BIAcore data. This approach, called global fitting analysis, was first used to evaluate protein-DNA interactions (30-32). Today, this is clearly the best method for understanding the molecular mechanisms of biological interactions and for describing it in terms of its kinetics and thermodynamics. Several software packages are now available that are using this approach for BIAcore data evaluation. The SPRevolution (http://www.bri.nrc. ca/csrg/equip.htm) can be used for analysis of only one surface density at a time and uses a Runge-Kutta integrator (a method for integrating differential equations) for the fitting analysis. The Clamp software (http: / / www.hci.utah.edu/cores/biacore/docs/clamp.html) can simulate and analyse BIAcore data. It allows several surfaces to be analysed simultaneously and includes common BIAcore models and simple modelling dialogue which allows new ones to be edited and created. This program also permits an evaluation of mass transport effects. Finally, recently BIAcore introduced a new product—software BIAevaluation 3.0. (http: //www. biacore.com/products/eval.html). This software, in our opinion, is the easiest and most efficient way to analyse BIAcore data.

3. Conclusion A detailed description of the kinetic and thermodynamic parameters at work on steroid/nuclear receptors are a key to understanding how these proteins ultimately regulate gene transcription. BIAcore is ideally designed to examine such parameters. Although real time interaction analysis, in some cases, generates more questions than answers, we believe that this complexity reflects the reality of biological processes.

References 1. Yamamura, H. L, Enna, S. J., and Kuhar, M. J. (ed.) (1985). Neurotransmitter receptor binding, 2nd edn. Raven Press, NY. 2. Lohman, T. and Bujalowski, W. (1991). In Methods in enzymology (ed. R. T. Sauer) Vol. 208, p. 258. Academic Press, London. 3. Parkhurst, K., Brenowitz, M., and Parkhurst, L. (1996). Biochemistry, 35, 7459. 4. Raney, K. D., Sowers, L. C., Millar, D. P., and Benkovic, S. J. (1994). Proc. Natl. Acad. Sci. USA, 74, 193. 5. Bjornson, K., Moore, K., and Lohman, T. (1996). Biochemistry, 35, 2268. 6. Bloom, L. B., Otto, M. R., Beechem, J. M., and Goodman, M. F. (1993). Biochemistry, 33, 7576. 7. Lundblad, J., Laurance, M., and Goodman, R. (1996). Mol. Endocrinol, 10, 607. 8. Kersten, S., Dawson, M., Lewis, B., and Noy, N. (1996). Biochemistry, 35, 3816.

116

4: Kinetic analysis of nuclear receptor interactions 9. Lefstin, J. A., Thomas, J. R., and Yamamoto, K. R. (1994). Genes Dev., 8, 2842. 10. Stenberg, E., Person, B., Roos, H., and Urbansky, C. (1991). J. Colloid Interface Set., 21, 1526. 11. Lofas, S. and Johnsson, B. (1990). J. Chem. Soc. Chem. Commun., 21, 1526. 12. Granzow, R. and Reed, R. (1992). BioTechnology, 10, 390. 13. Karlsson, R. and Stahberg, R. (1995). Anal. Biochem., 228, 274. 14. Hard, T., Kellenbach, E., Boelens, R., Maler, B. A., Dahlman, K., Freedman, L. P., et al. (1990). Science, 249, 157. 15. Luisi, B. F., Xu, W., Otwinowski, Z., Freedman, L. P., Yamamoto, K. R., and Sigler, P. B. (1991). Nature, 352, 497. 16. Renaud, J.-P., Rochel, N., Ruff, M., Vivat, V., Chambon, P., Gronemeyer, H., et al. (1995). Nature, 378, 681. 17. Wagner, R. L., Apriletti, J. W., McGrath, M. E., West, B. L., Baxter, J. D., and Fletterick, R. J. (1995). Nature, 378, 690. 18. Cheskis, B. and Freedman, L. (1996). Biochemistry, 35, 3309. 19. Kersten , S., Gronemeyer, H., and Noy, N. (1997). /. Biol. Chem., 272, 12771. 20. Noda, M., Vogel, R. L., Craig, A. M., Prahl, J., DeLuca, H. F., and Denhardt, D. T. (1990). Proc. Natl. Acad. Sci. USA, 90, 9995. 21. Cheskis, B. J., Karathanasis, S. K., and Lyttle, C. R. (1997). J. Biol. Chem., 272, 11384. 22. Klein-Hitpass, L., Schorpp, M., Wagner, U., and Ryffel, G. (1986). Cell, 46, 1053. 23. Umesono, K., Murakami, K. K., Thompson, C. C, and Evans, R. M. (1991). Cell, 65, 1255. 24. Sundstrom, S. A., Komm, B. S., Ponce-de-Leon, H., Yu., Z., Teuscher, C, and Lyttle, C. R. (1989). J. Biol. Chem., 264, 16941. 25. Karlson, R., Michaelson, A., and Mattson, L. (1991). J. Immunol. Methods, 145, 229. 26. O'Shannessy, D. J., Brigham-Burke, M., Sonenson, K. K., Hensley, P., and Brooks, I. (1993). Anal. Biochem., 212, 457. 27. O'Shannessy, D. J. (1994). Curr. Opin. Biotechnol., 5, 65. 28. Fagerstam, L. G., Frostel-Karlsson, A., Karlsson, R., Person, B., and Ronneberg, I. (1992). J. Chromatogr., 597, 397. 29. Karlsson, R., Roos, H., Fagerstam, L., and Persson, B. (1994). Companion to Methods in enzymology, 6, 199. 30. Fisher, R. J., Fivash, M., Casas-Finet, J., Erickson, J. W., Kondoh, A., Bladen, S. V., et al. (1994). Protein Sci., 3, 257. 31. Fisher, R. J. and Fivash, M. (1994). Curr. Opin. Biotechnol., 5, 389. 32. Fisher, R. J., Fivash, M., Casas-Finet, J., Bladen, S. V., and McNitt, K. L. (1994). Companion to Methods in enzymology, 6, 121.

117

This page intentionally left blank

5

Functional characterization of co-activators using mammalian cell microinjection D. W. ROSE, T.-M. MULLEN, M. G. ROSENFELD, and C. K. GLASS

1. Introduction The past few years have led to an appreciation that the transcriptional activity of the nuclear receptor family is regulated by binding of the receptor, in a ligand-dependent manner, to a multicomponent complex of co-activator proteins, many of which possess histone acetyltransferase activity. The activity of these co-activators is also further regulated by a variety of signal transduction pathways. Many of these co-activator molecules have now been isolated, often on the basis of their interaction with the nuclear receptors. The functional characterization of these protein complexes has been facilitated by the use of a series of powerful single cell microinjection assays that employ the use of specific anti-co-activator antibodies used in conjunction with reporter constructs specific for each of the nuclear receptors and expression constructs for wild-type and mutant forms of the co-activators.

1.1 Why microinjection? Many techniques have become available for the introduction of exogenous molecules into mammalian cells. Among these are transfection of DNA, liposome-mediated fusion, electroporation, scrape-loading, virus-mediated transfer, and microinjection. Of these choices, microinjection is arguably the most difficult to perform, the most tedious and demanding from a technical standpoint, and the most dependent upon relatively expensive equipment. Why then, use such a demanding approach? 1.1.1 Advantages of the approach The microinjection technique has in general proven to be a very useful approach to the functional characterization of cellular proteins, particularly those involved in signal transduction and gene expression (1-4). When

D. W. Rose et al. specific antibodies raised against a protein are available, their injection into mammalian cells allows the functional 'knockout' of the protein without direct genetic manipulation. Microinjection also allows for the direction of the neutralizing antibody or other reagent to a specific compartment of the cell, with minimum perturbation of cellular chemistry, morphology, or metabolism. This method is also very sparing of valuable reagents, as the amounts of material required for successful microinjection is extremely small relative to many other approaches. The efficiency of successful introduction is also very high. Perhaps the most significant advantage to this technique, however, is the level of control it affords with respect to timing of events. In the case of the experiments involving transcriptional regulation described here, one can control the levels of expression of a given protein, and also the time of expression, so as to prevent secondary adaptive changes from occurring. 1.1.2 Disadvantages of the approach The primary difficulty with microinjection entails the initial expense of setting up the technique in the laboratory. At the very least, it requires the purchase of a semi-automated microinjection apparatus mounted on a quality inverted microscope, and access to an epifluorescence photomicroscope. Microinjection is also a relatively difficult technique to learn and employ successfully. It typically requires at least a month of training and practice before it can be used to perform experiments. It is a labour-intensive and tedious technique that requires meticulous attention to detail. From a technical standpoint, it also suffers the disadvantage that metabolic and biochemical studies are usually not possible on the injected population, because of the small number of cells involved. Fortunately, in the case of transcriptional regulation it has often proven possible to devise single cell assays employing reporter constructs which allow an unambiguous evaluation of the effects of microinjection.

2. Equipment and materials 2.1 Equipment Listed here (see Protocol 5 and Figure 2) are the specifications of the equipment currently in use in our laboratory, for information purposes only. There are many configurations available from a variety of manufacturers and we are not recommending any particular supplier, although we are quite satisfied with this equipment.

2.2 Antibodies We have used a wide variety of antibodies in microinjection studies and have found that in general it is necessary to empirically evaluate each antibody for its ability to achieve the desired effect in the cell. This is true whether one is using highly characterized affinity purified commercial antibodies, or the 120

5: Functional characterization of co-activators

Figure 1. Schematic overview of the microjunction approach (13). This diagram represents a simplified description of experiments in which the regulation of the activity of specific nuclear receptors can be studied. Receptor specificity is achieved by careful selection of reporter construct. Using well-characterised antibodies raised against co activators and co-repressors of transcriptional activity, the roles of these molecules can be characterised (8 10, 12) using a functional 'knockout' approach. The addition of expression vectors for either wild-type or mutant forms of the target protein allows confirmation of the specificity of the antibody, as wed as a functional 'rescue'. The alter has piovided an opportunity to study structure-function relationships of the co r e g u l a t o r y molecules (1, 8, 11).

homegrown v a r i e t y , in g e n e r a l . p o l y c l o n a l amihodios tend to be more advantageous t h a n monoclonal antibodies, probably due to the presence of di Here in lu( i speeies direeted ap.nl n^t m u l t i p l e sites on the tared protein. The re;ison t l u i l it is not possible lo predict w h e l h e r or not a given antibody w i l l he e l l e e l i v e in those experiments, even if the epitope is well uiulersluod, is deeause it is no! u s u a l l y possihle to prediet the accessibility of the epitome in i he cell u n d e r t i n y s>iven set of clrcumslanees. or w h e t h e r the eliecl of the a n t i h o d y b i n d i n g to its inrcol w i l l be n e u t n i H / a l i n i i of r u n c l i n n . M L I H V antihodies raised against IrLinscriptionul co-act i valors appear in he elTcelive because they prcveiii the physical i n i e n i c l i o n of proteins, and probabh' i n h i b i t the normal nssemhly of a en a c t i v a t o r complex. A neiiaiive phenoiype does not. however, necessarily mean t h a t the target protein has no functional sitmilicancc. 1'erhaps the best predictor of usefulness for a siven antibody is its 121

D. W. Rose et al. ability to immunoprecipitate the target protein under conditions where the protein is expected to be 'active'. Purification of antibodies using chromatography with a protein A bead matrix is an effective and relatively simple approach for purifying antibodies from a crude preparation or from serum (see Protocol 1). Protein A (or protein G for monoclonal antibodies) conjugated to agarose beads is available from a variety of commercial sources. Antibodies are diluted and bound to the matrix in a neutral low salt buffer and then subsequently eluted from the column by lowering the pH. To avoid denaturation, the antibodies are collected in a buffer which neutralizes the pH. Antibodies originally purchased from a commercial source are generally already affinity purified, avoiding the necessity for this purification step. Protocol 1. Antibody purification Reagents • Equilibration buffer: 20 mM Tris-HCI pH 7.5, 1 mM EDTA, 50 mM KCI

• Elution buffer: 100 mM glycine pH 2.0 • Neutralization buffer: 1 M Tris-HCI pH 8.8

Method 1. Dilute the crude antibody preparation (or serum) in 10 volumes of equilibration buffer. 2. Prepare protein A-agarose column of appropriate size for the amount of antibody being purified, bearing in mind that each protein A molecule is capable of binding two molecules of IgG. Manufacturer's instructions should be useful as a guide for column volume. In general, protein A-agarose binds approx. 10-20 mg of IgG for each ml of swollen beads. 3. Pass the dilute antibody/buffer mixture over the protein A-agarose column. 4. For best results, wash the beads with at least 100 volumes of cold equilibration buffer. 5. Prepare collection tubes for the elution of antibody. To collect 1 ml fractions, prepare 1.5 ml tubes (with tight caps) by adding 200 ul of neutralization buffer to each tube. 6. Elute the IgG from the column by the addition of 15-20 column volumes of elution buffer to the beads. This elution method removes bound antibody from the column rapidly, and the IgG will typically be found only in the early fractions. 7. Collect the eluate in the tubes containing neutralization buffer. Immediately mix the contents of each tube gently. 122

5: Functional characterization of co-activators 8. Because the eluate contains essentially homogeneous IgG, fractions containing antibody can be identified by their absorbance at 280 nm. 9. Antibodies can be stored briefly in this form, but should be further concentrated and exchanged into microinjection buffer (see Protocol 2) as soon as possible.

Following purification of the antibodies to be injected, they must be concentrated and exchanged into a buffer which can be injected harmlessly into cells (see Protocol 2). Many buffers are suitable for this purpose; we have adhered to the premise that simpler is better. If more complex buffers are used, the appropriate control injections of pre-immune IgG in the same buffer become necessary. The simple procedure described in Protocol 2 has also been used to prepare commercial antibodies for microinjection, with minimal losses as in dialysis and other methods. Many commercial suppliers of antibodies will provide 'special order' preparations in which both preservatives and stabilizing carrier proteins have been omitted, preventing the need to first purify the IgG. Protocol 2. Concentration of antibodies and preparation for injection Equipment and reagents • Microinjection buffer: 5 mM sodium phosphate pH 7.2, 100 mM KCI; sterilize by filtration

Centricon 30 centrifugal concentration device (Amicon/Millipore)

Method 1. Prior to concentration of purified antibodies, rinse Centricon with sterile microinjection buffer to remove preservatives from filter. To upper chamber of Centricon add approx. 1 ml of pre-immune IgG from the appropriate species at 10 mg/ml in microinjection buffer. Incubate on ice for 15 min to block non-specific interaction with filter. This procedure prevents loss of specific IgG. After incubation, pour off preimmune IgG and rinse several times with microinjection buffer. 2. Add purified antibody to Centricon. Centrifuge according to manufacturer's directions in a fixed angle rotor. Do not exceed recommended r.p.m. Continue centrifugation until volume of retentate is reduced below 100 ul. 3. Refill upper chamber with approx. 2 ml of chilled microinjection buffer. Empty filtrate from bottom chamber of Centricon. 4. Centrifuge as before until volume of retentate is approx. 100 ul. 5. Repeat steps 3 and 4. Buffer is now exchanged.

123

D. W. Rose et al. Protocol 2.

Continued

6. Check absorbance at 280 nm of 1 ml of antibody diluted in microinjection buffer. For pure IgG, 1 OD is equivalent to 0.8 mg/ml. Adjust concentration to 10 mg/ml with microinjection buffer if necessary. Many antibodies, particularly when affinity purified, will be effective at much lower concentrations. A concentration curve for activity in the microinjection assay should be done to determine whether the antibody concentration can be reduced prior to freezing aliquots of the preparation. 7. Aliquots of antibody (30 ul is a convenient size) are frozen at -70°C, where they are stable for at least a year. Once thawed, aliquots should never be refrozen. Antibodies are typically stable at 4°C for at least a month, particularly if precautions are taken to maintain their sterility.

2.2.1 Additional purification Further purification of antibodies by affinity chromatography against the antigen used for their production is sometimes necessary. The requirement for additional purification depends upon a number of factors, including the titre and avidity of the active IgG species, as well as the intracellular concentration of the target protein and the availability of active epitopes in the cell. It is therefore usually worthwhile to attempt the microinjection experiments with purified IgG before proceeding with additional purifications in which yield of antibody is often low. If affinity purification proves necessary, it is accomplished with a procedure similar to the one described in Protocol 1, using a matrix to which the immunogen is covalently bound. The preparation of these affinity columns is accomplished using a variety of methods which vary depending upon the antigen. Many excellent sources of information exist for the preparation of these columns, including earlier books in this series (5, 6).

2.3 Choice of cells Many different mammalian cell lines have been successfully used in microinjection studies. A general characteristic of cells to be used in these studies is that the cell line should grow in an adherent manner in a monolayer. Handling of the cells during and after microinjection is dramatically simplified if the cells will grow on glass (see Protocol 3 for preparation of glass coverslips), and the quality of immunonuorescence imaging is also improved. Fortunately, most cells, which will grow in adherent culture, will also grow on glass, although it may in some cases be necessary to pre-coat the coverslips with a matrix of protein. In general, it is much easier to work with large, flat cell types, such as fibroblasts, although there will be cases when more specialized cell types produce results that are more biologically relevant. We have used Rat-1 fibroblasts extensively in these experiments because they are easy 124

5: Functional characterization of co-activators to grow and work with and because they are fairly forgiving when injected. These cells do not express all types of nuclear receptors, however, and it is often necessary to use another cell type when studying a receptor which is not expressed. Protocol 3. Preparation of acid washed coverslipsa Reagents • Acid wash solution: concentrated nitric acid and hydrochloric acid

Method 1. Gently transfer glass coverslips from supplier's container to a large (i.e. 1 litre or greater) beaker. 2. Place flask in fume-hood area on absorbent material. 3. Add sufficient nitric acid to just cover the coverslips. 4. Carefully add one-half of this volume of hydrochloric acid. Carefully mix with glass rod. Mixture will fume at this point. 5. After 2 h, transfer the acid mixture to waste container, and wash coverslips with large volumes of distilled water with occasional mixing. Removal of all traces of acid is imperative. 6. Pour off water and replace with 100% ethanol. Four to five rinses may be necessary to remove all water. 7. Store coverslips in tightly screw-capped jars. Use of small containers (less than 1000 coverslips) prevents absorption of excessive moisture during use. a Handling of cells during the microinjection process and subsequent staining and analysis is most easily achieved when cells are grown on glass coverslips. Various manufacturers sell 10 mm circular coverslips, which should be carefully cleaned prior to use. It is suggested that a six month or greater supply of coverslips be prepared at a time, as it is an unpleasant task at best, and the coverslips can be stored indefinitely in a sealed container.

2.4 Preparation of cells Cells to be used in microinjection studies should be grown and passaged normally in the medium in which they grow best. In preparation for an experiment, the cells are passaged into dishes containing glass coverslips. It is of particular importance that every effort be made to synchronize the cells prior to use in these experiments. Evidence is currently accumulating that the regulation of nuclear receptors and their co-regulatory molecules is acutely affected by interaction with cytoplasmic signal transduction cascades (7). For most cell lines, quiescence can be achieved by removal of serum from the growth medium for 24-48 h, although some transformed cell lines will 125

D. W. Rose et al. continue to grow unabated at least for several days in the absence of growth factors.

2.5 Preparation of samples to be injected The preparation of samples for injection purposes is in general best if done immediately prior to the injection experiment. In all cases, marker IgG must be included in the sample to allow the detection of injected cells. This IgG may be the specific antibody being tested, or it may be a pre-immune IgG obtained from the same species of animal. When available, pre-immune IgG from the same animal used for antibody production is the best possible negative control. Total volumes of samples may be as small as 1-2 ul, because of the small amount of material actually used in filling the needle. Reporter constructs and expression vectors are added directly to the IgG for coinjection purposes. We have used as many as three different DNA species mixed with two different antibodies in our experiments (8). The optimum concentration of DNA must be determined empirically and will dramatically affect the level of expression of the encoded protein. In general, a concentration of 100 ug/ml provides a high level of expression from an expression vector employing a viral promoter and is appropriate for nearly all reporters.

2.6 Choice of DNA to inject 2.6.1 Preparation of DNA We have used the procedure described in Protocol 4 for the routine preparation of plasmids for use in microinjection experiments. This method works equally well for reporter plasmids or for other expression constructs for use in rescue experiments. There are obviously other methods for purifying DNA, including the widely used commercial kits for that purpose, but we include this method here because it has proven to be quite reliable and relatively inexpensive. Protocol 4. Preparation of plasmids for microinjection Reagents • GTE: 25 mM Tris-HCI pH 8.0, 10 mM EDTA, • High salt solution: 3 M K acetate, 1.8 M 50 mM glucose (sterile filter and store at formic acid 4°C) • 3 M Na acetate pH 5.2 . NaOH-SDS: 0.2 N NaOH, 1% (w/v) SDS . TEN: 10 mM Tris-HCI pH 8.0, 0.15 M NaCI, (make fresh every time) 10 mM EDTA . 10 M LiCI

Method 1. Grow 50 ml of plasmid-containing bacteria in Luria broth overnight. 2. Centrifuge in conical tubes at 4500 g for 30 min. Pour off supernatant (cells can be frozen at this point if desired).

126

5: Functional characterization of co-activators 3. Add 4.25 ml of GTE buffer, resuspend gently. 4. Add 250 ul of lysozyme (10 mg/ml in TEN), gently mix, leave at room temperature until viscous. 5. Add 9 ml of fresh NaOH-SDS, mix gently, and leave at room temperature until solution is semi-transparent (about 15 min). 6. Add 6.75 ml of high salt solution. Invert tube immediately several times to mix and incubate on ice for 15 min. 7. Centrifuge for 1 h at 4500 g at 4°C. 8. Pour supernatant into clean 50 ml conical tube and add 11 ml of room temperature isopropanol. Mix and let stand at room temperature for 15 min. 9. Centrifuge to pellet DNA at 3000 g for 5 min. Wash with 70% ethanol, dissolve in 490 ul water, and add 10 ul of 10 mg/ml RNase. Incubate for 30 min at 37°C. 10. Transfer to an Eppendorf tube and add 200 ul of 10 M LiCI. Gently mix, do not vortex. Incubate on ice for 5 min. Centrifuge at 14000 r.p.m. in a table-top microcentrifuge for 5 min at 4oC. 11. Transfer supernatant to a clean tube, add 70 ul of 3 M Na acetate pH 5.2. Precipitate nucleic acids with 460 ul of room temperature isopropanol. 12. Centrifuge 10 min at 14000 r.p.m., discard supernatant. 13. Wash pellet with 70% ethanol. Air dry and resuspend in 200 ul of microinjection buffer (see Protocol 2). Determine concentration spectrophotometrically.

2.6.2 Reporter plasmids The design of reporter constructs for use in these studies has been described in a number of published sources (9-11). When designing a reporter, there are several considerations. First, there are several potentially useful reporter molecules which can be used including chloramphenicol acetyltransferase (CAT), luciferase, and b-galactosidase (LacZ). Because detection is possible using a histochemical staining technique instead of indirect immunofluorescence, and because background staining is non-existent or minimal in mammalian cells, LacZ has proven to be a very useful choice of reporter. Expression of LacZ is detected in injected cells by staining with a chromogenic b-galactosidase substrate, 5-bromo-4-chloro-3-indolyl-p-D-galactopyranoside (X-Gal), which produces an insoluble blue colour in cells which express the gene product (see Protocol 7). When constructing reporter plasmids, multiple copies of the transcription factor recognition element normally are needed to drive high level expression of the reporter. The number of elements required varies, however, with each 127

D. W. Rose et al. nuclear receptor or other transcription factor. It is desirable to use the number of elements whieh provides a low background (less than 10% expressing cells in the absence of ligand) and a f u l l transcriptional response (varies from 35-80% in the presence of ligand). This must be evaluated experimentally. In general, between three and eight copies will usually provide an optimum signal.

2.6.3 Expression plasmids There are no hard and fast rules for the construction of expression plasmids for use in microinjection experiments, but several useful features may be kept in mind. Expression of the encoded protein should he driven by a strong promoter, which will provide rapid, high level expression. This provides the user with maximum flexibility in the manipulation of timing of events. In most types of cells, the cytomegalovirus (CMV) promoter is the best choice. Secondly, it may he necessary at limes to demonstrate that the encoded protein is in fact being expressed in injected cells, and that it is stable during the time course of the experiment. One method to address the expression issue is to co-inject a second plasmid in which the same promoter is used to drive the expression of green fluorescent protein (1). This approach does not address the issue of stability of the protein of interest, however. To directly demon-

Figure 2. Microinjection apparatus. (A) Eppendorf semi-automated microinjection system. Movement of stage-mounted motors in three dimensions is accomplished with the joystick on the right. Equipment is installed here on a Zeiss Axiovert 100 inverted microscope. Controller on left allows adjustment of injection pressure and time. System shown here also includes a microscope-mounted video camera (top) and monitor (view obscured on the left). The addition of video permits easier demonstration and training for new initiates. With practice, it is possible to conduct the injection process while watching the monitor, which is ergonomically advantageous to using the microscope directly. (B) Close up view of the motor-mounted stylus which holds the needle and delivers air pressure from behind the sample. (C) View of cells during the injection process taken directly from the monitor.

128

5: Functional characterization of co-activators strate stable expression, it is good general practice to always use epitopetagged versions of the encoded protein. Several such epitope-tagged vectors are available commercially which permit detection of the expressed protein by immunofluorescence staining with a specific antibody.

