UNIVERSITY OF CALIFORNIA, SAN DIEGO

Structural and Functional Analysis of the Chemokine CCL27 and the Expression and Purification of Silent Chemokine Receptors D6 and DARC

A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy

in

Chemistry and Biochemistry

by

Ariane L. Jansma

Committee in charge:

Professor Tracy M. Handel, Chair Professor Patricia Jennings, Co-Chair Professor William Gerwick Professor Susan Taylor Asst. Professor Faik A. Tezcan

2009

This Dissertation of Ariane L. Jansma is approved, and it is acceptable in quality and

form for publication on microfilm and electronically:

______

______

______

______Co-Chair

______Chair

iv

TABLE OF CONTENTS

Signature Page ...... iii

Table of Contents...... iv

List of Figures...... x

List of Tables...... xiii

List of Abbreviations...... xiv

Acknowledgements...... xvi

Curriculum Vitae...... xvii

Abstract of the Dissertation ...... xx

INTRODUCTION...... 1

CHAPTER 1...... 7

Introduction to Chemokines ...... 7

1.1 Chemokines and Chemokine Receptors...... 7

1.2 Chemokine Oligomerization and GAG Interaction ...... 12

1.3 Chemokines and Disease ...... 17

1.4 Biological Diversity of Chemokines ...... 19

1.5 Acknowledgements...... 23

CHAPTER 2...... 24

Optimization of the Expression and Purification of CCL27 and CCL28 ...... 24

2.1 Introduction to CCL27 and CCL28 ...... 24

2.2 Results ...... 26

2.2.1: Protein expression and purification ...... 26

2.2.2: Protein expression of CCL28, ubiquitin fusion ...... 26

2.2.3: Initial purification of CCL28 ...... 28

iv

2.2.4: Ubiquitin cleavage and HPLC purification...... 29

2.2.5: Preparation of MOPS media ...... 31

2.2.6: Confirmation of CCL27 and CCL28 function...... 33

CHAPTER 3...... 35

NMR Analysis of the Structure, Dynamic and Unique Oligomerization Properties of the

Human Chemokine CCL27 ...... 35

3.1 Summary...... 35

3.2 Introduction ...... 37

3.3 Results ...... 41

3.3.1: CCL27 oligomerizes, forming a tetramer at high concentrations ...... 41

3.3.2: NMR assignment and structure determination of the monomeric form...... 51

3.3.3: Characterization of internal dynamics ...... 55

3.3.4: Attempts to characterize the CCL27 dimer interface ...... 56

3.3.5: Analysis of mutants of CCL27 by PFG diffusion NMR ...... 67

3.3.6: The interaction with glycosaminoglycans induces oligomerization ...... 70

3.4 Discussion...... 75

3.5 Conclusions...... 79

3.6 Materials and Methods...... 80

3.6.1: Protein expression and purification ...... 80

3.6.2: NMR spectroscopy...... 80

3.6.3: Heparin binding assay...... 87

3.6.4: Solubility assays...... 88

3.6.5: Chemical cross-linking ...... 88

3.7 Acknowledgements...... 89

CHAPTER 4...... 90

v

Characterization of the Interaction of CCL27 with Glycosaminoglycans and with the

Human CCR10 Chemokine Receptor ...... 90

4.1 Summary...... 90

4.2 Introduction ...... 92

4.3 Results ...... 96

4.3.1: Comparison of CCL27 and CCL28 ...... 96

4.3.2: Addition of an amino-terminal Phenylalanine to CCL27...... 97

4.3.3: Amino-terminal modifications of CCL27 produce partial agonists...... 100

4.3.4: The chemical nature of the N-terminus is not critical for signaling...... 102

4.3.5: Deletion of N-terminal residues results in antagonist...... 103

4.3.6: C-terminal truncation of CCL27 results in a partial agonist...... 106

4.3.7: CCL27 interactions with Heparin tetrasaccharide ...... 109

4.3.8: Residue specific identification of the Heparin binding sites for CCL27 .....113

4.3.9: Analysis of transendothelial migration...... 116

4.4 Discussion...... 118

4.5 Conclusion ...... 123

4.6 Materials and Methods...... 124

4.6.1: Expression and purification of WT and mutant CCL27 and CCL28 ...... 124

4.6.2: Cell culture ...... 124

4.6.3: Chemotaxis assays ...... 124

4.6.4: Calcium flux and desensitization assays...... 125

4.6.5: In vitro Heparin binding assays ...... 125

4.6.6: NMR analysis, 15N-1H HSQC and chemical shift perturbation ...... 126

4.7 Acknowledgements...... 128

vi

CHAPTER 5 ...... 129

Preliminary Functional Analysis of CCL28 Suggests a Role for the C-terminal Region

Independent of Leukocyte Migration ...... 129

5.1 Summary...... 129

5.2 Introduction ...... 130

5.3 Results ...... 132

5.3.1: The C-terminal region of CCL28 is not necessary for cellular migration...132

5.3.2: Chemical cross-linking experiments...... 134

5.4 Discussion and Future Directions ...... 136

CHAPTER 6...... 137

Investigation of the Effect of Glycosaminoglycans on...... 137

CXCL11 Oligomerization...... 137

6.1 Summary...... 137

6.2 Introduction ...... 138

6.3 Results ...... 140

6.3.1: 1H-15N HSQC spectra reveal conformational heterogeneity in CXCL11 ...140

6.3.2: PFG diffusion analysis indicates CXCL11 is dimeric ...... 141

6.3.3: GAG binding analysis by HSQC chemical shift perturbation ...... 143

6.3.4: Solubility analysis of CXCL11 with Heparin octasaccharide ...... 146

6.4 Discussion...... 147

6.5 Materials and Methods...... 149

6.5.1: NMR spectroscopy...... 149

6.5.2: Solubility assays...... 149

6.6 Acknowledgements...... 150

CHAPTER 7...... 151

vii

Expression and Purification of Silent Chemokine Receptors D6 and DARC ...... 151

7.1 Summary...... 151

7.2 Introduction to Silent Chemokine Receptors...... 153

7.2.1: The D6 chemokine receptor...... 153

7.2.2: The Duffy Antigen Receptor for Chemokines (DARC) ...... 154

7.2.3: Tetracycline-inducible mammalian cell expression system...... 155

7.3 Results ...... 157

7.3.1: Expression and purification of D6 ...... 157

7.3.2 Expression and purification of DARC...... 160

7.4 Discussion and Future Directions ...... 164

CHAPTER 8...... 167

Conclusions...... 167

APPENDIX I...... 170

PROTOCOLS...... 170

A. PFG diffusion NMR by Diffusion Ordered Spectroscopy (DOSY) ...... 170

A.1: Background...... 170

A.2: Calibrations...... 171

A.3: Parameter optimization...... 172

A.4: Data acquisition ...... 172

A.5: Data processing, pseudo-2D using DOSY plot...... 173

A.6: Data processing, calculation of the self-diffusion coefficient, Ds...... 174

A.7: Applications ...... 177

B: Migration assay, CCL27 with L1.2 cells expressing CCR10...... 180

C: Transendothelial migration assay...... 182

C.1 Human Umbilical Endothelial Cell (HUVEC) Maintenance...... 182

viii

C.2 Trans Endothelial Migration Assay ...... 183

D: Calcium flux and receptor desensitzation assays...... 186

E: Ubiquitinase purification...... 187

E.1: Culture ...... 187

E.2: Harvest ...... 187

E.3: Purification...... 187

F: D6 and DARC purification from HEK Cells ...... 190

APPENDIX II...... 192

Preliminary Transendothelial Migration Assay ...... 192

APPENDIX III...... 194

Transendothelial Migration Assay Control Experiments ...... 194

REFERENCES...... 196

ix

LIST OF FIGURES

Figure 1.1 Examples of monomeric chemokines ...... 8

Figure 1.2 Model cartoon of a ligand bound to a receptor ...... 9

Figure 1.3 Illustration of leukocyte migration...... 10

Figure 1.4 Chemokine oligomeric structures...... 13

Figure 1.5 CCL2 tetramer with dodecasaccharide...... 14

Figure 1.6 Chemokine receptor binding epitopes ...... 20

Figure 1.7 Examples of GAG binding regions for oligomeric chemokines ...... 22

Figure 2.1 Sequence alignment of CCL27 and CCL28...... 24

Figure 2.2 Schematic illustration of ubiquitin fusion expression system ...... 25

Figure 2.3 Expression of ubiquitin-CCL28 ...... 27

Figure 2.4 FPLC chromatogram of ubiquitin-CCL28 after concentration ...... 29

Figure 2.5 Ubiquitin cleavage reaction and purified HPLC fractions for CCL28 ...... 30

Figure 2.6 HPLC chromatograms of purified CCL28 and CCL27 ...... 31

Figure 2.7 ESI mass spectrometry results for 15N and 13C-CCL27 ...... 32

Figure 2.8 Migration assay for CCL27 and CCL28 with CCR10 ...... 34

Figure 3.1 Examples of chemokine oligomerization...... 38

Figure 3.2 Assigned HSQC for CCL27 ...... 42

Figure 3.3 PFG diffusion analysis for standard proteins and CCL27 ...... 44

Figure 3.4 Buffer conditions for CCL27 FPG diffusion ...... 45

Figure 3.5 R1 values for 1.0 mM CCL27 ...... 49

Figure 3.6 Monomeric structure of CCL27 by NMR ...... 54

Figure 3.7 Order parameters...... 56

Figure 3.8 HCNH-NOE analysis of CCL2 and CCL27 at 3.0 mM ...... 58

x

Figure 3.9 HCNH-NOE of CCL27 at 2.0 mM ...... 59

Figure 3.10 Chemical shift perturbation analysis of CCL2 and CCL27...... 61

Figure 3.11 Results from CSP mapped to dimer structures and models ...... 63

Figure 3.12 Exchange contributions for 1.0 mM CCL27 ...... 65

Figure 3.13 PFG diffusion analysis of mutant CCL27 ...... 68

Figure 3.14 Effects of GAGs on CCL27 and CCL2 oligomerization...... 71

Figure 3.15 Chemical crosslinking of CCL27 ...... 73

Figure 4.1 Comparison of CCL27 and CCL28 ...... 97

Figure 4.2 Functional analysis of F-CCL27...... 99

Figure 4.3 Functional analysis of N-terminal mutants ...... 101

Figure 4.4 CCL27 mutation analogous to CCL28 ...... 103

Figure 4.5 Functional analysis of N-terminal truncation mutants ...... 105

Figure 4.6 HSQC spectra of WT and [1-73] CCL27...... 107

Figure 4.7 Functional analysis of [1-73] CCL27 ...... 108

Figure 4.8 Chemical shift perturbation analysis of CCL27 with tetrasaccharide ...... 110

Figure 4.9 GAG binding sites mapped to monomeric CCL27 structure ...... 112

Figure 4.10 Heparin affinity assay...... 115

Figure 4.11 Transfilter and transendothelial migration assays...... 117

Figure 4.12 Transendothelial migration assays of WT and K25A CCL27...... 117

Figure 4.13 GAG binding sites mapped to dimer models of CCL27 ...... 121

Figure 5.1 Migration assay of WT and [1-81] CCL28...... 132

Figure 5.2 Limited proteolysis analysis of CCL28 ...... 133

Figure 5.3 Chemical crosslinking of WT and [1-81] with decasaccharide...... 134

Figure 6.1 HSQC of CXCL11 ...... 141

Figure 6.2 Oligomerization profile for CXCL11 by PFG diffusion NMR...... 142

xi

Figure 6.3 Effect of phosphate on CXCL11 ...... 143

Figure 6.4 CXCL11 titration with Heparin tetrasaccharide ...... 144

Figure 6.5 CXCL11 titration with Arixtra...... 145

Figure 6.6 Solubility analysis of CXCL11 with Heparin octasaccharide...... 146

Figure 7.1 Schematic illustration of the chemokine receptor DARC ...... 154

Figure 7.2 Schematic representation of protein expression...... 156

Figure 7.3 Surface expression of D6 on HEK293 cells ...... 158

Figure 7.4 Fluorescence microscopy using an Anti-D6 antiboty ...... 159

Figure 7.5 Expression and purification of D6 from HEK293 cells ...... 160

Figure 7.6 Western blot analysis of DARC with anti-His antibody ...... 160

Figure 7.7 Western blot analysis of DARC with anti-HA antibody...... 161

Figure A.1 Gradient calibration window by “gradpar”...... 171

Figure A.2 FIDs acquired with increasing gradient amplitude, 2 – 95%...... 173

Figure A.3 Example of DOSY parameters ...... 173

Figure A.4 Example of a pseudo 2D spectrum processed by DOSY plot ...... 174

Figure A.5 DOSY processing window ...... 175

Figure A.6 Spectrum resultsing from xf2 processing ...... 175

Figure A.7 Display for calculating the diffusion coefficient ...... 176

Figure A.8 Example of a fit decay curve ...... 176

Figure A.9 Results from polymer-assisted DOSY analysis ...... 178

Figure A.10 Overlaid pseudo 2D spectra for CCL27 at several concentrations ...... 179

Figure A.11 Oligomerization profile for CCL27 ...... 179

Figure C.1 Evan’s blue transendothelial migration assay ...... 185

Figure II.1 Priliminary transendothelial migration assay...... 192

Figure III.1 Resistance control experiments for HUVEC treated filters ...... 195

xii

LIST OF TABLES

Table 1.1 Chemokines and disease...... 17

Table 1.2 Chemokines and chemokine receptors...... 19

Table 2.1 Buffer conditions for protein purification...... 28

Table 2.2 Final yield of proteins ...... 33

Table 3.1 Hydrodynamic radii calculated by HdyroNMR and HycroPro...... 47

Table 3.2 Results from fitting rotational diffusion tensor ...... 50

Table 3.3 Experiment restraints and structural statistics...... 53

Table 3.4 Principle components for anisotropic diffusion tensors ...... 67

Table A.1 Optimized PFG diffusion pulse sequences...... 171

xiii

LIST OF ABBREVIATIONS

Ala………………………………………………………………………………………….Alanine

Arg…………………………………………………………………………………………Arginine

ARIA………………………………………….Ambiguous Restraints for Iterative Assignment

Asp…………………………………………………………………………………….Asparagine

AUC……………………………………………………………….Analytical Ultracentrifugation

CMV……………………………………………………………………………..Cytomegalovirus

CPMG……………………………………………………………….Carr-Purcell-Meibloom-Gill

CSA…………………………………………………………………..Chemical Shift Anisotropy

CTACK……………………………………………….Cutaneous T-cell Attracting Chemokine

Cys………………………………………………………………………………………..Cysteine

Ds………………………………………………………………………..Self diffusion coefficient

DARC…………………………………………………Duffy Antigen Receptor of Chemokines

DOSY…………………………………………………………Diffusion Ordered Spectroscopy

EGS…………………………………….Sulfo-ethylene glycolbis(sulfosuccinimidylsuccinate

ESI……………………………………………………………………….Electrospray Ionization

FPLC………………………………………………Fast Performance Liquid Chromatography

GAG…………………………………………………………………………Glycosaminoglycan

Gln………………………………………………………………………………………Glutamine

Glu………………………………………………………………………………………Glutamate

Gly………………………………………………………………………………………….Glycine

GPCR……………………………………………………………..G-Protein Coupled Receptor

HEK293………………………………………………………..Human Embryonic Kidney cells

His………………………………………………………………………………………...Histidine

xiv

HIV………………………………………………...…………..Human Immunodeficiency Virus

HPLC……………………………………………...High Performance Liquid Chromatography

HSQC…………………………………………….Heteronuclear Single Quantum Coherence

Ile………………………………………………………………………………………..Isoleucine

L1.2 cells……………………………………………………………….Mouse Lymphoma cells

Leu…………………………………………………………………………………………Leucine

Lys…………………………………………………………………………………………..Lysine

MALDI…………………………………………...Matrix Assisted Laser Desorption Ionization

MEC……………………………………………………………..Mucosal Epithelial Chemokine

MS……………………………………………………………………………...Multiple Sclerosis

NaCl……………………………………………………………………………..Sodium Chloride

NOE …………………………………………………………………Nuclear Overhauser Effect

NMR……………………………………………………………...Nuclear Magnetic Resonance

PFG………………………………………………………………………..Pulsed Field Gradient

Phe…………………………………………………………………………………Phenylalanine

Pro…………………………………………………………………………………………..Proline

RMSD…………………………………………………………….Root Mean Square Deviation

Ser…………………………………………………………………………………………..Serine

Sulfo-EGS……………………………...Sulfo-ethylene glycolbis(sulfosuccinimidylsuccinate

Thr………………………………………………………………………………………Threonine

Trp……………………………………………………………………………………..Tryptophan

Ub-“Chemokine”……………………………………………………Ubiquitinated “Chemokine”

WT………………………………………………………………………………………Wild Type

xv

ACKNOWLEDGEMENTS

Chapter 1.2, in part, was published in Methods in Enzymology: Jansma, A.;

Handel, T.; M. Hamel, D. Homo- and hetero-oligomerization of chemokines. Methods

Enzymol. 2009, 461, 31 - 50

Chapter 3 is currently being prepared for the submission of publication of the material. The following are the contributing authors: Jansma, A.; Kirkpatrick, J.; Hsu, A.;

Handel, T. M.; Nietlispach, D.

Chapter 4 is currently being prepared for the submission of publication of the material. The following are the contributing authors: Jansma, A.; Hsu, A.; Handel, T.