3. Microinjection technique 3.1 Basic microinjection technique A general guideline for the basic microinjection technique is given in Protocol 5. One of the advantages to this approach is the flexibility it presents in terms of timing and the amount of control of levels of expression that the user has by varying the timing and the concentration of materials injected into the cells. Because there are so many variations on the same theme that are possible, there is not a single protocol that can be written, which will cover all of these variations. Protocol 5. Microinjection Equipment and reagents • Automated or semi-automated microinjection system • Semi-automated microinjection system with micromanipulator (5171) and microinjector (5242) (Eppendorf) (see Figure 2) • Capillary puller: horizontal needle puller (Sutter Instruments) (see Figure 3) • Acid washed glass coverslips (see Protocol 3)

• Air suspension table (Technical Manufacturing Corp.) • Axiophot epifluorescence microscope (Zeiss) with Photometrix charge-coupled device (CCD) camera SenSys (see Figure 4) • Antibody/DNA-containing samples to be injected (see Protocols 2 and 4)

Method 1. At least two days prior to the actual microinjection, set up cells on glass coverslips: (a) Transfer sufficient number of acid washed coverslips to a sterile tissue culture dish in the tissue culture hood, draining as much of the 100% ethanol as possible, (b) Partially cover the dish with the lid, and allow the coverslips to air dry in a sterile environment for about 15 min. (c) Using sterile jeweller's forceps, transfer 8-12 coverslips from original dish to a sterile 60 mm tissue culture dish. If any trace of ethanol remains, briefly pass coverslips through flame to evaporate. (d) Arrange coverslips to avoid overlap; gently score small area in centre of each coverslip with diamond tip pen to ease subsequent localization of injected cells.

129

D. W. Rose et al. Protocol 5. Continued (e) Seed trypsinized cells on coverslips in appropriate complete medium. Aim for a density of approximately 70% confluency on the day of injection. 2. One day prior to injection, render cells quiescent by serum deprivation: (a) Remove complete medium by sterile aspiration. (b) Wash three times with sterile PBS. (c) Replace complete medium with appropriate serum-free or serum-poor medium, depending upon cell type. 3. On day of injection, prepare sufficient number of 35 mm dishes to hold coverslips to be injected. Fill dishes with 3 ml each of prewarmed serum-free medium. 4. Using sterile jeweller's forceps, transfer one coverslip to centre of each dish, maintaining orientation of cells. 5. Centrifuge antibody/DNA samples at 14000 r.p.m. for at least 15 min in refrigerated microcentrifuge. 6. 'Backload' needle by placing in sample on ice with point of needle facing skyward. Allow sample to fill tip of needle by capillary action. This method of loading needle will use approx. 0.1-0.5 ul of sample, depending upon length of time in contact with sample. 7. Deploy needle on stylus and mount on microscope. 8. Microinject 200-500 cells/coverslip. Take care to inject only the nuclei of cells. Adjust pressure and time of injection to allow the appropriate volume to enter each injected cell. Properly injected cells react with a very slight change in phase contrast. Perceptible changes in cell or nuclear volume indicate excessive material entering cell. Inject cells within the scored area on the coverslip to facilitate their localization during analysis. 9. Following injection, promptly return cells to incubator. 10. If expression constructs are being injected, allow time for expression (typically from 6 h to overnight). 11. Stimulate cells when appropriate with nuclear receptor ligand. 12. Allow expression of reporter for 4-6 h.

3.1.1 Preparation of needles The characteristics of the needle used to do the actual injection is one of the most important factors for a successful experiment. Whether or not it is possible to deliver the correct volume of sample into the nuclei of cells 130

K Functional

characterization

of

co-activators

Figure 3. Horizontal capillary puller apparatus. (A) Sutter Instruments needle puller which has sophisticated programming features that allow adjustment of the needle length and of the diameter of the needle opening. Needles are prepared in-house from glass capillaries. (B) x 64 microscopic view of an ideal needle configuration for nuclear injection of DNA and antibodies. Note the extremely fine tip.

depends largely on the shape, size, and thickness of the tip of the needle. In general, for nuclear microinjcction it is desirable to use a needle which has an extremely fine tip, so that one can control the injection volume precisely by variations in injection pressure and time. Equally important, the needle should also widen from the tip very gradually, because this helps to optimize the quality of the phase-contrast image seen during the injection process (see Figure 3B). Needles which widen rapidly from the tip are in general suitable for use in terms of delivery, hut create a difficult optical situation, making the injection process more tedious. Many commercially available needles suffer from this latter characteristic, in addition to being prohibitively expensive.

3.2 Monitoring expression Following incubation to allow expression of reporter activity, injected cells are stained for b-galactosidase expression as well as for co-injected IgG (see Protocols 6-8). This allows the unambiguous identification of injected cells, which can then be individually scored as either positive or negative for reporter expression. Indirect immunofluorcscence staining is accomplished simply by diluting the fluorescent secondary antibody in a diluent which simultaneously permeabilizes the cells and also contains a blocking agent to reduce background fluorescence. 131

D. W. Rose et al. Protocol 6. Preparation of X-Gal staining solution Reagentsa • 100 mg/ml X-Gal in N,N-dimethylformamide (DMF) (100 x stock): store at -20°C in amber glass container • 100 x potassium ferrocyanide solution: dissolve 221 mg in 1 ml water

• 100 x potassium ferricyanide solution: dissolve 160 mg in 1 ml water • 1 M MgCI2 in water (500 x stock solution)

Method 1. Add appropriate amounts of each of the four stock solutions to sterile PBS. For example, to make 50 ml of staining solution, add: • 500 ul X-Gal stock solution • 500 ul potassium ferrocyanide solution • 500 ul potassium ferricyanide solution • 100 ul of 1 M MgCI2

• 48.4 ml PBS 2. Gently heat to 37°C to completely dissolve X-Gal. 3. Cool to room temperature. 4. Filter to remove crystals. 5. Use within 24 h period. "X-Gal stock solution and MgCI2 are stable when stored as indicated. Other solutions are made fresh immediately prior to use.

Protocol 7. Staining of injected cells Reagents • Staining diluent solution: 50 mg/ml BSA, 0.5% Nonidet P-40 (or other similar nonionic detergent), dissolved in sterile PBS

• Fixative: 3.7% formaldehyde in PBS • X-Gal staining solution (see Protocol 6) • Mounting medium (see Protocol 8)

Method 1. Remove cells from incubator, and without removing coverslip from tissue culture dish, remove medium by aspiration and gently wash once with PBS; add sufficient fixative to cover cells completely. 2. Incubate at room temperature for 15 min. 3. Aspirate fixative, and wash three times with PBS. 4. Add sufficient X-Gal staining solution to cover cells, incubate at 37°C. Incubation time will vary depending upon conditions of experiment

132

5: Functional characterization of co-activators and reporter construct being used. Overnight staining at 37°C is acceptable and often convenient. 5. Remove cells from incubator, aspirate X-Gal solution, and wash three times with PBS. 6. Dilute fluorescent-conjugated secondary antibodya in staining diluent solution at dilution recommended by manufacturer. Remove microprecipitates from the diluted antibody by centrifugation in a refrigerated microcentrifuge for 15 min immediately prior to staining. 7. Aspirate remaining PBS, add approx. 50 ul of diluted fluorescent antibody to each coverslip. Incubate for 1 h at room temperature. 8. Remove stain, wash three times with PBS. Wash a final time in distilled water to remove salts. 9. For each coverslip used, place a small drop of mounting medium on glass slide. 10. Using jeweller's forceps, dab excess water from each coverslip, and mount with cells facing the glass slide. Take precautions to avoid air bubbles under the coverslip. Aspirate excess liquid from edges of coverslip to seat efficiently. 11. Mounting medium will dry in about 1 h. If desired, seal edges of coverslips with clear nail polish. a Either fluorescein- or rhodamine-conjugated antibodies may be used, but rhodamine may be preferable because it is less prone to photobleaching during analysis. In certain cell lines, background fluorescence may be a problem. If this proves to be the case, It is possible to dilute the secondary antibody in a suspension of fixed cells in the normal diluent. Allow the cell suspension to incubate on ice for 15 min with the antibody before clearing by centrifugation as indicated in step 6.

Protocol 8.

Preparation of mounting mediuma

Reagents • Glycerol . 200 mM Tris-HCI pH 8.5 • 1,4-diazobicyclo-{2.2.2}-octane

• Airvol polyvinyl alcohol 205 (Air Products and Chemicals, Inc.)

Method 1. Add 48 g of polyvinyl alcohol powder to 120 g of glycerol in a glass container. 2. Stir gently and thoroughly. Add 120 ml distilled water. Stir at room temperature. 3. Add 240 ml of 200 mM Tris-HCI pH 8.5 and heat with gentle stirring to 50°C for 10 min. 133

D. W, Rose et al. Protocol 8.

Continued

4. After solid completely dissolves, centrifuge at 5000 g for 30 min. 5. Add 12.1 ml of diazobicyclo-octane solution and mix gently. 6. Aliquot in small, tightly-sealed containers. Store frozen. Thawed aliquots are stable at 4°C for several weeks. 7. After mounting and drying, coverslips can be permanently sealed with a small amount of clear nail polish at the edges. a We have used this medium for permanent mounting of coverslips on glass slides for analysis, photography, and storage. The medium is resistant to invasion by air bubbles, and possesses little or no intrinsic fluorescence. Slides stored in a cool dry container in the dark are stable for long periods of time and can be reviewed even after years of storage. The medium can be prepared in large volumes and stored at -20 C C in small aliquots.

3.3 Analysis of results Analyses of the results of the completed experiments are conducted using a fluorescence microscope. Because all injected samples contain IgG. which is detected during the staining procedure, first locate the injected cells using the ultraviolet light source. Locating the injected cells is simplified by first looking inside the scored area of the coverslip under low magnification. Once the

Figure 4. Epifluorescence photomicroscope with CCD camera and digital imaging capability. Shown is a Zeiss Axiophot microscope used in our studies. Analysis and photography can be greatly enhanced by addition of digital capture equipment. Many such configurations are possible from a wide variety of manufacturers.

134

5: Functional characterization of co-activators area of injected cells is located, increase the magnification and begin the quantitation, taking care not to count any injected cell more than once. Each injected cell is scored as positive or negative for LacZ expression by switching on the phase-contrast light source, with any trace of blue staining representing a positive cell. If the fluorescent staining is sufficiently bright, it may be possible to view the cells simultaneously with the UV light source and the phase-contrast light source turned down to dim. This procedure makes the quantitation process less tedious and time-consuming.

References 1. Alberts, A. S., Frost, J. A., and Thorburn, A. M. (1993). DNA Cell Biol., 12, 935. 2. Bar-Sagi, D. (1995). In Methods in enzymology (ed. W. E. Balch, C. J. Der, and A. Hall), Vol. 255, p. 436. Academic Press, London. 3. Paterson, H., Adamson, P., and Robertson, D. (1995). In Methods in enzymology, (ed. W. E. Balch, C. J. Der, and A. Hall), Vol. 256, p. 162. Academic Press, London. 4. Lamb, N. J. C. and Fernandez, A. (1997). In Methods in enzymology (ed. W. G. Dunphy), Vol. 283, p. 72. Academic Press, London. 5. Harlow, E. and Lane, D. (ed.) (1988). Antibodies: a laboratory manual. Cold Spring Harbor Laboratory Press, NY. 6. Catty, D. (ed.) (1988). Antibodies: a practical approach. IRL Press, Oxford. 7. Xu, L., Lavinsky, R. M., Dasen, J. S., Flynn, S., Mclnerney, E., Heinzel, T., et al. (1998). Nature, 395, 301. 8. Korzus, E., Torchia, J., Rose, D. W., Xu, L., Kurokawa, R., Mullen, T.-M., et al. (1998). Science, 279, 703. 9. Kamei, Y., Xu, L., Heinzel, T., Torchia, J., Kurokawa, R., Gloss, B., et al. (1996). Cell, 85, 403. 10. Torchia, J., Rose, D. W., Inostroza, J., Kamei, Y., Glass, C. K., and Rosenfeld, M. G. (1997). Nature, 387, 677. 11. Lavinsky, R. M., Jepsen, K., Heinzel, T., Torchia, J., Mullen, T.-M., Schiff, R., et al. (1998). Proc. Natl. Acad. Sci. USA, 95, 2920. 12. Heinzel, T., Lavinsky, R. M., Mullen, T.-M., Soderstrom, M., Laherty, C. D., Torchia, J., et al. (1997). Nature, 387, 43. 13. Rose, D., McCabe, G., Feramisco, J. R., and Adler, M. (1992). J. Cell. Biol., 119, 1405.

135

This page intentionally left blank

6

Analysis of steroid/nuclear receptor phosphorylation B. G. ROWAN and N. L. WEIGEL

1. Introduction As with other transcription factors, phosphorylation of nuclear receptors has been shown to be important in regulating receptor function. Studies in this area have utilized a variety of approaches and methodology to analyse receptor phosphorylation and the relation of phosphorylation to receptor function. This chapter will outline these different approaches and the current methodologies available to identify phosphorylation sites, identify candidate kinases, and assay the functional significance of nuclear receptor phosphorylation. The advantages and disadvantages of each methodology are presented as well as a discussion of general considerations for examining receptor phosphorylation. This information should enable the beginning investigator to choose the most appropriate method to study receptor phosphorylation in different model systems.

2. Phosphorylation of the nuclear receptor superfarnily At this time, all the members of the superfarnily that have been examined have been shown to be phosphoproteins and most members are multiply phosphorylated. Most receptor phosphorylation occurs on serine residues and to a lesser degree on threonine residues. Thus far, phosphorylation on tyrosine residues has been detected in only two receptor types: the oestrogen receptor alpha form (ER) (1, 2) and the retinoic acid receptor (RAR) (3). There are numerous examples that show regulation of receptor function by phosphorylation. Phosphatase treatment of both the chicken oestrogen receptor (cER) and human oestrogen receptor (hER) reduces steroid binding capacity in in vitro assays (1, 4, 5). Regulation of DNA binding by phosphorylation is also suggested. Using various in vitro-based assays involving ligand, kinase, or phosphatase treatment, it was suggested that phosphorylation enhanced DNA binding for cPR (6), hPR (7), hER (2, 8), thyroid hormone

B. G. Rowan and N. L. Weigel receptor (TR) (9, 10), retinoic acid receptor a (RARa), and retinoid X receptor a (RXRa) (11). For other receptors including rabbit PR (12), vitamin D receptor (13), the TR variant TRa2 (14), and the orphan receptor Nur77 (15, 16), phosphorylation at specific sites decreased DNA binding. Most work examining the relation of phosphorylation to receptor function has focused on transactivation by using site-directed mutagenesis to delete specific phosphorylation sites and ectopic expression of active and inactive kinases. Using this approach it has been suggested that changes in phosphorylation at specific sites in the cPR (17-19), hER (20-25), GR (26-28), and human androgen receptor (hAR) (29) results in altered transcriptional activity in a cell- and promoter-specific manner. There is also evidence that some of the phosphorylation sites in the glucocorticoid receptor regulate receptor stability (30). Although other phosphorylation sites in these receptors and others receptors have little or no effect on receptor transactivation there are many more phosphorylation sites that remain to be examined. In summary, despite the amount of work in this area, the role of phosphorylation in nuclear receptor function remains incompletely described.

3. General guidelines for studying nuclear receptor phosphorylation Few investigators will wish or need to embark on a comprehensive identification of receptor phosphorylation sites. However, the approaches used to identify sites can be adapted to applications such as determining whether receptors are substrates for specific kinases, whether specific signalling pathways alter receptor phosphorylation at specific sites, and to determining the role of specific phosphorylation sites in receptor function. In cases where novel sites are detected, approaches described below can be used to identify the new sites. There are several general considerations when beginning a study of receptor phosphorylation. One consideration is the source of the receptor to be used for the study (i.e. endogenous receptor in cells or tissue, receptor expressed in cells using transfection, in vitro translated receptor, baculoviralor bacterially-expressed receptor systems). Some of these sources will provide large amounts of receptor necessary for phosphorylation site identification (e.g. some tissues or cells with endogenous receptor and baculoviral expression systems) whereas other sources provide much less receptor and hence, limit the types of studies that can be performed. Another general consideration is the method of receptor purification and the recovery after purification. One should determine whether there are high affinity antibodies available for immunopurification of the receptor and the relative amount of receptor that can be purified. Alternatively, antigen-tagged or fusion proteins of the receptor can be prepared in expression systems and subsequently purified using various column procedures. One other general consideration when 138

6: Analysis of steroid/nuclear receptor phosphorylation using [32P]H3PO4 labelling is whether the study requires uniform or activated receptor phosphorylation. It is best to have uniform labelling of receptors (> 8 h) when identifying phosphorylation sites. However, short-term labelling of receptors (1 h) is sufficient when studying activated phosphorylation of receptors following different treatment conditions. The following is a list of some phosphorylation studies that can be performed including some general considerations for each study: (a) Identification of phosphorylation sites in vivo. If there is little or no information in the literature about the site-specific phosphorylation of a nuclear receptor, then one may wish to identify the sites (29, 31, 32). This will require a large amount of purified receptor (ng range), possibly a large amount of [32P]H3PO4, and it is very time-consuming. (b) Regulation of previously identified phosphorylation sites. If some phosphorylation sites have already been identified, then one could study the regulation of phosphorylation at a specific site under chosen conditions. Some examples of studying phosphorylation site regulation include the effects of hormone, growth factor, or drug treatment (23, 24), cell cycle dependency (33), tissue- or ligand-specific effects (6, 7), and the effect of blockage or activation of signalling pathways (23, 24, 34). These studies would require far less protein and [32P]H3PO4 than is needed to identify phosphorylation sites. In general, a comparison to a published phosphopeptide map would be sufficient to study the regulation of specific phosphorylation sites. (c) In vitro phosphorylation of receptors. In vitro phosphorylation of receptors can be performed to identify kinases and hence, cellular pathways that control receptor phosphorylation (23, 26, 32, 35). These studies require purified receptor and low levels of [y-32P]ATP. Furthermore, phosphorylation site identification can also be performed. In general, if there is significant receptor phosphorylation in vitro, then it is reasonable to try and identify sites in vivo. One caution: because there are many potential kinases that can phosphorylate certain consensus phosphorylation sites in vivo, do not pursue phosphorylation of the receptor with a kinase that only marginally phosphorylates the receptor in vitro. The authentic in vivo kinase for a particular site should induce significant phosphorylation of the site in vitro. (d) Phosphorylation of receptors for functional studies. As an extension of in vitro phosphorylation, receptor phosphorylated in vitro can also be used to examine different receptor functions such as hormone binding (1), DNA binding (13, 14), and others. The results can be compared to phosphatase-treated receptor. Again, these studies require purified receptor and low levels of [y-32P]ATP. (e) Stoichiometry of receptor phosphorylation. The stoichiometry of receptor phosphorylation at specific sites can be determined using double labelling 139

B. G. Rowan and N. L. Weigel procedures in cells with [32P]H3PO4 and [35S]methionine (36). These studies require far less protein and [32P]H3PO4 than needed for site identification. (f) Site-directed mutagenesis of phosphorylation sites. Individual phosphorylation sites can be mutated to alanine (to mimic loss of a phosphorylation site) and the effects of the mutation on receptor functions can be studied by comparing wild-type and mutant receptor following transfection into cell lines (17-19, 23, 24). Transactivation, subcellular localization, and protein-protein interactions are some of the functions that can be studied. These experiments do not require labelled receptor or purified receptor. The following protocols will describe many of the procedures used in the examples given above.

4. Identification of phosphorylation sites in nuclear receptors 4.1 Approaches to identify phosphorylation sites in nuclear receptors and other proteins There are several approaches that can be used to identify phosphorylation sites in receptors, each of which has its advantages and disadvantages. In the conventional approach, cells or tissue minces are labelled with [32P]H3PO4 in vivo, followed by receptor purification and enzymic or other digestion of the receptor to generate phosphopeptides. The phosphopeptides are individually isolated and then subjected to direct gas phase protein sequencing to identify the phosphorylated amino acids (37). This method is very accurate but requires a large amount of purified receptor (ug range), high levels of [32P]H3PO4, and is costly and time-consuming. A second approach is similar to the conventional approach but requires less purified protein (pg to ng range). In this approach, individual receptor phosphopeptides are prepared in the same way as with the conventional approach. However, since there is not enough phosphopeptide generated for direct protein sequencing, each peptide is subjected to manual Edman degradation in combination with secondary protease digestion of phosphopeptides. This information, along with phosphoamino acid analysis of the protein, is sufficient for identification of most phosphorylation sites. A disadvantage of this procedure is that, although it requires less protein, it does require high levels of [32P]H3PO4. Also, manual Edman degradation and secondary protease digestion data may be insufficient to unequivocally identify the phosphorylation site(s) for some phosphopeptides. Finally, some large and/or hydrophobic peptides may not be suitable for manual Edman degradation. A third approach is an indirect approach that avoids the necessity for large 140

6: Analysis of steroid/nuclear receptor phosphorylation amounts of purified receptor, high levels of [32P]H3PO4, and the need to generate and resolve phosphopeptides. Phosphorylation sites are first sublocalized to different regions of the receptor by expressing various receptor deletion mutants in cells. Using this information along with phosphoamino acid analysis, site-directed mutagenesis is then used to mutate candidate phosphorylation sites in the different regions. Following mutagenesis of the different deletion mutants, each is then analysed for the presence or absence of phosphorylation. There are several disadvantages to this procedure. Expression of receptor deletion mutants can result in inappropriate subcellular localization of the mutants where non-physiological phosphorylation can occur. Also, improper folding of receptor mutants may expose nonauthentic phosphorylation sites and/or mask authentic sites. Furthermore, point mutation of a phosphorylation site can result in phosphorylation on an alternative, non-authentic site within the same region (38). A fourth approach to identifying phosphorylation sites is the use of mass spectrometry. This is a very powerful method that has the potential to identify every phosphorylation site in a protein without the use of [32P]H3PO4. The biggest disadvantage to this method is the prohibitive cost of a mass spectrometer. The analysis also depends on a high molar stoichiometry of protein phosphorylation and about 1 pmol of purified protein. In summary, the choice of the approach to use to identify phosphorylation sites in receptors will depend, in general, on the amount of purified receptor that can be obtained, equipment availability, and cost and time considerations.

4.2 Conventional method The following protocols describe the methodology used to identify phosphorylation sites by the conventional method. This method involves in vivo labelling of receptors in cells or tissue minces with [32P]H3PO4, purification of receptors, generation and separation of receptor phosphopeptides, and gas phase protein sequencing of the peptides to determine the phosphorylated amino acids. A detailed description of peptide sequencing is given elsewhere (37). The conventional approach has been used to identify phosphorylation sites in the chick PR (6), the human GR (39), and the human ER (40).

4.2.1 In vivo labelling of nuclear receptors Receptor in both cells and tissue minces can be labelled with [32P]H3PO4 in vivo. The source of receptor can be endogenous, or receptor from transfected expression vectors. For transfection into cells, it is usually best to find a procedure that will give maximal protein expression. For phosphorylation site determination, as many as 108 cells may be required when receptor expression is low and/or the stoichiometry of phosphorylation is low. To simply visualize a 32P-labelled receptor much less receptor protein is required. The amount of [32P]H3PO4 to add and the duration of incubation with [32P]H3PO4 will 141

B. G. Rowan and N. L. Weigel depend on the objectives of the experiment. To identify phosphorylation sites in the receptor, up to 3 mCi/ml of medium may be necessary. Similarly, the duration of labelling can vary. To achieve steady state labelling of receptors, it is desirable to incubate the cells with [32P]H3PO4 for as long as 12-15 h. However for activated phosphorylation, labelling as short as 1 h may be sufficient. The duration of reagent treatment should be chosen carefully. Some fast acting agents such as activators of some protein kinases may affect protein phosphorylation in as little as a few minutes followed by a rapid reversal of the effect. Other agents result in a sustained alteration in receptor phosphorylation over a long period allowing for longer treatments. A good starting point for choosing treatment duration is to first determine the minimum amount of time necessary for an agent to have an end-point functional effect on the receptor. If the agent alters receptor phosphorylation and this alteration contributes to the end-point effect, then the phosphorylation change should occur at some time before the end-point effect is seen. In general, it is best to do a time course of agent treatment and subsequently assess receptor phosphorylation. Extra precaution should be taken when labelling in vivo with high levels (greater than 1 mCi) of [32P]H3PO4. Adequate shielding, eye protection, ring badges (in addition to standard radiation badges), beta ray-blocking apron, double gloves, barrier tips, and disposal procedures should be employed. In addition, all bench space, centrifuges, floors, pipettes, door handles, and anything else that the user comes in contact with should be monitored with a Geiger counter immediately after use. Users should contact their radiation safety department before beginning experiments for appropriate license, instructions, and training. Protocol 1. In vivo labelling of nuclear receptors with [32P]H3P04 Reagents • [32P]H3P04, 0.5 mCi/ml (ICN) • Dialysed fetal calf serum (FCS) (Hyclone)

• DMEM without sodium pyruvate and sodium phosphate (Gibco/BRL)

Method 1. Culture cells at 37°C, 5% C02, 95% O2 in medium containing 5% FCS for 24 h prior to labelling with [32P]H3PO4.a 2. When cells are in log phase growth remove the medium and replace it with phosphate-free medium (no serum). For suspension cultures, centrifugation at 1000 g for 5 min is usually sufficient to pellet the cells. Incubate the cells at 37°C, 5% C02, 95% O2 for 1 h to deplete endogenous phosphate stores.b 3. Remove the medium from step 2 and replace it with phosphate-free medium containing 1% dialysed FCS. For adherent cells, use just enough

142

6: Analysis of steroid/nuclear receptor phosphorylation medium to completely cover the cells. For suspension cultures, use just enough medium to maintain the cells at a log phase density. 4. Add [32P]H3PO4 to the medium for a final concentration of 0.1-3 mCi/ml and incubate the cells at 37°C, 5% CO2, 95% O2 for 1 h.c 5. Add steroid or other reagents to the cells and continue incubation for the desired treatment time. Alternatively, for very short treatment duration, cells can be cultured in [32P]H3PO4-containing medium for 1-15 h prior to reagent addition. 6. Remove the [32P]H3P04-containing medium from the cells and add an equal volume of PBS (4°C) to the cells to remove traces of radioactive medium. Remove the PBS wash. For adherent cells, add a small volume of PBS to the dishes and use a disposable scraper to release cells from the plate. Transfer cells to a 50 ml centrifuge tube and centrifuge at 1000 g for 5 min to pellet the cells. 7. Remove the PBS supernatant. The pelleted cells are ready for receptor purification (see Protocol 2). " FCS must be stripped of endogenous steroids by treatment with dextran-coated charcoal (see Chapter 11). b To obtain the highest level of labelled receptor, it is important to add [32P]H3PO4 when the cells are in log phase growth. Adding [32P]H3PO4 to cells that are overgrown will result in poor incorporation of label. c High levels of [32P]H3PO4 can result in genomic damage to cells, which may affect cell growth.

4.2.2 Receptor purification When performing in vivo labelling, the purification procedure should result in receptor that is free of major protein contaminants. Certain one-step column purification procedures taking advantage of tagged proteins (e.g. His-Tag, GST fusion) may result in too much [32P] background due to non-specific protein binding to column beads (B. Rowan, unpublished observation). In general, immunoaffinity purification of receptors results in very little nonspecific background. The optimal extraction buffer will have to be determined empirically. For immunoprecipitation, the detergent and ionic strength of the buffer will be dictated by the affinity of the antibody for the receptor antigen (i.e. high affinity antibodies will tolerate higher NaCl and detergent concentrations without disruption of antibody-antigen interaction). However, all extraction buffers should contain protease inhibitors, phosphatase inhibitors, EDTA, EGTA, and reducing agent (see below). Most nuclear receptors can be extracted from cells using a Tris-based buffer (pH 8.0) containing 0.4 M NaCl. In initial experiments, a control lacking primary antibody should be included to check for co-purification of non-receptor phosphoproteins of the same molecular weight. 143

B. G. Rowan and N. L. Weigel Protocol 2. Immuoprecipitation of nuclear receptors Reagents • General extraction buffer: 10 mM Tris-HCI pH 8.0, 400 mM NaCI, 1 mM EDTA, 1 mM EGTA, 25 mM NaH2P04, 50 mM NaF,a 10 mM sodium vanadate,a 10 mM DTT, protease inhibitor cocktail (1 ul/ml of each of the following from a 1 mg/ml stock in DMSO: leupeptin, antipain, aprotinin, benzamidine-HCI, chymostatin, pepstatin; 2.5 ul/ml of PMSF from a 200 mM stock in dioxane)

• Protein A-Sepharose CL-4B (Pharmacia) • Antibody to receptor • Secondary antibody that reacts with the primary antibody if needed

Method b 1. Add the general extraction buffer to the cell pellet (about 1 ml/107 cells).c 2. Vortex the cell pellet well. 3. Centrifuge the suspension at 100 000 g for 30 min. 4. Purify receptors using a general immunoprecipitation procedure. 5. Prior to elution of receptors from protein A-Sepharose beads, wash the beads extensively with wash buffer.d 6. Rinse Sepharose beads with a small volume of water (equal to the volume of the beads) to remove detergents and salts. 7. Transfer beads to a fresh siliconized tube prior to adding 1 x SDSPAGE sample buffer. Add 1 x SDS-PAGE sample buffer and heat at 100°C for 5 min. The purified receptor is now ready for SDS-PAGE. "Phosphatase inhibitors. b All steps should be carried out on ice or at 4°C. c Small amounts of detergent (e.g. 0.1% Triton X-100) can be added to the general extraction buffer to facilitate receptor extraction from cells. d Inadequate washing will result in high [32P) background on SDS-PAGE gels. To reduce high background, add NaCI and/or detergent (e.g. Triton X-100) to the wash buffer. Only high affinity antibodies will withstand high levels of NaCI and detergent without disruption of antibody-receptor interaction.

4.2.3 Generation of phosphopeptides To examine protein phosphorylation, it is necessary to digest the protein into smaller phosphopeptides that can be separately analysed. The ideal condition is to find a proteolytic method that generates phosphopeptides, each of which contain no more than one phosphorylation site. Digestion of proteins with the protease trypsin is the most commonly used method to generate phosphopeptides. Trypsin has been the protease of choice because it cleaves proteins at multiple sites (on the C terminal side of every arginine [R] and lysine [K]

144

6: Analysis of steroid/nuclear receptor phosphorylation residue), and it is inexpensive. A minor disadvantage of trypsin is that it cleaves inefficiently at R/K-proline sequences. Other common proteases include Asp-N, Glu-C, Arg-C, and Lys-C each of which have a different specificity than trypsin. There are also several non-proteolytic methods to cleave proteins such as treatment with cyanogen bromide, which cleaves proteins at methionine residues. The choice of proteolytic method should be based upon the predicted phosphopeptides generated and the ease with which the peptides can be analysed. For example, generation of phosphopeptides using cyanogen bromide treatment would not be the method of choice for proteins that contain few methionine residues as this would result in only very large phosphopeptides that are more difficult to analyse. The protocol below will describe the generation of phosphopeptides by trypsin digestion. For a list of enzymes and reagents used for proteolysis see van der Geer and Hunter (41). Protocol 3. Primary enzyme digest in gel slice Equipment and reagents • SpeedVac concentrator with refrigerated trap (Savant or comparable) . X-OMAT AR film (X-ray film) (Kodak)

• Trypsin, treated with n-tosyl-L-phenylalanine chloromethyl ketone (TPCK) (Worthington Biochemicals or other company)

Method 1. Electrophorese the purified receptor by SDS-PAGE (receptor identity should be confirmed by Western blot). 2. Wrap the wet gel in cellophane and expose it to X-ray film (be sure to mark the orientation and position of the gel). 3. Cut out the phosphorylated receptor band and place the gel slice in a siliconized 1.5 ml microcentrifuge tube. 4. Add 1 ml of 50% methanol to the gel slice and rock tube for 1 h at room temperature.a 5. Remove the methanol and add 0.5 ml of HPLC grade water to the gel slice. Rock for 5 min at room temperature. Repeat once and remove the water.b 6. Record the total counts in the gel slice by Cerenkov counting.c 7. Add 0.5 ml of freshly prepared 50 mM ammonium bicarbonate (in HPLC grade water) to the gel slice. 8. Add 1-10 ng of trypsin (1 mg/ml prepared in 50 mM ammonium bicarbonate) to the gel slice and incubate the tube in a 37°C waterbath for 4 h.d 9. Add an additional 1-10 ug of trypsin every 4 h over a 12 h period. 10. Remove the supernatant to another siliconzied microcentrifuge tube.

145

B. G. Rowan and N. L. Weigel Protocol 3.

Continued

Add 0.5 ml of HPLC grade water to the gel slice, invert the tube several times, and combine this wash with the original supernatant. 11. Dry down the eluted tryptic phosphopeptides using a SpeedVac. Record the counts released in the dried supernatant and the counts remaining in the gel slice.e 12. The released phosphopeptides are now ready to be separated by reversed-phase HPLC or two-dimensional phosphopeptide mapping (see Section 4.2.5; Protocol 7). a 50% methanol removes the SDS, Tris, and glycine from the gel slice. b This removes the methanol which would inhibit trypsin activity. c Cerenkov counting is performed by placing the microcentrifuge tube containing the gel slice in a large scintillation vial and then counting the vial in a scintillation counter. d 'The amount of trypsin used should be determined empirically. Too little trypsin will result in incomplete digestion whereas excessive amounts of trypsin will interfere with subsequent separation of phosphopeptides by two-dimensional phosphopeptide mapping and can also interfere with Edman degradation and secondary protease digestion (see Section 4.3; Protocol 9). If phosphopeptides are to be sequenced, trypsin must be limited to 5-10% (mol/mol) of the total receptor protein. Because digestion of the receptor protein in a gel slice may require in excess of this amount of trypsin, the receptor should first be eluted from the gel slice prior to digestion with trypsin (41). In solution, trypsin is needed at no more than 5% (mol/mol) of the receptor protein to achieve complete digestion. e At least 70% of the total counts in the gel slice should be released in the dried supernatant. If significantly fewer counts were released, repeat the trypsin digestion and/or check the quality of the trypsin.

4.2.4 Phosphoamino acid analysis Phosphoamino acid analysis of receptors may be performed on intact protein, or individual phosphopeptides. Essentially, an acid hydrolysis step is used to digest the 32P-labelled protein into individual amino acids. The phosphoamino acids are separated by two-dimensional electrophoresis and then visualized by autoradiography and compared to the migration of standard phosphoserine, phosphothreonine, or phosphotyrosine (see Figure 7). Protocol 4. Phosphoamino acid analysis Equipment and reagents • HTLE 7000 two-dimensional peptide gel apparatus (CBS Scientific) • Phosphoserine, phosphothreonine, and phosphotyrosine standards (Sigma) . pH 1.9 buffer: 2.5% (v/v) formic acid (88% stock), 7.8% (v/v) glacial acetic acid in dH2O

• pH 3.5 buffer: 5% (v/v) glacial acetic acid, 0.5% (v/v) pyridine in dH2O • 20 X 20 cm X 100 um thin-layer cellulose plates (EM Science) • Ninhydrin (Pierce Chemical Co.) • Atomizer (General Glassblowing Inc.)

Methoda,b 1. Resuspend either purified 32P-labelled receptor protein or phosphopeptides (at least 500 c.p.m.) with 100-200 ul of 6 M HCI in a 1.5 ml 146

6: Analysis of steroid/nuclear receptor phosphorylation siliconized microcentrifuge tube. Add 2 ul each of 2 mg/ml phosphoserine, phosphothreonine, and phosphotyrosine standards to the tube. Secure the cap and incubate the tube at 110°C for 1 h in a heating block.c 2. Dry the sample using a SpeedVac that has been equipped with a NaOH acid trap. 3. Resuspend the dried sample with 15 ul of pH 1.9 buffer and spot the sample on a 20 x 20 cm cellulose plate (see Figure 1). After the sample is dry, wet the plate uniformly with pH 1.9 buffer.d 4. Electrophorese the sample in the first dimension for 20 min at 1500 V.e 5. Dry the plate for 30 min at room temperature. Wet the plate uniformly with pH 3.5 buffer. 6. Rotate the plate 90° counterclockwise and electrophorese the sample in the second dimension for 15 min at 1300 V using pH 3.5 buffer. 7. Dry the plate for 1 h and use an atomizer to spray the plate with ninhydrin (0.5 g in 100 ml acetone). Visualize the phosphoamino acid standards by heating the plate in a 80°C oven for about 1-5 min.f 8. Use a marker to outline the standards on the back of the plate and mark the orientation of the plate. Wrap the plate with plastic wrap and expose the plate to X-ray film.g " Perform this procedure in a fume-hood because of volatile organic buffers. b This procedure has been optimized to work with the HTLE 7000 two-dimensional peptide gel apparatus. Van der Geer and Hunter (41) give a more detailed description of this procedure. c This reaction breaks most of the peptide bonds without unacceptable levels of hydrolysis of the phosphoamino acids. d To wet the plate, use an atomizer or two pieces of Whatman 3MM filter paper soaked in the buffer. Van der Geer and Hunter (41) give a detailed description of plate wetting with Whatman paper. 'Phosphoamino acids are negatively charged and will migrate toward the anode (red, +) in both dimensions. f The standards will appear in purple colour. g In addition to the phosphoamino acids seen on the autoradiograph, both free [32P]phosphate and incompletely digested peptides will also be detected (Figure 1).

4.2.5 Separation of phosphopeptides The methodologies that are used to separate phosphopeptides are, in general, based on the size and charge of the phosphopeptide. In reversed-phase HPLC, peptides are separated on the strength of their interaction with a hydrophobic matrix in an aqueous solution. The peptides are then eluted from the column with a hydrophobic solution. Two-dimensional phosphopeptide mapping separates peptides in the first dimension (electrophoresis) based on charge/mass ratios and then in the second dimension (chromatography) based on hydrophobicity and size. Phosphopeptides can also be 147

B. G. Rowan and N. L. Weigel

Figure 1. Diagram of an autoradiograph of a hypothetical protein subjected to phosphoamino acid analysis. A protein that was hydrolysed with 6 M HCI to break the peptide bonds was spotted onto a 20 x 20 cm cellulose plate and electrophoresed in two dimensions using the HTLE 7000 two-dimensional electrophoresis system (CBS Scientific) as described in Protocol 4. Shown are the position of the free phosphate, the phosphoarnino acids, and the undigested peptides.

separated using alkaline polyacrylamide gels (no SDS) that also separate peptides based on charge/mass ratios. There are advantages and disadvantages to each separation procedure. Separation of phosphopeptides by reversed-phase HPLC is fast and can accommodate large amounts of protein but an HPLC system is costly and phosphopeptides with similar properties may elute in the same fraction. This problem can be overcome by electrophoresing individual HPLC fractions by alkaline gel electrophoresis as a second separation step. Two-dimensional phosphopeptide maps provide separation of phosphopeptides in two dimensions and most, if not all, of the phosphopeptides can be individually separated. These maps can be generated with equipment that is much less expensive than an HPLC system. The disadvantages of two-dimensional phosphopeptide mapping is that it is not as reproducible as separation by HPLC and there is a limit to the total amount of protein (phosphopeptides plus trypsin) that can be spotted onto the cellulose plates (usually about 30-40 ug in our experience). The final separation method, alkaline peptide gel electrophoresis, is not by itself sufficient to completely separate a complex pool of phosphopeptides (although it may be sufficient for a protein with only a few sites). It is most 148

6: Analysis of steroid/nuclear receptor phosphorylation useful as a second separation step to HPLC as described above. This method is inexpensive and is very sensitive for visualizing low amounts of 32P-labelled phosphopeptides since the dried gels can be exposed to X-ray film for up to several weeks at -70°C. It is also useful for comparing a large number of samples since each gel has multiple lanes. Protocol 5.

Reversed-phase HPLC

Equipment and reagents • HPLC system (Beckman or comparable) with a C18 reversed-phase column (Vydac): have a guard column in-line before the C18 column to protect the column from paniculate matter a • Radioactive flow detector (Packard) in-line after C18 columnb

• Fraction collector • Solution A: HPLC grade water containing 0.1% (w/v) trifluoroacetic acid (TFA) • Solution B: acetonitrile containing 0.1% TFA • SpeedVac (Savant or comparable)

Method 1. Dry the phosphopeptides in a SpeedVac. Resuspend the dried phosphopeptides in 50% formic acid in HPLC grade water (100-200 ul), vortex well, and centrifuge the sample at 12000 g. 2. Run solution A through the HPLC system at a flow rate of 1 ml/min. 3. Inject the sample into the HPLC using either a manual or automatic injector. 4. After the sample has been injected, immediately begin a 0-100% gradient of solution B (1%/min). Begin collecting 1-2 ml fractions.c 5. Dry down individual fractions in a SpeedVac and count the dried fractions by Cerenkov counting.d a The C18 column is sufficient for separation of most phosphopeptides. Very small hydrophilic peptides may elute in the flow-through with free [32P]. For very large or hydrophobic phosphopeptides, a C4 column may be more appropriate. b The radioactive flow detector should be configured to measure [32P] by Cerenkov counting c Most phosphopeptides will elute within a 0-50% acetonitrile gradient. The most hydrophilic peptides will elute early and the most hydrophobic peptides and some of the very large peptides will elute later. The elution gradient can be modified for optimal separation of phosphopeptides. An example of an HPLC elution profile is shown in Figure 2. d Dried HPLC fractions can be used for gas phase sequencing, manual Edman degradation (Section 4.3; Protocol 8), secondary protease digestion (Section 4.3; Protocol 9), or phosphoamino acid analysis (Section 4.2.4; Protocol 4). HPLC fractions that contain more that one phosphopeptide can be separated further by alkaline peptide gel electrophoresis (see Protocol 6).

Generation of phosphopeptides by either chemical or enzymic means as described in Section 4.2.3, will result in phosphopeptides of various lengths and charge. Any given separation procedure may not be able to resolve every phosphopeptide. Consequently, some phosphopeptides in the pool may have to be resolved using a different separation procedure. For example, phos149

B. G. Rowan and N. L, Weigel

Figure 2. Reversed-phase HPLC profile of tryptic phosphopeptides of the chicken progesterone receptor A form (cPRA). cPRA was expressed in CV-1 cells and 24 h later, the cells were labelled with [32P]H3PO4. After 1 h of labelling, the cells were treated with progesterone for an additional 12 h. The 32P-labelled cPRA was immunopurified and the receptor band was isolated by SDS-PAGE. The gel slice containing the receptor was subjected to trypsin digestion for 16 h and the released tryptic phosphopeptides of cPRA were separated by HPLC with a Vydac C18 reversed-phase column using an acetonitrile gradient of 0.5% acetonitrile/min as described in Protocol 5. The [32P] peaks are shown for the tryptic phosphopeptides containing phosphorylation sites at serines 260, 211, 367, and 530.

phopeptides longer than 40 amino acids may have to be resolved using tricine gels (42). One should note the size and the predicted charge of the expected phosphopeptides that will be generated by the methods outlined in Section 4.2.3 and from this information decide on the most appropriate separation procedure(s). Regardless of the separation procedure chosen, phosphopeptides containing cysteine and methionine residues may exist as several charged species that will be differentially separated. To eliminate any potential charge isoforms, phosphopeptides can be treated with performic acid prior to separation (41). Protocol 6. Alkaline peptide gel electrophoresis8 Equipment and reagents • Polyacrylamide gel apparatus with 15 x 18 cm glass plates, 0.5 mm spacers, and 0.5 mm combs • Resolving gel (final concentrations): 40% (w/v) acrylamide (from a 60% (w/v) stock), 0.037% (w/v) bis-acrylamide (from a 2.5% (w/v) stock), 0.75 M Tris-HCI pH 8.8 (from a 3 M stock), 0.035% (v/v) TEMED, 0.1% (w/v) ammonium persulfate (from a 10% (w/v) stock), deionized water to volumeb • Reservoir buffer (final concentrations, pH not titrated): 0.05 M Tris (from solid), 0.4 M glycine (from solid)

• Stacking gel (final concentrations): 3.3% (w/v) acrylamide (from a 60% (w/v) stock), 0.16% (w/v) bis-acrylamide (from a 2.5% (w/v) stock), 0.125 M Tris-HCI pH 6.8 (from a 2.5 M stock), 0.05% (v/v) TEMED, 6 M urea, 0.1% (w/v) ammonium persulfate (from a 10% (w/v) stock), deionized water to volume • Sample buffer (final concentrations): 0.125 M Tris-HCI pH 6.8 (from a 2.5 M stock), 6 M urea, 0.01% (w/v) methylene blue (sample buffer should be prepared fresh)

150

6: Analysis of steroid/nuclear receptor phosphorylation A. Alkaline peptide gel electrophoresis 1. Prepare the resolving gel followed by the stacking gel. 2. Resuspend the dried phosphopeptides with sample buffer, vortex well, and add the sample to dry wells. Overlay the samples in the wells with reservoir buffer. 3. Electrophorese at 5-10 mA per gel in reservoir buffer for 10-16 h or until the methylene blue tracking dye migrates about two-thirds the length of the gel. 4. Pry the glass plates apart and place a piece of Whatman 3MM filter paper on the gel with firm pressure. Pull the Whatman paper and the attached gel away from the glass plate. 5. Dry the gel using a gel dryer for 1.5 h with heat. Mark the orientation of the dried gel and expose the gel to X-ray film for autoradiography.c B. Elution of phosphopeptides from alkaline peptide gels 1. Cut out the band of interest from the dried gel with a razor blade and place the gel slice and the Whatman paper backing into a 1.5 ml siliconized microcentrifuge tube. 2. Wash the gel slice 2 x 5 min with 0.5 ml of 50% methanol in HPLC grade water d and then 1 x 5 min in 0.5 ml of HPLC grade water. Remove the final wash and count the gel slice by Cerenkov counting. 3. Add 0.5 ml of HPLC water and elute the peptide by rocking the gel slice for 8-16 h at room temperature. 4. Dry the supernatant in a SpeedVac and count the dried fraction by Cerenkov counting. At least 50% of the total gel slice counts should have been released in the supernatant.e a The following protocol and recipes are taken from West, Wu, and Bonner (43). b The recipe is for a 40% acrylamide gel. For larger peptides that do not migrate far into the 40% gel, a lower percentage resolving gel (20-30%) can be prepared. c Following autoradiography, individual phosphopeptide bands can be eluted from the gel for further analysis as described in Protocol 6B. d This wash removes the Tris and glycine from the gel slice. e The dried peptide can be used for gas phase sequencing (if present at high enough levels), manual Edman degradation (Section 4.3; Protocol 8), secondary protease digestion (Section 4.3; Protocol 9), or phosphoamino acid analysis (Section 4.2.4; Protocol 4).

Protocol 7. Two-dimensional phosphopeptide mapping Equipment and reagents • HTLE 7000 two-dimensional peptide gel apparatus (CBS Scientific) • Atomizer (General Glassblowing Inc.)

. pH 1.9 buffer: 2.5% (v/v) formic acid (88% (v/v) stock), 7.8% (v/v) glacial acetic acid in dH2O

151

B. G. Rowan and N. L. Weigel Protocol 7.

Continued

• Chromatography buffer: 37.5% (v/v) nbutanol, 25% (v/v) pyridine, 7.5% (v/v) glacial acetic acid in dH20 • 20 x 20 cm x 100 um thin-layer cellulose plates (EM Science)

• Chromatography tank with lid about 30 cm length x 5 cm width x 30 cm height or larger • Whatman 3MM paper

A. Two-dimensional phosphopeptide mappinga,b 1. Start with a pool of dried phosphopeptides (at least 1000 c.p.m.) in a 1.5 ml siliconized microcentrifuge tube. 2. Resuspend the phosphopeptides with 0.5 ml of HPLC grade water, vortex well, and dry the sample in a SpeedVac. Repeat this step once more.c 3. Add 200 ul of pH 1.9 buffer to the tube, vortex well, and centrifuge at 12 000 g for 15 sec. Transfer the supernatant to another tube and dry the supernatant in a SpeedVac. 4. Resuspend the dried sample in 10-15 ul of pH 1.9 buffer and vortex well. 5. Apply the sample as a single spot (approx. 1 mm diameter) on a 20 X 20 cm cellulose plate (Figure 3).d 6. Uniformly wet the plate and electrophorese the sample in the first dimension at 1000 V for 20-50 min.e,f 7. Dry the plate (approx. 1 h at room temperature) and place it vertically in a Chromatography tank containing Chromatography buffer approx. 1.5 cm deep. Place the plate in the tank with the phosphopeptides down (Figure 3). Allow the buffer to rise on the plate until the buffer front is 1-2 cm from the top of the plate (approx. 6-8 h).g 8. Dry the plate (approx. 1 h), mark its orientation, and then expose it to X-ray film for autoradiography.h' B. Extraction of phosphopeptides from cellulose plates 1. Following autoradiography, use the exposed film to mark the positions of individual phosphopeptides on the cellulose plate. 2. Scrape the cellulose off the plate with a small spatula. Place the cellulose particles in a 1.5 ml siliconized microcentrifuge tube and count by Cerenkov counting. 3. Add 100-200 ul of 50% acetonitrile, 0.1% TFA (in HPLC water) and vortex for 30 sec. Centrifuge the tube for 15 sec to pellet the cellulose. Transfer the supernatant to a fresh tube. 4. Repeat step 3, combine supernatants, and dry the combined volume in

152

6: Analysis of steroid/nuclear receptor phosphorylation a SpeedVac. Count the dried sample by Cerenkov counting. At least 70% of the total counts should be released in the supernatant. i a This procedure has been optimized to work with the HTLE 7000 two-dimensional peptide gel apparatus. For a more detailed description on this procedure see van der Geer and Hunter (41). b Perform this procedure in a fume-hood because of volatile organic buffers. c This step removes traces of ammonium bicarbonate from the tryptic digest. d The sample should be applied 1 ul at a time with drying in between. e To wet the plate, use an atomizer or two pieces of Whatman 3MM filter paper soaked in pH 1.9 buffer. Van der Geer and Hunter (41) give a detailed description of plate wetting with Whatman paper. The choice of electrophoresis buffer will depend on the properties of the phosphopeptides to be separated and will have to be determined empirically. In general, pH 1.9 buffer works well for the tryptic phosphopeptides of most proteins. There are also several other chromatography buffers to choose from (41). f At pH 1.9, most phosphopeptides are positively charged and will migrate toward the cathode (black, -). g The phosphopeptides on the cellulose plate should be about 1.5 cm above the level of the buffer. The inside walls of the tank should be lined with Whatman 3MM paper that is saturated with chromatography buffer. The tank should be prepared and kept with the lid on for 2 h prior to placing the cellulose plate inside. h 'Streaking of phosphopeptides in the first dimension or failure of the majority of phosphopeptides to leave the origin may indicate that the sample was overloaded, or that the pH of the sample was significantly different from pH 1.9. Phosphopeptides can be eluted from the cellulose plates following autoradiography as described in Protocol 76. i The dried peptide can be used for gas phase sequencing (if at high enough levels), manual Edman degradation (Section 4.3; Protocol 8), secondary protease digestion (Section 4.3; Protocol 9), or phosphoamino acid analysis (Section 4.2.4; Protocol 4).