Chapter 6 is, in part, has been submitted for publication of the material. This chapter reflects my contributions to this work. The following are the contributing authors:

Sielaff, I.; Gaudry, J. P.; Johnson, Z.; Kungl, A.; Gesslbaur, B.; Mulloy, B.;Power, C.;

Proudfoot, A. I. E.; Handel, T. M.,

Special thanks also goes to present and past members of the Handel Laboratory,

Marie-Hèléne Tremblay, Susan Crown, Andro Hsu, Samantha Allen, Melinda Hanes,

Rina Salanga, and Morgan O’Hayre. Thanks also goes to Gouri Ghosh, Jim Esko, and

Elizabeth Komives from UCSD, Additionally, thanks to Nicole Kruse from Bruker- biospin, and David Jones from GNF. This work was funded by the NIH through the

Molecular Biophysics Training Grant (GM08326) awarded to AJ, NIH (RO1-AI37113) and NIH (RO1- AI37113) awards to TMH.

xvi

CURRICULUM VITAE

Education

PhD, Biochemistry University of California, San Diego San Diego, CA 2004 – 2009 Handel Laboratory

Dissertation: Structural and Functional Analysis of the Chemokine CCL27  NMR analysis of both Monomeric and Oligomeric forms of WT and mutant CCL27 o Multi-dimensional 15N-1H, 13C-1H NMR experiments on single and double- labeled samples o PFG diffusion experiments on unlabeled samples to measure diffusion coefficients o Development of 13C-edited PFG diffusion experiment to measure GAG- induced oligomerization  Functional analysis o WT and mutant CCL27 to determine receptor binding epitopes and design of antagonist forms o Chemotaxis, transendothelial migration, and calcium flux with chemokine receptor CCR10 o GAG binding studies using affinity chromatography, solubility assays, as well as HSQC chemical shift perturbation analysis and edited PFG diffusion by NMR

MS Analytical Chemistry San Diego State University San Diego, CA 2001 – 2004

Thesis: Development and Implementation of a Micro-Coil Capillary Flow NMR System in a Drug Discovery Environment

BS Chemistry Pepperdine University Malibu, CA 1996 – 2001

BA Spanish, Cum Laude Pepperdine University Malibu, CA 1997 – 2002

xvii

Research Experience

UCSD School of Pharmacy San Diego, CA 2007 – 2009 Acting NMR Facilitator  Coordinated set-up of two 600 MHz systems o Bruker 5.0mm Cryo and 1.7mm CryoProbes  Implemented IconNMR Automation Software for natural product chemists  Set up and optimize new experiments as necessary  Train all users on both IconNMR and Topspin  Troubleshooting and routine maintenance of hardware and software  Interface with both Bruker and Varian engineers

Genomics Institute of the Novartis Foundation (GNF) San Diego, CA 2002 – 2004 Research Associate, NMR Spectroscopy  Facilitator for three Bruker NMR systems o 600 MHz NMR with 5.0mm CryoProbe o 400 MHz NMR system with room temperature QNP probe and automation platform o 400 MHz NMR system with micro-coil flow probe and automation platform  Small molecule structure elucidation support for medicinal chemistry  Train new users in IconNMR and XwinNMR  Implement and optimize new experiments as necessary  Troubleshooting and routine maintenance of hardware and software  Development and implementation of automated micro-coil flow NMR system  Metabolomics studies of low MW components in blood serum

DuPont Pharmaceuticals/Deltagen Research San Diego, CA 2001 – 2002 Contracted NMR Spectroscopist  Facilitator for two Varian NMR systems o 500 MHz NMR with interchangeable 5.0mm, 3.0mm and flow probes o 300 MHz NMR with automation platform  Small molecule structure elucidation in support of medicinal chemistry  Troubleshooting and routine maintenance of hardware and software  Quantitative analysis of compound libraries via automated flow NMR

Fellowships and Awards

 Molecular Biophysics Training Grant, NIH 2006 - 2008  Teaching Assistant Excellence Award o Dept. of Chem/Biochem., UCSD 2006  Acting NMR Facilitator for chemistry, Pacific Hall o Dept. of Chem/Biochem., UCSD 2004 – 2005

xviii

Bibliography

1. Jansma, A.; Kirkpatrick, J.; Hsu, A.; Handel, T. M.; Nietlspach, D. NMR Analysis of the Structure, Dynamics, and Unique Oligomerization Properties of the Human Chemokine CCL27, 2009, Submitted to Journal of Biological Chemistry 2. Jansma, A.; Hsu, A.; Handel, T. M. Characterization of the Interaction of CCL27 with Glycosaminoglycans and the Human Chemokine Receptor CCR10, 2009, Manuscript in advanced preparation 3. Severin, I.C.; Gaudry, J. P.; Johnson, Z.; Kungl, A.; Jansma, A.; Gesslbauer, B.; Mulloy, B.; Power, C.; Proudfoot, A. I. E.; Handel, T. M., CXCL11 requires interactions with glycosaminoglycans in vivo and has an unusually high affinity for Heparin. 2009, Submitted to Journal of Biological Chemistry 4. Jansma, A.; Handel, T.; M. Hamel, D. Homo- and hetero-oligomerization of chemokines. Methods Enzymol. 2009, 461, 31 - 50 5. Winter, J. M.; Jansma, A.; Handel, T. M.; Moore, B. S. Formation of the Pyridazine Natural Product Azamerone by Biosynthetic Rearrangement of an Aryl Diazoketone. Angewandte Chemie, 2008, 48 (4), 767-770 6. Jansma, A.; Zhang, Q.; Li, B.; Ding, Q.; Uno, T.; Bursrlaya, B.; Liu, Y.; Furet, P.; Gray, N.; Geierstanger, B. Verification of a Designed Intramolecular Hydrogen Bond in a Drug Scaffold by Nuclear Magnetic Resonance Spectroscopy. J. Med. Chem. 2007, 50 (24), 5875 - 5877 7. Jansma, A.; Chuan, T.; Albrecht, R. W.; Olson, D. L.; Peck, T. L.; Geierstanger, B. H. Automated Microflow NMR: Routine Analysis of Five-Microliter Samples Anal. Chem. 2005, 77, 6509 – 6515

xix

ABSTRACT OF THE DISSERTATION

Structural and Functional Analysis of the Chemokine CCL27 and

Optimization of the Expression and Purification of Silent

Chemokine Receptors

by

Ariane L. Jansma

Doctor of Philosophy

University of California, San Diego, 2009

Tracy M. Handel, Chair

Patricia Jennings, Co-Chair

Chemokines are small chemoattractant proteins that function by binding to G- protein coupled receptors (GPCRs) on a wide variety of cell types, triggering cascades of intracellular signaling pathways. While they are best known for their role in leukocyte migration, both in standard immune surveillance and maintenance, as well as in response to inflammation, chemokines are involved in a variety of physiological and pathophysiological processes. In addition to their complex function, the chemokine network itself is highly complex, with some chemokines being specific to one receptor, while others activate multiple receptors expressed on different cell types and in some

xx

cases, resulting in very different cellular responses. This project in part involves the chemokine CCL27, which is expressed in skin and selectively chemoattracts CLA+ memory T cells expressing the chemokine receptor, CCR10. The first set of aims for this project involved a comprehensive analysis of the structural and functional mechanism contributing to the biological diversity of CCL27. This was accomplished through a biophysical characterization of the oligomerization properties, residues targeting receptor activation, as well as sites of glycosaminoglycan (GAG) interactions for this chemokine. The results suggest that CCL27 exists in multiple oligomeric states, and its unique oligomerization patterns appear to play a role in the diversity of its multiple binding partners. The second major project aim involves optimization of the expression and purification of the silent chemokine receptors D6 and the Duffy Antigen

Receptor for Chemokines (DARC). D6 and DARC are termed silent receptors because they bind many chemokines with high affinity and specificity, but unlike other chemokine receptors, they do not signal through G-proteins. Instead, they act as regulators, either by targeting their chemokine ligands for degradation, or by shuttling them from one location to another. One of the rate-limiting steps in the biophysical characterization of

7-transmembrane helical chemokine receptors is obtaining sufficient amounts of purified, functional protein. In this project, a tetracycline-indicuble mammalian cell expression system is applied to both D6 and DARC. Results from the initial test expressions and purifications indicate that this method was successful in generating solubilized receptor protein.

xxi

INTRODUCTION

Project Overview

Structural and Functional Analysis of the Chemokine CCL27

CCL27 is a member of the chemokine family, composed of small chemoattractant proteins that function by binding to G-protein coupled receptors

(GPCRs) on virtually all cell types, triggering cascades of intracellular signaling pathways. While they are involved in a variety of physiological and pathophysiological processes, chemokines are best known for their role in leukocyte migration, as part of standard immune system maintenance as well as in response to inflammation. The chemokine network is highly complex, with some chemokines being specific to one receptor, while others activate multiple receptors expressed on different cell types and in some cases, elicit very different cellular responses. CCL27 is expressed in skin and selectively chemoattracts CLA+ memory T cells expressing the chemokine receptor,

CCR10. Overexpression of both CCL27 and CCR10 has been implicated in melanoma

metastasis to secondary sites within the skin. The aims of this project involved gaining a

better understanding of the biological diversity of CCL27 through a biophysical analysis

of its oligomerization properties, receptor activation, and interactions with

glycosaminoglycans (GAGs).

1 2

NMR Analysis of the Structure, Dynamics, and Unique Oligomerization

Properties of the Human Chemokine CCL27

The first aim of this project was to investigate the structure and dynamics of

CCL27 using NMR spectroscopy. In addition, because many chemokines oligomerize as part of their in vivo function, and the different types of oligomeric structures contribute to diversity in terms of GAG specificity, the goal was to study the oligomeric behavior of

CCL27. The first step involved developing a method, described in Chapter 2, to express, purify, and refold milligram quantities of both unlabeled and isotopically labeled functional protein. The oligomeric properties of CCL27 were then examined using pulsed field gradient (PFG) NMR diffusion experiments, and the results indicated that

CCL27 transitions between a monomer, dimer and tetramer species over a relatively narrow concentration range. This data, in conjunction with 15N-1H HSQC chemical shift perturbations, filtered (HC)NH-NOEs, and site-directed mutagenesis, suggest that several equilibria involving different interfaces seem to be simultaneously at work. The overall results of this study indicate that while the monomeric form of CCL27 has the standard chemokine motif, it oligomerizes up to a tetramer by itself in solution, albeit weakly. The presence of the GAG heparin, however, shifts the equilibrium to the oligomeric state, suggesting an important role for the oligomeric form in GAG binding. It is possible that CCL27 is able to adopt different oligomeric structures, depending on its

GAG (or other) binding partners, thereby increasing the functional diversity of this chemokine. This study, presented in Chapter 3, forms the basis of a manuscript currently in advanced preparation. In addition, this work resulted in an invitation to write a chapter focusing on chemokine oligomerization for the 2009 Methods in Enzymology

volume on chemokines, part of which is presented in Chapter 1.2.

3

Characterization of the Interaction of CCL27 with the Human

CCR10 Chemokine Receptor

An interesting area of chemokine function is receptor specificity. As stated previously, this represents a very complex field, with some chemokines specific to one receptor, while others activate multiple receptors, expressed on diverse cells types.

Because CCL27 is highly specific to the receptor CCR10, the second aim of this project was to determine the structural properties of CCL27 that induce activation of CCR10.

Through targeted mutagenesis, coupled to cellular migration assays, along with calcium influx, and receptor desensitization assays, several mutants were isolated with both partial agonist and antagonist properties. In addition, the results indicate that the N- terminal phenylalanine residue is necessary for optimal function, with the addition of a second phenylalanine at the N-terminus resulting in a super-agonist, with a 10-fold increase in activity. The results also indicate that the C-terminal region of CCL27 plays a significant role in receptor activation. The latter is interesting since traditionally, it is the N-terminal region of chemokines that is responsible for receptor activation.

Comparing these results to CCL28, the nearest sequential homolog to CCL27, and the only other known ligand for CCR10, it appears that chemokine receptor specificity cannot be determined by the primary sequence alone. These results are presented here in Chapter 4 and form the basis of a manuscript currently in preparation. In addition, this project led to a preliminary functional analysis of CCL28, discussed in Chapter 5.

4

Identification of the Glycosaminoglycan Binding Sites for the

Human Chemokine CCL27

Interaction with GAGs is crucial for chemokines to successfully direct in vivo

leukocyte migration. In addition, it has been shown with several chemokines that GAG

binding stabilizes the oligomeric form. Given the heterogeneous expression and

structural diversity of GAGs, these interactions are believed to contribute to chemokine

tissue specificity and biological diversity. As such, the third aim of this project involved

analysis of the GAG binding regions of CCL27. Previous studies using a binding assay,

with Heparin as the representative GAG, revealed that single alanine mutations to

several surface exposed, positively charged residues, particularly K25A, significantly

compromised Heparin affinity. The K25A mutant also showed decreased activity in

terms of transendothelial cellular migration, suggesting GAG binding is directly related to

CCL27 function. Finally, the GAG binding regions were confirmed by a more global

approach utilizing 15N-1H HSQC chemical shift perturbation analysis, combined with titration of the GAGs Heparin tetrasaccharide and Heparin hexasaccharide. Mapping the results from these titrations onto the surface of CCL27 dimer models suggests that multiple oligomeric structures may be stabilized by interaction with different GAGs.

Based on the analysis of CCL27 presented in Chapter 3, which suggests that this chemokine exists in multiple oligomeric states, these results lend credence to the proposed hypothesis that CCL27 may adopt different oligomeric forms depending on its

GAG binding partner. The results of this study are presented in Chapter 4, and form the second part of the manuscript currently in preparation detailing the interaction of CCL27 with the receptor CCR10. In addition, some of the methods developed for this project were used as part of a GAG-binding study of the chemokine CXCL11 (ITAC). The

5 details of my contributions to this analysis, some of which are included in a manuscript currently in advanced preparation, are discussed in Chapter 6.

6

Expression and Purification of the Silent Chemokine Receptors

D6 and DARC

The final aim of this project involved the expression and purification of the silent chemokine receptors D6 and the Duffy Antigen Receptor for Chemokines (DARC). D6 and DARC are termed silent receptors because they bind many chemokines with high affinity and specificity, but unlike other chemokine receptors, they do not signal through

G-proteins. They instead act as regulators, either by targeting their chemokine ligands for degradation, or by shuttling them from one location to another. This project involved optimizing protein expression by transfecting both D6 and DARC into a tetracycline- inducible mammalian cell expression system. This technique, similar to bacterial expression, allows the cells grow to confluency prior to the induction of protein expression. This helps to minimize cellular toxicity normally associated with the expression of these large seven trans-membrane helical proteins. Both D6 and DARC were purified from these mammalian cells and subsequently solubilized, demonstrating several advantages of this method over other expression systems. These results are presented in Chapter 7.

CHAPTER 1

Introduction to Chemokines

1.1 Chemokines and Chemokine Receptors

Chemokines are small (~8-12 kDa) chemoattractant proteins involved in a multitude of physiological and pathophysiological processes. They are most commonly known for their role in directing the movement of leukocytes as part of routine immune surveillance as well as in response to inflammation. It has been shown that inappropriate expression of these ligands or their receptors contributes to a broad spectrum of diseases, including a wide variety of inflammatory disorders, cancer metastasis, and HIV [1-5]. Chemokines function by binding to seven transmembrane G- protein coupled receptors (GPCRs) expressed on the surface of virtually all immune cell types. This binding event, results in conformational changes to the receptor that trigger cascades of intracellular signaling pathways involved primarily in G-protein activation and cellular movement [1-6].

Chemokines are categorized into four subfamilies (CC, CXC, CX3C, and C) based on the relative position of their conserved N-terminal cysteine residues, which form disulfide bonds with conserved cysteines found elsewhere in the primary sequence

[7-9]. There are currently 50 known human chemokines and 30 known receptors. While chemokines overall have relatively low sequence homology, their monomeric structures are very similar, consisting of an unstructured N-terminal region, containing the

7 8

“signaling domain,” involved in receptor activation. This is followed by a long loop region ending in a 310 helix, three anti-parallel β-strands, and a C-terminal α-helix. The disulfide bonds help to stabilize the overall tertiary structure (Figure 1.1) [7, 9-11].

A

NT 180º

B

180º

NT

Figure 1.1 Examples of monomeric chemokines. The structural motif is the same for both CC and CXC chemokines. A: Monomeric CCL2, B: Monomeric CXCL10

Despite their conserved monomeric structures, many chemokines have also been shown

to form dimers and higher order oligomers, which are considerably more varied. The

structural details and functional relevance of chemokine oligomerization will be

discussed in Chapter 1.2.

Chemokines bind and activate chemokine receptors in the monomeric form. It

has been postulated that the process of receptor activation involves a two-step process

[7, 12, 13]. The basic residues on the chemokine surface interact with acidic residues

9 on the extracellular loop regions of the receptor, enabling the N-terminus of the chemokine to be inserted into the receptor’s intermembrane helical bundle (Figure 1.2).

N-term.

Chemokine

Extracellular Membrane TM5 TM1

TM3 TM7

TM6 TM2 Receptor

TM4 Cytosol

Figure 1.2 Hypothetical model of a monomeric chemokine (pink) bound to a chemokine receptor (blue), illustrating the interaction with the acidic loop regions of the receptor, enabling the N-terminus of the chemokine to be inserted into the helical bundle. Figure reprinted with permission from Lau, et. al. [12].

Many studies have shown that activation occurs between the receptor and the N-

terminal region of the chemokines, and deletion of N-terminal residues frequently results

in variants that do not induce signaling, but retain their high affinity interactions [12, 14,

15]. Deletion of the first seven residues from the N-terminus of CCL2, for example, results in an antagonist to the receptor CCR2 [16]. Although the mechanism of receptor activation is currently unknown, the chemokine ligands induce conformational changes which allow the transmission of a signal across the membrane, resulting in the activation of G-protein (Figure 1.2) [6, 7, 9, 14, 15, 17].

10

In addition to their structural classification, chemokines are also classified based

on their functional roles, as either homeostatic or inflammatory. Homeostatic

chemokines are constitutively expressed and coordinate cell migration required for

proper function of the immune system. Inflammatory chemokines are upregulated in

response to inflammatory mediators such as tumor necrosis factor and interferon-,

microbial invasion, or trauma. Many chemokines, however, fall into both categories and

as stated previously, they are best known for their role in leukocyte migration. The

process of migration is illustrated in Figure 1.3.

Figure 1.3 Illustration of the role of chemokines in leukocyte migration through the endothelial layer. This cartoon also serves to demonstrate the role of GAGs in leukocyte migration. Figure reprinted with permission from Lau, et. al. [12]

To summarize, during an inflammatory response, selectins on the endothelial surface interact transiently with mucin receptors on the leukocyte surface. This results in a rolling motion of the leukocyte along the endothelial surface. Chemokines secreted in

11 response to signals such as proinflammatory cytokines, form gradients from the site of inflammation to the tissue surface, crossing the endothelial layer into the bloodstream.

Under such sheer flow condition, these gradients are believed to be maintained through interactions with glycosaminoglycans (GAGs) expressed on the endothelial cell surface.

This localized concentration of chemokines provides directional cues for the leukocyte cells, bringing the chemokines in contact with their receptors expressed on the leukocyte surface. This chemokine:receptor binding event transmits a signal across the cell membrane, triggering cascades of intracellular signaling, resulting in firm adhesion between the migrating leukocyte and integrins on the tissue surface. The activated leukocyte is then extravasated through the endothelial layer and into the tissue, where it continues to migrate along the chemokine gradient to the source of inflammation [13, 18-

20]. While there are many interactions involved in the process of leukocyte migration, this study examines the structural and mechanistic details of three main interactions – chemokines with their receptors, with GAGs, and chemokine oligomerization. All three of these interactions are necessary for chemokine function. As will be discussed in subsequent sections, in many cases, these interactions are mechanistically connected to one another.