4.3 Approach for use with low protein levels When there is not enough receptor protein for gas phase sequencing of phosphopeptides (sub-microgram range), an alternative approach can be used to identify phosphorylation sites in receptors. This procedure follows the steps outlined for the conventional method through the point where individual 32Plabelled phosphopeptides are separately isolated (Section 4.2.5). At this point, instead of gas phase sequencing, individual phosphopeptides are subjected to a modified manual Edman degradation (MED, Protocol 8). Briefly, the peptide is coupled to a membrane through carboxyl groups and subjected to an Edman degradation procedure. Each cycle of MED releases one amino acid, sequentially, from the N terminal of a phosphopeptide. Because the peptide itself is coupled to the disc, the released amino acid or phosphoamino acid can be extracted with acid rather than with an organic solvent as is usual for Edman degradations. This ensures high efficiency extraction of the phosphoamino acid or phosphate released. By [32P] Cerenkov counting of the released amino acid after each cycle, one can determine the position of the 32P-labelled amino acid relative to the N terminal of the peptide. MED can be performed manually, or automated machines can be modified to perform this procedure. The advantages to MED are its low cost and the ability to analyse multiple phosphopeptides in one experiment. MED has the added advantage over peptide sequencing in that phosphopeptides need only be purified from other 153

B. G. Rowan and N. L. Weigel

Figure 3. Two-dimensional phosphopeptide map of the hPRB phosphorylated in vitro with cyclinA/cdk2. hPRB and cyciinA/cdk2 were separately expressed using a baculovirus expression system. Purified hPRB and cyclinA/cdk2 were incubated together along with [y-32P]ATP, non-radioactive ATP, and 10 x cyelirtA/cdk2 kinase reaction buffer for 30 min at 30'C as described in Protocol 10. 32P-labelled hPRB was purified by SDS-PAGE and the gel slice containing the receptor was subjected to trypsin digestion for 16 h. An aliquot of the hPRB tryptic phosphopeptides containing about 10 ug of total protein (about 4 ug of receptor plus 6 ug of trypsin) was spotted onto a 20 x 20 cm cellulose plate as shown. The sample was electrophoresed in the first dimension at 1000 V for 45 min using the HTLE 7000 electrophoresis system, The plate was dried and then placed vertically in a chromatography tank with the side containing the phosphopeptides at the bottom of the tank. Chromatography in the second dimension proceeded for 7 h and then the plate was dried and exposed to film for autoradiography (2 h). Shown are the major hPRB tryptic phosphopeptides induced by cyclinA/cdk2. Note that a small amount of the sample precipitated at the site where it was spotted and was not resolved.

phosphopeptides. For peptide sequencing, phosphopeptides must be purified from all other peptides, both phosphorylated and non-phosphorylated. In addition to MED, phosphopeptides are also subjected to secondary digestion with sequence-specific proteases (Protocol 9) followed by electrophoresis on alkaline peptide gels to help identify specific residues within the phosphopeptide. The combined information is sufficient to identify the majority of phosphorylation sites in receptors. Purified phosphopeptides from HPLC fractions, alkaline peptide gels, or two-dimensional peptide maps can be used with these procedures. This approach has been used to identify 154

6: Analysis of steroid/nuclear receptor phosphorylation phosphorylation sites in the human PR (32, 44-46) and the human vitamin D receptor (38). Protocol 8.

Manual Edman degradation

Reagents • Sequelon™ AA Reagent Kit (Millipore) a

Method 1. Place a Sequelon-AA membrane on a Mylar sheet that is placed on top of a heating block set at 55°C. 2. Dissolve the phosphopeptide in 30 ul of 50% (v/v) acetonitrile, 0.1% (w/v) TFA in HPLC grade water and spot the solution on the membrane.b 3. After the membrane is dry (10-15 min) remove the membrane from the heating block. 4. Add 5 ul carbodiimide3 (10 mg/ml in coupling buffer,a prepared fresh) to the membrane and let stand at room temperature for 30 min.c 5. Place membrane in a 1.5 ml microcentrifuge tube and wash the disk with 1 ml of HPLC grade water five times (5 min each wash with rocking). 6. Add 0.5 ml of 100% (w/v) TFA to the disk, invert the tube several times, and then remove and discard the TFA. Repeat this four more times.d 7. Wash the disk three times with 1 ml methanol. Remove the last wash. 8. Add 0.5 ml PITC reagent (methanol: HPLC grade water: triethylamine: phenylisothiocyanate; 7:1:1:1) and place the tube in a heating block (55°C) for 10 min.e 9. Remove the reagent and wash disk five times with 1 ml methanol. 10. Dry the disk in a SpeedVac for 5 min. Add 0.5 ml TFA and place the tube in a heating block (55°C) for 6 min to cleave the derivatized amino acid. 11. Save the TFA wash and extract the disk with 1 ml of TFA:42.5% phosphoric acid (9:1, v/v). 12. Combine the washes from steps 10 and 11 and count by Cerenkov counting. Count the disk also. 13. Wash the disk five times with 1 ml methanol. 155

B. G. Rowan and N. L. Weigel Protocol 8.

Continued

14. Start the next cycle at step 8 or store at -20°C until ready to begin again. a The kit contains Sequelon-AA membranes, Mylar sheets, coupling buffer, and carbodiimide. "Because only about 50% of the total c.p.m. of the purified phosphopeptide will be coupled to the disk, there should be at least 500-600 c.p.m. of phosphopeptide to start with. Some phosphopeptides may not show a [32P] release for a variety of reasons including blocked N termini, or a position within the peptide beyond the limits of repetitive degradation (e.g. > 20 residues). c This step covalently couples the peptide to the membrane. d TFA extracts unbound peptides from the membrane. e This step couples phenylisothiocyanate to the N terminal amino acid sensitizing the peptide bond to cleavage with anhydrous TFA in step 10. The PITC reagent should be prepared fresh for each cycle. Phenylisothiocyanate is subject to oxidation and the bottle must be flushed with a stream of nitrogen upon opening and closing.

Protocol 9. Secondary digestion of phosphopeptides with Asp-N and Glu-Ca Reagents • Endoproteases Asp-N and Glu-C (Boehringer Mannheim)

Method 1. Resuspend the dried phosphopeptide in 50% acetonitrile, 0.1% TFA (in HPLC grade water). Divide the solution into three equal volumes (in 1.5 ml siliconized microcentrifuge tubes) and dry in a SpeedVac. 2. Resuspend one sample in 90 ul of 50 mM sodium phosphate buffer pH 8.0 and add to this 5-10 ul of 10 ug/ml Asp-N prepared in the same buffer. Incubate for 4 h at 37°C. Resuspend the second sample in 99 ul of 25 mM ammonium bicarbonate and add to this 1 ul of 0.5 ug/ul GluC prepared in the same buffer. Incubate for 8 h at 37°C. Resuspend the third sample in either sodium phosphate buffer or ammonium bicarbonate buffer and incubate for 4-8 h at 37°C (control sample). 3. Dry digests in a SpeedVac. 4. Electrophorese Asp-N, Glu-C, and control digests side by side on an alkaline peptide gel. Dry the gel and expose to X-ray film. a

Asp-N cleaves on the N terminal side of most aspartate residues, Glu-C cleaves on the C terminal side of most glutamate residues (47, 48). An altered migration of the Asp-N and/or Glu-C digested sample compared to the control sample would indicate that the phosphopeptide contains an aspartate and/or a glutamate residue, respectively. See Figure 4 for schematic representation.

156

6: Analysis of steroid/nuclear receptor phosphorylation

Figure 4. Diagram of a hypothetical tryptic phosphopeptide subjected to secondary protease digestion with Asp-N and Glu-C. The hypothetical tryptic peptide contains an aspartate residue (D), a glutamate residue (E), a phosphorylated serine residue (S), and either an arginine or lysine residue at the C terminal (R/K). (A) Digestion with either Asp-N or Glu-C (Protocol 9) will result in two phosphopeptides with different characteristics. (B) Hypothetical diagram of 32P-labelled phosphopeptide migration on an alkaline peptide gel (Protocol 6). Alkaline peptide gel electrophoresis separates phosphopeptides based on their charge/mass ratio. In this example, the phosphopeptide digested with Asp-N migrates further on the gel compared to the undigested phosphopeptide presumably because it is smaller (and assuming that the three amino acids lost from the N terminal of the peptide do not have a significant effect on the overall charge of the peptide). In contrast, the phosphopeptide digested with Glu-C migrates less distance on the gel than the undigested phosphopeptide. Even though the Glu-C digested phosphopeptide is smaller than the undigested peptide, the loss of two negative charges (from the D and the E residues) may overcome the effect of the smaller size and impede the migration of this peptide in the gel. It should be noted that these figures are based on generalizations. The inherent charge of every amino acid residue in a phosphopeptide will contribute to the overall migration pattern on alkaline peptide gels.

4.4 Additional experiments to provide more sequence information Phosphopeptide maps of receptor protein digested with either trypsin (cuts at both arginine and lysine residues), Arg-C (cuts only at arginine residues), or Lys-C (cuts only at lysine residues) can be compared. This will indicate whether a peptide is bounded by only arginine residues, only lysine residues, or both arginine and lysine residues. A phosphorylation site identified by the methodologies described in Section 4.3 can be confirmed by mutating the site to alanine 157

B. G. Rowan and N. L. Weigel and then comparing the phosphopeptide map of the point-mutated receptor with the phosphopeptide map of the wild-type receptor. Absence of the [32P]phosphopeptide containing this site in the point-mutant receptor would confirm the identification of the site.

4.5 Indirect method The indirect method relies on identification by loss of a phosphopeptide through mutation of the candidate amino acid to alanine, a non-phosphorylatable amino acid. In some cases, the site is first localized to a region of the receptor through expression and analysis of receptor domains. Retention of a phosphopeptide places the peptide within the expressed domain. Loss of a phosphopeptide should be interpreted with caution since removal of a functional domain, such as the DNA binding domain, may cause mislocalization. The [32P]H3PO4 labelling, receptor purification, and SDS-PAGE procedures are the same as that described in Section 4.2 with the exception that much less purified receptor protein and [32P]H3PO4 is usually needed. There are several methods for site-directed mutagenesis and creation of deletion mutants (49). The indirect method has been used to localize phosphorylation sites in the human ER (20, 22) and the human androgen receptor (29).

5. Identifying candidate kinases 5.1 Introduction Several pieces of information are important when trying to identify candidate kinases for receptors. Most importantly, identification of authentic phosphorylation sites in vivo will reveal consensus sequences for various kinases. Other indirect evidence includes assessment of receptor phosphorylation after treatment of cells with inhibitors and activators of known kinases and/or phosphatases. Various growth factors that are known to regulate receptor function and that also selectively activate specific kinases and/or phosphatases can also be used in this type of experiment. In the absence of information about the identity of the in vivo phosphorylation sites, the sequence of the protein can be scanned for candidate consensus sites. However, the number of kinases and uncertainty about sequence requirements makes this a very complicated problem. For example, human PR-B contains about a dozen casein kinase II consensus sequences. However, only one of these is actually phosphorylated in vivo (44). This site is also the only site phosphorylated by casein kinase II in vitro. Taken together, this information will provide a starting point from which to investigate the role of specific kinases and phosphatases in receptor phosphorylation. Many authentic phosphorylation sites are found in sequences that are consensus sequences for several kinases. Indeed, a number of sites in tran158

6: Analysis of steroid/nuclear receptor phosphorylation scription factors are phosphorylated both in vivo and in vitro by multiple kinases. Because of the complex interactions between various signalling pathways, a combination of in vivo and in vitro studies will usually be most fruitful. The problems that occur with in vitro phosphorylation of receptors using purified kinases are phosphorylation at non-physiological sites and lack of phosphorylation at authentic, identified in vivo sites. Phosphorylation of receptors at non-physiological sites is a property of the purified kinase and protein substrate that are free from the constraints of a cellular environment. Typically, sites that can be phosphorylated with good stoichiometry in vitro are the most likely in vivo candidates. To determine the authenticity of phosphorylation sites identified by in vitro phosphorylation, it is necessary to compare the phosphopeptide maps of in vitro and in vivo phosphorylated receptor. The failure of a kinase to phosphorylate in vitro authentic, consensus phosphorylation site(s) that were identified in vivo suggests that other kinases are normally involved in the phosphorylation. However, some kinases will phosphorylate a consensus site only after an adjacent phosphorylation event has occurred (38). Another procedure to identify candidate kinases, the use of kinase expression vectors, will determine whether the kinase can affect the overall level and the site-specific phosphorylation of receptors in cells when it is co-expressed with receptor. Because of endogenous kinase, this procedure is usually not sufficient by itself for identification of candidate kinases and should be used in conjunction with in vitro phosphorylation. Both of these procedures have been used to identify candidate kinases in nuclear receptors (2, 23, 32, 50-53).

5.2 In vitro phosphorylation of nuclear receptors In vitro phosphorylation of receptors differs from in vivo phosphorylation in several ways. Fewer phosphorylation sites will be detected by in vitro phosphorylation because only one kinase is being used as compared to multiple kinases that are available to phosphorylate the receptor in vivo. In vitro phosphorylation is easier to perform because the starting material is usually purified protein and [y-32P]ATP can be used directly. Also, more [32P] is incorporated into the receptor during in vitro phosphorylation due to higher specific activities making site identification easier. Identification of phosphorylation sites using in vitro phosphorylated receptor can be performed using the procedures described in Sections 4.2.3 to 4.3. An example of an in vitro phosphorylation experiment is shown in Figure 5. Many purified kinases can now be obtained commercially. Alternatively, kinase can be purified from cell or tissue extracts for use in in vitro phosphorylation. The source of receptor can be from bacterial or baculoviral expression systems, in vitro translation, or purification from cell or tissue extracts. The most common method for purifying kinase and receptor from cells or tissue minces is immunoprecipitation because of its ease and speed. 159

B. G, Rowan and N. L. Weigel Protocol 10. In vitro phosphorylation of nuclear receptors with purified kinasea Reagents • [y-32P]ATP, 10 mCi/ml (DuPont NEN)

Method 1. Prepare 40 mM ATP and a 10 X kinase reaction buffer ahead of time. Store in aliquots at -70°C. 2. Dilute 40 mM ATP to 0.4 mM with dH2O. 3. Mix 1-10 ul of 0.4 mM ATP with 1-3 ul of 10 mCi/ml [y-32P]ATP. 4. In a separate tube, mix 4 ul of 10 X kinase reaction buffer, 0.1-10 pmol purified receptor, purified kinase,b and add dH2O to make a total volume of 32 ul. 5. Mix the solution from step 3 with the solution from step 4 and incubate at 30°C for 1 h (the time and temperature will vary for different kinases). 6. Stop the reaction by adding SDS-PAGE sample buffer and heating at 100°C for 5 min. 7. Follow Protocol 3, steps 1-6 to isolate the phosphorylated receptor band. Record the Cerenkov counts in the band. Use these counts to determine the specific activity of [y- 32 P]ATP incorporated into the receptor (see Protocol 11). 8. The phosphorylated receptor band can be used for phosphorylation analysis as described in Sections 4.2.3 to 4.4. a This protocol is for one 40 ul reaction; the amount of ATP, [y-32P]ATP, and the composition of the 10 x kinase buffer will depend on the kinase. The choice of final specific activity of the ATP in the reaction will depend upon the amount of receptor used. For example, if 1 pmol receptor is used, then a specific activity of 3000 c.p.m./pmol ATP would allow the investigator to detect 10% phosphorylation of one site or 300 c.p.m. The control samples to be included are: (-) kinase (+) receptor; and (+) kinase (-) receptor. It should be noted that some kinases will autophosphorylate resulting in a kinase band on the autoradiograph. In vitro phosphorylation can be performed directly on protein A-Sepharose beads that are used to immunopurify either receptor or kinase. However, it is not possible to perform this experiment by mixing beads from two immunoprecipitations (i.e. one for receptor and one for kinase). b The amount of kinase should be titrated so that the level of kinase used gives maximal receptor phosphorylation. Below this level of kinase, there should be a roughly linear relationship between the level of kinase used and the specific activity of [y-32P]ATP incorporated into the receptor (see Section 5.3).

160

6: Analysis of of steroid/nuclear recepto

Figure 5. In vitro phosphorylation of hPRB with cyclinA/cdk2. Different combinations of purified hPRB, histone H1, and purified cyclinA/cdk2 were incubated together along with [y-32P]ATP, non-racfioactive ATP, and 10 x cyclinA/cdk2 kinase reaction buffer for 30 min at 30°C as described in Protocol 10. Lane 1 was a mixture of the two substrates without cyclinA/cdk2 kinase to determine whether there was any contaminating kinase activity in either of the substrate preparations. Lane 2 is the cyclinA/cdk2 kinase alone and shows phosphorytation of cyclin A by cdk2. Lanes 3-5 are mixtures of hPR B with increasing levels of cyclinA/cdk2, Note that phosphorylation of hPR B increased with increasing levels of kinase. Lanes 6 and 7 were used as a control to determine whether there were any contaminating phosphatases or kinase inhibitors in the hPRB preparation. In lane 6 both histone H1 and hPRB were added together whereas in lane 7 only histone H1 was added (histone H1 is commercially available and free of any contaminating phosphatases or kinase inhibitors). The level of histone H1 phosphorylation by cyclinA/cdk2 in lanes 6 and 7 was roughly equivalent indicating the absence of any contaminating phosphatases or kinase inhibitors in the hPRe preparation. Protein molecular weight standards are indicated on the left side of the figure in kDa.

Protocol 11,

Calculations for in vitro phosphorylationa

1. Calculate the specific activity of [y- 32P]ATP: (a) Moles of ATP: 5 ul of 0.4 mM ATP = 2000 pmol ATP.b (b) Total d.p.m.: 3 (J of 10 uCl/ul [y-32P]ATP = 30 uCi • 2.2 x 106 d.p.m./uCi c - 6.6 x 107 d.p.m. (c) Specific activity of [y- 32 P]ATP: 6.6 x 107 d.p.m./2000 pmol = 33000 d.p.m./pmol ATP. 2. Estimate the pmol of receptor. The mass of purified receptor can be estimated from a silver stained gel by comparing the intensity of the

161

B. G. Rowan and N. L. Weigel Protocol 11.

Continued

receptor band to the intensity of a known mass of standard protein. From this, the pmol of receptor can be determined. For this example assume 1.5 pmol receptor. 3. Specific activity incorporated in receptor. For this example, assume that Cerenkov counting of the gel slice gave 15 000 c.p.m. Multiply this by a correction factor to account for the efficiency of the counter. For this example, we will use a correction factor of 2: Gel slice: 15000 c.p.m. x 2 = 30000 d.p.m./1.5 pmol receptor = 20000 d.p.m./pmol receptor. 4. Per cent incorporation of [32P]PO4 in the receptor (for this example, we will assume that only one site in the receptor is phosphorylated by the kinase): % incorporation = specific activity incorporated in receptor/specific activity of ATP 20000 d.p.m./pmol receptor = x 33000 d.p.m./pmol ATP

1 0 0 = 6 1 % o r 0.61 mole/moled

a As an example, calculations are performed for the example shown in Protocol 10 with 1.5 pmole of receptor. b Because the molar amount of [y-32P]ATP is far less than the molar amount of unlabelled ATP, the [y- 32 P]ATP does not significantly contribute to the total moles of ATP. C 2.2 x 106 d.p.m./uCi is a constant. d This means that 0.61 moles of ATP is incorporated into every 1.0 moles of receptor. For receptors phosphorylated at multiple sites, one should first determine the total number of phosphorylation sites in the receptor (e.g. by 2D phosphopeptide mapping) and then calculate the proportion of the [32P] radioactivity that each site contributes to the total [32P] radioacitvity. Then, per cent incorporation should be performed for each phosphorylation site. Low per cent incorporation (a few per cent) may be due to technical problems such as too little kinase, endogenous receptor phosphorylation, contaminating kinase inhibitors, or contaminating phosphatases. If these possibilities are eliminated and the stoichiometry is still low, (a few per cent) it is unlikely that the receptor is a substrate for the kinase.

6. Assaying the functional significance of selected phosphorylation sites 6.1 Introduction Phosphorylation of nuclear receptors has been shown to play a role in several receptor functions (see Section 2). To determine the functional significance of a newly identified phosphorylation site, it is important to first establish in vitro and in vivo functional assays for receptors that are testable (e.g. transactivation, nuclear translocation, DNA binding, ligand binding, interaction with other proteins, turnover, etc.). These assays can then be tested with wildtype receptor and receptor that is point mutated at specific phosphorylation 162

6: Analysis of steroid/nuclear receptor phosphorylation site(s) to determine if phosphorylation plays some role in these receptor functions. Another approach used for some in vitro-based assays, is the use of general and specific phosphatases and kinases to alter receptor phosphorylation (54). The consequence of the altered receptor phosphorylation to receptor function can then be assayed (1, 4, 5, 8, 14). One difficulty with this approach is that the phosphatase will likely dephosphorylate all sites complicating the interpretation of the results. Still another approach is to co-express kinase or phosphatase expression vectors with receptor in cells and characterize the effect of these constructs on receptor function(s).

6.2 Site-directed mutagenesis of nuclear receptors A serine or threonine phosphorylation site should be mutated to an alanine residue and to a glutamic acid residue to evaluate the importance of the site to receptor function. Alanine is a small, uncharged, non-phosphorylatable residue that is less likely to result in disruption of protein secondary structure when compared to other residues. Hence, any alterations of receptor function with an alanine point mutation could be related, predominantly, to the loss of phosphorylation at this site. In addition to the alanine point mutation, a glutamic acid point mutation should also be made at the same phosphorylation site. Glutamic acid is a negatively charged residue at physiological pH and might be considered to partially mimic the doubly or triply negatively charged phosphate group at this site. Tyrosine phosphorylation sites are typically mutated to phenylalanine. These constructs can then be separately expressed in cell lines or in vitro and the activity of the constructs (in the functional assay chosen) can be compared to wild-type receptor. These experiments should be accompanied by an examination of receptor levels (by immunoblotting) to determine whether the mutations have an effect on protein levels. Similarly, if assaying transactivation, an examination of receptor ligand binding, DNA binding, and nuclear translocation should also be performed. Mutagenesis of receptor phosphorylation sites can be performed directly in receptor expression vectors using commercially available mutagenesis kits or by polymerase chain reaction technology (49). Both procedures require the purchase of custom oligonucleotides for each site to be mutated.

References 1. Castoria, G., Migliaccio, A., Green, S., Di Domenico, M., Chambon, P., and Auricchio, F. (1993). Biochemistry, 32, 1740. 2. Arnold, S. F., Obourn, J. D., Jaffe, H., and Notides, A. C. (1995). Mol. Endocrinol., 9, 24. 3. Rochette-Egly, C., Gaub, M. P., Lutz, Y., Ali, S., Scheuer, I., and Chambon, P. (1992). Mol. Endocrinol., 6, 2197. 4. Dayani, N., McNaught, R. W., Shenolikar, S., and Smith, R. G. (1990). Biochemistry, 29, 2691. 163

B. G. Rowan and N. L. Weigel 5. Raymoure, W. J., McNaught, R. W., and Smith, R. G. (1985). Nature, 314, 745. 6. Denner, L. A., Schrader, W. T., O'Malley, B. W., and Weigel, N. L. (1990). J. Biol. Chem., 265, 16548. 7. Beck, C. A., Weigel, N. L., and Edwards, D. P. (1992). Mol. Endocrinol., 6, 607. 8. Denton, R. R., Koszewski, N. J., and Notides, A. C. (1992). J. Biol. Chem., 267, 7263. 9. Lin, K. H., Ashizawa, K., and Cheng, S. Y. (1992). Proc. Natl. Acad. Sci. USA, 89, 7737. 10. Sugawara, A., Yen, P. M., Apriletti, J. W., Ribeiro, R. C., Sacks, D. B., Baxter, J. D., et al. (1994). J. Biol. Chem., 269, 433. 11. Lefebvre, P., Gaub, M. P., Tahayato, A., Rochette-Egly, C., and Formstecher, P. (1995). J. Biol. Chem., 270, 10806. 12. Bailly, A., Le Page, C., Rauch, M., and Milgrom, E. (1986). EMBO J., 5, 3235. 13. Hsieh, J. C., Jurutka, P. W., Galligan, M. A., Terpening, C. M., Haussler, C. A., Samuels, D. S., et al. (1991). Proc. Natl. Acad. Sci. USA, 88, 9315. 14. Katz, D., Reginato, M. J., and Lazar, M. A. (1995). Mol. Cell. Biol., 15, 2341. 15. Hirata, Y., Kiuchi, K., Chen, H.-C, Milbrandt, J., and Guroff, G. (1993). /. Biol. Chem., 268, 24808. 16. Davis, I. J. and Lau, L. F. (1994). Mol. Cell. Biol., 14, 3469. 17. Bai, W., Tullos, S., and Weigel, N. L. (1994). Mol. Endocrinol., 8, 1465. 18. Bai, W., Rowan, B. G., Allgood, V. E., O'Malley, B. W., and Weigel, N. L. (1997). J. Biol Chem., 272, 10457. 19. Bai, W. and Weigel, N. L. (1996). J. Biol Chem., 271, 12801. 20. Ali, S., Metzger, D., Bornert, J. M., and Chambon, P. (1993). EMBO J., 12, 1153. 21. Trowbridge, J. M., Rogatsky, I., and Garabedian, M. J. (1997). Proc. Natl. Acad. Sci. USA, 94, 10132. 22. Le Goff, P., Montano, M. M., Schodin, D. J., and Katzenellenbogen, B. S. (1994). J. Biol. Chem., 269, 4458. 23. Kato, S., Endoh, H., Masuhiro, Y., Kitamoto, T., Uchiyama, S., Sasaki, H., et al. (1995). Science, 270, 1491. 24. Bunone, G., Briand, P. A., Miksicek, R. J., and Picard, D. (1996). EMBO J., 15, 2174. 25. Lahooti, H., White, R., Hoare, S. A., Rahman, D., Pappin, D. J., and Parker, M. G. (1995). J. Steroid Biochem. Mol. Biol., 55, 305. 26. Krstic, M. D., Rogatsky, I., Yamamoto, K. R., and Garabedian, M. J. (1997). Mol. Cell. Biol., 17, 3947. 21. Rogatsky, L, Waase, C. L. M., and Garabedian, M. J. (1998). J. Biol. Chem., 273, 14315. 28. Rogatsky, L, Logan, S. K., and Garabedian, M. J. (1998). Proc. Natl. Acad. Sci. USA, 95, 2050. 29. Zhou, Z. X., Kemppainen, J. A., and Wilson, E. M. (1995). Mol. Endocrinol., 9, 605. 30. Webster, J. C., Jewell, C. M., Bodwell, J. E., Munck, A., Sar, M., and Cidlowski, J. A. (1997). J. Biol. Chem., 272, 9287. 31. Poletti, A. and Weigel, N. L. (1993). Mol. Endocrinol., 7, 241. 32. Zhang, Y., Beck, C. A., Poletti, A., Clement, J. P., Prendergast, P., Yip, T. T., et al. (1997). Mol. Endocrinol, 11, 823. 33. Hsu, S. C., Qi, M., and DeFranco, D. B. (1992). EMBO J., 11, 3457. 164

6: Analysis of steroid/nuclear receptor phosphorylation 34. Goldberg, Y., Glineur, C., Gesquiere, J. C, Ricaurt, A., Sap, J., Vennstrom, B., et al. (1988). EMBO J., 7, 2425. 35. Glineur, C., Bailly, A., and Ghysdael, J. (1989). Oncogene, 4, 1247. 36. Mendel, D. B., Bodwell, J. E., and Munck, A. (1987). J. Biol. Chem., 262, 5644. 37. Roach, P. J. and Wang, Y. H. (1991). In Methods in enzymology (ed. T. Hunter, and B. M. Sefton), Vol. 201, p. 200. Academic Press, London. 38. Milliard, G. M., Cook, R. G., Weigel, N. L., and Pike, J. W. (1994). Biochemistry, 33, 4300. 39. Bodwell, J. E., Orti, E., Coull, J. M., Pappin, D. J., Smith, L. I., and Swift, F. (1991). J. Biol. Chem., 266, 7549. 40. Arnold, S. F., Obourn, J. D., Jaffe, H., and Notides, A. C. (1994). Mol. Endocrinol, 8, 1208. 41. van der Geer, P. and Hunter, T. (1994). Electrophoresis, 15, 544. 42. Schagger, H. and Jagow, G. (1987). Anal. Biochem., 166, 368. 43. West, M. H. P., Wu, R. S., and Bonner, W. M. (1984). Electrophoresis, 5, 133. 44. Zhang, Y., Beck, C. A., Poletti, A., Edwards, D. P., and Weigel, N. L. (1994). J. Biol. Chem., 269, 31034. 45. Zhang, Y., Beck, C. A., Poletti, A., Edwards, D. P., and Weigel, N. L. (1995). Mol. Endocrinol., 9, 1029. 46. Beck, C. A., Zhang, Y., Altmann, M., Weigel, N. L., and Edwards, D. P. (1996). J. Biol. Chem., 271, 19546. 47. Drapeau, G. R. (1980). J. Biol. Chem., 255, 839. 48. Sorensen, S. B., Sorensen, T. L., and Breddam, K. (1991). FEBS Lett., 294, 195. 49. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. A., Smith, J. A., and Struhl, K. (ed.) (1994). Current protocols in molecular biology, Vol 1, Chapter 8. John Wiley & Sons, Inc. 50. Arnold, S. F., Obourn, J. D., Yudt, M. R., Carter, T. H., and Notides, A. C. (1995). J. Steroid Biochem. Mol. Biol., 52, 159. 51. Jurutka, P. W., Hsieh, J. C, Nakajima, S., Haussler, C. A., Whitfield, G. K., and Haussler, M. R. (1996). Proc. Natl. Acad. Sci. USA, 93, 3519. 52. Rochette-Egly, C., Oulad-Abdelghani, M., Staub, A., Pfister, V., Scheuer, I., Chambon, P., et al. (1995). Mol. Endocrinol., 9, 860. 53. Davis, I. J., Hazel, T. G., Chen, R. H., Blenis, J., and Lau, L. F. (1993). Mol. Endocrinol., 7, 953. 54. Shenolikar, S. (1997). In Current protocols in protein science (ed. J. E. Coligan, B. M. Dunn, L. Hidde, H. L. Ploegh, D. W. Speicher, and P. T. Wingfield). John Wiley & Sons, Inc.

165

This page intentionally left blank

7

Ligand- and cofactor-regulated transcription with chromatin templates W. L. KRAUS and J. T. KADONAGA

1. Introduction Nuclear hormone receptors function as DNA binding, ligand-, and cofactorregulated transcription factors in the chromatin environment of the cell nucleus (1). Experiments aimed at understanding the detailed mechanisms of transcriptional activation by nuclear hormone receptors should use a biochemical approach and should include both the receptor and non-receptor (i.e. co-activators, co-repressors, chromatin, ligands) components of the nuclear hormone receptor signalling pathway. This chapter describes biochemical approaches for the analysis of ligand- and cofactor-regulated transcription by nuclear hormone receptors. Methods are presented for the synthesis and purification of nuclear hormone receptors and co-activators (Protocols 1 and 2), assembly of chromatin templates in vitro (Protocols 3 and 4), and ligand- and co-activator-regulated transcription of the assembled chromatin templates in vitro (Protocols 5-7).

2. Synthesis and purification of recombinant nuclear hormone receptors and cofactors 2.1 Choice of expression systems Biochemical studies of nuclear hormone receptors and their associated cofactors require purified receptor and cofactor proteins. Although these proteins can be purified from native sources, the yields of protein and ease of purification, as well as the possibility of examining mutant versions of the proteins, are usually much better with recombinant expression systems. These expression systems include bacteria (e.g. Escherichia coli), yeast (e.g. Saccharomyces cerevisiae or Pichia pastoris), baculovirus-infected cultured insect cells (e.g. Spodoptera frugiperda [Sf9] cells), and cultured mammalian cells

W. L. Kraus and J. T. Kadonaga (e.g. COS cells or HeLa cells). When choosing an expression system, one must consider the amount of protein required, the intended use of the expressed protein, and the solubility of the protein in a particular expression system. For example, with some nuclear hormone receptors, the level of expression in bacteria may be considerable, but the transcriptional activity of the recombinant protein may be low due to improper folding or lack of appropriate post-translational modifications. For other receptors synthesized in bacteria, the solubility or yield of the full-length protein may be problematic. With mammalian cell expression systems, the recombinant protein may closely resemble the native protein with regard to post-translational modifications, but the yield may be low. We have found that baculovirus-infected cultured insect cells work well for synthesizing full-length receptors and cofactors, providing a good balance between yield and activity. However, bacterial systems can be very useful for synthesizing isolated receptor and cofactor domains.

2.2 Purification of recombinant receptors and cofactors 2.2.1 Conventional versus affinity chromatography The initial biochemical studies of nuclear hormone receptors, especially the steroid hormone receptors, used conventional protein chromatography techniques to purify the receptors from native sources. Most conventional purification protocols involve the incubation of the receptors with their ligands prior to purification to increase the stability of the receptors through multiple chromatographic steps, but this approach limits subsequent analysis of the ligand-dependent properties of the receptors. The cloning of nuclear hormone receptor cDNAs has allowed the addition of affinity tags to recombinant expressed receptor proteins, which facilitates their purification. Lengthy, multiple step purification protocols that can reduce the integrity or activity of the purified proteins can now be replaced by rapid, single step affinity chromatography protocols that yield ligand-free receptors. With the oestrogen receptor (ER), we have found that a single step purification protocol for the receptor expressed in baculovirus-infected Sf9 cells yields highly purified (> 90% purity) and highly active (ligand binding, DNA binding, and transcriptional activities) protein (2) (Figure 1A). The same principles and methodology can be applied to nuclear hormone receptor cofactors (2) (Figure 1B). 2.2.2 Affinity tags A number of affinity tags, which consist of amino acid residues added to the protein, are now available, and all can readily be engineered into the receptor cDNAs by simple PCR and cloning techniques. The affinity tags include epitope tags (e.g. FLAG, haemagglutinin, and c-myc epitopes recognized by specific commercially available antibodies), ion chelate affinity tags (e.g. polyhistidine tracts that bind nickel-agarose), and substrate affinity tags (e.g. 168

7: Ligand- and cofactor-regulated in vitro transcription glutathione-S-transferase, which binds glulathione-agarose). The choice of affinity tags depends on the intended use of the protein. In general, using smaller affinity tags increases the chance of preserving the biochemical activities found in the native protein. Single or tandemly repeated tags can be placed either at the amino or carboxyl terminus of the protein, or a different tag can be placed at each end. It is useful to note that carboxyl terminal tags will not be present in truncated products resulting from incomplete mRNA synthesis and/or translation. With large proteins (~ 100 kDa or larger), we have found that the use of carboxyl terminal tags reduces the presence of truncated polypeptides in the purified protein preparation,

2.2.3 Receptor and cofactor purification In Protocol 1, we describe methods for the purification of an epitope-tagged nuclear hormone receptor (i.e. FLAG-tagged ER) produced in baculovirusinfected Sf9 cells. In Protocol 2, we describe methods for the purification of a polyhistidine-tagged receptor cofactor (i.e. His6-tagged p300 co-activator) also produced in baculovirus-infected Sf9 cells (see Figure 1). The protocols

Figure 1. Purification of recombinant ER and p300 synthesized in Sf9 cells. (A) FLAGtagged human ER was synthesized in Sf9 cells by using a baculovirus vector and purified by anti-FLAG immunoaffinity chromatography as described in Protocol 1. (B) His E -tagged human p300 was synthesized in Sf9 cells by using a baculovirus vector and purified by Ni-NTA affinity chromatography as described in Protocol 2. The purified proteins were subjected to 8% or 6% (w/v) polyacrylamide gel electrophoresis, respectively, and were stained with Coomassie Blue. From ref. 2. Used with permission.

169

W. L. Kraus and J. T. Kadonaga follow the general outline of infection of the cells, preparation of a whole cell extract under high salt conditions, incubation of the extract with the affinity resin, washing the resin, and elution of the purified protein. These methods should be directly applicable to the expression and purification of other nuclear hormone receptors, co-activators, and co-repressors containing epitope tags and/or polyhistidine tags. Protocol 1. Purification of recombinant oestrogen receptor synthesized in Sf9 cells by immunoaffinity chromatography Reagents • TMN-FH medium: Grace's insect cell culture medium, pH 6.2 with KOH, containing 10% (v/v) fetal bovine serum, 3.3 mg/ml lactalbumin hydrolysate, 3.3 mg/ml yeastolate, 10 ug/ml penicillin, 10 ug/ml streptomycin (all reagents from Gibco/BRL) • Recombinant baculovirus for expressing epitope (FLAG)-tagged ER (or other FLAGtagged factor) . Phosphate-buffered saline (PBS) • Homogenization buffer: 20 mM Tris-HCI pH 7.5, containing 500 mM NaCI, 1.5 mM MgCI2, 0.2 mM EDTA, 20% (v/v) glycerol, 2 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), 20 ug/ml leupeptin, 20 ug/ml aprotinin

. Dilution buffer: 20 mM Tris-HCI pH 7.5, containing 1.5 mM MgCI2, 0.2 mM EDTA, 10% (v/v) glycerol, 0.5% (v/v) NP-40, 2 mM DTT, 1 mM PMSF, 20 ug/ml leupeptin, 20 pig/ml aprotinin . Wash buffer: 20 mM Tris-HCI pH 7.5, containing 150 mM NaCI, 1.5 mM MgCI2, 0.2 mM EDTA, 10% (v/v) glycerol, 0.2% (v/v) NP-40, 2 mM DTT, 1 mM PMSF . Elution buffer: 20 mM Tris-HCI pH 7.5, containing 100 mM NaCI, 0.2 mM EDTA, 15% (v/v) glycerol, 0.1% (v/v) NP-40, 0.2 mg/ml FLAG peptide (Kodak/IBI), 0.5 mg/ml purified recombinant insulin (Boehringer Mannheim), 2 mM DTT, 1 mM PMSF • Anti-FLAG M2 affinity resin (Kodak/IBI)

Methoda 1. Dilute a log phase suspension culture of Sf9 cells with TMN-FH medium to a final volume of 500 ml and a density of 1 x 106 cells/ml. 2. Infect the culture with freshly amplified recombinant FLAG-ER baculovirus at a multiplicity of infection (m.o.i.) of 5-10 and incubate the infected culture for three days at 26°C. 3. Collect the cells by centrifugation at 1000 g (e.g. 2500 r.p.m. in a GSA rotor) for 5 min at 4°C, wash the cell pellet with 50 ml of ice-cold PBS, and centrifuge again.b 4. Decant the PBS and resuspend the cell pellet in 6.5 ml of homogenization buffer. 5. Homogenize the cells with a Wheaton Dounce homogenizer (12 strokes with the A pestle in a 7 ml vessel) and then incubate on ice for 15 min. 6. Centrifuge the resulting whole cell lysate at 190 000 g(e.g. 46000 r.p.m. in a Ti70.1 rotor) for 30 min. 7. Collect the supernatant into a 15 ml conical tube and dilute with an equal volume of dilution buffer.

170

7: Ligand- and cofactor-regulated in vitro transcription 8. Add 250 ul of a 50% (v/v) slurry of anti-FLAG M2 affinity resin previously equilibrated in dilution buffer. 9. Incubate the lysate with the resin for 4 h at 4°C with gentle mixing. 10. Pellet the resin by centrifugation at 1000 g in a refrigerated bench-top centrifuge, and remove the supernatant by aspiration. 11. Wash the resin in batch four times with 10 ml of wash buffer per wash, pelleting the resin by centrifugation at 1000 gin a refrigerated bench-top centrifuge after each wash. 12. After the final wash, transfer the resin in 1 ml of wash buffer to a 1.5 ml microcentrifuge tube. Pellet the resin by centrifugation for 20 sec in a microcentrifuge, and then aspirate as much of the wash buffer as possible from the resin. 13. Elute the purified receptor in batch by incubating the resin in 150 ul of elution buffer for 10 min. Pellet the resin by centrifugation for 20 sec and remove the purified receptor supernatant to a new tube. Repeat the elution one or two more times. 14. Freeze the purified receptor in aliquots in liquid N2 and store at -80°C.C 15. Check the receptor preparation for purity and yield by SDS-PAGE versus bovine serum albumin standards. d Check the ligand binding and DNA binding activities of the receptor preparation by conventional radiolabelled ligand binding assays and gel mobility shift assays. a

All volumes listed in this protocol are based on an initial starting culture volume of 500 ml. b All procedures from this point on should be performed on ice or at 4°C with ice-cold buffers. c The receptor preparation should be stable for up to six months when stored at -80°C with limited thawing and refreezing. d Typical yields are 5-10 ug of purified receptor per 500 ml culture.

Protocol 2. Purification of recombinant p300 synthesized in Sf9 cells by nickel-NTA affinity chromatography Reagents • TMN-FH medium (see Protocol 1) • Recombinant baculovirus for expressing His6-tagged p300 (or other His6-tagged factor) . PBS • Homogenization buffer: 10 mM Tris-HCI pH 7.5, containing 500 mM NaCI, 10% (v/v) glycerol, 0.1% (v/v) NP-40, 15 mM imidazole, 2 mM 2-mercaptoethanol, 2 mM PMSF, 20 ug/ml leupeptin, 20 ug/ml aprotinin

. Wash buffer: 10 mM Tris-HCI pH 7.5, containing 200 mM NaCI, 10% (v/v) glycerol, 0.2% (v/v) NP-40, 15 mM imidazole, 2 mM 2-mercaptoethanol, 2 mM PMSF . Elution buffer: 10 mM Tris-HCI pH 7.5, containing 100 mM NaCI, 10% (v/v) glycerol, 0.1% (v/v) NP-40, 250 mM imidazole, 2 mM 2-mercaptoethanol, 2 mM PMSF • Nickel-NTA affinity resin (Qiagen)

171

W. L. Kraus and J. T. Kadonaga Protocol 2.

Continued

a

Method 1. Plate 2 X 107 Sf9 cells from a log phase culture in a 15 cm diameter plate in TMN-FH medium. After the cells have attached firmly, replace the medium with 25 ml of fresh medium. 2. Infect the culture with freshly amplified recombinant p300His6 baculovirus at a m.o.i. of 5-10 and incubate the infected culture for three days at 26°C. 3. Remove the cells from the plate by pipetting medium over the loosely attached monolayer. Transfer the cell suspension to a 50 ml conical tube and collect the cells by centrifugation at 1000 g in a bench-top centrifuge for 5 min at 4°C.b 4. Resuspend the cell pellet in 1 ml of PBS and transfer the cells to a 1.5 ml microcentrifuge tube. Pellet the cells by centrifugation for 20 sec in a microcentrifuge, and then remove the PBS by aspiration. 5. Resuspend the cell pellet in 1 ml of homogenization buffer. Homogenize the cells with a Wheaton Dounce homogenizer (ten strokes with the A pestle in a 2 ml vessel) and then incubate on ice for 15 min. 6. Centrifuge the resulting whole cell lysate in a microcentrifuge for 10 min. 7. Collect the supernatant into a new 1.5 ml microcentrifuge tube and add 30 ul of a 50% (v/v) slurry of nickel-NTA affinity resin previously equilibrated in homogenization buffer. 8. Incubate the lysate with the resin for 2 h at 4°C with gentle mixing. 9. Pellet the resin by centrifugation in a microcentrifuge for 20 sec, and remove the supernatant by aspiration. 10. Wash the resin in batch four times with 1 ml of wash buffer per wash, pelleting the resin by centrifugation in a microcentrifuge for 20 sec after each wash. 11. After the final wash, aspirate as much of the wash buffer as possible from the resin. 12. Elute the purified p300 in batch by incubating the resin in 100 ul of elution buffer for 10 min. Pellet the resin by centrifugation and remove the purified p300 supernatant to a new tube. Repeat the elution. 13. Freeze the purified p300 in aliquots in liquid N2 and store at -80°C.C 14. Check the p300 preparation for purity and yield by SDS-PAGE versus bovine serum albumin standards.d "All volumes listed in this protocol are based on initial starting material from one 15 cm diameter plate. Human p300 is synthesized particularly well in Sf9 cells. However, for other proteins that are not synthesized as well, this preparation could be scaled up to more plates or to a large volume suspension culture. b All procedures from this point on should be performed on ice or at 4°C with ice-cold buffers. c The p300 preparation should be stable for up to one year when stored at -80°C with limited thawing and refreezing. d Typical yields are 20-40 ug of purified p300 per 15 cm diameter plate.

172

7: Ligand- and cofactor-regulated in vitro transcription

2.3 Functional analyses of purified receptors and cofactors After the expression and purification of nuclear hormone receptors and cofactors, the purity of the proteins should be assessed by SDS-polyacrylamide gel electrophoresis (Figure 1). In addition, the biochemical activities of the purified proteins should be tested, especially for those proteins whose transcriptional activity will be examined using in vitro transcription assays. For example, with purified receptors, the ligand binding and DNA binding activities should be tested by radioligand binding assays and electrophoretic mobility shift assays. For purified cofactors with enzymatic activities, enzyme assays should be performed (e.g. histone acetyltransferase assays with [3H]acetyl coenzyme A for p300). These assays provide additional information regarding the integrity of the purified proteins.

3. In vitro chromatin assembly 3.1 Chromatin assembly systems Chromatin is the physiological template for many nuclear processes, including ligand- and cofactor-regulated transcription by nuclear hormone receptors (3). Thus, biochemical analyses of receptor transcriptional activity should be performed with chromatin templates. A number of systems for the assembly of DNA templates into chromatin are currently available. These systems use either purified components or crude extracts prepared from Drosophila embryos or Xenopus oocytes (see ref. 4 for a review). The available purified systems allow the production of chromatin templates with a defined composition, but the chromatin typically lacks regularly spaced nucleosomes and physiological nucleosome repeat lengths. Conversely, the extract-based systems allow for the assembly of chromatin templates with properties similar to those of native chromatin, but the composition is not as well defined as with the purified systems. Importantly, however, chromatin assembled with the extracts contains the ATP-dependent chromatin remodelling factors that are necessary for efficient transcriptional activation. In addition, chromatin assembled with the extracts can be purified by sucrose gradient sedimentation and/or size exclusion chromatography after the assembly reaction is complete (5). We have found that chromatin assembled with an extract prepared from Drosophila embryos (the S190) (5, 6) works well for analysing the dynamic processes that occur during transcription factor binding to chromatin and transcription activation (see ref. 7 for a review).

3.2 Assembly of plasmid DNA templates into chromatin For the assembly of plasmid DNA templates into chromatin (Protocol 3), the S190 chromatin assembly extract is incubated with purified core histones to allow the binding of the histones by histone chaperone proteins (e.g. CAF-1, 173

W. L. Kraus and J. T. Kadonaga

Figure 2. Schematics for setting up chromatin assembly (A) and mock assembly (B) reactions as described in Protocols 3 and 4, respectively. MNase, micrococcal nuclease.

NAP-1, nucleoplasmin). Template DNA, ATP, and an ATP-regeneration system are then added, and the reaction is incubated until chromatin assembly is complete (see Figure 2A). Purified nuclear hormone receptors, ligands, and cofactors can be added during the chromatin assembly reaction or after chromatin assembly is complete. In addition, other chromatin proteins, such as histone H1 and HMG proteins, can be incorporated into the chromatin by the addition of purified proteins during the chromatin assembly reaction. Protocol 3. Assembly of plasmid DNA templates into chromatin Reagents • Chromatin assembly extract (S190) pared from Drosophila embryos a • Core histories (~ 1 mg/ml) prepared Drosophila embryos or HeLa cells b • Histone H1 (~ 2 mg/ml) prepared Drosophila embryos or HeLa (optional)c

prefrom from cells

. Buffer R: 10 mM Hepes (K+) pH 7.5, containing 10 mM KCI, 1.5 mM MgCI2, 0.5 mM EGTA, 10% (v/v) glycerol, 10 mM bglycerophosphate, 1 mM DTT, 0.2 mM PMSF; store in aliquots at -20°C (this buffer can be thawed and refrozen many times)

174

7: Ligand- and cofactor-regulated in vitro transcription . Purified ER (from Protocol 1) or other nuclear hormone receptor (optional)d • Elution buffer from ER preparation (from Protocol 7; optional)d • Oestrogen receptor (or other nuclear hormone receptor) ligands: for ER, 17poestradiol (2 uM) or an anti-oestrogen such as hydroxytamoxifen (400 uM) prepared by diluting stock solutions first in ethanol and then 1:10 to the desired final concentration in buffer R

Plasmid DNA template (~ 1 mg/ml) prepared by successive CsCI gradient centrifugation ATP mix: 20 mM ATP, 0.2 M phosphocreatine, 650 ug/ml creatine phosphokinase, 30 mM MgCI2—prepare just before use by diluting the following stock solutions in water: 0.5 M ATP, 0.5 M creatine phosphate, 5 mg/ml creatine phosphokinase, 0.1 M MgCI2 Purified p300 (from Protocol 2, optional)

Methode 1. Mix the following in a 1.5 ml microcentrifuge tube in the indicated order with gentle mixing after the addition of each component: buffer R (the amount needed to give a final volume of 55 ul), S190 (25 ul), core histones (0.7 ug), and histone H1 (0.45 ug; if desired). Incubate for 30 min at room temperature. 2. In a separate 1.5 ml microcentrifuge tube, mix the following components in the indicated order with gentle mixing after each addition: buffer R (the amount needed to give a final volume of 30 ul), ER or corresponding buffer control (for a final ER concentration of 10-30 nM in the completed assembly reaction), oestrogen receptor ligand (for a final oestradiol concentration of 50-100 nM and/or a final antioestrogen concentration of 10-20 uM [a 200-fold molar excess relative to oestradiol] in the completed assembly reaction), and plasmid template DNA (500 ng). Incubate on ice for 20 min.f 3. Combine the S190-histone mix (55 ul), ER-template mix (30 ul), and ATP mix (15 ul) with gentle mixing after each addition. 4. Incubate this chromatin assembly mix for 4 h at 27°C. If desired, cofactors such as p300 can be added at this point. If cofactors are added, incubate the reaction for an additional 30 min at 27°C to allow interaction of the cofactors with the pre-assembled chromatin.g 5. Use the chromatin templates immediately for transcriptional or structural analyses. * Prepared as described in refs 5 and 6. The S190 can be frozen and thawed twice. b Prepared as described in ref. 5 (Drosophila core histones) or in ref. S (HeLa cell core histones). The core histones can be frozen and thawed numerous times. c Prepared as described in ref. 9. The histone H1 can be frozen and thawed numerous times. d Receptor, ligand, and corresponding buffer controls can be added either before chromatin assembly is initiated or after chromatin assembly is complete. If the receptor is added after chromatin assembly is complete (i.e. after the 4 h incubation), incubate the reaction for an additional 30 min at 27°C to allow interaction of the receptor with the pre-assembled chromatin. e This protocol describes a 100 (J assembly reaction with 500 ng of template DNA. This reaction can be scaled up if necessary. f This same basic approach can be used with other purified nuclear hormone receptors and their ligands. g Alternatively, cofactors such as p300 can be added directly to the transcription reactions (see Protocol 5).