12

1.2 Chemokine Oligomerization and GAG Interaction

Although chemokine monomeric structures are very similar throughout all four subfamilies, their quaternary structures are more varied. Two primary structural types of dimers have been observed: CC dimers which interact via a two-stranded antiparallel beta-sheet near the N-terminus, and CXC dimers formed by the first strand of the beta- sheet from each monomer (Figure 1.4b, c) [8, 21-23]. More recently, a third type of dimer was identified for the C chemokine, lymphotactin/XCL1, that consists of an all- beta-sheet arrangement with no similarity to other known protein structures and which rapidly interconverts with the canonical monomeric chemokine fold (Figure 1.4d) [24]. In addition, while the known quaternary structure of dimers are similar among members of each subfamily, some chemokines, (e.g. MPC-1/CCL2, PF4/CXCL4, IP-10/CXCL10) have been shown to form tetramers that in some cases have both CC and CXC interfaces [21], while others adopt completely novel folds (e.g. IP-10/CXCL10) (Figure

1.4e, f) [25]. Finally, several chemokines form heterodimers, and it has been shown that heterodimerization can occur within members of a given subfamily, as well as between subfamilies [26-28].

13

A B C D

E F

Figure 1.4 Examples of Chemokine Structures (A) CCL2/MCP-1 monomer (PDB code 1DOL) (B) CCL2 dimer, an example of a “CC dimer” (PDB code 1DOM). The interface is composed of a small antiparallel beta-sheet formed from residues at the N-terminus of both subunits, colored yellow and blue. (C) IL-8/CXCL8 dimer, an example of a “CXC dimer” (PDB code 1IL8). The interface is formed by the first beta strand in each subunit as well as interactions between the C-terminal end of the helix of one subunit and the beta-sheet of the opposing subunit, colored red and blue. (D) The Lymphotactin/XCL1 dimer, with each subunit colored cyan and orange, it contains a unique all-β-sheet structure that exists in equilibrium with a canonical chemokine monomer (PDB code 2JP1). (E) The CCL2 tetramer has characteristics from both CC (interface colored in yellow and blue) and CXC (interface colored in red and blue) dimer interfaces (adapted from PDB code 1DOL). (F) H-form of IP-10/CXCL10 tetramer, with each subunit colored red, blue, orange, and cyan, associating through the third β-strands forming a twelve- stranded antiparallel β-sheet with a sharp kink in the middle (PDB code 1O80).

While, the current understanding of the functional relevance of these various oligomeric forms is far from complete, there is ample evidence suggesting that oligomerization is important to the overall mechanism of cell migration. Previous studies have shown that mutant forms of chemokines that are unable to oligomerize, are generally still fully functional with respect to receptor binding and cell migration in vitro, suggesting that receptor binding occurs through the monomeric form [20, 21, 29, 30].

Nevertheless, these monomeric mutants are inactive in vivo when tested in an intra-

14 peritoneal recruitment assay [20, 31]. The prevailing explanation for these apparently anomalous results is related to the fact that as part of the mechanism for providing directional cues, chemokines are maintained near sites of production by localization on cell surfaces through interactions with glycosaminoglycans (GAGs). These interactions often involve oligomerization of chemokines on the GAGs. Indeed, biochemical and biophysical studies indicate that some chemokine:GAG interactions are facilitated by chemokine oligomerization, and also that chemokine oligomerization can be facilitated by interactions with GAGs (Figure 1.5) [13, 20, 32].

Figure 1.5 GAG binding sites mapped to the tetrameric structure of CCL2, shown next to heparin dodecasaccharide in order to illustrate the relationship between GAG binding and stabilization of the chemokine tetramer, adapted from Lau, et. al. [13].

Furthermore, chemokine variants that are incapable of binding GAGs can also be functional in vitro but not in vivo in much the same way as mutants that are unable to

oligomerize [20]. The functional coupling between GAG binding and oligomerization is

15 perhaps best exemplified by the [44AANA47]-RANTES/CCL5 mutant, which is unable to bind GAGs and blocks the activity of the WT protein through a dominant negative effect by forming nonfunctional heterodimers with the WT protein [13, 33]. While chemokine:GAG interactions are not discussed further here, many reviews are available on this topic [7, 13, 19, 34-39].

In addition to playing a role in cell migration, oligomerization may also contribute to the functional regulation of chemokines. For example, the disordered N-termini are the key signaling domains in all chemokines, and thus proteolytic processing of the N- terminus represents a natural mechanism for modulating chemokine function [40]. Most frequently, agonist activity is reduced or abolished completely upon N-terminal proteolysis, however there are examples of increased activity, and even alterations in receptor-binding specificity with N-terminal processing [40, 41]. While interactions with

GAGs have clearly been shown to protect chemokines from proteolysis [42, 43], in principle, oligomerization alone could also be protective both directly, and indirectly by facilitating interactions with GAGs.

Finally, oligomerization has been shown to influence cellular signaling. In some cases, oligomerization promotes additional signaling pathways not induced by the monomeric chemokine. For example, although a non-aggregating variant of

RANTES/CCL5 was able to induce chemotaxis via Gi coupling, it was unable to activate

T cells, monocytes, and neutrophils through protein tyrosine kinase (TK) pathways; this contrasts with wild type (WT) CCL5, which forms large oligomers nucleated by a CC-like dimer, and induces TK pro-inflammatory pathways [21]. Oligomerization can also inhibit signaling as demonstrated by an obligate SDF-1/CXCL12 dimer, which was engineered

16 by introduction of an inter-molecular disulfide. It was shown to flux calcium, but unlike

WT CXCL12, could not induce all of the pathways required for chemotaxis; instead, it

blocked migration of cells to the WT chemokine [44]. Recently, several chemokines

have been shown to heterodimerize with other chemokines and in some cases, such as

MCP-1/CCL2 and MCP-2/CCL8, their presence could be correlated with altered or

amplified functional responses relative to signaling by only one of the chemokines [28,

45, 46].

While the relevance of oligomerization is now well-established as the above

examples suggest, there are 50 human chemokines and little is known about the vast

majority. Clearly, determining whether chemokines oligomerize and the structural details

of the oligomers are crucial to understanding the mechanisms of chemokine-mediated

processes. Furthermore, non-oligomerizing forms can have therapeutic value as

demonstrated for a monomeric form of MCP-1/CCL2 which had anti-inflammatory

properties in animal models of experimental autoimmune encephalomyelitis and arthritis

[47, 48].

17

1.3 Chemokines and Disease

Chemokines are involved in a broad spectrum of diseases, ranging from inflammatory and autoimmune disorders to cancer metastasis and HIV. Table 1.1 lists some of the diseases associated with several chemokine receptors.

Table 1.1 Examples of diseases associated with several chemokine receptors [49-53]. Chemokine Examples of Disease Connection Receptor

CCR1 Multiple Sclerosis, Rheumatoid Arthritis, Cancer, Kidney Disease

CCR3 Allergic Asthma, Rhinitis

CCR5 HIV, Cancer, Transplant Rejection

CCR10 Melanoma Metastasis, Psoriasis

CXCR4 HIV, Cancer

Due to their critical role in the immune system response to inflammation, through

directed migration of leukocytes, chemokines and chemokine receptors are implicated in

a wide variety of inflammatory diseases. Inflammation associated with Rheumatoid

arthritis, for example, involves the recruitment of monocytes and T cells into synovial

tissue, believed to be caused by upregulation of the inflammatory chemokines CCL2,

CCL3, and CCL5 [3]. In addition, inappropriate expression of chemokines and

chemokine receptors is often associated with the progression and severity of

autoimmune disorders such as multiple sclerosis (MS), psoriasis, and rheumatoid

arthritis [3]. For example, patients with active MS show elevated levels of the

inflammatory chemokine CXCL10, while at the same time CCL2 is consistently down-

18 regulated in the cerebral-spinal fluid compared to patients with non-inflammatory

neurological disorders [54]. Chemokines and chemokine receptors not only influence

inflammatory and autoimmune diseases, but have also been shown to play a critical role

in cancer progression. In highly metastasizing cancers, such as breast cancer and

melanoma, it has been shown that chemokines are involved in directing the migration of

tumor cells expressing their receptors to organ-specific sites [52, 55, 56].

19

1.4 Biological Diversity of Chemokines

It is evident that the chemokine network is highly complex. For example, it is the monomeric forms of chemokines that are responsible for receptor activation, and while their monomeric structures are highly homologous, some chemokines are very specific to one receptor on one cell type, while others activate multiple receptors on a variety of cell types eliciting very different cellular responses. This complex “redundancy” is illustrated in Table 1.2.

Table 1.2 Chemokine Receptors and Chemokine Ligands

20

In addition to structural homology, chemokines have been shown to have similar

receptor binding motifs. Figure 1.6 shows the CCL2 residues responsible for binding

and activating the receptor CCR2 mapped to the monomeric structure, compared to the

murine CXCL10 residues responsible for interaction and activation of the receptor

CXCR3 [14, 57].

A B

R24 K47 K46 K49 R22

N20 K38 Y13

K35 Q1-P8 R8

Figure 1.6 Examples of chemokine receptor binding epitopes, shown in yellow. A: CCL2 monomer (PDB code 1DOL) demonstrating the residues necessary for binding and activating the chemokine receptor CCR2 [14]. B: Monomeric structure of murine CXCL10 (PDB code 2r3z) illustrating the activation and binding regions for the receptor CXCR3 [31, 57].

The variety of chemokine binding partners is not limited to chemokine receptors, but includes a multitude of surface expressed GAGs, as discussed in Section 1.2.

Unlike the receptor binding motifs, the GAG binding regions of chemokines are highly varied. Initial studies of chemokine:GAG interactions led to the hypothesis that chemokines could bind simultaneously to GAGs at the tissue surface and to their receptors on the leukocyte cell surface [19, 58].

21

However, subsequent studies have shown that this relatively simple explanation

is not adequate. There are many cases where the chemokine GAG-binding sites

overlap with the chemokine receptor binding epitopes. Mutagenesis studies of CCL5 for

example, isolated a basic region, 44RKNR47, located on the surface-exposed 40s loop.

The GAG-deficient mutant CCL5-44AANA47 experienced a 200-fold decreased in affinity

for its receptor CCR1, demonstrating that the GAG binding site can overlap with the

receptor binding site [59, 60]. Additionally, studies of CCL2 identified residues in the 20

and 40s loop, Arg18, Lys19, Arg24, and Lys49, as being critical for GAG-binding and, as

figure 1.6 shows, these residues also contribute to interaction with CCR2 [13, 14].

These results have led to the idea that chemokines do not bind simultaneously to their

receptors and to GAGs, but the interactions involve more of sequential hand-off

mechanism. Ultimately, this suggests that there are many possible sites for GAG

interaction on the chemokine surface, which leads to the idea that chemokines are

capable of binding a variety of different GAGs.

The final level of complexity to these interactions discussed here, also presented

in Section 1.2, is the idea that GAGs enhance chemokine oligomerization and

chemokine oligomerization in turn promotes GAG binding. This idea has been

supported by previous studies showing that while monomeric variants of functionally

oligomeric chemokines are able to interact with their receptors in vitro due to lack of

shearing forces, their in vivo function as well as their GAG binding in vitro is significantly

compromised [18, 20, 31, 32]. In addition, it has been shown that GAGs binding sites

often overlap with the oligomeric interface of the chemokine. Figure 1.7 demonstrates

that different oligomeric structures have different patterns of GAG binding sites,

suggesting that different GAGs are able to stabilize different oligomers.

22

A BC

D E

Figure 1.7 Examples of oligomeric chemokine structures with the specific GAG binding regions mapped to the surface. The individual monomeric subunits are colored in different shades of blue and the GAG binding residues are highlighted in red. A: CCL2 dimer (PDB code 1DOM) [13], B: CXCL12 dimer (PDB code 2KEC) [61], C: CXCL8 dimer (PDB code IL8) [58], D: CCL2 tetramer [13], E: CXCL10 tetramer, H-form (PDB code 1O80) [25, 57].

Due to differential expression and localization of GAGs in different tissues, and the fact that several chemokines have shows specificity for certain GAGs compared to others, it has been postulated that GAG interaction and oligomerization play a critical role, along with receptor specificity, in the functional diversity of chemokines [13, 18-20].

Given the significant role chemokines and chemokine receptors play in physiological and pathophysiological processes, understanding their structural and functional mechanisms ultimately represents a crucial step in the design of novel therapeutics for the treatment of disease.

23

1.5 Acknowledgements

Chapter 1.2, in part, was published in Methods in Enzymology: Jansma, A.;

Handel, T.; M. Hamel, D. Homo- and hetero-oligomerization of chemokines. Methods

Enzymol. 2009, 461, 31 - 50

CHAPTER 2

Optimization of the Expression and Purification of

CCL27 and CCL28

2.1 Introduction to CCL27 and CCL28

CCL27, also known as the Cutaneous T-cell Attracting Chemokine (CTACK), is a

CC chemokine that selectively attracts CLA+ memory T cells expressing the chemokine receptor CCR10 [62]. CCL27 is constitutively expressed in keratinocytes predominantly in the skin. However, it is also upregulated in response to pro-inflammatory signals [63,

64]. Both CCL27 and CCR10 are believed to play a role in melanoma metastasis to secondary skin sites [52, 60, 65]. CCL27 is ~40% sequentially homologous to another

CC chemokine CCL28 (the Mucosal Epithelial Chemokine, MEC), the only other known ligand for CCR10 (Figure 2.1).

CCL27 FLLPPSTACCTQLYRKPLSDKLLRKVIQVELQEADGDCHLQAFVLHLAQRSICIHPQNPS 60 CCL28 -ILPIASSCCTEVSHH-ISRRLLERVNMCRIQRADGDCDLAAVILHVKRRRICVSPHNHT 58

CCL27 LSQWFEHQERKLHG------TLPKLNFGMLRKMG----- 88 CCL28 VKQWMKVQAAKKNGKGNVCHRKKHHGKRNSNRAHQGKHETYGHKTPY 105

Figure 2.1 Sequence alignment of CCL27 and CCL28, showing relative position of cysteine residues and potential disulfide bonds. CCL28 has one additional pair of cysteine residues, forming a third disulfide bond.

CCL28 was discovered more recently and unlike CCL27, which is specifically expressed in skin, CCL28 is constitutively expressed in the epithelial cells of virtually all

24 25 mucosal tissues [66-68]. In addition to CCR10, CCL28 also binds and activates another

chemokine receptor CCR3, which has nine other chemokine ligands, CCL5/RANTES,

CCL7/MCP-3, CCL8/MCP-2, CCL11/Eotaxin, CCL13/MCP-4, CCL15, CCL16, CCL24,

and CCL26 [7]. A crucial first step to understanding the structural and functional

properties of CCL27 and CCL28 was to optimize an expression system that would yield

milligram quantities of protein for such studies.

In many cases, overexpression of chemokines in bacterial cells is toxic and

makes the production of high levels of protein extremely difficult. In order to avoid the

toxicity effects, these proteins are often expressed as insoluble inclusion bodies. Their

conserved disulfide bonds are advantageous in that many chemokines are easily

solubilized and refolded following initial purification. In order to ensure that the majority

of protein is produced as inclusion bodies, a His-tagged ubiquitin fusion expression

system was utilized, known as pHUE. Figure 2.2 illustrates the ubiquitin system, with an

N-terminal His tag, fused to CCL28 [69]. Because chemokines signal their receptors

through N-terminal residues, it is critical to maintain an intact and function N-terminus.

This system has the advantage of a very specific enzyme, ubiquitinase that cleaves

exactly at the ubiquitin-protein interface.

His6 Ubiquitin CCL28

LRLRGGILPIASSC

Figure 2.2 Schematic illustration of ubiquitin fusion, pHUE [69]

26

2.2 Results

2.2.1: Protein expression and purification

The following is an example of a typical expression and purification method for

an unlabeled sample, using CCL28 as the representative chemokine. This method was

used to express and purify CCL27, CCL28, and CCL2. The protocols for all three

chemokines are very similar. The differences are detailed below, as well as a

description of the media for 15N and 13C-labeled protein.

2.2.2: Protein expression of CCL28, ubiquitin fusion

CCL28 was first cloned into the pHUE vector (kindly provided by Baker) [69].

The protein, His6-Ubiquitin-CCL28 (Ub-CCL28), was overexpressed by induction with isopropyl β-D-1-thiogalactopyranoside in BL21(DE3)pLysS Escherichia coli. His6-

Ubiquitin-CCL28 expresses predominantly as an inclusion body in the insoluble fraction,

with a molecular weight approximately equal to 20 kDa (Figure 2.3) [69]. Final yields

were significantly increased when the entire growth was performed in one day,

beginning with inoculation of 2.5 mL media, which was used to inoculate 25 mL, which

was grown until an A600 absorbance of ~0.8. 2.5 mL of this culture were then added to

1 L of media. Following induction, the cells were grown for approximately 4 hours,

harvested by centrifugation, resuspended in Resuspension Buffer (Table 2.1), and

stored at -80ºC until purification.

27

Figure 2.3 Example of CCL28 expression from the pHUE vector. Ub-CCL28 expresses predominantly in the pellet as an inclusion body, with a molecular weight of approximately 20 kDa.

28

2.2.3: Initial purification of CCL28

Table 2.1 Buffer conditions for the purification of CCL2, CCL27, and CCL28. CCL28 CCL27/CCL2 Resuspension Buffer 10 mM Tris, pH 8 50 mM Hepes, pH 7.2 20 mg Lysozyme 20 mg Lysozyme DNAse I, pinch DNAse I, pinch 2 x Protease Inhibitor Tablets 2 x Protease Inhibitor Tablets

1 mM PMSF 5 mM MgCl2 Solubilization Buffer 10 mM Tris, pH 8 50 mM Hepes/KPhos, pH 7.5 6 M Guanidine, HCl 6 M Guanidine, HCl Hampton Fold-it Buffer, 1L #2 #13 Dialysis Buffer, 20 L 10 mM Tris, pH 8 20 mM Hepes, pH 7.2 300 mM NaCl 200 mM NaCl 10 mM Imidazole FPLC Concentrating Buffers 10 mM Tris, pH 8 50 mM Hepes, pH 7.2 300 mM NaCl 200 mM NaCl 20 mM Imidazole 20 mM Imidazole 1 M Imidazole, Elution 1 M Imidazole, Elution

To initiate purification, cells were lysed by sonication and lysozyme treatment, and the insoluble fraction was pelleted by centrifugation. The insoluble fraction was washed once with buffered detergent before being solubilized with the denaturing

Solubilization Buffer (Table 2.1), and passed over a gravity Ni-NTA column (Qiagen).

29

The fractions containing Ub-CCL28 were eluted with low pH and pooled. Ub-CCL28 was refolded by dilution into Hampton Fold-it Buffer #2 (refolding of CCL27 and CCL2 is performed in Hampton Fold-it Buffer #13, with 550 mM Arginine-HCl). After dialyzing into 20 L Dialysis Buffer, the protein was concentrated by passing again over a Ni-NTA

(Qiagen) column using an FPLC system (GE Healthcare) and the protein fractions were eluted with imidazole (Figure 2.4, Table 2.1).