175

W. L. Kraus and ]. T. Kadonago

3.3 Analysis of chromatin templates assembled in vitro 3,3,1 Chromatin template and factor integrity After the assembly reaction incubations arc complete, the efficiency of chromatin assembly and the integrity of the chromatin template should be analysed. Highly efficient chromatin assembly is necessary because factor binding and transcriptional activation of incompletely assembled chromatin could be predominantly from nucleosome-frec regions. Also, the addition of any factor that disrupts chromatin assembly will non-specifically relieve chromatin-mcdiated Iranscriptional repression and thus give an apparent activation of transcription. DNA supercoiling and micrococcal nuclease digestion analyses, described in detail elsewhere (4), can be used to analyse the efficiency of chromatin assembly and the integrity of the chromatin template under a variety of experimental conditions (e.g. ± receptor, ± ligand, ± cofactor). Figure 3 illustrates the use of these assays to demonstrate that the addition of ER ± oeslradiol (E 2 ) does not affect chromatin assembly. In addition to assessing the integrity of the chromatin template, the integrity of receptors and cofactors added during the ehromatin assembly reaction should be assayed after the reaction is complete. Alterations in receptor or

Figure 3. Micrococcal nuclease and supercoiling analyses of chromatin. The plasmid template pERE, which contains four copies of an oestrogen-response element upstream of the adenovirus E4 promoter (see Figure 5, bottom), was assembled into chromatin as described in Protocol 3 in the presence or absence of ER (15 nM) or oestradiol (E,; 100 nM) as indicated. The templates were then subjected to micrococcal nuclease (MNase) digestion (A) or DNA supercoiling analyses (B), as described previously (4). From ref. 2. Used with permission.

176

7: Ligand- and cofactor-regulated in vitro transcription cofactor stability under various experimental conditions (e.g. ± ligand) during the 4 h assembly reaction could lead to problems in subsequent structural analyses and transcription assays. Western blot analyses of small aliquots from the assembly reactions with anti-receptor or anti-cofactor antibodies can be used to ensure that receptor and cofactor levels remain constant under all experimental conditions (Figure 6B). 3.3.2 Structural analyses In addition to analysing the transcriptional activity of nuclear hormone receptors with chromatin templates (described in Section 4.2), it may also be of interest to analyse how the receptors interact with and alter the structure of the chromatin templates, and whether these effects are influenced by coactivators or co-repressors. A number of assays, described in detail elsewhere (4), can be used to address these questions. For example, the binding of receptors to the chromatin templates can be examined by DNase I footprinting; receptor- and cofactor-dependent localized alterations in chromatin structure can be examined by nucleosome array disruption assays; and nucleosome positioning can be examined by indirect end-labelling assays. These structural assays can be coupled with transcriptional analyses of the same chromatin preparations, which allows direct comparison of chromatin structure with transcriptional activation using identical chromatin samples.

3.4 Mock chromatin assembly of plasmid DNA templates For some transcription experiments, it may be useful to compare the activity of a receptor or cofactor with chromatin and non-chromatin templates. Although naked plasmid DNA could be used for comparison with chromatin templates assembled by using the S190, the additional proteins contributed by the S190 would make the final reaction conditions quite different and direct comparison difficult. We have developed a method for preparing nonchromatin templates in the presence of the S190 and core histones under reaction conditions similar to those used for assembling chromatin templates (2) (Protocol 4). In this protocol, a competitor plasmid template lacking an RNA polymerase II promoter (e.g. pUC118, added at a threefold mass excess relative to the experimental template DNA) is incubated with the S190 and core histones in the presence of ATP for 30 min prior to the addition of the experimental plasmid template (see Figure 26). Under these conditions, the competitor DNA is assembled into chromatin and the reaction is depleted of available core histones. Thus, when the experimental plasmid template is added, it is not assembled into chromatin. As an additional control to check for non-specific effects of the competitor DNA on in vitro transcription reactions, the competitor DNA can be added to the assembly reaction after the incubation is complete (the experimental template should be fully assembled into chromatin in this control experiment). 177

W. L. Kraus and J. T. Kadonaga Protocol 4. Mock chromatin assembly of plasmid DNA templates Reagents • Chromatin assembly extract (S190) prepared from Drosophila embryosa • Core histories (~ 1 mg/ml) prepared from Drosophila embryos or HeLa cellsb • Histone H1 (~ 2 mg/ml) prepared from Drosophila embryos or HeLa cells (optional)c • Purified ER (from Protocol 7) or other nuclear hormone receptor (optional)d • Elution buffer from ER preparation (see Protocol 1; optional)d

• Buffer R (see Protocol 3) • Oestrogen receptor (or other nuclear hormone receptor) ligands (see Protocol 3) • Plasmid DNA template (~ 1 mg/ml) prepared by successive CsCI gradient centrifugation • Promoterless competitor plasmid DNA (e.g. pUC118) • ATP mix (from Protocol 3) • Purified p300 (from Protocol 2; optional)

Methode 1. Mix the following in a 1.5 ml microcentrifuge tube in the indicated order with gentle mixing after the addition of each component: buffer R (the amount needed to give a final volume of ~ 55 ul; adjust the volume to compensate for the addition of the competitor DNA in step 2, below), S190 (25 ul), core histones (0.7 ug), and histone H1 (0.45 ug; if desired). Incubate for 30 min at room temperature. 2. Combine the S190-histone mix (~ 55 ul), ATP mix (15 ul), and 1.5 ug competitor DNA (a threefold mass excess relative to the specific template DNA) with gentle mixing after each addition. Incubate at 27°C for 30 min. 3. In a separate 1.5 ml microcentrifuge tube, mix the following components in the indicated order with gentle mixing after each addition: buffer R (the amount needed to give a final volume of 30 ul), ER or corresponding buffer control (for a final ER concentration of 10-30 nM in the completed assembly reaction), oestrogen receptor ligand (as described in Protocol 3), and plasmid template DNA (500 ng). Incubate on ice for 20 min.f 4. Combine the S190-histone-competitor DNA-ATP mix (70 ul) with the ER-template mix (30 ul) and mix gently. 5. Incubate the reaction for 4 h at 27°C. If desired, cofactors such as p300 can be added at this point. If cofactors are added, incubate the reaction for an additional 30 min at 27°C to allow interaction of the cofactors with the templates.g 6. Use the templates immediately for transcriptional analyses. * Prepared as described in refs 5 and 6. The S190 can be frozen and thawed twice. b Prepared as described in ref. 5 (Drosophila core histones) or in ref. 8 (HeLa cell core histones). The core histones can be frozen and thawed numerous times. c Prepared as described in ref. 9. The histone H1 can be frozen and thawed numerous times.

178

7: Ligand- and cofactor-regulated in vitro transcription d

Receptor, ligand, and corresponding buffer controls can be added either before the reaction is initiated or after it is complete. If the receptor is added after the 4 h incubation, incubate the reaction for an additional 30 min at 27°C to allow interaction of the receptor with the mock assembled template. e This protocol describes a 100 ul mock assembly reaction with 500 ng of specific template DNA. This reaction can be scaled up if necessary. f This same basic approach can be used with other purified nuclear hormone receptors and their ligands. g Alternately, cofactors such as p300 can be added directly to the transcription reactions (see

Protocol 5).

4. In vitro transcription of chromatin templates 4.1 In vitro transcription systems Many studies have now shown that in addition to RNA polymerase II, basal transcription from RNA polymerase Il-dependent promoters requires a set of transcription factors including TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH (see ref. 10 for a review). Activated transcription requires additional factors, including DNA binding activators (such as nuclear hormone receptors) and a variety of co-activators, some of which are known (such as p300) and others that have yet to be identified and purified. All these factors can be supplied to in vitro transcription reactions in crude nuclear extracts prepared from HeLa cells and Drosophila embryos. When RNA polymerase II, the basal factors, activators, and co-activators are combined with a DNA template containing transcription factor binding sites in the presence of ribonucleoside 5'-triphosphates (rNTPs) under the appropriate conditions in vitro, activated transcription is observed. The RNA transcripts produced can be analysed by the inclusion of a radiolabelled nucleotide or by extension of an end-labelled primer using reverse transcriptase, with subsequent analysis on denaturing polyacrylamide gels.

4.2 Ligand- and cofactor-regulated in vitro transcription with chromatin templates Since chromatin is the physiological template for transcription by nuclear hormone receptors, the use of chromatin templates instead of naked DNA templates for in vitro transcription reactions should give results that more closely mimic the transcriptional activity of the receptors in vivo. In this section, we describe protocols for analysing ligand- and p300-regulated transcription by ER with chromatin templates in vitro. The basic protocols presented herein should be applicable to a variety of other nuclear hormone receptors and cofactors. For the in vitro transcription reactions, chromatin templates are assembled as described above (Protocol 3) under various experimental conditions (± receptor, ± ligand, ± cofactor). The templates are incubated with a HeLa cell nuclear extract (as a source of the basal 179

W. L. Kraus and J. T. Kadonaga

Figure 4. Schematics for setting up multiple round (- Sarkosyl) and single round (+ Sarkosyl) transcription reactions as described in Protocols 5 and 6, respectively.

transcriptional machinery) to allow the formation of transcription preinitiation complexes. rNTPs are subsequently added to initiate transcription, and the resulting RNA products are detected by primer extension analysis (Figure 4) (Protocols 5 and 6). Protocol 5. In vitro transcription of chromatin templates with a HeLa cell nuclear extract Equipment and reagents a • Chromatin or mock assembled template, freshly prepared (from Protocol 3 or 4) • Buffer H: 66 mM Hepes (k+) pH 8.0, containing 0.66 mM EDTA, 234 mM KCI, 15 mM MgCI2 • 10% (w/v) polyvinyl alcohol (PvOH; M, 10000) • HeLa nuclear extract b • Ribonucleoside 5'-triphosphate (rNTP) mix: 10 mM Hepes (K+) pH 7.6, containing 5 mM each of rATP, rCTP, rGTP, rTTP • Proteinase K solution: 10 mM Tris-HCI pH 7.5, containing 0.5 mM EDTA and 2.5 mg/ml proteinase K (USB, No. 20818)

• Purified p300 (from Protocol 2; optional) • Transcription stop solution: 20 mM EDTA, 0.2 M NaCI, 1% (w/v) SDS, 0.25 mg/ml glycogen (this solution should be stored at room temperature); add 5 ul proteinase K solution per 100 ul transcription stop solution just before use • 3 M sodium acetate • Phenol:chloroform:isoamyl alcohol mix (25:24:1, by vol.) (PCIAA), equilibrated with Tris-HCI pH 7.5 • 100% ethanol • 75% ethanol • Rotary evaporator (e.g. SpeedVac)

Method 1. Thaw a sufficient number of aliquots of HeLa cell nuclear extract for the experiment quickly in an ice/water-bath. Also thaw aliquots of 10% PvOH and rNTP mix. Keep all reagents on ice once thawed. 2. Combine the following reagents in a 1.5 ml microcentrifuge tube on

180

7: Ligand- and cofactor-regulated in vitro transcription ice for each sample in the order listed with gentle mixing after each addition:c,d • 15 ul of chromatin (containing 75 ng DNA) • 7.5 ul of buffer H • 12.5 ul of 10% PvOH • 10 ul of HeLa nuclear extract • 0.5 ul of purified p300 (50-100 nM in the completed reaction; optional)e 3. Incubate at room temperature (c. 21°C) for 15 min. 4. Initiate transcription by the addition of 5 ul of the rNTP mix and then incubate at 30°C for 30 min. 5. Stop the reactions by the addition of 100 ul of transcription stop solution. Mix well and incubate for 10 min at 37°C. 6. Extract the samples with 200 ul of PCIAA, centrifuge for 5 min in a microcentrifuge, and transfer the upper aqueous phase to a new 1.5 ml microcentrifuge tube. 7. Add 15 ul of 3 M sodium acetate and 400 ul of 100% ethanol to each sample. Vortex the samples to mix, and then centrifuge them for 15 min at room temperature in a microcentrifuge. 8. Carefully aspirate the liquid from the small, white pellets and add 400 ul of 75% ethanol to each tube. Centrifuge for 5 min at room temperature in a microcentrifuge. 9. Carefully aspirate all the liquid from the pellets. Dry the pellets in a rotary evaporator. Store the samples at -20°C, or proceed immediately with the primer extension analyses. " All reagents used for this protocol should be RNase-free. Likewise, all plasticware should be RNase-free, and great care should be used to minimize contamination by RNases. b Prepared as described in refs 11 and 12. Avoid repeated thawing and refreezing. c The buffer H and the PvOH can be combined before addition to the reaction to reduce pipetting. d The reaction for each set of experimental conditions should be performed in duplicate to compensate for any pipetting errors or loss of sample. e Alternately, cofactors such as p300 can be added to the chromatin assembly or mock assembly reactions (see Protocols 3 and 4).

Figure 5 shows a typical in vitro transcription result for E2-regulated transcription by ER with chromatin templates. For these experiments, a plasmid template containing four copies of an oestrogen-response element (ERE) upstream of the adenovirus E4 promoter was used. Note that the receptor can be added either during chromatin assembly or after chromatin assembly is complete. The E2-stimulated transcriptional activation can be inhibited by anti-oestrogens (Figure 6A, left panel). In addition, a single point mutation in 181

.W.L, Kraus and J. T. Kadonaga

Figures. Oestrogen-regulated transcription by ER with chromatin templates. The plasmid template pERE (bottom) was assembled into chromatin as described in Protocol d in the presence or absence of ER or oestradiol (E2), which were added either during or after chromatin assembly as indicated. The templates were then subjected to in vitro transcription and primer extension analyses in duplicate (top) as described in Protocols 5 and 6, respectively. The final concentrations of ER and E, in the transcription reactions were 4.5 nM and 30 nM, respectively. The relative fold transcriptional activation is indicated. From ref. 2. Used with permission.

the ligand-dependenl activation function (AF-2) of FR can render the receptor transcriptionally inactive (2). Addition of the co-activator p300 dramatically enhances the ligand-dependent transcriptional activity of ER (Figure 7, left panel). These results illustrate the appropriate ligand-, activation domain-, and co-activator-dependent responses that are observed with this system. Protocols.

Primer extension analysis of in vitro transcription products

Equipment and reagents DNA oligonucleotide ( • • • • 25-30 nucleotides in length) at 200 pmol/ul2 10 x T4 polynueleotide kinase buffer: 0.5 M Tris-HCI pH 7.6, 0.1 M MgCl2, 1 mM spermidine, 1 mM EOT A, 5 mM DTT T4 polynucleotide kinase (10 U/ul) England Biolabsl

(New

182

(•y-32P]ATP (25 pmol/ul. 6000 Ci/mmol) (New England Nuclear) 2.5 M ammonium acetate 10 mg/ml glycogen 100%ethanol TE: 10 mM Tris-HCI pH 7.5, 1 mM EDTA 3 M sodium acetate

7: Ligand- and cofactor-regulated in vitro transcription PCIAA (25:24:1, by vol.), equilibrated with Tris-HCI pH 7.5 Primer extension annealing buffer: 10 mM Tris-HCI pH 7.8, 0.25 M KCI, 1 mM EDTA Primer extension mix (mix the following, in order): Tris-HCI pH 8.3 (2.5 ml), 0.1 M MnCI2 (0.5 ml), 2.5 mg/ml actinomycin D dissolved in water (2 ml), 0.1 M solution of each of the four deoxyribonucleoside 5'triphosphates (dNTPs; 140 ul), 0.5 M DTT (1 ml), and water to a final volume of 40 ml— aliquot and store at-20°C 8% (w/v) polyacrylamide-urea sequencing gel

Geiger counter Reverse transcriptase-primer extension mix (RT-PE mix): immediately before using, combine 60 U of Moloney murine leukemia virus reverse transcriptase (Promega; 0.3 ul) and 40 ul of primer extension mix per sample Formamide loading solution (FLB): 80% (v/v) formamide, 10 mM EDTA, 1 mg/ml xylene cyanol, 1 mg/ml bromophenol blue FLB-NaOH mix: immediately before using, combine two parts of FLB with one part of 0.1 M NaOH Rotary evaporator (e.g. SpeedVac)

A. Radiolabelling the DNA oligonucleotide 1. Make a working dilution of the primer at 2.5 pmol/ul. Radiolabel 5 pmol of primer in a 1.5 ml microcentrifuge tube by using T4 polynucleotide kinase according to the manufacturer's specifications in the presence of 2 ul (50 pmol) of [y-32P]ATP. The reaction should have a final volume of 20 ul and should be incubated at 37°C for 1 h.b 2. Add 100 ul of 2.5 M ammonium acetate. Heat the reaction to 70°C for 15 min (to inactivate the T4 polynucleotide kinase) and allow the tube to cool to room temperature. Add 2 ul of 10 mg/ml glycogen and mix. Then add 375 ul of ethanol to precipitate the radiolabelled primer. Mix by vortexing. 3. Centrifuge the ethanol precipitate in a microcentrifuge for 15 min. Carefully remove the ethanol supernatant from the pellet, add 100 ul of TE to the tube, and dissolve the pellet by vortexing. 4. Add 100 ul of PCIAA and mix by vortexing. Centrifuge the tube in a microcentrifuge for 5 min to separate the organic and aqueous phases. Remove the aqueous phase to a new 1.5 ml microcentrifuge tube. 5. Add 10 ul of 3 M sodium acetate and mix. Then add 300 ul of ethanol to precipitate the radiolabelled primer. Mix by vortexing and centrifuge for 15 min in a microcentrifuge. Carefully remove the ethanol supernatant from the pellet and discard. Dry the pellet for 10 min at room temperature.c 6. Dissolve the pellet to a final concentration of 33 fmol/ul in 150 ul of TE. Store the radiolabelled primer at -20°C.d B. Setting up the primer extension reactions 1. Make a mix containing 10 ul of primer extension annealing buffer and 0.25 ul of labelled primer (~ 8 fmol) for each sample to be analysed in the assay. 2. Add 10 ul of the annealing buffer-primer mix to each tube of RNA product. Dissolve the RNA by vortexing.

183

W. L. Kraus and J, T. Kadonaga Protocol 6.

Continued

3. Heat the samples to 70°C for 2 min, and then incubate for 40 min at 58°C to anneal the primer to the RNA product.6 4. Collect the condensation from the lids of the tubes by a 20 sec centrifugation in a microcentrifuge, and allow the samples to cool to room temperature. Add 40 ul of RT-PE mix and mix by gentle vortexing. Incubate for 40 min at 37 °C. 5. Add 225 ul of ethanol. Mix the samples by vortexing and centrifuge for 15 min in a microcentrifuge. 6. Remove the ethanol supernatant and dry the pellets in a rotary dehydrator. Dissolve the pellets in 6 ul of FLB-NaOH mix. Store the samples at-20°C or proceed immediately with denaturing polyacrylamide gel electrophoresis. C. Analysing the samples by gel electrophoresis 1. Pour a standard 8% (w/v) polyacrylamide-urea sequencing gel. Boil the primer extension samples for 4 min in a water-bath. Run the samples on the gel in 1 x TBE until the bromophenol blue tracking dye is 3-5 cm from the bottom of the gel.f 2. Dry the gel on a piece of Whatman 3MM paper, followed by Phosphorlmaging analysis or autoradiography. "The oligo should be complimentary to the RNA product from the in vitro transcription reactions and should be designed to yield a primer extension product -100 nucleotides in length (from the 5' end of the annealed oligo to the transcription start site). The oligo should be gel purified on high percentage (e.g. 10%, w/v) polyacrylamide gels run in 1 x TBE, eluted, PCIAA extracted, ethanol precipitated, dissolved in water, and quantified. bGreat care should be taken and appropriate shielding used when working with [-y-32P]ATP. All bench surfaces, pipettors, tube racks, microcentrifuges, hands, and forearms should be carefully monitored with a Geiger counter. c Do not dry the pellets under vacuum or they will be very difficult to dissolve. "'Primers labelled in this manner are good for two to three weeks for primer extension analyses. "The optimal annealing temperature will vary depending on the length and G + C content of the primer. fA 28 base primer will run in the vicinity of the bromophenol blue tracking dye on an 8% (w/v) sequencing gel. Reverse transcription products of 100-150 nucleotides will run slightly above the xylene cyanol tracking dye on an 8% (w/v) sequencing gel.

Control experiments should be performed to confirm the specificity of the responses as well as to optimize the experimental conditions. For example, receptor dependence and specificity of the transcriptional effects can be assessed by performing parallel experiments with a similar template lacking the receptor binding sites (EREs in the experiments described herein; Figure 6A, right panel). As described in Section 3.3.1, the integrity of the receptor under the various experimental conditions (i.e. different ligands in Figure 6A) should be assessed by Western blotting (Figure 6B). Receptor and coactivator dose-response studies should also be performed to determine the optimal amounts of protein for each experiment. 184

7; Ligand- and cofactor-regulated

in vitro transcription

Figured. (A) Anti-oestrogen-regulated and binding site-dependent transcription by ER with chromatin templates. The plasmid template pERE (left), or the same template lacking oestrogen-response elements (EREs), was assembled into chromatin as described in Protocol 3 in the presence or absence of ER, E2, trans-hydroxytamoxifen [TOT), or ICI-164384 (ICI) as indicated. The templates were then subjected to in vitro transcription and primer extension analyses in duplicate as described in Protocols Sand 6, respectively. The final concentrations of ER, E2, and anti-oestrogens [TOT and ICIJ in the transcription reactions were 4.5 nM, 15 nM, and 3 uM, respectively. Note that the antioestrogens were added at a 200-fold molar ratio relative to the E2. The relative fold transcriptional activation is indicated, (B) ER is not degraded under various ligand conditions during chromatin assembly. Aliquots of the chromatin assembly reactions from panel A, were subjected to Western blot analysis with an anti-ER antibody. From ret 2. Used with permission.

4.3 Chromatin versus non-chromatin transcription experiments For some transcription experiments, it may be useful to compare the activity of a receptor or cofactor with chromatin versus non-chromatin templates. In vitro transcription reactions using HeLa cell nuclear extracts can be performed with the mock assembled templates described in Protocol 4 in side-byside experiments with chromatin templates under identical reaction conditions (see Protocol 5). In most cases, basal transcription will be higher and activated transcription will be lower with non-chromatin templates than with chromatin templates (Figure 7). In addition, with RR we have not observed significant amounts of E2 and p300-stimulated transcription with non-chromatin templates (2) (Figure 7), Comparing the activity of a receptor or cofactor with chromatin versus non-chromatin templates provides useful 185

W. L. Kraus and J. T. Kadonagu

Figure 7. Oestrogen- and p300-regulated transcription by ER with chromatin templates, but not with non-chromatin templates. The plasmict template pERE was assembled into chromatin (left) as described in Protocol 3 or subjected to mock assembly (right) as described in Protocol 4 in the presence or absence of ER, E2 or p300 as indicated. The templates were then subjected to in vitro transcription and primer extension analyses in duplicate as described in Protocols 5 and 6, respectively. The final concentrations of ER, E2r and p300 in the transcription reactions were 4.5 nM, 30 nM, and 15 nM, respectively. The relative fold transcriptional activation is indicated. From ref. 2. Used with permission.

insights into the role of chromatin in ligand- and cofactor-regulatcd transcriptional activation by nuclear hormone receptors.