Figure 2.4 CCL28 is concentrated by passage over a Ni-NTA column using an automated FPLC (GE Healthcare) with an external peristaltic pump for sample loading. The chromatogram shows Ub-CCL28 and the linear Imidazole gradient from 20 mM – 1.0 M.

2.2.4: Ubiquitin cleavage and HPLC purification

Protein fractions were then dialyzed overnight into 10 mM Tris, pH 8 (50 mM

Hepes, pH 7.2 for CCL27 and CCL2) in order to remove imidazole. The His-Ubiquitin tag was cleaved by the specific protease, ubiquitinase (this protease does not function in the presence of imidazole). The cleavage reaction was quenched by lowering the pH to

30

2.0, immediately before the last purification step, which involved reverse phase HPLC with a semi-prep C4 column. As demonstrated in Figure 2.5, HPLC separates Ubiquitin,

CCL28, and residual Ub-CCL28 fusion. The HPLC fractions were lyophilized and stored at -80ºC until use. Purity was assessed by MALDI Mass Spectrometry.

Figure 2.5 Ubiquitin cleavage of CCL28 showing time 0, 1, and 2 hours. The HPLC fractions show separation of Ubiquitin and CCL28.

For CCL28, reverse phase HPLC with a semi-prep column was sufficient to separate cleaved protein from residual fusion. However, for CCL27 and CCL2, the

Ubiquitin fusion proteins eluted in overlapping conditions with the cleaved products. As a result, immediately following the cleavage reaction, CCL27 and CCL2 were passed over a small (5 mL) gravity Ni-NTA column. CCL27 and CCL2 were eluted from this column with 60 mM Imidazole. The His6-Ubiquitin, His6-Ubiquitin-CCL27/2, and His6-

Ubiquitinase remain bound to the column. The protein elutions were then purified via semi-prep HPLC, as described for CCL28. Figure 2.6 illustrates the difference between

31 the HPLC chromatograms for CCL28, immediately following the cleavage reaction, and

CCL27 after the additional Ni-NTA elution.

Figure 2.6 HPLC chromatograms, recorded at 280 nm A: CCL28 peak elutes first, followed by Ubiquitin, and Ub-CCL28 fusion, all three peaks are baseline separated. B: Residual Ubiquitin peak from Ni-NTA purification, followed by purified CCL27.

2.2.5: Preparation of MOPS media

MOPS based media was prepared for the expression of labeled protein. 10x

MOPS was prepared with 0.1 mM FeSO4, 2.76 mM K2SO4, 5.0 mM CaCl2, 5.28 mM

15 13 13 MgCl2, 0.5 M NaCl, 10 mL Micronutrients per liter. N/ C- and C-labeled CCL27 were

grown using 98% 15N ammonium sulfate, and 99% 13C glucose, and 15N-labeled CCL27 was grown with 15N-labeled ammonium sulfate and 13C-depeleted glucose (99.98% 12C).

The protein growth in E. coli was accomplished in one day in order to maximize the final yield. Initially, 2.5 mL LB were inoculated and grown for 2.5 hours. This culture was then used to inoculate 300 mL LB, which was grown to an A600 of ~0.7. 50 mL of this LB

32 culture was then used to inoculate 1 L of the MOPS media. This process of “LB doping” resulted in significantly more protein and had no apparent affect on incorporation of the isotope label. The protein samples were then purified by the same method described previously. Figure 2.7 shows the ESI Mass Spectrometry results, indicating that the proteins are pure and the 15N and 13C labels incorporated at 99.8% and 99.7% respectively.

A B 10254 10572 Intensity

Mass, Da

Figure 2.7 A: 15N-labeled, 13C-depleted, 99.8% incorporation. B: 13C-labeled, 99.7% incorporation

The final yields for each protein, unlabeled and labeled, are listed in Table 2.1.

33

Table 2.2 Standard yields for CCL2, CCL27, and CCL28 Protein Approximate Yield (mg/L)

CCL27 15-18

15N and 13C CCL27 10-12

CCL28 10-12

CCL2 5

15N CCL2 3-5

2.2.6: Confirmation of CCL27 and CCL28 function

Before beginning structural analysis, it is necessary to confirm protein function.

In order to confirm that CCL27 and CCL28 are functional, migration assays were employed to measure their ability to induce cells expressing CCR10 to migrate across a filter. The migrated cells are counted by flow cytometry and plotted as a function of chemokine concentration, normalized to the total number of cells in the absence of a filter. Figure 2.8 shows the results for the migration assay of CCL27 and CCL28 with

L1.2 cells expressing CCR10.

34

Chemotaxis with L1.2 cells expressing CCR10 0.70

CCL27WT 0.60 CCL28MEC

0.50

0.40

0.30

0.20

Cells Migrated/Total Cells Migrated/Total Cells 0.10

0.00 0 0.1 1 5 10 50 100 250 500 1000 Chemokine Concentration (nM)

Figure 2.8 Migration assay as a test of function for CCL27 and CCL28

As the chemokine concentration is increased, the number of migrated cells also initially increases. However, excess chemokine results in receptor desensitization and internalization, resulting in fewer migrated cells. The results seen in Figure 2.8 represent a classic migration curve and indicate that CCL27 and CCL28 are active to migration.

CHAPTER 3

NMR Analysis of the Structure, Dynamic and Unique Oligomerization Properties of the Human Chemokine CCL27

3.1 Summary

Chemokines are involved in a variety of physiological processes, but are most

commonly known for their role in leukocyte migration. Cutaneous T cell-attracting

chemokine (CTACK), also known as CCL27, is expressed in skin and selectively

chemoattracts CLA+ memory T cells that express the receptor, CCR10. In this study, we investigated the structure and dynamics of CCL27 by solution NMR spectroscopy. While it has been shown that the monomeric forms of chemokines are capable of receptor activation in vitro, many chemokines oligomerize and their oligomeric forms are required for function in vivo, for interactions with glycosaminoglycans (GAGs), as well as for signaling. 15N relaxation and translational self-diffusion rates over a range of concentrations indicate that CCL27 has a strong tendency to oligomerize. However, in contrast to many other chemokines that form discrete dimers, CCL27 transitions between monomer, dimer and tetramer species over a relatively narrow concentration range. Binding to the glycosaminoglycan Heparin, avidly promotes oligomerization under conditions in which CCL27 is monomeric by itself. A 3D structure determination was pursued under the conditions that gave the best experimental sensitivity and revealed the standard chemokine motif. Based on the analysis of 15N relaxation and

translational diffusion data, the main species under these solution conditions

corresponds to a dimer. Nevertheless, attempts to find intermolecular NOEs were not

35 36 conclusive. Analysis of chemical shift perturbations of 15N-1H HSQC spectra as a

function of CCL27 concentration, and filtered (HC)NH-NOEs with a mixed isotope

sample (50% 13C + 50% 15N), suggest that the oligomeric form of CCL27 does not adopt a discrete CXC or CC dimer motif. Similarly, five µs-ms exchange-broadened regions involving surface exposed residues were located by T2 cpmg relaxation-dispersion measurements. While the regions from all of these experiments include residues that would be expected to be part of a putative CC or CXC type dimer, their concerted presence is mutually exclusive and indicative of an uncommon oligomerization behavior for this chemokine, where several equilibria involving different interfaces seem to be simultaneously at work. The overall results of this study indicate that while the monomeric form of CCL27 has the standard chemokine motif, it oligomerizes up to a tetramer by itself in solution, albeit weakly. The presence of Heparin, however, shifts the equilibrium to the oligomeric state, suggesting an important role of the oligomers in GAG binding. We hypothesize that the plasticity of the oligomerization state of CCL27 allows it to adopt different oligomeric structures, depending on its GAG (or other) binding partners, thereby increasing the functional diversity of this chemokine.

37

3.2 Introduction

The chemokine family consists of small (~8-12kDa) chemoattractant proteins

responsible for controlling the migration of leukocytes during both routine immune

surveillance and inflammation [1-5]. Inappropriate expression of chemokines and

chemokine receptors has been connected to a variety of pathophysiological processes

[3, 5, 7], including autoimmune disorders such as rheumatoid arthritis and multiple

sclerosis [47, 70, 71], pulmonary diseases such as asthma [72], cancer metastasis [73],

and the chemokine receptors CCR7 and CXCR4 help facilitate HIV entry into host cells

[74]. To date, there are approximately 50 known human chemokines and 20 receptors

[7]. While some chemokines are specific to one chemokine receptor, other chemokines

bind multiple receptors and many receptors bind multiple chemokines. Despite this

apparent redundancy, there is a considerable degree of specificity within the chemokine

network in terms of tissue expression and receptor activation in vivo [19].

Chemokines are structurally categorized into two major subfamilies (CC, CXC)

and two minor subfamilies (CX3C, C) based on the relative position of their conserved

N-terminal cysteine residues, which form intermolecular disulfide bonds with two other

cysteines. Chemokines adopt a conserved monomeric structural motif consisting of a

disordered N-terminal region, followed by a 310 helix, three antiparallel β-strands, and a

C-terminal α-helix [8, 21-23]. Despite their similar monomeric structures, multiple dimeric and tetrameric interfaces have been reported. There are two predominant patterns of chemokine dimerization, known as the CC and the CXC dimer [75]. The disordered N-terminal residues within each monomeric subunit adopt a two-stranded antiparallel β-sheet in the CC dimer interface, while the more globular CXC dimers

38 interact primarily through the first strand of their β-sheets (Figure 3.1a and b) [8, 21-23].

In addition, several different tetrameric species have been reported. For example, the

CCL2/MCP-1 tetramer has properties of both CC and CXC dimer interfaces (Figure

3.1c) [21, 76], while CXCL10/IP-10 adopts multiple tetrameric forms, one similar to

CCL2, as well as two entirely different conformations [25].

AB

C

Figure 3.1 Examples of chemokine oligomerization. A: Dimer structure for CCL2 showing a standard CC chemokine dimer with two small β-strands at the interface formed by the N-terminal regions of both subunits, monomeric subunits colored in yellow and red (PDB code 1DOM). B: Dimer structure for CXCL8, a CXC chemokine dimer with the interface predominantly formed between resides on the first β-strand of each subunit. Monomeric subunits are colored red and green (PDB code IL8). C: Two views of the tetramer structure for CCL2. The tetrameric form of CCL2 has characteristics of both a CC dimer, shown in the first view, and a CXC dimer, shown in the second view. The monomeric subunits are colored in cyan, yellow, red, and green, to emphasize CC and CXC interfaces.

39

Although the functional relevance of these oligomeric forms is not entirely understood, there is considerable evidence that chemokine oligomerization is important for in vivo cellular migration [10, 18, 21, 31, 44, 77]. For example, monomeric variants of

CCL2, CCL4, and CCL5 maintained the ability to activate their receptors to induce cellular migration in vitro but failed to induce cellular migration when injected in mice

[20]. The predominant explanation for these results is that interactions between the oligomeric form of the chemokine and surface expressed Glycosaminoglycans (GAGs) tether the chemokines to the tissue surface [12, 18-20, 36, 37, 78]. This aids in the accumulation of localized chemokine gradients from the tissue surface to the site of inflammation, despite the shear forces associated with blood flow. The connection between chemokine oligomerization and GAG interactions are further supported by studies showing that the oligomeric forms of chemokines have a higher affinity for specific GAGs and GAGs aid in stabilizing chemokine oligomerization [13, 21, 32, 37,

77-80]. Although chemokines bind their receptors as monomers and the monomeric structures are homologous, chemokines appear to bind GAGs as oligomers and their oligomeric forms are considerably more diverse. The immense diversity of GAGs and their differential temporal and localized expression suggest that chemokine:GAG interactions may play an important role in chemokine tissue specificity [12]. It is also possible that chemokine diversity may not simply be a function of the different oligomeric structures, but also the ability of these oligomers to exchange between monomer, dimer, and tetramer.

This study focuses on the structure of the chemokine CCL27 (also known as the

Cutaneous T-cell Attracting Chemokine, CTACK), which is constitutively expressed in the skin and has only one known chemokine receptor, CCR10 [63]. CCL27 has been

40 implicated in inflammatory skin diseases such as psoriasis, and is believed to play a role

in melanoma metastasis by directing cells to secondary tumor sites within the skin [41,

52]. The solution NMR structure of monomeric CCL27 is presented here, along with

NMR data which suggest that its oligomerization behavior is unusual and involves

multiple simultaneous equilibria with weak interactions between monomers, dimers and

tetramers that can't be definitively assigned to currently known chemokine

oligomerization states.

41

3.3 Results

3.3.1: CCL27 oligomerizes, forming a tetramer at high concentrations

To ascertain optimal solution conditions for characterizing the structure and dynamics of CCL27, a uniformly-labeled 15N/13C sample was prepared at 2.1 mM and

15N-1H HSQC spectra were recorded. The linewidths were significantly broader and the

sensitivity much lower than would be expected for a monomeric protein of approximately

10 kDa, or even for a 20 kDa dimer. Because many chemokines form dimers and higher

order oligomers, the CCL27 sample was gradually diluted and the spectral quality

assessed, until at 1.0 mM, the experimental sensitivity was deemed to be optimal

(Figure 3.2).

42

112 36G 82G 74G

114 49Q

11T 75T 62S 60S 116 38C 39H 24R 50R 51S 6S 88G 67H 57Q 19S 43F 7T 25K 28Q 118 73H 35D 21K

N (ppm) 23L 40L 80N 66E 68Q 15 63Q 52I 83M 37D 70R 22L 78K 61L 120 71K 26V 46H 69E 72L 64W 13L 81F 12Q 48A 87M 85R 10C 18L 84L 34A 58N 122 54I 44V 20D 79L 53C 33E 86K 14Y 42A 2L 3L/76L 27I/30E 31L 41Q 47L 8A 124 45L 55H 32Q 10 9 8 7 1 H (ppm)

Figure 3.2 Assigned 15N-1H HSQC of CCL27 acquired on a Bruker Avance II 600 MHz NMR with a 5 mm TCI CryoProbe. Sample concentration 0.5 mM, prepared in 50 mM acetate buffer, pH 5.6 at 302.3K.

The HSQC spectrum showed excellent dispersion indicating that the protein is

folded; the linewidths were also much sharper and the sensitivity significantly greater

than for the 2.1 mM sample. Concentrations of 1.0 mM CCL27 were therefore used for

structural and dynamic studies (discussed later). However, in light of the direct

relevance of oligomerization to chemokine function, the oligomerization behavior of

CCL27 was further evaluated.

Oligomerization properties of CCL27 by Pulsed Field Gradient (PFG) diffusion NMR

Translational diffusion of CCL27 was measured by PFG 1H NMR spectroscopy to assess the dependence of the oligomerization state on concentration and solution

43

conditions. The self diffusion coefficient, Ds, is related to the average protein size, and can be extracted from a series of 1D 1H spectra following the attenuation of signal with

increasing gradient field strength. Upon oligomerization, the theoretical change in Ds

can be calculated from the Stokes-Einstein equation. For example, by approximating the

interaction with hard-sphere molecular contacts, the dimer:monomer ratio of the

diffusion coefficients was estimated to be 0.75, a value shown to be in good agreement

with experimental data for the monomer-dimer equilibrium of CCL2 [81]. One can also

estimate the hydrodynamic radius and molecular weight from the measured Ds value

(see below).

However, since interpretation of diffusion coefficients requires certain

assumptions, Ds values were first compared for several proteins of similar molecular weight to CCL27 and known oligomerization patterns [44, 77, 81] (Figure 3.3a). At 0.1 mM, CCL27 (10.1 kDa) has a Ds value similar to that of 0.075mM CCL2 (8.9 kDa, monomeric at this concentration), P8A CCL2 (an 8.9 kDa monomeric variant of CCL2),

-10 2 -10 2 -10 and ubiquitin (8.6 kDa) (Ds values are 1.35 x 10 m /s, 1.39 x 10 m /s, 1.45 x 10

2 -10 2 m /s, and 1.38 x 10 m /s, respectively). The Ds for lysozyme was somewhat lower (1.2 x 10-10 m2/s), as expected from its molecular weight (15.9 kDa). However, when the

-10 CCL27 concentration was increased to 3.0 mM, the Ds value dropped to 0.80 x 10

m2/s, indicating an increase in apparent molecular weight relative to the 0.1mM sample due to oligomerization. This behavior is similar to CCL2 which also oligomerizes at higher concentrations, although the Ds of CCL27 suggest that it is larger than CCL2 at

3.0 mM (Figure 3.3a) [82]. Since buffer conditions are known to affect chemokine oligomerization [77, 83], the effect of acetate, phosphate, and pH for CCL27 were evaluated and found to have little effect on apparent molecular weight (Figure 3.4,).

44

A PFG Diffusion Analysis of Standard Proteins B Oligomerization Profile of CCL27 1.50 1.50 m2/s m2/s

-10 Monomer -10 1.25 1.30

1.00 1.10 Dimer

0.75 0.90

Tetramer Diffusion Coefficient x 10 x Coefficient Diffusion Diffusion Coefficient x 10 Diffusion Coefficient x 0.50 0.70 0.1mM0.1mM 3.0mM3mM 1.0mM1.0mM 0.075mM0.075mM 3.03mM mM UbiquitinUbiquitinLysozyme 0 0.5 1 1.5 2 2.5 3 3.5 CCL27CCL27 CCL27CCL27 P8AP8A CCL2 CCL2 CCL2 Concentration (mM) CCL2 C CCL27 Calculated from Translational Diffusion D Oligomerization Profile of CCL2 45 1.50 Tetramer 40 Monomer m2/s 35 -10 30 1.25 25 Dimer 20 Dimer 15 1.00 Monomer 10

Molecular Weight(kDa) 5 Tetramer

0 10 Diffusion Coefficient x 0.75 012340 0.5 1 1.5 2 2.5 3 3.5 Concentration (mM) Concentration (mM)

Figure 3.3 Pulsed Field Gradient (PFG) diffusion analysis of CCL27, performed on a Bruker Avance II at 600 MHz with a TCI CryoProbe at 302.3K. Assuming the diffusion coefficient at 0.075 mM corresponds to the monomeric form, the theoretical Ds value for the dimer was estimated using the Stoke’s-Einstein equation as 0.975 x 10-10 m2/s. A: PFG diffusion analysis of standard proteins with known molecular weights and oligomerization patterns. B: PFG diffusion analysis of CCL27 over a concentration range of 0.050 mM to 3.0 mM in 50 mM acetate, pH 5.6. C: Molecular weight of CCL27 at different concentrations, calculated by translational diffusion. D: PFG diffusion profile of WT CCL2 over a range of concentrations from 0.01mM to 3.0 mM, in 50 mM acetate, pH 5.6.