4.4 Single versus multiple round transcription experiments In the transcription experiments described in Protocol 5 and shown in Figures 5, 6A, and 7. multiple rounds of transcription (i.e. pre-initiation complex formation, initiation of RNA synthesis, promoter clearance, and elongation) were allowed to occur. However, experiments in which transcription is limited to a single round can also be informative. Factor-dependent effects that are observed in a single round of transcription often occur early in the transcription process (i.e. during pre-initiation complex formation, initiation of RNA synthesis, or promoter clearance, which can collectively be referred to as transcription initiation). Effects that are observed only after multiple rounds of transcription are most likely the result of enhanced transcription reinitiation. Thus, comparing the activity of a receptor or co-activator in single versus multiple rounds of transcription can help elucidate the mechanism(s) of transcriptional activation by those factors. By using this approach, we have found that liganded ER and p300 synergistically enhance transcription 186

7; Ligand- and cofactor-regulated in vitro transcription initiation with chromatin templates (2). In addition, liganded ER also stimulates transcription reinitiation with chromatin templates (2). A single round of transcription can be achieved through the use of reagents that selectively block transcription initiation, but not transcription elongation. These reagents include the detergent Sarkosyl (0.05-0.4%, w/v) (13, 14) and heparin (16 ug/ml) (15). The addition of Sarkosyl or heparin to preformed pre-initiation complexes in in vitro transcription reactions immediately after the addition of the rNTPs will allow only one round of transcription to occur (see Figure 4). Protocol 7 describes methods for setting up single round transcription experiments with chromatin templates. Protocol 7. Single round in vitro transcription of chromatin templates with a HeLa cell nuclear extract Equipment and reagents3 • Chromatin or mock assembled template, freshly prepared (from Protocol 3 or 4) Buffer H (see Protocol 5) 10% (w/v) PvOH (M, 10000) HeLa cell nuclear extract6 rNTP mix (see Protocol 5) 10% (w/v) Sarkosyl in H20

Transcription stop solution (see Protocol 5) 3 M sodium acetate PCIAA (25:24:1, by vol.), equilibrated with Tris-HCI pH 7.5 100%ethanol 75% ethanol Rotary evaporator (e.g. SpeedVac)

Method 1. Set up the in vitro transcription experiments as described in Protocol 5, steps 1 and 2.c 2. Incubate the reactions at 30°C for 30 min to allow the formation of transcription pre-initiation complexes.'* 3. Initiate transcription by the addition of 5 ul of rNTP mix. Mix gently and incubate for 10 sec at room temperature. 4. Immediately add 1 ul of 10% Sarkosyl with gentle mixing after addition. Incubate at 30°C for 30 min. 5. Stop the reactions by the addition of 100 ul of transcription stop solution. Mix well and incubate for 10 min at 37°C. 6. Finish the preparation of the samples as described in Protocol 5, steps 6-9. "All reagents used for this protocol should be RNase-free. Likewise, all plasticware should be RNase-free, and great care should be used to minimize contamination by RNases. 6 Prepared as described in refs 11 and 12. Avoid repeated thawing and refreezing. "The reaction for each set of experimental conditions should be performed in duplicate to compensate for any pipetting errors or loss of sample. "Multiple round transcriptions, when run side-by-side with the single round transcriptions, should include the 30 min incubation at 30°C to allow formation of transcription pre-initiation complexes.

187

W. L. Kraus and J. T. Kadonaga

5. Exploiting ligand- and cofactor-regulated in vitro transcription systems A fascinating and unique aspect of in vivo transcriptional regulation by members of the nuclear hormone receptor superfamily is the modulation that is achieved upon ligand binding by the receptors. Thus, in vitro biochemical systems used for studying the mechanisms of nuclear hormone receptor transcriptional activity should accurately recreate known ligand-dependent effects. Our studies indicate that chromatin is an integral component for achieving ligand-dependent transcriptional responses by nuclear hormone receptors in vitro (2) (Figure 7). Therefore, in vitro chromatin assembly and transcription systems should be useful for examining the roles that ligands and cofactors play in the transcriptional activity of nuclear hormone receptors. In addition, these systems have the potential to be used for screening novel ligands for orphan nuclear receptors, characterizing receptor and cofactor mutants, and identifying new co-activators and co-repressors. The use of biochemical approaches, like those described herein, should facilitate the study of transcriptional mechanisms and will increase our understanding of nuclear hormone receptors as ligand- and cofactor-regulated transcription factors.

Acknowledgements We thank Trish Willy, Alan Kutach, and Mark Levenstein for their helpful comments on this chapter. J. T. Kadonaga is supported by grants from the National Institutes of Health and the National Science Foundation. W. L. Kraus is supported by a Career Award in the Biomedical Sciences from the Burroughs Wellcome Fund.

References 1. Mangelsdorf, D. J., Thummel, C, Beato, M, Herrlich, P., Schiitz, G., Umesono, K.,et al. (1995). Cell, 83, 835. 2. Kraus, W. L. and Kadonaga, J. T. (1998). Genes Dev., 12, 331. 3. Kadonaga, J. T. (1998). Cell, 92, 307. 4. Pazin, M. J. and Kadonaga, J. T. (1998). In Chromatin: a practical approach (ed. H. Gould), p. 173. IRL Press, Oxford. 5. Bulger, M. and Kadonaga, J. T. (1994). In Methods in molecular genetics (ed. K. W. Adolph), Vol. 5, p. 241. Academic Press, San Diego. 6. Kamakaka, R. T., Bulger, M., and Kadonaga, J. T. (1993). Genes Dev., 7,1779. 7. Robinson, K. M. and Kadonaga, J. T. (1998). Biochim. Biophys. Acta, 1378, Ml. 8. Ausio, J. and van Holde, K. E. (1986). Biochemistry, 25,1421. 9. Croston, G. E., Lira, L. M., and Kadonaga, J. T. (1991). Protein Expression Purification, 2, 162. 188

7: Ligand- and cofactor-regulated in vitro transcription 10. Orphanides, G., Lagrange, T., and Reinberg, D. (1996). Genes Dev., 10, 2657. 11. Dignam, J. D., Lebovitz, R. M., and Roeder, R. G. (1983). Nucleic Acids Res., 11, 1475. 12. Kadonaga, J. T. (1996). In Strategies for protein purification and characterization: a laboratory course manual, p. 127. Cold Spring Harbor Laboratory Press, New York. 13. Hawley, D. K. and Roeder, R. G. (1985). /. Biol Chem., 260, 8163. 14. Hawley, D. K. and Roeder, R. G. (1987). J. Biol. Chem., 262, 3452. 15. Reinberg, D. and Roeder, R. G. (1987).J. Biol. Chem., 262, 3310.

189

This page intentionally left blank

8

Hormone-resistance syndromes M. GURNELL and V. K. K. CHATTERJEE

1. Introduction The nuclear receptors comprise a superfamily of ligand-inducible transcription factors which exhibit a conserved structural organization with distinct functional domains. Members of this family include the glucocorticoid (GR), androgen (AR), oestrogen (ER), mineralocorticoid (MR), and progesterone (PR) receptors, together with those for thyroid hormone (TR), vitamin D (VDR), retinoids (RAR, RXR), eicosanoids (PPAR), and an expanding number of 'orphan' receptors whose ligand and function remain to be elucidated. Mutations in nuclear receptor genes form the basis of a number of inherited human diseases and defects have been identified using two principal lines of investigation. First, in the 'candidate gene' approach, nuclear receptor defects have been anticipated in cases of hormone resistance characterized by reduction in target organ responsiveness to circulating hormone levels. Characterization of the various receptor genes has enabled direct sequencing approaches to dissect the molecular abnormality present in many cases of hormone resistance. The evidence for the pathological role of these defects has then been supported by studies documenting a decreased receptor number or impaired function in each case. Secondly, in a 'reverse genetic' approach, linkage studies have localized the molecular basis of a given disease to a chromosomal locus, and positional cloning has subsequently revealed a gene encoding a nuclear receptor. Table 1 summarizes the human disorders which are currently known to be associated with germline defects in nuclear receptor genes. Most inherited receptor defects are recessive—with disease manifesting when the defect is homozygous or compound heterozygote. Dominantly inherited disorders are due to heterozygous gene mutations which impair hormone action either as a consequence of cellular haploinsufficency of normal receptor or by inhibition of wild-type receptor action by the mutant counterpart in a dominant negative manner. This chapter focuses on the syndrome of resistance to thyroid hormone (RTH), a dominantly-inherited disorder, which is characterized by elevated circulating free thyroid hormones together with pituitary and peripheral refractoriness to hormone action. Following linkage of RTH to the TR|3 gene locus, over 70

M. Gurnell and V. K. K. Chatterjee Table 1. Clinical syndromes associated with mutations in nuclear receptors Receptor

Clinical features

Androgen

(i) Partial (PAIS) or complete (CAIS) androgen-insensitivity syndrome

Vitamin D

(ii) Spinal and bulbar muscular atrophy (SBMA) Hereditary vitamin D-resistant rickets

Glucocorticoid Elevated cortisol, hyperandrogenism. hypokalaemia Oestrogen Elevated gonadotrophins. tall stature, osteoporosis Thyroid (i) Elevated T4, normal TSH, deaf mutism, dysmorphic facies, hormone (3 receptor stippled epiphyses (ii) Elevated T4, normal TSH, goitre, variable peripheral resistance

Hepatocyte nuclear factor 4a DAX1

Inheritance

Nature of mutations

Autosomal recessive Autosomal recessive Autosomal recessive Autosomal recessive

Missense, premature stop Missense

Autosomal dominant

Maturity onset diabetes of the young type 1 (MODY1)

Autosomal dominant

Adrenal hypoplasia Hypogonadotrophic hypogonadism

X-l inked

Premature stop TRpgene deletion Missense, frame shift, in-frame deletion, premature stop Missense, premature stop Frame shift, Missense, premature stop

different missense, frame shift, and in-frame codon deletion receptor mutations have been documented, all localizing to the ligand binding domain (LBD) (1). Here, we describe the characterization of (3 receptor defects: the identification of receptor mutations by sequencing either the gene or the cDNA from an index case; genotyping of affected family members by restriction analysis; haplotyping families to determine whether identical mutations have arisen independently in different families; mutant receptor construction by site-directed mutagenesis; measurement of ligand binding affinity for mutant protein; functional characterization of mutant receptor by co-transfection assays using positively- or negatively-regulated reporter genes; quantitation of the dominant negative potential of a mutant receptor by co-transfection with wild-type receptor in transfection assays.

2. Receptor analysis 2.1 Gene sequencing The most direct method of receptor gene analysis involves PCR amplification of its coding exons from genomic DNA usually extracted from peripheral 192

8: Hormone-resistance syndromes blood leucocytes. All the TR|3 mutations identified to date are encompassed within exons 7-10 of the LBD. The genomic organization of TRB, including gene sequence at its intron/exon boundaries, is known. Each of these exons is amplified using a pair of forward and reverse sequencing primers (Table 2). The sequence of each primer corresponds to regions within flanking introns at least 50-100 nucleotides from intron/exon boundaries, enabling splice donor and acceptor sites and contextual sequence to be checked. Following amplification, excess primers and dNTPs are removed by polyethylene glycol precipitation. The PCR product is then purified by ethanol precipitation and subjected to automated sequence analysis (Protocol 1). Protocol 1. Amplification and sequencing of the coding exons of TRp1 Reagents Taq DNA polymerase (5 U/ul) 10 x PCR buffer: 100 mM Tris-HCI pH 8.3, 500 mM KCI 25 mM MgClz Polyethylene glycol (PEG) precipitation solution: 26.2% (w/v) PEG 8000, 6.6 mM MgCI2, 0.6 M KOAc

dNTP solution containing 1.25 mM of each deoxynucleotide Primer pairs (5 (oM) for each exon to be amplified (Table 2t Genomic template DNA (200 ng/n.1) Amplitaq dye terminator sequencing mix (Perkin-Elmer Applied Biosystems)

A. Exon amplification 1. For each exon to be amplified, prepare the following mix: 5 jxl of 10 x buffer, 5 ul dNTP, 4 ul MgCI2, 27 ul dH2O, 0.2 ul Taq DNA polymerase. For ease, a master mix for n + 1 reactions should be prepared and aliquoted. 2. Add 2 ul of both the forward and reverse primers for each exon, together with 5 ul of DNA (1 ug), to the PCR mix. 3. Amplify sequence as follows: initial denaturation at 94°C for 3 min, then 30 cycles at 94°C for 30 sec, 60°C for 30 sec, 72°C for 30 sec, with a final extension at 72°C for 5 min. 4. Clean-up PCR product: (a) Add 50 ul of PEG precipitation solution to each 50 ul PCR reaction (extracted if necessary from below oil) and allow to stand at room temperature for 15 min. (b) Place in microcentrifuge and spin at 15000 g for 15 min. (c) Aspirate off supernatant and wash pellet with 70% ethanol solution. (d) Microcentrifuge briefly, repeat aspiration, and allow pellet to air dry for 5 min. (e) Resuspend pellet in 10-15 ul of dH2O by vortexing briefly. The PCR products can be checked on a 2% agarose gel at this stage.

193

M. Gurnell and V. K. K. Chatterjee Protocol 1.

Continued

B. Automated sequencing 1. Add 2-5 ul of each PCR product to 1 (ul of primer (forward or reverse), 4 ul of sequencing mix, and make up to a final volume of 10 ul with dH20. 2. Cycle sequence as follows: 94°C for 30 sec, 60°C for 30 sec, 72°C for 30 sec, for 30 cycles. 3. Clean-up sequence product: (a) Make up to 100 ul with dH20 (if necessary extract from below oil). (b) Add 50 ul of 7.5 M NH4OAc, 375 (ul of absolute ethanol, and mix thoroughly. (c) Incubate on dry ice for 15 min. (d) Microcentrifuge at 15000 fir for 15 min. (e) Aspirate off supernatant and wash with 70% ethanol. (f) Microcentrifuge briefly, repeat aspiration, and allow pellet to air dry for 5 min. (g) Resuspend pellet in formamide EDTA or appropriate reagent according to manufacturer's instructions. 4. Run sample on automated sequencer.

2.2 cDNA sequencing If the genomic organization of a receptor gene is not known, an alternative approach to screening for mutations involves the amplification and sequencing Table 2. Primer pairs for amplification of coding exons 7-10 of human TR(31" Primer Exon 7 forward Exon 7 reverse Exon 8 forward Exon 8 reverse Exon 9 forward Exon 9 reverse ExonlOforward Exon 10 reverse

Sequence 5'TGT/AAA/ ACG/ACG/GCC/AGT/CAG/TGG/TCC/CAC/TCC/TGA/ GGC3' 5'GAT/TCT/AGA/AAT/TGA/GGT/AGA/AAA/CAC/TGG3' 5' TGT / AAA / ACG / ACG / GCC / AGT / GTT / CAG / AAG / ATG / ATT / TTC /TGC3' 5'GAT/CTG/CAG/ACC/CAG/TAT/TCC/TGG/AAA/CTG 3' 5'TGT/ AAA/ ACG/ACG/GCC/AGT/ACA/GAA/GGT/TAT/TCC/TAT /TGC3' 5'GAT/CTG/CAG/GCT/CTT/TGG/ATG/CCC/ACT/AAC3' 5'TGT/ AAA/ ACG/ ACG/GCC/ AGT /AGG/CCT/ GGA/ ATT /GGA/CAA/AGC3' 5' GGA / ATT / ATG / AGA / ATG / AAT / TCA / GTC / AGT 3'

"The forward primers are M13 tagged (bold type) to allow direct sequencing using the M13 dye primer system if preferred.

194

8: Hormone-resistance syndromes

Figure 1. Schematic representation of the strategy for reverse transcription PCR of the human TRB1 receptor as two overlapping fragments (I, II).

of the receptor cDNA. For example, in the case of TRB, the gene is transcribed in fibroblasts. Accordingly, total RNA is extracted from cultured fibroblasts obtained following a skin biopsy in an index case of RTH (Protocol 2). The mRNA encoding human TRB1 is then reverse transcribed as two overlapping fragments (I, II; see Figure 1) using two gene-specific primers (Table 3, primers B and D) for the initial cDNA synthesis step (Protocol 3). Following this, the cDNA is amplified by PCR using appropriate primer pairs (A and B, C and D; see Table 3) to yield products of predicted length. Indeed, comparison of their sizes by agarose gel electrophoresis with cDNAs amplified from normal control fibroblast RNA might reveal differences indicating an insertion or deletion due to aberrant splicing. It is also important to check that parallel control reactions performed without added reverse transcriptase yield no product, to exclude the possibility of amplification from contaminant cDNA. The PCR product is then purified and cycle sequenced as described in Protocol 1. Protocol 2. Extraction of total RNA from fibroblasts Reagents Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% fetal calf serum (FCS) and 1% penicillin/streptomycin/ fungizone (PSF) (all reagents supplied by Gibco BRL)

Trizol™ reagent (Gibco BRL) Nuclease-free chloroform, isopropanol, and 70% ethanol

Method 1. Add 1 ml of trypsin to one confluent 10 cm plate of fibroblasts and leave at 37°C for 2-3 min. Gently agitate to separate cells from plate. Add 2 ml of media. 2. Transfer to 15 ml Falcon™ tube and centrifuge at 250 g for 5 min. 195

M. Gurnell and V. K. K. Chatterjee Protocol 2.

Continued

3. Aspirate off supernatant, resuspend cell pellet in 0.5 ml of Trizol™, transfer to Eppendorf tube, and incubate at room temperature for 5 min. 4. Add 0.5 ml of chloroform, shake vigorously for 15-20 sec, and allow to stand at room temperature for 2-3 min. 5. Centrifuge at 10000 g for 15 min at 4°C. 6. Transfer the upper aqueous phase to a clean Eppendorf tube and precipitate the RNA with 0.25 ml of room temperature isopropanol. Leave to stand for 10 min. 7. Centrifuge at 10000 g for 15 min at 4°C. 8. Aspirate off supernatant, wash pellet with 1 ml of 70% ethanol (made up with nuclease-free dH20), and mix by vortexing. 9. Centrifuge at 6000 fir for 5 min at 4°C. 10. Aspirate off supernatant and air dry pellet for 5 min. 11. Resuspend in 30-50 ul of nuclease-free dH20 and incubate at 55°C for 10 min. 12. Determine concentration and check for integrity of 18S and 28S RNA bands on a 1% agarose gel. Store at -70°C. Avoid repeated freezethaw cycles.

Protocol 3.

RT-PCR amplification of TRB1 cDNA

Reagents Reverse transcriptase enzyme, e.g. Superscript™ Preamplification System (Gibco BRL) which also includes appropriate reagents and a control RNA sample—alternatively, the PCR buffer and reagents (nuclease-free) outlined in Protocol 1 can be used

Total RNA (2.5 ug) from both subject and control Primer pairs (Figure 1 and Table 3) at 2 uM and 5 uM 0.1 M DTT; RNase

A. cDNA synthesis 1. For each receptor fragment add 2.5 ng of total RNA to 1 ul of genespecific primer (2 uM) corresponding to the region to be amplified (primers B or D) and make up to a total volume of 12 ul with nucleasefree dH20. Prepare duplicate annealing reactions to act as controls without added reverse transcriptase. 2. Incubate at 70°C for 10 min and then transfer onto ice for 2-3 min. 3. Prepare the following mix for each reaction (for ease, make a master mix for n + 1 reactions and aliquot), adding the components in the order indicated: 2 ul of 10 x PCR buffer, 2 ul of 25 mM MgCI2, 1 ul of 10 mM dNTPs, 2 ul of 0.1 M DTT. 196

8: Hormone-resistance syndromes 4. Add 7 ul of this mix to each reaction and incubate at 45°C for 5 min. 5. Add 1 (ul of reverse transcriptase or 1 ul of nuclease-free dH2O as appropriate and leave at 45°C for a further 50 min. 6. Terminate reactions by heating to 70°C for 5 min and then return to ice for 2-3 min. 7. Add 1 (ul of RNase and incubate at 37°C for 20 min. B. PCR amplification 1. To 10 (ul of each cDNA sample add 5 ul of 10 x PCR buffer, 3 ul of 25 mM MgCI2, 1 ul of 10 mM dNTPs, 2 ul of both the forward and reverse primers (5 (uM) corresponding to the cDNA to be amplified, 0.2 ul Taq DNA polymerase, and 27 ul dH2O. 2. Amplify sequence as follows: initial denaturation at 94°C for 3 min, then 35 cycles at 94°C for 60 sec, 50°C for 60 sec, and 72°C for 90 sec, with a final extension at 72°C for 5 min. 3. Clean-up PCR products as outlined in Protocol 1. C. Automated sequencing 1. Follow Protocol 1. Additional internal primers may be required to allow complete sequencing of the receptor fragments.

2.3 Genotyping by restriction analysis An example of a typical mutation identified by direct sequencing of DNA from an index case of RTH is shown in Figure 2. To rule out the (unlikely) possibility of a sequence abnormality being artefactual due to nucleotide misincorporation, the mutation is verified by sequencing a second, independently generated, PCR reaction. Affected family members can also be screened for the same receptor abnormality by gene sequencing. However, in some cases, the nucleotide substitution corresponding to a particular TR(3 mutation may alter a restriction endonuclease recognition site in its vicinity, either creating or losing an additional site, thus providing a useful, rapid method of screening other family members for the presence of the mutation. Here, PCR Table 3. Primer pairs for amplification of TR(31 cDNA from mRNA Primer

Sequence

A B C D

5' AGGGATCCAGAATGATTACTAACCT 3' 5' TGGCTTGTGCCCGATGGACTT 3' 5' GATTTGGTGCTGGATGAC 3' 5' GAAATTATGAGAATGAATTCAGTCAGT 3'

197

M. Gurnell and V. K. K. Chatterjee

Figure 2. Typical automated sequence analysis demonstrating a heterozygous arginine to tryptophan missense mutation at codon 338 (R338W) in the human TRp51 receptor. The oligonucleotide sequence required for synthesis of the mutant protein is also shown.

amplification of the relevant exon containing the mutation is followed by digestion with the relevant restriction enzyme followed by gel electrophoresis. The digestion pattern from the mutant allele differs from that of its wild-type counterpart. Figure 3 illustrates such analyses in a kindred harbouring a glycine (GGT) to serine (AGT) substitution at codon 340 (G340S) in the (3 receptor. The mutation creates a second HinfI site, such that digestion with this enzyme results in two additional fragments of 183 and 125 bp from the mutant allele, whereas unaffected members show only 308 and 34 bp fragments corresponding to two wild-type receptor alleles.

2.4 Haplotype analysis of kindreds with identical mutations Many mutations in TR@ associated with RTH have been identified in more than one kindred, raising the question as to whether they have arisen as separate mutational events or reflect a shared genetic ancestry between different families. A technique, which can be used to address this, involves haplotyping affected and unaffected individuals in each family using markers closely linked to the TR(3 gene locus. A number of polymorphisms in the TR|3 gene have been identified, which alter the pattern of DNA digestion 198

8; Hormone-resistance syndromes

Figure 3. Genotyping by restriction enzyme digestion. A schematic representation of exon 9 amplified from either mutant (G340S) or wild-type alleles is shown. The position of a novel Hinfi restriction enzyme site is denoted by an asterisk. Following digestion with Hinfl, 308 bp and 34 bp products from two wild-type alleles are generated from normal controls, whereas affected cases exhibit two additional fragments of 183 bp and 125 bp from the mutant allele of the p-receptor gene. Reproduced from ref. 10 with copyright permission of Blackwell Science Ltd,

by restriction endonucleases, giving rise to so-called restriction fragment length polymorphisms (RFLPs). In individuals with a common ancestry, these polymorphisms and hence a particular RFLP pattern are highly likely to co-segregate with a TRp mutation. However, affected individuals from an unrelated family will usually exhibit a different RFLP pattern with the same mutation. An example of such analyses in two families (SC. GM) with an R320L receptor mutation is shown in Figure 4. In affected members of the SC kindred the mutant receptor allele exhibits a different RFLP pattern (M,d) compared with that seen in kindred GM (m,D), confirming that the receptor abnormality has occurred independently in the two families. 199

M. Gurnell and V. K. K. Chutterje?.

Figure 4. Haplotype analysis from two kindreds (G.M. and S.C.) harbouring the R320L mutation, using Mspl and Oral RFLPs. The area surrounding each polymorphic site was amplified by PCR of unaffected (open symbols) and affected (filled symbols) family members. The absence (M,D) or presence (m,d) of each restriction endonuclease site was assessed after agarose gel electrophoresis and was used to determine the haplotype of the mutant allele in each kindred. In each family the haplotype associated with the mutation is ringed. Reproduced from ref. 11 with copyright permission of The American Society for Clinical Investigation.

200

8: Hormone-resistance syndromes

3. Ligand binding and DNA binding studies 3.1 Site-directed mutagenesis to generate mutated receptor cDNA Following the identification of a TRB mutation, it is necessary to test mutant receptor function in a number of assays. The technique of site-directed mutagenesis is used to introduce a particular nucleotide substitution into a wild-type cDNA background. A number of different mutagenic techniques exist. The method described here is well validated (2) and commercially available. The principles of this technique are outlined in Figure 5. The receptor cDNA is cloned into an M13-based vector enabling generation of phage. The latter are passaged through a mutant (dut, ung) strain of E. coli, resulting in the synthesis of a uracil-substituted single-stranded DNA template. A mutagenic oligonucleotide containing the mutation of interest (an example is shown in Figure 2) is then annealed to this template and primes the synthesis of the complementary strand using T7 DNA polymerase and T4 DNA ligase. This double-stranded heteroduplex is transformed into a second strain of E. coli containing uracil glycosylase. The uracil-containing strand is inactivated such that the majority of progeny phage are derived from the newly synthesized strand containing the desired mutation. Single-stranded DNA prepared from the progeny phage is sequenced to verify the presence of the required mutation. Such phage are used to infect further bacterial cultures to prepare double-stranded replicative form DNA from which the mutagenized TRB cDNA can be excised and cloned into various vectors.