45

Analysis of Buffer Conditions for CCL27 1.50 50mM Acetate, pH 5.5 m2/s

-10 50mM Acetate, 100mM NaCl, 10mM Phosphate, 1.25 pH 5.5 50mM Acetate, 100mM NaCl, 10mM Phosphate, pH 6.5

1.00

Diffusion Coefficient x 10 Diffusion Coefficient 0.75 0.25 0.75 1.25 1.75 2.25 2.75 Concentration (mM)

Figure 3.4 PFG diffusion of WT CCL27 analyzing the effect of different buffer conditions.

The oligomerization profile for CCL27 was then determined over a more extensive concentration range, from 0.075 mM – 3.0 mM in 50 mM acetate buffer at pH

5.6 (Figure 3.3b). The diffusion coefficient for CCL27 is relatively constant from 0.075 –

-10 2 0.25 mM, with an average Ds value of ~1.30 x 10 m /s, but it decreases gradually

beyond 0.25 mM to 0.813 x 10-10 m2/s at 3.0 mM. This value is less than what would be expected for a dimer (0.975 x 10-10 m2/s), suggesting the presence of higher order oligomers, most likely tetramers by analogy to other chemokines (theoretical Ds value for

the CCL27 tetramer is 0.780 x 10-10 m2/s). Indeed, when the diffusion coefficient is converted into the hydrodynamic radius via the Stokes–Einstein relation, and used to estimate the molecular weight, the data suggests that CCL27 is tetrameric at 3.5mM (39 kDa estimated versus 40.4 kDa theoretical, Figure 3.3c and Table 3.1a). Comparing the hydrodynamic radius of CCL27, calculated from the translational diffusion (via

HYDRONMR), to hydrodynamic radii for other known chemokine tetramers, calculated

46 using their atomic structures (via HYDROPRO), the radius for CCL27 at 3.5 mM is almost identical to that of CCL2, Fractalkine/CX3CL1, and CCL14 tetramers (2.68 nm,

2.65 nm, 2.70 nm, and 2.73 nm respectively, Table 3.1b). This further suggests that

CCL27 forms a tetramer at 3.0 – 3.5 mM [84, 85].

The diffusion profile was then determined for CCL2 under the same conditions, in order to compare the results for CCL27 to a chemokine known to form a stable dimer in solution. Between 0.05 mM and 0.10 mM, a sharp decrease in the Ds value of CCL2 is observed, corresponding to the formation of a dimer; it then remains relatively constant up to 3.0 mM (Figure 3.3d). CCL2 has also been shown to form tetramers at higher concentrations and in the presence of GAGs [76]. The slight decrease in the Ds value at

1.5 mM suggests some population of the tetrameric form. Nevertheless, the overall

diffusion profile is what would be expected for a protein that forms a discrete dimer with

a Kd distinguishable from the tetrameric state [77, 82]. These results contrast with the

profile of CCL27 where the Ds value steadily decreases over the entire concentration range, suggesting the presence of several oligomeric species (monomer, dimer, tetramer), with similar dissociation constants (Figure 3.3c). Notably, the stability of the

CCL27 dimer also appears to be weaker than that for CCL2.

47

Table 3.1 A: Variation in the translational diffusion constant, hydrodynamic radius and estimated molecular weight with concentration of CTACK over the range 0.05 mM to 3.5 mM, using HydroNMR. B: Calculated Hydrodynamic ratios (rh) from atomic structures using HydroPro A

-10 2 Conc. (mM) Dt x 10 m /s rh (nm) M (kDa) 0.05 1.26 1.88 13.4 0.1 1.26 1.89 13.7 0.2 1.25 1.89 13.8 0.5 1.19 1.98 15.8 1.0 1.09 2.15 20.1 1.75 1.03 2.25 23.2 2.1 0.96 2.41 28.5 3.5 0.84 2.68 39.1

B

Tetrameric Proteins M (kDa) rh (nm) CCL2 34.3 2.65

CX3XL1 31.5 2.7 CXCL10-H 34.6 2.6 CXCL10-M 34.6 2.55 CXCL10-T (β-strands) 34.6 2.57 CCL14_1 34.8 2.75 CCL14_2 34.8 2.73 Dimeric Proteins MCP-1 17.4 2.51 CCL14 CCL4 (Mip1B) 15.7 2.33 CXCL8 (CXCL8) 16.6 2.33 CXCL10-H (CXCL10) 17.3 2.20 CXCL10-T (CXCL10) 17.3 2.05 Monomeric Proteins (controls) CCL27 10.2 1.87 CCL2 8.7 1.84 CXCL8 8.3 1.82 CXCL10 8.7 1.83

48

Rotational diffusion by NMR

As a complement to the PFG experiments, and ultimately for analysis of internal

15 dynamics (see below), N R1 and R2 rates, as well as heteronuclear NOE (HNOE), were

measured for the backbone amide groups on a 1.0 mM CCL27 sample at 500 and 600

MHz, and analyzed by the Model-free formalism (Figure 3.5). However, analysis of the

data was complicated by significant exchange contributions (> 5Hz) to the transverse

relaxation rates for the majority of residues, which were not suppressed under the fast

CPMG pulsing conditions applied. Initial attempts at fitting the rotational diffusion tensor

in TENSOR2 [86] showed that the experimental data could not be reproduced using the

calculated diffusion tensor in combination with the experimental uncertainties.

Therefore, an alternative approach was adopted, in which exchange-free transverse

0 15 15 1 relaxation rates, R2 , were computed from the measured N(CSA)/ N- H(dipole-dipole)

0 transverse cross-correlation rate constant, xy [87]. Comparison of R2 with the R2 rates

measured under fast CPMG pulsing confirmed that indeed the majority of residues

possess significant residual exchange contributions (Figure 3.5).

49

A 1.8 1.6 1.4 1.2 1 / Hz

1 0.8 R 0.6 0.4 0.2 0 1F 2L 3L 4P 5P 7T 64 6S 9C 8A 27I 52I 54I 11T 13L 30E 31L 33E 75T 76L 77P 79L 81F 84L 17P 18L 22L 23L 40L 43F 45L 47L 69E 72L 56P 65F 66E 59P 61L 10C 78K 80N 85R 86K 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 70R 71K 73H 50R 51S 53C 55H 67H 58N 60S 62S 87M 12Q 14Y 26V 28Q 29V 32Q 34A 36G 74G 82G 83M 41Q 42A 44V 48A 49Q 68Q B 57Q 63Q 30

25 CPMG R2 20 0 R2

/ Hz 15 2 R 10

5

0 1F 2L 3L 4P 5P 7T 9C 6S 27I 52I 54I 8A 11T 13L 17P 18L 22L 23L 30E 31L 33E 40L 43F 45L 47L 56P 59P 61L 65F 66E 69E 72L 75T 76L 77P 79L 81F 84L 10C 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 50R 51S 53C 55H 58N 60S 62S 67H 70R 71K 73H 78K 80N 85R 86K 12Q 14Y 26V 28Q 29V 32Q 34A 36G 41Q 42A 44V 48A 49Q 57Q 63Q 68Q 74G 82G 83M 87M 88G 64W C 1.0

0.5

0.0 N NOE 64 64 1F 2L 2L 3L 3L 4P 4P 5P 5P 7T 7T 9C 9C 6S 6S 52I 54I 27I 27I 8A 52I 54I 8A 59P 61L 56P 65F 66E 17P 18L 22L 23L 40L 43F 45L 47L 69E 72L 11T 13L 30E 31L 33E 75T 76L 77P 79L 81F 84L 59P 61L 56P 65F 66E 17P 18L 22L 23L 40L 43F 45L 47L 69E 72L 11T 13L 30E 31L 33E 75T 76L 77P 79L 81F 84L 58N 60S 62S 50R 51S 53C 55H 67H 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 70R 71K 73H 10C 78K 80N 85R 86K 58N 60S 62S 50R 51S 53C 55H 67H 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 70R 71K 73H 10C 78K 80N 85R 86K 57Q 63Q 41Q 42A 44V 48A 49Q 68Q 12Q 14Y 26V 28Q 29V 32Q 34A 36G 74G 82G 83M 57Q 63Q 41Q 42A 44V 48A 49Q 68Q 12Q 14Y 26V 28Q 29V 32Q 34A 36G 74G 82G 83M 87M 87M 15 H-

1 -0.5 Residue

-1.0

-1.5

CPMG Figure 3.5 A: R1, values for 1.0 mM CCL27. B: Comparison of R2 rates (blue) with 0 R2 = κηxy (pink) for 1.0 mM CCL27 at 500 MHz. The majority of residues possess significant residual exchange contributions to their transverse relaxation rates measured using fast CPMG pulsing. C: [88]15N heteronuclear NOE values for 1.0 mM CCL27 at 500 MHz.

0 Recalculation of the isotropic rotational correlation time using the R2 /R1 ratio of

50 residues without significant internal motion yielded values of 9.3 and 9.2 ns for the data

collected at 500 and 600 MHz, respectively. These correlation times translate into an

estimated molecular weight of ~21 kDa, almost exactly equal to that of a CCL27 dimer,

consistent with the PFG diffusion measurements. Furthermore, diffusion tensor fitting

0 with the new R2 /R1 ratios using TENSOR2 indicated that rotational diffusion properties of 1.0 mM CCL27 are best described by an axially symmetric (D||/D = 0.69), slightly oblate diffusion tensor (Table 3.2).

Table 3.2 Results from fitting the rotational diffusion tensor using TENSOR2. The principal components are listed for the two axially symmetric models (oblate and prolate) 2 2 2 and the fully anisotropic models.  exp is the experimental  value and  sim the MC- derived 95 % confidence limit. Model Parameter 500 MHz Data 600 MHz Data

7 -1 Axially Symmetric, Oblate Dperp./10 s 1.73 ± 0.02 1.74 ± 0.01

7 -1 Dpar/10 s 1.99 ± 0.03 2.09 ± 0.05

2  exp 156 172

2  sim 55 56

7 -1 Axially Symmetric, Prolate Dperp./10 s 1.71 ± 0.02 1.74 ± 0.01

7 -1 Dpar/10 s 1.99 ± 0.03 2.09 ± 0.05

2  exp 156 172

2  sim 55 56

7 1 Fully Anisotropic Dxy/10 s- 2.11 ± 0.05 2.10 ± 0.04

7 1 Dyy/10 s- 2.00 ± 0.05 2.03 ± 0.04

7 1 Dzz/10 s- 1.47 ± 0.05 1.42 ± 0.05

2  exp 128 131 2 53 55 sim

51

3.3.2: NMR assignment and determination of the

monomeric form

The results of the PFG diffusion measurements suggesting the presence of multiple species in equilibrium, including tetramers, is consistent with the broadening observed in the initial HSQC spectra of 2.1 mM CCL27. Experiments used for resonance assignment and structure determination were therefore recorded on 1.0 mM

15N and 15N/13C labeled samples, a concentration which showed translational diffusion

behavior consistent with the presence of a predominantly dimeric species. The

assignment procedure involved a relatively standard approach employing triple

resonance experiments. Distance restraints were derived from NOE-based

measurements, dihedral angle restraints, and H-bond restraints, and the structure

calculations were performed with ARIA (ambiguous restraints for iterative assignment)

[89, 90] (see Materials and Methods for full details).

In the assigned 2D 15N-1H HSQC spectrum, the signals are well dispersed and all

backbone amide resonances are visible with the exception of Arg15 and Lys16,

indicating that these sites are undergoing chemical exchange on an intermediate

timescale (Figure 3.2). The linewidths for residues C-terminal to Pro77 suggest that this

region is unstructured and exhibits significant flexibility due to fast ps internal motion.

These residues display a second set of amide peaks in addition to the principal peaks,

presumably due to cis-trans isomerization of the peptide bond between Leu76 and

Pro77. Several residues, such as Asp35 and Asp37 exhibit unusually broad 15N-

linewidths, indicative of large exchange contributions to the 15N transverse relaxation

rate. In addition, a number of peaks are significantly overlapped, in particular

Leu3/Leu76 and Ile27/Glu30. Since Leu3 and Leu76 are also both N-terminal to Proline

52 residues and lie in unstructured regions, they were ultimately assigned by reference to methyl-selective TOCSY experiments.

The structure is well defined by the NMR ensemble, with a root-mean-square- deviation (RMSD) ~0.5 Å over the backbone atoms of the ordered residues. The calculated structures are in good agreement with the experimental restraints, with no

NOE restraints violated by more than 0.5 Å and no dihedral angle restraints by more than 5° (Table 3.3).

53

Table 3.3 Experimental restraints and structural statistics for CCL27 structure calculation. [SA] represents average values for the ensemble, for which the RMSD (root mean square deviation) values are from the mean structure. [SA]c represents the corresponding values for the structure closest to the mean. EL–J is the Lennard–Jones energy (the Lennard–Jones potential was only used during the water refinement stage). Number of experimental restraints

Distance restraints from NOEs intra seq med long Unambiguous 1364 814 339 103 108 Ambiguous 1204 Dihedral 44 H-bond 22

Coordinate precision (residues 8-71) SA SAc RMSD of backbone atoms / Å 0.55 ± 0.10 0.38 RMSD of heavy atoms / Å 0.97 ± 0.07 0.78 RMSD from experimental restraints NOE distances / Å 0.0207 ± 0.0025 0.0181 TALOS dihedral angles / º 0.54 ± 0.11 0.61 RMSD from idealized geometry Bonds / Å 0.00393 0.00386 Angles / Å 0.548 ± 0.019 0.53 Improper / Å 1.58 ± 0.019 1.57 Final energy

-1 EL-J / kJ mol -3340 ± 50 -3350 Ramachandran analysis Residues in most favored regions 74.3% 72.7% Residues in additionally allowed 21.3% 24.7% Residues in generously allowed 2.3% 2.6% Residues in disallowed regions 2.1% 0.0%

54

The ensemble of structures of the ordered core of CCL27 is shown in Figure

3.6a, along with the corresponding ribbon diagrams (Figure 3.6b), and the primary sequence colored to reflect the secondary structure (Figure 3.6c).

Figure 3.6 Solution NMR structure of monomeric CCL27, displaying the standard chemokine structural motif – an N-loop (green) followed a 310 (purple), three anti-parallel β-strands (yellow), and a C-terminal α-helix (purple). A: stereo views of the overlaid backbone traces for the 30 water-refined structures (residues Ala-8 to His-71). B: corresponding ribbon representation. C: Sequence of CCL27 showing unstructured and loop regions in green, α-helices in purple, and β-strands in yellow.

55

CCL27 adopts a typical chemokine fold, where the flexible N-terminus (Phe1 to

Thr7) is followed by an extended series of bends forming the N-loop, which leads into a

310 helix (Asp20 to Arg24), immediately followed by the first of three anti-parallel β- strands. The loop between the first and second β-strands, the 30s loop, stretches from

Glu33 to Gln41 and is anchored to the N-loop by the disulfide bridge between Cys38 and

Cys9. In contrast, the 40s loop, linking the second and third β-strands, is a short type I turn comprising only Gln48 and Arg49. The end of the third β-strand is also attached to the N-loop by the second disulfide bond between Cys53 and Cys10. A short loop from

His55 to Asn58 connects the end of the third β-strand to a long C-terminal α-helix, which extends to Arg70 and lies across one face of the β-strand, while the other side of the β- strand faces onto the N-terminus and early part of the N-loop. The C-terminal α-helix is followed by a long, disordered C-terminal region, a feature that distinguishes CCL27 from most other chemokines.

3.3.3: Characterization of internal dynamics

0 The exchange-free R2 values, described above, enabled the local motion of the majority of amide sites to be accurately described using the classical Model-Free formalism [91]. While the majority of the residues are in well structured regions, the profile of order parameters confirms the flexibility of the N- and C-termini, with increased internal mobility in the C-terminus becoming evident towards the end of the α-helix

(Figure 3.7). The affected amide groups exhibit internal correlation times on the order of

1 ns, with S2 values that decrease from 0.6–0.7 at the end of the helix to close to zero for

Gly88. The first eight residues in the N-terminus also exhibit significant internal mobility,

which is a characteristic of most if not all chemokines where this region plays an

important role as a signaling trigger. The extended N-loop, despite its lack of regular

56 secondary structure, is well ordered. For the remainder of the structured core of the

protein, the most obvious feature is the relative mobility of the 30s loop, where the order

parameters drop to ~ 0.6 at Gly36.

A 1.2

1

0.8

0.6 2 S 0.4

0.2

0 64 64 1F 2L 2L 3L 3L 7T 7T 8A 8A 4P 4P 5P 5P 6S 6S 9C 9C 27I 27I 52I 52I 54I 54I 65F 65F 72L 72L 75T 75T 76L 76L 79L 79L 81F 81F 84L 84L 11T 11T 13L 13L 14Y 14Y 18L 18L 22L 22L 23L 23L 26V 26V 29V 29V 31L 31L 34A 34A 40L 40L 42A 42A 43F 43F 44V 44V 45L 45L 47L 47L 48A 48A 61L 61L 66E 66E 69E 69E 71K 71K 77P 77P 78K 78K 86K 86K 16K 16K 17P 17P 19S 19S 21K 21K 25K 25K 30E 30E 33E 33E 51S 51S 56P 56P 59P 59P 60S 60S 62S 62S 67H 67H 68Q 68Q 70R 70R 73H 73H 74G 74G 80N 80N 82G 82G 83M 83M 85R 85R 87M 87M 88G 88G 10C 10C 12Q 12Q 15R 15R 20D 20D 24R 24R 28Q 28Q 32Q 32Q 35D 35D 36G 36G 37D 37D 38C 38C 39H 39H 41Q 41Q 46H 46H 49Q 49Q 50R 50R 53C 53C 55H 55H 57Q 57Q 58N 58N 63Q 63Q B 3 2.5 2

/ns 1.5 e/ τ 1 0.5

0 Residue 64 1F 2L 3L 7T 8A 4P 5P 6S 9C 27I 52I 54I 1F 2L 3L 4P 5P 7T 64 6S 9C 65F 72L 75T 76L 79L 81F 84L 11T 13L 14Y 18L 22L 23L 26V 29V 31L 34A 40L 42A 43F 44V 45L 47L 48A 61L 8A 27I 52I 54I 66E 69E 71K 77P 78K 86K 16K 17P 19S 21K 25K 30E 33E 51S 56P 59P 60S 62S 67H 68Q 70R 73H 74G 80N 82G 83M 85R 87M 88G 10C 12Q 15R 20D 24R 28Q 32Q 35D 36G 37D 38C 39H 41Q 46H 49Q 50R 53C 55H 57Q 58N 63Q 11T 13L 17P 18L 22L 23L 30E 31L 33E 40L 43F 45L 47L 56P 59P 61L 65F 66E 69E 72L 75T 76L 77P 79L 81F 84L 10C 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 50R 51S 53C 55H 58N 60S 62S 67H 70R 71K 73H 78K 80N 85R 86K -0.5 12Q 14Y 26V 28Q 29V 32Q 34A 36G 41Q 42A 44V 48A 49Q 57Q 63Q 68Q 74G 82G 83M 87M 88G -1 Residue

Figure 3.7 Overall order parameters and internal correlation times from TENSOR2 analysis of relaxation data recorded on 1.0 mM CCL27 at 500 MHz. Where the extended model-free formalism was required to fit the relaxation data, the order 2 2 2 parameter shown is given by S = S xS f.

3.3.4: Attempts to characterize the CCL27 dimer interface

Filtered NOE analysis of interfacial contact residues

Despite significant effort, it was not possible to unambiguously identify intersubunit NOEs from the 15N-separated and 13C-separated NOESY spectra

57 experiments used to calculate the monomer structure of CCL27. In order to calculate a

dimer structure, we therefore resorted to a 13C/15N-separated and filtered NOESY

experiment on a sample containing 50% 13C-CCL27 and 50% 15N-CCL27 (13C-depleted).

In this experiment, NOEs are detected between 1H -13C and a 1H -15N pairs. The mixed isotope sample favors detection of intersubunit NOEs, as previously demonstrated for

CCL2 [82]. To validate the experiment, and for a direct comparison to a well-defined dimer, we first recorded the 15N-1H plane of a 2D version of this (HC)NH-NOE experiment on a 3.0 mM sample containing 50% 13C-labeled CCL2 and 50% 15N-labeled

CCL2, and compared it with the 15N-1H HSQC (Figure 3.8a and b).

58

A HSQC CCL2 B HCNH-NOE CCL2, 12-hour acquisition 14N 14N 114 114 10T 16T 15F 36C 116 116 66H 65D 14N 118 118 23Q 57Q 9V 70Q N (ppm) 120 120 51I 11C N (ppm) 76T

15 40A 15 30R 19K 122 122 68D 16R 39E 4A 124 7A 38K 124 52C 47V 17N 10 9 8 7 10 9 8 7 1H (ppm) 1H (ppm) C HSQC CCL27 D HCNH-NOE CCL27, 4-day acquisition 112 112 82G 114 114 49Q 75T 116 38C 116 51S 6S 88G 50R 118 118 73H

N (ppm) 80N 83M N (ppm) 15 120 120 37D 81F 15 85R 13L 87M 84L 122 122 34A 79L 53C 86K 76L 124 124 31L

10987 10 9 87 1H (ppm) 1H (ppm)

Figure 3.8 Filtered NOE analysis of 50/50 15N labeled, 13C labeled sample of CCL2 at 3.0 mM and 50/50 15N labeled (13C-depleted), 13C labeled sample of CCL27 at 3.0 mM. Experiments were acquired with a Bruker Avance II 600 MHz, 5 mm TCI CryoProbe at 302.3K, and samples were prepared in 50 mM acetate, pH 5.6. A: Assigned 15N-1H HSQC of CCL2. B: 2D 15N-1H (HC)NH-NOE, acquired in 12 hours and assigned based on the HSQC. This experiment was previously employed to determine the dimeric structure of CCL2 in solution and used here in order to validate the method and compare to CCL27 [82]. C: Assigned 15N-1H HSQC of CCL27. D: 2D 15N-1H (HC)NH-NOE CCL27 at 3.0 mM, acquired in 4 days and assigned based on the HSQC.

The resulting spectrum was well dispersed, with intense signals corresponding to structured regions of the protein that were previously identified as comprising the dimer interface [82]. A similar experiment was then collected on the 3.0 mM CCL27 sample and showed positive signals from four different regions of the protein, Ser6 and Leu13 in the N-terminus, Leu31, Ala34, Asp37, and Cys38 in the first β-strand/30s loop, Gln49-

Ser51, and Cys53 in the third β-strand/40s loop, and His73, Thr75, Leu76, and Leu79-

59

Gly88 in the C-terminus (the positive signals from the unstructured C-terminus are likely due to intra-residue NOEs resulting from residual natural abundant 13C in the 15N/13C- depleted sample that appear due to favorable relaxation properties and close distance).

Residues were assigned based on comparison to the 15N-1H HSQC (Figures 3.8c and

d). When the experiment was repeated on a 2.0 mM CCL27 sample, similar results

were observed, with additional signals from the first β-strand and 310 helix that were not

present at the higher concentration (Figure 3.9).

112 82G

49Q 114

116 24R 51S 88G

118 83M

N (ppm) 80N 22L

15 37D 81F 120 85R 13L 87M 84L 20D 79L 122 34A 86K 76L 30E 124 31L

9 8 7 1H (ppm)

Figure 3.9 15N-1H (HN)CH-NOE of CCL27 at 2.0 mM, 50/50 15N labeled (13C-depleted), 13C labeled sample of CCL27, prepared in 50 mM acetate, pH 5.6, acquired at 302.3K, additional peaks indicated by boxes.

At 1.0 mM where the dimer dominates, the NOE signals were no longer

detectable due to the insensitivity of the experiment (data not shown). In general, the

results do not define a discrete CC or CXC dimer interface. A CXC dimer interface

would include only the first β-strand, while a CC dimer interface would involve

predominantly the N-terminal region and the third β-strand, which is linked to the N-

terminus by a disulfide bond between the conserved cysteine residues [77, 82], In

60 addition, the signal intensity from this experiment was considerably weaker for CCL27

than CCL2 at all concentrations; CCL2 required only a 12-hour acquisition time in order

to reproduce all known interfacial contacts, while CCL27 required an acquisition time of

4 days and ultimately resulted in fewer signals. This is consistent with the formation of a

lower affinity oligomer of CCL27 compared to CCL2, as indicated by the PFG diffusion

studies. The data is also consistent with the idea that CCL27 does not form a discrete

dimeric species, even at the lower concentration where the dimer dominates, but rather

has multiple interfaces of similar affinity that ultimately contribute to the tetramer.

Analysis of chemical shift perturbations (CSP)

As a second method to identify residues that may be involved in the

oligomerization interface(s), and which covers a broader concentration range than was

feasible for the (HC)NH-NOE experiment, 15N-1H HSQC chemical shift changes were monitored as the concentration of CCL27 was incrementally diluted from 1.0 mM, where the dimeric species should predominate, to 0.025 mM, where CCL27 is monomeric. The residues that experienced a change in chemical shift corresponded to several regions of the protein, including Leu2, Leu3, Ala8, Thr11, and Leu13 in the N-terminus, Ser19,

Asp20 and Leu22 in the 310 helix, Gln28 and Leu31 in the first β-strand, Ala34 – Asp37 and His39 – Gln41 in the 30s loop, Ala48 – Gln49 in the 40s loop, Ile54 in the third β- strand, Leu61, Trp64, and His67 – Leu72 from the C-terminal α-helix, and Gly74, Leu76, and Lys78 from the C-terminus (Figure 3.10a).

61

A 0.3

0.25

0.2

0.15 ppm

∆ 0.1

0.05

0 64 2L 3L 7T 1F 4P 5P 6S 8A 9C 52I 54I 27I 31L 40L 45L 47L 61L 11T 13L 18L 22L 23L 75T 76L 72L 43F 65F 29V 30E 33E 34A 35D 37D 38C 39H 42A 44V 46H 48A 50R 51S 53C 55H 56P 58N 59P 60S 62S 10C 14Y 15R 16K 17P 19S 20D 21K 24R 25K 26V 77P 78K 66E 67H 69E 70R 71K 73H 64W 28Q 32Q 36G 41Q 49Q 57Q 63Q 12Q 68Q 74G

B 0.3

0.25

0.2

ppm 0.15 ∆ 0.1

0.05

0 5I 4A 7A 9V 2P 8P 1Q 3D 6N 20I 31I 42I 46I 51I 10T 13Y 15F 16T 22V 25L 26A 28Y 32T 40A 41V 43F 45T 47V 48A 53A 60V 67L 71T 73T 76T 19K 21S 27S 33S 34S 35K 37P 38K 39E 44K 49K 50E 55P 56K 58K 63S 69K 74P 75K 11C 12C 14N 17N 18R 23Q 24R 29R 30R 36C 52C 54D 57Q 61Q 62D 64M 65D 66H 68D 70Q 72Q 59W Residue

Figure 3.10 15N-1H HSQC chemical shift perturbation analysis of CCL27 and CCL2 performed with a Bruker Avance II at 600 MHz using a 5 mm TCI CryoProbe at 302.3K. A: Histogram of the change in chemical shift for each residue of CCL27, comparing 1.0 mM with 0.025 mM. The monomeric secondary structural elements are displayed above, β-strands in yellow and α-helices in purple. B: Histogram of the change in chemical shift for each residue of CCL2, comparing 1.0 mM with 0.05 mM. The secondary structural elements are displayed above, with monomeric β-strands in yellow, monomeric α-helices in purple, and the N-terminal β-strand formed between the two monomeric subunits in the dimer structure shown in orange. The red line indicates changes in shift above the baseline, based on line broadening.

The histogram in Figure 3.10a shows that the most significant changes clustered to the 30s loop, where Cys38 forms a disulfide bond with Cys9 in the N-terminus, and the C-terminal α-helix/C-terminus. As was the case for the (HC)NH-NOE experiment,

CCL27 does not readily fit the patterns expected for either type of chemokine dimer.

62

In order to compare these results to a known CC dimer, chemical shift changes were also monitored for CCL2 from 1.0 mM, where CCL2 forms a stable dimer, to

0.05mM (Figure 3.10b). As expected, CCL2 shows some of the most significant changes in the N-terminus, in and around the region forming the small β-strand between the monomeric subunits in the dimer structure. The other highly shifted regions include the 310 helix, leading into the first β-strand, the second β-strand leading into the 40s loop

and the third β-strand, where a disulfide bond connects Cys52 to Cys12 in the N-

terminus. A comparison of the two histograms in Figure 3.10 shows that CCL27 and

CCL2 both have similar shift changes in the N-terminus. However, unlike CCL27, CCL2

has almost no changes in the 30s loop and C-terminal α-helix. In a similar analysis,

Veldkamp, et. al. used chemical shift perturbations to evaluate the dimeric interface for

CXCL12 [77]. Their results demonstrated that the regions experiencing the greatest shift

changes involved the first β-strand and the C-terminal α-helix [77], which is to be expected for a standard CXC chemokine dimer [25, 92]. Comparison to CCL2 and

CXCL12 suggests that CCL27 has characteristics of both types of chemokine dimers.

Because CCL2 forms a discrete dimer at 1.0 mM, but forms a tetramer at higher concentrations, this experiment was repeated for CCL27 from 3.0 mM to 0.025 mM, and the results were identical (data not shown).

In order to better visualize the chemical shift changes in Figure 3.10, Figure 3.11 shows model structures of both a CC and CXC chemokine dimer of CCL27, alongside the previously solved dimer structure for CCL2, and the CXC dimer interface of the

CCL2 tetramer.

63

A B

D C

Figure 3.11 Results from chemical shift perturbation mapped to dimeric forms of CCL27 and CCL2, using a color gradient to depict the intensity of chemical shift change – 0.06 – 0.09 ppm in pink, 0.10 – 0.15 in maroon, and >0.16 in red. A: Model of CCL27 forming a CC chemokine dimer. B: Dimeric solution structure of CCL2. C: Model of CCL27 forming a CXC chemokine dimer. D: The CXC dimer-like interface of the tetrameric form of CCL2.

64

The residues experiencing changes in chemical shift were plotted as a color

gradient with pink representing the smaller changes and red indicating the most

significant. In the context of a CC dimer structure, CCL27 has some significant changes

at the dimer interface, including the N-terminus and the 30s loop. However, the very

large changes to the C-terminal α-helix/C-terminus do not fit this interface, and are not

connected to the N-terminus by loops or disulfide bonds (Figure 3.11a). The results for

CCL2, on the other hand, match the interface for the CC dimer quite well. The greatest

shift changes exist at the N-terminus, or are connected to the N-terminal region by either

disordered loops, which would cause movement of the first β-strand, or the disulfide

bond between Cys52 in the third β-strand and Cys12 (Figure 3.11b). When these same

results are mapped to the CXC dimer structures, changes to CCL2 at 1.0 mM do not

appear to correspond to this interface, which makes sense given the dimeric species

predominates at this concentration and the tetramer, which contains the CXC interface,

forms at higher concentrations (Figure 3.11c) [76]. Interestingly, the chemical shift changes to the C-terminal α-helix/C-terminus of CCL27 map very well to the CXC dimer interface (Figure 3.11d). In addition, the changes to the 30s loop also make sense in the context of this dimer. While the patterns for CCL2 at 1.0 mM correspond to the expected

CC dimer interface, CCL27 appears to have characteristics of both CC and CXC dimers.

These results, together with the gradient diffusion and filtered NOEs, suggest that

CCL27 appears to exist as a mixture of oligomeric states throughout the entire concentration range, forming a relatively lose tetramer, with characteristics of both dimer interfaces.

Exchange broadening by relaxation dispersion experiments

The initial comparison of the transverse cross-correlation rate values with the R2

65 values recorded under CPMG fast-pulsing conditions revealed many residues containing

an exchange contribution (see above); consequently we decided to characterize these

Rex contributions using CPMG relaxation-dispersion experiments. Information on

oligomerization may be extracted from an analysis of exchange contributions to the

transverse relaxation. In situations where multiple oligomeric forms are present in

solution, exchange of subunits between the different forms will be manifest as increased

transverse relaxation rates for those sites whose chemical shifts are dependent on the

oligomeric state, such as sites typically present at the interface between the monomeric

subunits in a dimer.

A

40 35 30 25 20

R / Hz 15 10 5 0 64 1F 2L 3L 4P 5P 7T 9C 6S 52I 54I 27I 8A 59P 61L 56P 65F 66E 17P 18L 22L 23L 40L 43F 45L 47L 69E 72L 11T 13L 30E 31L 33E 75T 76L 77P 79L 81F 84L 58N 60S 62S 50R 51S 53C 55H 67H 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 70R 71K 73H 10C 78K 80N 85R 86K 57Q 63Q 41Q 42A 44V 48A 49Q 68Q 12Q 14Y 26V 28Q 29V 32Q 34A 36G 74G 82G 83M 87M 88G Residue B

180º

Figure 3.12 A: Exchange contributions for 1.0 mM CCL27 at 600 MHz from fitting the TROSY-selected spin-echo experiment. B: Two orthogonal views of the monomeric CCL27 structure showing the five exchange “hotspots” identified from the TROSY- selected spin-echo exchange experiment colored red.

66

Exchange contributions, Rex, for 1.0 mM CCL27 were characterized using a

TROSY-selected spin-echo experiment at 600 MHz (3.12) [68]. The results indicate that

Asp35, Asp37, His39 and Gln41 in the 30s loop have exchange contributions approaching 30 Hz, constituting approximately 75% of their overall transverse relaxation rates. In addition, four other exchange “hot spots” include, Cys9, Cys10, Leu13 in the N- loop, Ile27 – Gln32 in the first β-strand, Ile52 – His55 in the third β-strand, and Gln57,

Ser62, Phe65, and Gln68 – Arg70 in the C-terminal α-helix, as well as Leu72 and His73 in the C-terminus. Because residues experiencing significant exchange broadening may be forming the oligomeric interface, these results were mapped onto the monomeric structure of CCL27 in order to compare with the chemical shift perturbation data, and to determine whether they correspond to a known chemokine dimer pattern (Figure 3.12b).

These results, similar to chemical shift analysis, can only be reconciled with more than one interface in equilibrium, and are qualitatively consistent with the presence of both

CC and CXC chemokine dimer types.

Comparison of the rotational diffusion tensor to theoretical

CXC and CC dimer models

Finally, since the rotational diffusion tensor (calculated above by the 15N relaxation

rates) is determined by the size and shape of the molecule, it can in principle, provide

information on both the oligomeric state, as well as distinguish between types of

oligomers/dimers, if the complexes are significantly different in shape. However,

simulated tensor properties of putative CC and CXC type dimer models (constructed

using an in-house docking program based on the monomer structure) did not show

sufficient agreement with the experimental data, even though the simulated tensors

showed clear differences between the CC and CXC models (Table 3.4). These results

67 are consistent with the concept that CCL27 does not adopt a discrete chemokine dimer.

Table 3.4 Magnitudes of the principal components of the anisotropic diffusion tensors for CC-type and CXC-type dimer models of CCL27, as determined using the program HYDRONMR. The first and second CC-type dimers were modelled from the monomeric structure with the flexible C-terminus removed (1) and intact (2). CC-dimer (1) CC-dimer (2) CXC-dimer

7 –1 Dxx / 10 s 0.97 0.80 1.55

7 –1 Dyy / 10 s 0.98 0.82 1.65 D / 107 s–1 2.11 1.63 1.93 zz

One caveat to methods such as chemical shift perturbation analysis and analysis of exchange contributions is that they reflect overall conformational changes, and the signals are therefore not restricted to residues at the protein-protein interface, but instead can include contributions from the entire protein. Additionally, the dimer models constructed for comparison of theoretical and experimental diffusion tensors may not be sufficiently accurate. Nevertheless, the results from all of the above methods reach a consistent picture that at 1.0 mM CCL27, where the dimer form predominates, there are interactions between multiple interfaces that ultimately contribute to the tetramer structure, a behavior which contrasts with many chemokines that adopt discrete oligomerization states at a given concentration.

3.3.5: Analysis of mutants of CCL27 by PFG diffusion NMR

Since the interfacial contacts could not be definitively determined by the above

NMR methods, targeted mutagenesis was performed on regions known to be important for the dimerization of other chemokines. A mutant truncated in the C-terminal region was also investigated, as this region is extended in CCL27 relative to most chemokines.

68

The Ds values were determined at several concentrations for the mutants using PFG

diffusion experiments (Figure 3.13).

A B 1.50 FLL PPSTACCTQLYRKPLSDKLLRKVIQVELQEADGDCHLQAFVSTACCTQLYRKPLSDKLLRKVIQVELQEADGDCHLQAFV WT LHLAQRSICIHPQNPSLSQWFEHQERKLHGTLPKLNFGMLRKMG 1.40 P4A /s 2 P5A

, m 1.30 P4A P5A -10 Series5 1.20

1.10

1.00 Dimer 0.90

Diffusion Coefficient x 10 0.80

0.70 0.25 0.5 1 3 C D Concentration (mM) 1.50 1.50 WT 1.40 WT 1.40 [6-88] /s /s

2 [1-73] 2 [3-88]

, m 1.30 Series3 , m 1.30 Series5 -10 -10 1.20 1.20

1.10 1.10

1.00 1.00 Dimer Dimer 0.90 0.90 Diffusion Coefficient x 10 Diffusion Coefficient Coefficient x 10 x 10 0.80 0.80

0.70 0.70 0.25 0.5 1 3 0.25 0.5 1 3 Concentration (mM) Concentration (mM)

Figure 3.13 PFG diffusion analysis of WT and mutant CCL27, acquired with a Bruker Avance II, 600 MHz, 5 mm TCI CryoProbe, acquired at 302.3K. A: Monomeric structure of CCL27, N-terminal truncation mutant residues shown as lines, [3-88] CCL27 and [6- 88] CCL27 (red), Proline mutations (green), C-terminal truncation mutant [1-73] CCL27 (blue), truncated two residues to the C-terminal side of the α-helix. B: PFG diffusion results for mutations of the two N-Terminal Proline residues to alanine (P4A, P5A, and P4AP5A). C: PFG diffusion results for N-Terminal Truncations, resulting in monomeric variants of CCL7 based on theoretical calculations. D: PFG diffusion analysis of the C- Terminal truncation mutant [1-73] CCL27.