3.2 Synthesis of mutant receptor protein and ligand binding assays The wild-type and mutant cDNAs are cloned into a plasmid (e.g. pGEM, Promega) that enables the generation of receptor proteins by coupled transcription and translation in vitro. The basis of this method is illustrated in Protocol 4. [35S]methionine is included in the reaction to generate radiolabelled protein. This is then analysed by SDS-PAGE to verify that the product is of authentic size, excluding the possibility that additional frame shift or premature stop mutations have been introduced during the mutagenesis procedure. This step also provides a measure of the efficiency of protein synthesis. Ligand binding assays are usually performed with unlabelled receptor proteins. Essentially, duplicate receptor aliquots are incubated with radiolabelled thyroid hormone ([125I]T3) alone and in the presence of increasing concentrations of unlabelled T3. The liganded receptor is then immobilized by vacuum filtration and specific radioligand binding measured to generate a displacement curve, which is then transformed by Scatchard analysis to derive an affinity constant (Ka) for hormone binding. A subset of mutant receptors may 201

M. Gurnell and V. K. K. Chatterjee

Figure 5. Strategy for site-directed mutagenesis of wild-type human TRB1 cDNA in M13mp18.

202

8: Hormone-resistance syndromes be markedly impaired for ligand binding: some may exhibit no specific binding with excess (KF6 M) cold T3; others show very reduced specific binding precluding quantitation of an affinity constant; a third group requires the use of more receptor protein in the binding assay to derive a Ka. Protocol 4.

In vitro [35S]methionine labelled transcription and translation of TR(31 proteins3

Equipment and reagents In vitro expression vectors containing wildtype or mutant TRB1 cDNA 10 uCi/ul [35S]methionine (NEN Life Science Products) SDS-PAGE equipment

Reticulocyte lysate (RL), T7 RNA polymerase, TnT buffer, RNasin, amino acid minus methionine mix, nuclease-free dH20 (TnT kit from Promega)

Method 1. To 1 ug of plasmid DNA, add 25 ul RL, 2 ul TnT buffer, 1 ul RNasin, 1 u1 amino acid minus methionine mix, 4 ul [35S]methionine, 1 ul T7 RNA polymerase, and make up to a final volume of 50 ul with nuclease-free dH2O. 2. Incubate at 30°C for 60 min. 3. Check products by running 2-4 ul of the translations (in duplicate) on SDS-PAGE and analyse by autoradiography. 8

If [35S]methionine labelling is not required, this should be substituted by a 'cold' amino acid mixture including methionine.

Protocol 5.

[125I]T3 binding assay

Equipment and reagents 0.2 uCi/ul [125I]T3 (Amersham); unlabelled ('cold') 10-2 M T3 stock (Sigma) Vacuum or equivalent filter system and filter membranes (0.45 um HA filters from Millipore)

T3 assay buffer: 20 mM Tris-HCI pH 8.0, 50 mM KCI, 1 mM MgCI2, 10% glycerol Wild-type and mutant TRB1 in vitro expression vectors

Method 1. In vitro translate wild-type and mutant receptor protein using 1 ug of plasmid template in a 50 ul reaction as outlined in Protocol 4.a 2. Prepare serial 'cold' T3 dilutions from 10-2 M stock solution. A typical range of concentrations might begin at 10~10 M and increase in a stepwise progression by a factor of two (e.g. 10-10 M, 2 x 10-10 M, 4 x 10-10 M, etc.) through to 10-8 M.b

203

M. Gurnell and V. K. K. Chatterjee Protocol 5.

Continued

3. Incubate in vitro translated wild-type or mutant protein with [125I]T3 in T3 assay buffer. For each receptor prepare a master mix containing 2.5 ul of 1 M DTT, 1 (ul of [125I]T3 (equating to a final concentration of 0.2 nM per reaction), and T3 assay buffer to a final volume of 450 (ul for n + 1 reactions. Reactions should be performed in duplicate for each receptor at each T3 concentration. Into each of two Eppendorf tubes aliquot 450 ul of this mix together with a further 50 ul of assay buffer to serve as total count controls. Then add the 50 ul protein translations to their respective master mixes. 4. For each concentration of 'cold' T3 add 450 ul of the receptor-[125l]T3 mix to 50 ul of the T3 dilution and incubate at 30°C for 60 min. 5. Set up vacuum and pre-wash filter papers with 2 ml of T3 assay buffer. 6. Carefully apply each receptor incubation to the centre of a filter paper and allow to drain by vacuum. 7. Wash with 3 x 2 ml of T3 assay buffer. 8. Transfer filter papers to appropriate tubes and count in a 7 counter. •For mutant receptors with a low T3 affinity it may be necessary to increase the amount of protein which is added to the final incubation mix. This does not however interfere with calculation of the receptor Ka,which is derived from the slope of the Scatchard plot analysis. bA +/- assay in the presence and absence of saturating cold T3 (10-6 M) to screen for nonbinding mutations should be performed prior to undertaking a full binding curve.

3.3 DNA binding properties of mutant receptors In vitro synthesized receptor proteins can also be used to assay the DNA binding properties of mutant receptors. Electrophoretic mobility shift assays (EMSA) are performed following incubation of 32P-labelled oligonucleotide duplexes corresponding to thyroid hormone-response elements (TREs) with wild-type and mutant receptor proteins (see also Chapter 2). With direct repeat and palindromic TREs, we find that TR binds optimally as a heterodimer with the retinoid X receptor (RXR). However, with an everted repeat TRE, it is possible to generate both TR homodimer and TR-RXR heterodimer complexes, with the addition of a submaximal amount of RXR (Figure 6). Although all TR mutants identified in RTH exhibit preserved DNA binding (3), some mutant receptors do exhibit altered function: a subset of mutants form homodimers less well than wild-type receptor (e.g. R316H; Figure 6). Wild-type TR homodimer complexes dissociate with the addition of ligand, whereas some mutant receptor homodimers dissociate less well or not at all, depending on their degree of impaired T3 binding (e.g. V264D; Figure 6). It has been suggested that mutants with impaired homodimer formation exhibit weaker dominant negative inhibition (4) and conversely that strong dominant 204

8: Hormone-resistance syndromes

Figure 6. Electrophoretic mobility shift assays (EMSAs) on an everted repeat TRE (F2 chick lysozyme promoter) with wild-type (WT) and mutant (V264D; R316H) receptors. RL denotes the non-programmed control lysate and RXR the retinoid X receptor. The position of RXR-TR heterodimers and TR-TR homodimers are shown.

negative activity of other mutants reflects their eonstitutive homodimeric DNA binding which is not reversible with T3 (5).

4. Assaying mutant receptor function by transient transfection 4.1 T3-dependent transactivation In keeping with other members of the nuclear receptor family. TR modulates transcription by binding to regulatory sequences (TREs) in the promoter regions of target genes. For those genes that are positively regulated in a T3dependent manner, consensus TREs consisting of two or more copies of the hexanucleotide motif AGGTCA arranged in direct repeat, palindrornic, or everlcd repeat configurations have been identified. In contrast, TREs within genes that are negatively regulated by TR in a hormone-dependent manner (e.g. TSHcx and TSHfl suhunits and TRH) are more complex and do not 205

M. Gurnell and V. K. K. Chatterjee appear to contain readily discernible repeats of the consensus sequence, although various half-sites have been identified. In order to study the functional properties of TR mutants, transient transfection assays are performed in a human choriocarcinoma cell line (JEG-3), which has previously been shown to be virtually devoid of functional endogenous TR (3). Receptor expression is driven by a vector which contains the TR3 cDNA cloned downstream of the Rous sarcoma virus (RSV) enhancer and promoter. This is co-transfected with a reporter gene: positively regulated reporters can consist of one or more copies of direct or everted repeat or palindromic TREs, cloned upstream of the thymidine kinase promoter and the firefly luciferase gene; negatively regulated reporter genes contain either the human TSHa subunit (-846 to +44) or TRH (-900 to +55) promoters linked to luciferase. Also included in each transfection is a plasmid encoding (3-galactosidase coupled to a powerful constitutive promoter (elongation factor la). This serves to correct for variations in the efficiency of transfection of different wells of cells. Prior to examining the transcriptional activity of a mutant receptor, a range finding experiment is performed to determine the optimal concentration of wild-type receptor expression vector required to elicit maximal T3-mediated transcriptional responses. Subsequent experiments are then performed using this quantity of wild-type and mutant expression vectors, ensuring that differences in reporter gene responses are not due to variations in relative expression of wild-type versus mutant receptor proteins. The methodology of transient transfection is summarized in Protocol 6 (see also Chapters 2 and 3). The typical transcriptional responses obtained with receptor mutants are shown in Figure 7 and fall into three categories: (a) Mutants with a slightly reduced hormone binding affinity exhibit type I profiles that are right-shifted but attain a maximum comparable to wildtype receptor at T3 concentrations sufficient to overcome the ligand binding defect. (b) Non-hormone binding mutants show virtually no transcriptional response (type III) except at high T3 concentrations. (c) Type II responses are characterized by reduced maximal transcriptional activation which is disproportionate to the impaired hormone binding affinity of the mutant receptor. Such a profile may signify altered receptor interaction with cofactors (see below).

4.2 Dominant negative inhibition by mutant receptors Consonant with the dominant mode of inheritance of RTH, affected individuals are heterozygous for mutations in the TRp gene. However, receptor haploinsufficency cannot account for the disorder as both humans and mice harbouring a deletion of one TRp allele are completely normal. Accordingly, 206

8: Hormone-resistance syndromes

Figure 7. T3-dependent transactivation curves for wild-type TR31 and three different types of mutant receptor. Values are expressed as the percentage of the maximal response achieved by the wild-type receptor.

Figure 8. Mutant thyroid receptor proteins inhibit the function of their wild-type counterparts in a dominant negative manner. Examples of the typical transcriptional responses of wild-type TRB1 (WT) at 1 nM or 1000 nM T3 are shown in the presence of equimolar concentrations of different types of mutant or additional wild-type receptor. Values are expressed as the percentage of the maximal response achieved by the wildtype receptor alone.

207

M. Gurnell and V. K. K. Charter/eg it was hypothesized that the mutant receptors in RTH were not simply transcriptionally impaired but also capable of inhibiting the function of their wild-type counterparts. Transient transfection experiments have successfully demonstrated such dominant negative inhibition (3). In such experiments a reporter gene is co-transfected with an optimal amount of wild-type receptor together with an equal quantity of either wild-type or mutant receptor expression vectors. Transcriptional responses are measured at low (1 nM) or high (1000 nM) T3 concentrations. Figure 8 illustrates schematically the responses observed with the different types of receptor mutants. At low T3 levels, all mutants exhibit dominant negative inhibition of wild-type receptor action. At high T3 concentrations, this inhibition is reversible with type I mutants, consistent with their normal transcriptional response. In contrast, inhibition by type III mutants is poorly reversible and type II mutants exhibit intermediate responses. Protocol 6. Transient transfection studies Reagents JEG-3 cell line (ATCC) Optimem supplemented with 2% fetal calf serum (FCS) and 1% penicillin/streptomycin/ fungizone (PSF) (all reagents supplied by Gibco BRU Charcoal treated PCS (see Chapter 11, Protocol 2) Trypsin/EDTA (Gibco BRL) Hepes-buffered saline (HBS): 21 mM Hepes pH 7.1, 136 mM NaCI, 5 mM KCI, 0.7 mM Na2HP04.2H20, 6 mM dextrose; adjust to pH 7.05 with 2 M NaOH and filter sterilize Gly-Gly buffer: 25 mM Gly-Gly pH 7.8, 15 mM MgSCv 4 mM EGTA

Cell lysis buffer: Gly-Gly buffer with 1% Triton X-100 and 1 mM DTT Luciferase assay buffer: Gly-Gly buffer with 16.5 mM KH2P04, 2.2 mM ATP, 0.1 mM DTT Luciferin solution: Gly-Gly buffer with 0.2 mM luciferin and 10 mM DTT p-Galactosidase assay buffer: 63 mM Na2HPO4.2H20, 14 mM NaH2P04.2H20, 1 rnM MgCI2, 14 mM p-mercaptoethanol, 0.9%(w/v)ONPG Eukaryotic expression vectors encoding wild-type and mutant receptors, reporter construct, and internal control plasmid (e.g. Bos-p-gal)

A. Cell plating 1. Trypsinize one 10 cm plate of JEG-3 cells (80-90% confluent) and resuspend in culture medium. Seed 1 cm wells of 24-well plates with 0.5 ml of resuspended cells to approx. 50% confluency. 2. Return to incubator for 4-6 h to allow attachment to plate. 3. Change to medium containing 2% charcoal treated PCS and 1% PSF, return to incubator, and leave overnight. B. Calcium phosphate-DNA precipitation 1. For each 0.5 cm well, prepare a mix containing plasmid DNAa encoding the RSV receptor expression vector, luciferase reporter construct, and Bos-p-gal, together with 50 ul of HBS (this is most conveniently done by preparing a master mix for the total number of 208

8: Hormone-resistance syndromes wells requiring transfection with any given plasmid combination). Mix by vortexing. 2. Slowly add 3 ul of 2 M CaCI2 per well to the mix drop by drop and shake gently. Leave at room temperature for 45 min. 3. During this period, wash the cells by replacing the overnight medium with fresh medium containing 1% PSF. 4. Immediately prior to transfection remove the wash medium. Then add 50 [i,\ of the calcium phosphate-DNA mixture dropwise to each well. 5. Add 0.5 ml of culture medium (with 2% charcoal treated PCS and 1% PSF) to each well and return to the incubator. Leave for 4-6 h. 6. Following this incubation, gently shake the plate to dislodge remaining precipitate from the cells and aspirate off the medium. Replace with 0.5 ml of culture medium (with 2% charcoal treated PCS and 1% PSF) containing T3 as required. 7. Return to incubator for a further 40 h. C. Cell lysis, luciferase and $-galactosidase assays 1. Remove culture medium and wash each well with 500 |xl of PBS. Carefully aspirate off PBS and replace with 200 ul of lysis buffer. Leave at room temperature for 10 min. 2. Transfer cell lysates to 1.5 ml Eppendorf tubes and microcentrifuge at 15000 g for 30 sec to pellet cell debris. 3. Add 300 ul of luciferase assay buffer to 100 ul of each cell extract in a 5 ml LPStube. 4. Immediately transfer to luminometer and measure light emission in the 10 sec period following addition of 100 ul of luciferin solution. 5. Add 150 ul of B-galactosidase buffer to 20 ul of each cell extract in a flatbottomed microtitre plate, cover, and incubate at 37°C for 30-60 min. 6. Determine relative activity in each well by measuring absorbance at 414 nm in a microtitre plate reader. a Ideally prepare plasmid stocks to a concentration of x ng/ul where V is the amount of plasmid required per well for the transfection. By minimizing the volumes added to the buffer, the HBS remains at approx. 1 x concentration, which is important in promoting transfection efficiency.

5. Interaction of mutant receptors with cofactors Most recently, it has become apparent that a subset of mutant receptors in RTH are functionally impaired principally as a consequence of altered receptor interaction with cofactors. In particular, some mutants which exhibit a type II transactivation profile (Figure 7), have been found to bind co-activators 209

M. GurneUand V. K. K. Chatterjee

Figure 9. Differential recruitment of the co-activator ACTR by wild-type TR01 (WT) and the mutant T277A when bound to DNA as a heterodimer with RXR. Using an electrophoretic mobility shift assay, in vitro translated TR (wild-type or T277A) and RXR were incubated with the malic enzyme DR F 4 TRE, along with bacterially expressed GSTACTR in the presence of increasing concentrations of T3. Reproduced from ref, 7 with copyright permission of the Oxford University Press.

less well using protein-protein interaction (pulldown) assays with GST-receptor fusions and radiolabelled co-activator proteins (6, 7), This methodology is described in detail in Chapters 3 and 4. Such interactions can also be examined by modifying the EMSA. When bacterially expressed and purified co-activator is added to TR, RXR, and a labelled TRE, it is possible to generate a higher order ternary complex in a hormone-dependent manner (Figure 9). Conversely, some mutants have been found to dissociate less readily than wild-type TR from co-reprcssors following the addition of ligand, in either pulldown (8) or gel mobility supershift analyses (9). It has been suggested that this is the minimal functional abnormality required for a mutant receptor to be associated with the RTH phenolypu.

References 1. Refetoff, S.. Weiss, R. E.. and Usala. S. J. (1TO). Endocr Rev., 14, 348. 2. Kunkel, T. A., Roberts. J. D., and Zakour. R, A. (1987). In Methods in enzymnlngy (ed. R. Wu and L, Grossman), Vol. 154, p. 367. Academic Press, London. 210

8: Hormone-resistance syndromes 3. Collingwood, T. N., Adams, M., Tone, Y., and Chatterjee, V. K. (1994). Mol Endocrinol, 8, 1262. 4. Kitajima, K., Nagaya, T., and Jameson, J. L. (1995). Thyroid, 5, 343. 5. Yen, P. M., Wilcox, E. C., Hayashi, Y., Refetoff, S., and Chin, W. W. (1995). Endocrinology, 136, 2845. 6. Collingwood, T. N., Rajanayagam, O., Adams, M., Wagner, R., Cavailles, V., Kalkhoven, E., etal. (1997). Proc. Nad. Acad. Sci. USA, 94, 248. 7. Collingwood, T. N., Wagner, R., Matthews, C. H., Clifton-Bligh, R. J., Gurnell, M., Rajanayagam, O., et al. (1998). EMBO J., 17, 4760. 8. Yoh, S., Chatterjee, V. K., and Privalsky, M. (1997). Mol. Endocrinol, 11, 470. 9. Clifton-Bligh, R. J., de Zegher, F., Wagner, R. L., Collingwood, T. N., Francois, I., Van Helvoirt, M., et al. (1998). Mol. Endocrinol., 12, 609. 10. Adams, M., Nagaya, T., Tone, Y., Jameson, J. L., and Chatterjee, V. K. (1992). Clin. Endocrinol, 36, 281. 11. Adams, M., Matthews, C., Collingwood, T. N., Tone, Y., Beck-Peccoz, P., and Chatterjee, V. K. (1994). J. Clin. Invest., 94, 506.

211

This page intentionally left blank

9

In vitro assembly of steroid receptor-Hsp90 complexes D. F. SMITH

1. Introduction 1.1 Background Receptors for the steroid hormones—oestrogens, androgens, progesterone, glucocorticoids, and mineralocorticoids—typically exist in heteromeric complexes prior to binding hormone (for a recent, comprehensive review see ref. 1). After hormone binding or certain non-hormonal signals, the receptor monomer dissociates from complexes and undergoes a series of biochemical changes that lead to its activation as a transcription factor. In early studies of steroid receptors, the unactivated receptor was commonly observed to migrate as an 8-10S complex by density gradient centrifugation while the activated receptor migrated at 4-5S, the approximate size of a receptor monomer or dimer. Initial efforts to purify unactivated receptor complexes were thwarted by the lack of rapid purification procedures and the instability of receptor complexes. Advances came with the use of steroid affinity resins and the discovery that molybdate and related oxyanions would stabilize receptor complexes against dissociation. Taking advantage of these reagents, several groups independently found that steroid receptor complexes contain a common 90 kDa non-receptor protein (2-4) that was later identified (5-7) as the major heat shock protein Hsp90. The development of specific monoclonal antibodies recognizing native receptor complexes allowed a more complete characterization of receptor complexes. In one of the first reports of immunoaffinity purified progesterone receptor (PR) complexes (8), five non-receptor proteins were identified that specifically co-precipitated with PR from chicken oviduct cytosol. These were Hsp90, the additional major heat shock protein Hsp70, two proteins of about 50-55 kDa, and a 23 kDa protein. Subsequent studies have shown that the 50 kDa proteins are two members of the FK506 binding family of immunophilins, FKBP51 and FKBP52. The 23 kDa receptorassociated protein, commonly termed p23, is a unique Hsp90 binding protein. Attempts for several years to assemble unactivated receptor complexes

D. F. Smith from isolated components were futile. For example, receptor complexes were immunoaffinity purified and treated with 0.3 M or greater ionic strengths to dissociate complexes; however, after lowering the ionic strength by dialysis, receptor complexes were not re-formed. The breakthrough in cell-free assembly of receptor complexes followed cloning of cDNAs encoding vertebrate steroid receptors. Pratt and his colleagues showed that glucocorticoid receptor (GR) expressed in a rabbit reticulocyte lysate-based in vitro synthesis system was associated with Hsp90. His interpretation was that Hsp90 only establishes interactions with the nascent receptor chain. Paradoxically, the 30 °C temperature used for in vitro synthesis of receptor and formation of Hsp90-GR complexes was known to cause dissociation of native receptor complexes in tissue extracts. Toft, reasoning that reticulocyte lysate (RL) may provide factors that favour assembly and maintenance of receptor complexes without the need for incompletely folded nascent receptor chains, subsequently showed that addition of purified PR monomers to RL under conditions used for cell-free translation did indeed reconstitute PR complexes (9). With a cell-free system available that assembles receptor complexes, the laboratories of David Toft, Bill Pratt, and myself have performed extensive analyses over the last few years to identify factors participating in assembly and to define biochemical mechanisms that underlie the assembly process. The story that has emerged reveals a highly co-ordinated, ordered pathway of assembly that involves multiple components in the cytoplasmic/nuclear molecular chaperone machinery.

1.2 Proteins associating with unactivated steroid receptors Before describing the assembly process, it will perhaps be useful to identify the protein players involved. Table 1 contains a list of proteins that have been identified in unactivated steroid receptor complexes. None of the proteins on this list appears to be a specific regulator or partner of steroid receptors, since all are expressed in many tissue types, often at high levels, and most have been identified in complexes with target proteins unrelated to steroid receptors. Notably, all of these proteins have general chaperone functions or are partner proteins in multi-chaperone complexes.

1.3 Basic pathway for assembly of steroid receptor complexes As illustrated by the model depicted in Figure 1, the initial protein interaction observed after addition of receptor to RL is the ATP-dependent binding of Hsp70. Genetic studies and more recent biochemical evidence points to a role for the Hsp70 co-chaperone DnaJ at this initial assembly step; however, since only substoichiometric amounts of DnaJ are detected in receptor Hsp70 com214

9: In vitro assembly of steroid receptor—Hsp90 complexes Table 1. Known proteins in unactivated steroid receptor complexes Protein (alternative name)

Notes on function

Hsp70 (Hsc70) DnaJ (Hsp40) Hip(p48) BAG-1 (Rap46, Hap46) Hop(Sti1,p60) Hsp90

Mediator of protein folding processes Co-chaperone with Hsp70; folding activity Hsp70 partner Hsp70 partner Partner of Hsp70 and Hsp90 Holding misfolded proteins in a folding-competent state; regulation of various signalling proteins Hsp90 partner; holding activity Hsp90 partner; PPIase;" holding activity Hsp90 partner; PPIase Hsp90 partner; PPIase; holding activity Hsp90 partner; protein phosphatase

P23

FKBP52 (p59, hsp56) FKBP51 (p54, FKBP54) Cyp40 (Cpr7, Cpr6) PP5 • Peptidylprolyl isomerase.

plexes, it may be that DnaJ functions in a more catalytic manner to promote Hsp70 interactions. Subsequent to formation of the early receptor Hsp70 complex, three additional chaperone components are incorporated to form an intermediate complex. Two of these, Hip and Hop, are proteins that interact with Hsp70 through respective tetratricopeptide repeat (TPR) domains. Hip binds the ATPase domain of Hsp70, only recognizing the ADP-bound form of Hsp70 (10,11). Since Hsp70 binding to a substrate—in this case the receptor— stimulates its ATPase activity, the delayed binding by Hip probably reflects substrate-dependent hydrolysis of ATP to ADP. Hop also binds Hsp70 in an ADP-dependent manner, but Hop's interaction with Hsp70 is probably localized to the C terminal one-third of Hsp70 that contains the substrate binding domain and downstream regulatory sequences (12). Interestingly, mutant forms of Hip have been generated that act as dominant inhibitors of PR assembly, arresting assembly after binding of Hip to Hsp70 PR complexes but prior to binding of Hop (13). Through two separate TPR domains, Hop can bind concomitantly to both Hsp70 and Hsp90 (14-16). If Hop function is blocked, Hsp90 association with receptor complexes is prevented (15, 17), so Hop may act as an adaptor to promote Hsp90 binding to pre-existing Hsp70 receptor complexes (18). There is a poorly understood transition from intermediate complexes containing Hsp70, Hip, Hop, and an Hsp90 dimer to mature complexes that contain Hsp90 and two additional Hsp90-associated proteins. One of the associated proteins is p23 which binds Hsp90 in an ATP-dependent manner (19, 20) and stabilizes Hsp90s interaction with receptor (21, 22). In the case of PR and GR complexes, formation of a mature complex with Hsp90 and p23 is required at physiological temperatures to maintain the receptor's ligand binding domain in a conformation that is competent for high affinity and high efficiency 215

D, F. Smith

Figure 1. Chaperone-mediated assembly of unactivated steroid receptor complexes. In vitro assembly studies have been used to define an ordered, dynamic, multistep pathway for assembly of functional receptor complexes. The components illustrated in this pathway are a receptor monomer (Rec.; dashed border indicates lack of hormone binding), Hsp70 (70), DnaJ (J), an Hsp90 dimer (90), Hip, Hop, p23 (23), and any one of the three immunophilins