The first set of variants involved mutations P4A and P5A near the N-terminus of

CCL27, since Proline to Alanine substitutions in the N-termini of CCL2 and CCL4 rendered these chemokines incapable of dimerizing even at very high protein concentrations [18]. While individual mutations of P4A and P5A had no effect on

69

oligomerization (Figure 3.13b), the double mutant, P4A/P5A CCL27, had a smaller Ds

value at 0.25 mM compared to WT, suggesting that unlike WT, which most likely has

some oligomer present at 0.25 mM, this mutant is fully monomeric. P4A/P5A also

showed delayed oligomerization, requiring up to 1.0 mM protein concentration before a

decrease in Ds was observed.

As a less residue-biased approach, we also investigated two N-terminal truncation

mutants, since the interface of standard CC chemokine dimers involves residues from

the N-terminal region [82], and the contribution of Pro4 and Pro5 to oligomerization could

not be probed by the above NMR methods because they detect only backbone NH’s.

Two mutants were prepared in which the first two residues, [3-88] CCL27, and the first

five residues, [6-88] CCL27, were deleted. According to the PFG results, neither of

these N-terminal truncation mutants is able to efficiently oligomerize, as only a slight

decrease in Ds is observed with increasing concentration, probably due to weak

interactions at other interfaces (Figure 3.13c) [81]. Finally, because CCL27 has a longer

C-terminal region than most chemokines, the C-terminal truncation mutant, [1-73]

CCL27 was evaluated, but showed little effect on the overall oligomerization pattern

(Figure 3.13d). These results confirm the role of the N-terminus in the oligomerization of

CCL27, consistent with its classification as a CC-chemokine. However, the results do

not rule out the involvement of other regions such as those important for CXC dimers;

since the associations appear to be weak, interfering with a single interface may inhibit

the overall transition to a tetramer.

70

3.3.6: The interaction of CCL27 with glycosaminoglycans

It has become clear that the ability of certain chemokines to oligomerize is relevant to their function. Although some roles of oligomeric forms in signaling have been reported [44], many chemokines oligomerize upon binding to glycosaminoglycans

(GAGs), which causes the accumulation and localization of chemokines on cell surfaces, and the formation of haptotactic gradients for cell migration. We therefore investigated whether the oligomerization behavior of CCL27 might be relevant to interaction with

GAGs.

Equilibrium competition binding on immobilized Heparin

In this assay, Heparin Sepharose beads are incubated with the radioactively labeled chemokine of interest, and the labeled is then competed off with increasing concentrations of the cold chemokine [37]. In general, it would be expected that addition of competitor would simply compete off the radiolabeled chemokine. However, Figure

3.14a shows that when a fixed amount of 125I-CCL27 is incubated with heparin beads,

addition of increasing concentrations of unlabeled WT CCL27 causes the recruitment of

additional 125I-CCL27, reflecting oligomerization on the heparin. This result suggests that heparin increases the propensity for CCL27 to oligomerize, as observed for other chemokines like CCL2, CCL5, and CXCL8 [32] [20, 32].).

71

A B 3 1.2 4

1.0 2 0.8

0.6

1 CCL27 0.4 CCL27 CCL2 CCL2 Specific cpm x bound 10 0.2 Weighted Abs. (280 nm) (280 Weighted Abs.

0-7-6-5-4-3 0 1:4 1:8 1:12 1:16 1:20 Log [WT CCL27 (M)] Chemokine:Octasaccharide (molar Ratio) D 4.5 C Ratio 0 018:1 4:1 2:1 1:1 1:2 :4 Monomer /s 2 3.5 Tetramer m -10 Dimer Dimer x 10 s 2.5 Monomer D Tetramer

1.5 EGS - + ++ + + ++ 0 1:2 1:4 1:8 1:10 CCL2:Octasaccharide (molar Ratio)

Figure 3.14 GAG binding analysis of CCL27 and CCL2. A: Radioactive Heparin- sepharose binding assay of WT CCL27. Concentration of radioligand was fixed at 2 nM 125I-CTACK. Radioactivity specifically bound to heparin, determined by subtracting sepharose-bound from heparin-sepharose-bound radioactivity. B: Solubility analysis of 0.1 mM CCL2 and CCL27 in the presence of increasing amounts of heparin octasaccharide. Solubility was determined by measuring absorbance at 280 nm and normalizing to the value at 0 mM octasaccharide. C: Chemical cross-linking of CCL27 using Sulfo-EGS in the presence of increasing molar amounts of the GAG Heparin decasaccharide. The molar ratios of CCL27:Heparin decasaccharide are indicated above the gel and presence of Sulfo-EGS indicated below. D: 13C-edited PFG diffusion profile for 0.1 mM CCL2 with increasing concentration of heparin octasaccharide.

Soluble Heparin octasaccharide induces oligomerization of CCL27

Ideally, more quantitative information on CCL27:GAG interactions could be determined with solution methods such as analytical ultracentrifugation (AUC), dynamic light scattering, PFG diffusion, and/or fluorescence polarization. Accordingly, we initially attempted PFG diffusion studies with CCL27 and heparin octasaccharide. However,

72 upon GAG addition, a significant attenuation of signal was observed due to precipitation.

This same result was seen with heparin hexasaccharide. Therefore, as a prelude to any further studies, solubility tests of CCL27 with heparin octasaccharide were conducted to profile its behavior in comparison to CCL2. The results, shown in Figure 3.14b, suggest that at 1.5:1 chemokine:octasaccharide (0.1 mM chemokine), the minimum amount of each protein remains in solution. For CCL27, the majority is insoluble, with only 4.5% left in solution. CCL2 only decreases by 40% of its original concentration, but is completely resolubilized at a 1:4 molar ratio of CCL2:octasaccharide. CCL27 on the other hand, only resolubilizes up to approximately 50% of its original concentration and requires a 20-fold excess of octasaccharide. The results at 25 µM showed an identical pattern (data not shown), suggesting that CCL27 oligomerizes more avidly in the presence of GAGs than CCL2.

At 25 and 50 µM, CCL27 was able to remain in solution up to approximately 20% in the presence of Heparin octasaccharide. Therefore, in order to determine if the lack of solubility is due to oligomerization, chemical cross linking was performed using Sulfo-

EGS in the presence of increasing amounts of Heparin decasaccharide. The gel in

Figure 9C suggests that in the absence of GAGs, CCL27 is predominantly monomeric at

50 µM. However, as the concentration of Heparin decasaccharide increases, the intensity of bands corresponding to dimeric, tetrameric, and higher oligomeric forms are apparent. When the GAG is in excess and CCL27 stabilized between monomer and dimer, as GAG competes for CCL27 binding sites. This dissociation from higher order oligomers to monomer/dimer is most likely a consequence of the relatively small size of the GAG and the excess GAG concentration. In contrast, cross linking studies performed with increasing concentration of CCL27 in the absence of GAGs also indicate

73 the presence of dimer and tetramer, however not nearly to the extent seen in the

presence of GAGs (Figure 3.15A). Similarly, the monomeric mutant [6-88] CCL27

showed a decreased propensity to oligomerize on GAGs (Figure 3.15b)

A mM 0.01 0.05 0.1 0.25

Tetramer Dimer

Monomer

EGS - +-+-+-+

B Ratio 0 0 8:1 4:1 2:1 1:1 1:2 1:4

Tetramer

Dimer

Monomer

- +++++++ EGS

Figure 3.15 Chemical cross linking of CCL27 using Sulfo-EGS. A: Concentration analysis in the absence of GAGs. Concentration is indicated above the gel and the presence of Sulfo-EGS is indicated below. B: The monomeric mutant [6-88] CCL27 with increasing amounts of Heparin decasaccharide. The [6-88] CCL27:Heparin decasaccharide ratio is indicated above the gel and the presence of Sulfo-EGS is indicated below.

Finally, because of the lack of solubility at higher concentrations made it impossible to quantify the extent of CCL27 oligomerization in the presence of GAGs by

PFG diffusion NMR, we examined CCL2, which remains in solution up to 60% at 0.1 mM. In order to confirm that the insolubility at higher concentrations equates to

74 oligomerization, a 13C-edited gradient diffusion experiment was used to determine the effect of heparin octasaccharide on the diffusion of 13C-labeled CCL2. Because only 1H-

13C signals are detected in this experiment, the proton peaks from the octasaccharide do

not interfere with the analysis. The results in Figure 3.14c show that at 2:1

CCL2:octasaccharide, the Ds value decreases from monomer in the absence of GAG to the theoretical value of the tetramer in the presence of GAG. This is consistent with previous AUC data which also showed that MCP-1 forms a tetramer in the presence of octasaccharide, (not shown, [13]). Furthermore, it suggests that the observed insolubility is associated with increasing size of the complex. At a molar ratio of 1:1 and higher, the Ds value corresponds to approximately that of a dimer, as excess GAG

competes for binding sites on CCL2. This dissociation from tetramer to dimer is

probably non-physiological however, and is a consequence of the relatively small size of

the GAG and the excess GAG concentration. The main conclusion to be made is that

the insolubility correlates with oligomerization of CCL2; thus it is likely that GAG-binding

also stabilizes CCL27, at least in a tetramer state, but unfortunately its solubility behavior

precludes studies by AUC and PFG diffusion.

75

3.4 Discussion

The structure of CCL27 was solved by NMR and has the standard motif expected

for a monomeric chemokine. Analysis of the oligomeric properties of CCL27 confirms

that it oligomerizes on its own in solution. In addition, the results suggest that CCL27

exists as a mixture of oligomeric states with the higher order forms stabilized by the

presence of GAGs. While the oligomeric interface of CCL27 has some characteristics of

known chemokine dimers, this study suggests it is forming a tetramer with a potentially

novel interaction.

The functional role of chemokine oligomerization is not entirely understood.

There is however, a direct correlation between the ability of chemokines to oligomerize

and to induce in vivo cellular migration. Proudfoot, et. al. previously demonstrated that

monomeric variants of CCL2, CCL4 (MIP-1β), and CCL5 (RANTES) were inactive to

cellular migration when injected into mice [20]. More recently Handel, et. al. showed that the same monomeric form of CCL2 has anti-inflammatory activity in animal models of experimental arthritis [18]. It is widely believed that chemokine oligomerization is involved with in vivo GAG interaction, since it is the monomeric form responsible for binding and activating chemokine receptors to induce cellular migration in vitro. This idea is apparent in studies of GAG-deficient mutant chemokines, which behave similarly to the monomeric mutants in that they are also active in vitro but inactive in vivo [20]. An excellent example is the GAG-deficient mutant [44AANA47]-CCL5, which is not only

inactive to migration in vivo, but also acts as an inhibitor by forming non-functional

heterodimers with WT CCL5 [12, 19]. Given the sheer forces at the endothelial surface,

it is believed that GAG interaction aids in maintaining high localized chemokine

76 concentrations, allowing for receptor interaction and leukocyte extravasation. Due to the heterogeneity and specificity of GAG expression, the ability of chemokines to adopt multiple oligomeric forms greatly increases their biological diversity in terms of interaction with different GAGs at the tissue surface.

In this study, the oligomeric profile for CCL27 differs with the results seen for the chemokines CCL2 and CXCL12 [77], both of which form stable dimers at micromolar concentrations. In those cases, the Ds value rapidly decreases and remains constant,

approximately corresponding to that of a dimer. In contrast, the gradual linear decrease

in the diffusion coefficient for CCL27 suggests the presence of monomer, dimer, and

tetramer over the concentration range of 0.5 – 3.0 mM, with the larger species

predominating at the higher concentrations. This also indicates that the CCL27

tetramer forms at a lower concentration than that of CCL2, which appears to be mostly

dimeric up to 3.0 mM. The results for CCL27 are indicative of a more dynamic oligomer,

possibly leading to increased diversity in terms of GAG binding partners.

Attempts to characterize the oligomeric interface of CCL27 involved an extensive

analysis by multi-dimensional NMR spectroscopy, utilizing rotational diffusion, 15N-1H

HSQC chemical shift perturbation, relaxation dispersion experiments, and filtered

(HC)NH-NOEs. Several of these experiments were run in parallel with CCL2 in order to

compare CCL27 to a chemokine forming a stable dimer with a known structure. Taken

together, the results support the translational diffusion data in that they strongly suggest

CCL27 is forming a tetramer, displaying characteristics of both CC and CXC chemokine

dimers. This idea is most clearly illustrated by mapping changes to chemical shift onto

77

CC and CXC dimer models for CCL27, showing residues at the interface of both structures, as opposed to CCL2, which maps predominantly to the CC form.

Based on the behavior of other chemokines, this unique oligomerization pattern for CCL27 may play a role in both GAG and receptor specificity [12, 19]. For example, the chemokine CXCL10 has three tetrameric forms, each with a unique GAG-binding epitope. This suggests that different GAGs help to stabilize the different oligomeric forms of CXCL10. [25]. Lau, et. al. also demonstrated that the GAG heparin octasaccharide induces the formation of the CCL2 tetramer. A detailed analysis of the

GAG binding sites then revealed a continuous ring encircling the oligomeric interface, suggesting the octasaccharide binds near to the interface, stabilizing the tetrameric form

[13]. In addition, Veldkamp, et. al. has recently shown that dimerization of CXCL12 promotes binding to the sulfatyrosines of its receptor CXCR4, suggesting oligomerization plays a role in chemokine receptor specificity [44]. In this study, the heparin binding

assay suggests that immobilized heparin causes radiolabeled CCL27 to recruit

additional CCL27, as evidenced by the bell-shaped pattern of the binding curve [20]. At

the same time, the solubility assay and edited DOSY experiments suggest that heparin

octasaccharide stabilizes the oligomeric forms of both CCL27 and CCL2. Taken

together, these results suggest that the ability of CCL27 to readily convert between

oligomeric states may allow for multiple binding partners, thus increasing its functional

diversity.

Although the monomeric structure for CCL27 is very similar to other chemokines,

its oligomeric form is more unique. CCL27 appears to form a tetramer with some

characteristics of both CC and CXC chemokine dimers, similar to the CCL2 tetramer and

78 the M form of the CXCL10 tetramer [25, 76]. Although the monomeric forms of chemokines are structurally very similar to one another, they are highly varied in terms of receptor specificity and tissue localization. The unique oligomeric structure of CCL27 may therefore contribute to its specificity for CCR10 as well as its localization to skin cells, through interaction with different GAGs. Further studies of CCL27 will involve a detailed analysis of GAG-binding regions, as well as residues necessary for the activation of CCR10, in order to better understand how the structure of this chemokine influences its function.

79

3.5 Conclusions

The monomeric solution structure of the human chemokine CCL27 was solved by NMR spectroscopy. While the structural and dynamic properties of the CCL27 monomer reflect the standard chemokine motif, the oligomeric form does not directly correspond to that of other known chemokines. The combined results from gradient diffusion, relaxation studies, chemical shift perturbation analysis, filtered NOE experiments, and finally targeted mutagenesis, suggest that CCL27 exists as a mixture of monomer, dimer, and tetramer. The oligomeric interface appears to have characteristics of both CC and CXC chemokine dimers as well as some similarities to the CCL2 tetramer. The heparin binding assays also suggest that this unique oligomeric structure of CCL27 plays a role in GAG binding.

80

3.6 Materials and Methods

3.6.1: Protein expression and purification

All labeled and unlabeled proteins were expressed by the ubiquitin fusion

system, described in detail in Chapter 3.

3.6.2: NMR spectroscopy

PFG diffusion measurements

All experiments were run on a Bruker Avance II 600 MHz with a 5 mm TCI

CryoProbe at 302.3K. The gradient field strength was initially calibrated using a doped

water sample[81]. The 15N-labeled CCL27 and CCL2 samples were prepared in 50 mM

acetate-d6, pH 5.6. The experiments were carried out using the pulse sequence

ledbpgpprwg2s with the Bruker macro DOSY, which employed a linear increase in

gradient field strength from 2%-95%, running 16 successive 1D proton experiments.

Values for the diffusion time, d20 (∆) and the gradient pulse length, p30 (*0.5) were 150

ms and 2.0 ms respectively. Water suppression was achieved using a combination of

Watergate and pre-saturation sequences. The self-diffusion coefficients (Ds) were calculated using the Bruker program T1/T2 relaxation, detailed in Appendix I-A, with manual integration for peaks at 7.0 ppm, 3.0 ppm, 2.0 ppm, and 0.7 ppm for each proton spectrum.

Monomeric structure determination

The assignment experiments recorded were: 15N HSQC, HNCA, intra-HNCA,

HN(CO)CA, HNCACB, HN(CO)CACB, HNCO, HN(CA)CO and (H)NNH-NOESY, 13C

HSQC, (H)CC(CO)NH-TOCSY and H(CC)(CO)NH-TOCSY (14.3 ms DIPSI mixing),

81

TOCSY–15N HSQC (43.3 ms DIPSI-2rc mixing), H(C)CH-TOCSY, methyl-selective

H(C)CH-TOCSY and (H)CCH-TOCSY (18.6 ms FLOPSY-16 mixing). Data were processed using the Azara suite of programs (version 2.7, W. Boucher, University of

Cambridge). The assignment of 15N/13C-labeled CCL27 was performed by determining

sequential connectivities of backbone resonances using a pair-wise approach via

frequency-matching in the 13C dimension. Partial side-chain assignment and connection between the backbone and side-chains was accomplished using 3D TOCSY–15N HSQC,

H(CC)(CO)NH and (H)CC(CO)NH experiments. Side-chain assignment was completed

using the 3D H(C)CH-TOCSY experiment analyzed in conjunction with the constant-time

13C-HSQC spectrum. The assignment of the HCCH-TOCSY spectrum in the crowded

methyl regions was expedited by acquisition of methyl-selective implementations of both

the H(C)CH- and (H)CCH-TOCSY 3D experiments, in which only methyl-type protons

are detected during acquisition. Assignments for aromatic resonances were obtained

from the 13C-separated NOESY spectrum, aided by reference to a constant-time 13C

HSQC spectrum optimized for aromatic spin systems, as well as 2D NOESY spectra

acquired with and without 15N decoupling.

Generation of distance restraints for structure calculations proceeded with

assignment of 15N-separated and 13C-separated NOESY spectra. Identification and assignment of medium-range NOEs within helical secondary structure elements, and inter-strand NOEs within β-sheet regions were facilitated by the secondary structure prediction provided by the Chemical Shift Index (CSI) [93]. Structure calculations were performed with ARIA (ambiguous restraints for iterative assignment) [89, 90] interfaced to CNS (crystallography and NMR system) [94] using a combination of molecular dynamics in torsion-angle and Cartesian space. Dihedral angle restraints were

82 generated from the backbone chemical shifts using the program TALOS [88], and amide

hydrogen bonds were included in the structure calculation as pseudo-NOE distance

restraints. Amide hydrogen bond donors were determined using the CLEANEX

experiment, which identifies amide protons that are protected from solvent exchange

[95]. The hydrogen bond acceptors were inferred from the secondary structure

prediction in combination with the observed NOE patterns within the secondary structure

elements. In short helical regions and at the ends of long helices, the identity of the

hydrogen bond acceptor may be ambiguous depending on the type of helix (α-helix or

310 helix). In these regions, both possibilities for the hydrogen bond acceptor were incorporated by modeling the hydrogen bonds as ambiguous distance restraints.

CCL27 contains two disulfide bridges that stabilize the tertiary fold. The regiochemistry of these disulfide bonds was determined by running a complete structure calculation in which no information on the disulfide bridges was provided. The measured distances between the sulfur atoms in the calculated structures conclusively identified the disulfide bridges to be between Cys9 and Cys38, and between Cys10 and Cys53, which were then included as restraints.

Nine iterations were performed for the final structure calculation. Of the 100 structures calculated in the last iteration, the 30 lowest in energy were selected for water refinement and analysis. For resolution of ambiguous distance restraints, the total contribution of the accepted NOEs to an ambiguous cross-peak was reduced during the calculation from 1.0 to 0.8. The number of ambiguous restraints was reduced from 1204 to 562 during the course of the structure calculation. For the final calculation, a mean structure was generated from the ensemble of the 30 water-refined structures. The

83 ensemble and the structure closest to the mean were then analyzed with respect to their

coordinate precision, energy and deviations from the experimental restraints and

idealized geometry (Supplemental Table 1). Ramachandran analysis was performed

using the programs PROCHECK and PROCHECK-NMR [96, 97].

15N relaxation measurements

Relaxation experiments for determination of the rotational diffusion tensor and characterization of the internal dynamics were recorded on 1 mM 15N-labeled CTACK at

15 1 15 fields of 500 MHz and 600 MHz. N R1 and R2 relaxation rates, and [ H] N heteronuclear NOEs were measured using standard pulse sequences. The transverse cross-correlation rate constant, ηxy, was measured using three methods: via cross-

correlation-mediated interconversion of 15N in-phase and anti-phase magnetization (at

both 500 MHz and 600 MHz) [98], via selection of the two lines of the 15N doublet, and comparing their intensities (at 500 MHz only), and from the TROSY-selected spin echo experiment (at 600 MHz only) [20].

Relaxation data curve-fitting was performed using CcpNmr analysis. Errors in the relaxation rate constants were estimated using a boot-strapping method, which repeats the datafitting many times with data sets generated from the original data sets by resampling with replacement. Errors in the heteronuclear NOE values were calculated from the peak intensities and noise levels in the reference and saturated spectra.

Rotational diffusion tensor fitting and analysis of internal mobility were performed using the program TENSOR2 [86]. The rotational diffusion tensor was determined from

84

0 1 15 the 2 RR 1 ratios for those residues that exhibited [ H] N heteronuclear values greater

0 than 0.60 at 500 MHz. R2 rate constants were calculated from ηxy according to

0 R2   xy , using values for κ of 1.423 and 1.292 at 500 MHz and 600 MHz,

respectively. For the analysis at 500 MHz, η xy rate constants were calculated as the

averages of the two values obtained using the in-phase/anti-phase interconversion and

H1/H1 selection methods, with the uncertainties taken as the corresponding standard deviations. Similarly, η xy rate constants at 600 MHz were calculated as the averages of

the values obtained from the in-phase/anti-phase interconversion experiment and those

extracted from the TROSY-selected spin-echo exchange experiment. The uncertainties

in the R1 rate constants used in the TENSOR2 analysis were taken to be twice those

computed using the boot-strapping procedure in CcpNmr analysis. Ignoring negligible

contributions from high-frequency spectral density terms the measured xy values were

0 0 2 2 2 converted to exchange-free R2 rate constants using k = R2 / xy = (c + d ) / (cd(3cos 

– 1), The angles between the 15N CSA tensor and the N–H bond, θ, was set at –17°, and the amide bond length, rNH, was set at 1.02 Å. The errors in the magnitudes of the principal components of the diffusion tensor were computed from 500 simulated data- sets constructed through Monte-Carlo sampling of the Gaussian error distributions defined by the experimental uncertainties. These 500 data-sets were also used to generate the simulated 2 distribution, the value of which at the 95 % confidence interval was compared to the experimental 2 value to assess the agreement of the diffusion

tensor model with the experimental data. The fitting of the internal mobility was

0 performed for all residues with R1, R2 and NOE data using the fully anisotropic diffusion

tensor.

85

Dimer and hydrodynamic modeling

Hydrodynamic modeling of the rotational diffusion tensor was performed using

the program HYDRONMR [99], which employs bead-modeling to simulate hydrodynamic

properties of globular proteins [99, 100]. The diffusion tensor for the CC-dimer model is

significantly elongated, and would be closely approximated by a heavy prolate axially

symmetric tensor. In contrast, the diffusion tensor for the CXC dimer is almost isotropic

with a slightly prolate tensor the closest axially symmetric approximation. The primary

hydrodynamic models were generated using an atomic element radius of 3.2 Å.

Secondary shell modeling was performed for 6 values of the bead radius from 1.2 Å to

2.2 Å.

Exchange broadening

Relaxation-compensated CPMG relaxation-dispersion profiles were recorded for

1 mM 15N-labelled CTACK at 500 MHz and 600 MHz [101]. In addition to a reference spectrum acquired without a relaxation period, 16 sub-spectra were acquired from

 CPMG  7.7 Hz to  CPMG  1000 Hz. The data was fit using the program CPMGFit (A.

Palmer, Columbia University). R 1  cp2  values were calculated from the intensities of

the sub-spectra, I1  cp , and that of the reference spectrum, Iref:

1  I  R 1   ln ref  cp2     I1  cp  where  is the length of the constant-time period (80 ms).

Exchange contributions at 600 MHz were also measured using a version of the TROSY-

selected spin echo experiment described by Wang et al. [20]. Three TROSY-type 15N–

86

1H spectra are recorded such that the peak intensities are proportional to the relaxation

rates R(NxH ), R(Nx/2NyHz) and R(2NzHz). By rearranging the expression for the relaxation rate of the TROSY-line, Rex can be written as:

 1 0 ex  x HNRR )(  xy  2 z )(  RHR 2

0 For proteins with tm>4ns, R2   xy and z  zz  NRHNRHR z )()2()( and therefore:

 1 1 ex  x HNRR xy   2 zz  2 NRHNR z )()2(1)(

 Since  xy  zyx  x HNRHNNR )()2/( , values for xy can be calculated from the peak intensities in the corresponding sub-spectra via:

1  HNI  )(    ln x  xy   2  HNNI zyx )2/( 

and therefore Rex is given by:

1  HNI )2(   1  HNI  )(  NR )( R  ln zz   ln x   z ex      2  x HNI )(  2  HNNI zyx )2/(  2

1 The length of the relaxation delay is 2 in the R(NxH ) and R(Nx/2NyHz) sub-spectra, and η

in the R(2NzHz) sub-spectrum. R(Nz) values are measured in a separate experiment.

1H-15N HSQC chemical shift perturbation analysis and filtered (HC)NH-NOE

Experiments were performed using a Bruker Avance II 600 MHz NMR system

with a 5mm TCI CryoProbe. Samples were prepared in 50 mM acetate-d6, pH 5.6 over a concentration range of 0.025 mM – 3.0 mM. All experiments were run at 302.3K.

Spectra were processed and analyzed using the programs nmrPipe and Sparky.

Assignment of resonance peaks was made using the data from the monomeric structure

87 determination and chemical shift perturbation values were calculated using the following equation:

To determine NOEs corresponding to the oligomeric interface, 15N-1H 2D

versions of the 4D 13C/15N-separated (HC)NH-NOESY experiment were recorded on 3 mM, 2 mM and 1 mM samples which were 50% 15N CCL27 (13C-depleted) and 50% 13C

CCL27.

3.6.3: Heparin binding assay

125I-CCL27 was purchased from Perk-Elmer. 125I-CCL27 and unlabeled WT

CCL27 at different concentrations were added to wells of a 96-well plate. To labeled

and unlabeled chemokine were added 100 µg of Heparin-sepharose or Sepharose 6

Fast Flow, to a final volume of 60 µL. The 96-well plate was covered and incubated for 2

hours at room temperature with gentle agitation. During this time, a GF/C filter plate was

soaked in 0.5% polyethylinimine. After incubation, the filter plate was washed with 50

mM Hepes, pH 7.2, 5 mM MgCl2, 1 mM CaCl2, and 0.1% BSA. The contents of the 96- well plate were harvested onto the filter plate by aspiration, as were the contents of two subsequent washes of the 96-well plate with 200 µL of wash buffer. The filter plate was then immediately washed with 400 µL of wash buffer and allowed to dry under a heat lamp. After drying, 50 µL of scintillation fluid were added to each of the wells, the top and bottom of the filter plate were sealed, and the radioactivity counted with a scintillation plate reader.

88

3.6.4: Solubility assays

The protein concentration was determined using a nanodrop spectrophotometer, measuring the absorbance at 280 nm. The concentration was determined in the absence of heparin octasaccharide and these values used to determine the percent chemokine in solution in the presence of octasaccharide.

3.6.5: Chemical cross-linking

Reactions were performed in 50 mM Hepes, pH 7.2 with 5 mM sulfo-ethylene glycolbis(sulfosuccinimidylsuccinate (Sulfo-EGS, Pierce) for 1 hour at 25ºC, and quenched with 1 M Tris.

89

3.7 Acknowledgements

Chapter 3 is currently being prepared for the submission of publication of the material. The following are the contributing authors: Jansma, A.; Kirkpatrick, J.; Hsu, A.;

Handel, T. M.; Nietlispach, D.

CHAPTER 4

Characterization of the Interaction of CCL27 with

Glycosaminoglycans and with the Human CCR10

Chemokine Receptor

4.1 Summary

The chemokine network consists of highly complex interactions between chemokines and their receptors as well as chemokines and Glycosaminoglycans

(GAGs). One aspect of this complexity is that some chemokines are specific to one receptor and vice versa, while other chemokines activate several receptors expressed on a variety of cell types and in some cases, elicit different cellular responses. In this study, we focus on the human chemokine CCL27 and how its structural properties affect interaction with its receptor CCR10 as well as the GAGs Heparin, Heparin tetrasaccharide, and Heparin hexasaccharide. Through targeted mutagenesis coupled to cellular migration, calcium flux, and receptor desensitization assays, our results indicate that the N-terminal Phenylalanine residue is necessary for optimal function, deletion of the first three and five residues results in mutants with antagonist properties, and the C-terminal region plays a significant role in CCL27 activity. Comparing these results to CCL28, the nearest sequential homolog to CCL27, it is apparent that chemokine receptor specificity cannot be determined by the primary sequence alone.

Finally, looking at CCL27 interactions with Heparin from a residue-specific binding assay, as well as with tetra- and hexasaccharide, through a more global analysis of

90 91 chemical shift perturbations, the data suggests that GAGs interact with a region of basic surface-exposed residues that cluster mainly to the N-terminal 310 helix, the 30s loop

and the third β-strand. Using both CC and CXC dimer models for CCL27, it is possible

that different oligomerization states may be stabilized through interaction with different

GAGs, suggesting an additional role for oligomerization in the biological diversity of this

chemokine.

92

4.2 Introduction

CCL27 also known as the Cutaneous T-cell Attracting Chemokine (CTACK), is a member of the chemokine family, composed of small chemoattractant proteins responsible for controlling the migration of leukocytes during both routine immune surveillance and times of inflammation. It has been shown that inappropriate expression of these ligands or their receptors contributes to a broad spectrum of diseases, including a wide variety of inflammatory disorders, cancer metastasis, and HIV [1-5].

Chemokines function by binding to seven transmembrane G-protein coupled receptors

(GPCRs) expressed on immune cells resulting in conformational changes in the receptor that trigger cascades of intracellular signaling pathways involved primarily in G-protein activation and cell movement [1-6].

Chemokines are categorized into four subfamilies (CC, CXC, CX3C, and C) based on the relative position of their conserved N-terminal cysteine residues, which form disulfide bonds with conserved cysteines found elsewhere in the primary sequence

[7-9]. There are currently 50 known human chemokines and 30 known receptors. While chemokines overall have relatively low sequence homology, they are basic proteins and their monomeric structures are very similar, consisting of an unstructured N-terminal region, followed by three anti-parallel β-strands, and a C-terminal α-helix [7, 9-11]. It has

been postulated that the process by which chemokine receptors are activated by their

ligands involves first the interaction of basic residues on the chemokine surface with

acidic residues on the extracellular loop regions of the receptor. This enables the N-

terminus of the chemokine to be inserted into the receptor’s intermembrane helical

93 bundle and induce conformational changes which allows signaling across the membrane

and the in vivo activation of G-protein [6, 7, 9, 14, 15, 17].

In order to facilitate this interaction with their receptors, it is believed that high

localized concentrations of chemokines form gradients from the tissue surface to the site

of inflammation and these gradients are maintained through interactions between

chemokines and surface expressed carbohydrates known as glycosaminoglycans

(GAGs) [13, 18-20]. Chemokine:GAG interactions are quite complex in that GAGs are

believed to enhance chemokine oligomerization and chemokine oligomerization in turn

promotes GAG binding. This idea has been supported by previous studies showing that

while monomeric variants of functionally oligomeric chemokines are able to interact with

their receptors in vitro due to lack of shearing forces, their in vivo function as well as

their GAG binding in vitro is significantly compromised [18, 20, 31, 32]. Due to

differential expression and localization of GAGs in different tissues, and the fact that

some chemokines are able to bind certain GAGs and not others, it has been postulated

that chemokine tissue specificity is not only dependent on receptor binding but may also

be related to GAG interaction [13, 18-20]. This idea is further supported by the fact that

while some chemokines bind to one receptor and vice versa, there are many cases

where multiple ligands are able to activate multiple receptors with different tissue

specificities and cellular functions. Taken together, this functional redundancy suggests

that the relationship between chemokines, chemokine receptors, GAGs, and tissue

specificity is highly complex [3, 7, 15].

CCL27 is a 10 kDa CC chemokine with 88 residues that activates the chemokine

receptor CCR10. It is constitutively expressed in keratinocytes predominantly in the skin

94 and along with CCR10, is upregulated during times of inflammation[63, 64]. CCL27 adopts the standard chemokine monomeric structure and while the details of its oligomeric structure are currently unknown, it appears to have a novel higher order state, which we have shown previously to be enhanced by the presence of the GAG heparin

[102]. CCL27 is ~40% sequentially homologous to another CC chemokine CCL28 (the

Mucosal Epithelial Chemokine, MEC), the only other known ligand for CCR10 (Figure 1).

CCL28 was discovered more recently and unlike CCL27, which is specifically expressed in skin, CCL28 is constitutively expressed in the epithelial cells of virtually all mucosal tissues [66-68]. In addition to CCR10, CCL28 also binds and activates another chemokine receptor CCR3, which has nine other chemokine ligands, CCL5/RANTES,

CCL7/MCP-3, CCL8/MCP-2, CCL11/Eotaxin, CCL13/MCP-4, CCL15, CCL16, CCL24, and CCL26 [7]. CCL27 and CCL28 represent an excellent example of the apparent redundancy of the chemokine network in that CCL27 appears to be specific for the receptor CCR10 and localized to the skin, while CCL28 is broad in its tissue expression and binds CCR10, a receptor with only two known ligands, and the more promiscuous receptor CCR3. In order to elucidate the details of CCL27 interaction with CCR10, we identified the specific CCL27 residues responsible for CCR10 activation and how they compare with CCL28. In addition, we probed CCL27:GAG interaction from both a residue-specific as well as a global perspective, using Heparin and Arixtra as representative GAGs.

Our results suggest that CCL27 requires a Phenylalanine at its N-terminal position in order to maintain full functionality and surprisingly, the C-terminal region of

CCL27 but not CCL28 seems to play a major role in receptor activation. Comparison of the sequences for CCL27 and CCL28, along with results from several other specific

95 point mutations, suggest that receptor activation is not necessarily determined by the

sequence of the ligand’s N-terminus, but involves a considerably more complicated

mechanism. We have also shown previously that Heparin binding promotes

oligomerization of CCL27 [102]. The results from the GAG binding studies here suggest

the residues involved in GAG interaction cluster to several positively charged surface

exposed regions within the N-terminal α-helix, the loop between the first and second β-

strands, and the third β -strand. Our overall goal was to better understand some of the

driving forces behind the specificity of CCL27 both in terms of receptor activation and

GAG interaction.

96

4.3 Results

Analysis of CCL27:CCR10 interactions

In the present study, three assays were used to determine receptor activation: chemotaxis, calcium flux, and desensitization measured by calcium flux. All assays were performed using mouse lymphoma L1.2 cells stably transfected with the human chemokine receptor CCR10. In order to determine the effect of a mutation in terms of chemotaxis, the potency (the amount of chemokine required for maximal cellular migration) and efficacy (the maximum number of cells that migrate) were evaluated for each mutant and compared to WT. Calcium flux measures the amount of cellular calcium produced in response to receptor activation. Finally, because chemokine receptors are internalized following activation of G-protein, desensitization assays help to determine whether the mutant forms of CCL27 are inducing CCR10 activity that results in cellular internalization [7].

4.3.1: Comparison of CCL27 and CCL28

Alignment of CCL27 and CCL28 shows they are approximately 42% homologous and form two overlapping disulfide bonds (Figure 4.1a). CCL28 has an addition pair of cysteine residues which form a third disulfide bond. The monomeric structure of CCL27, shown in Figure 4.1b, was solved by NMR and has a standard chemokine motif [102].

Figure 4.1b also shows a homology model structure for CCL28, generated using the sequence alignment. This structure demonstrates the most likely position for the third disulfide bond, which appears to aid in stabilizing the C-terminal domain. The N-terminal regions of CCL27 and CCL28 are very similar, consisting mainly of the hydrophobic residues Ile, Leu, and Pro. Several differences include the Phe at position 1 (Phe1),

97 which is replaced by an Ile in CCL28 and a second Pro at position 5 of CCL27, which is

absent in CCL28.

A

CCL27 FLLPPSTACCTQLYRKPLSDKLLRKVIQVELQEADGDCHLQAFVLHLAQRSICIHPQNPS 60 CCL28 -ILPIASSCCTEVSHH-ISRRLLERVNMCRIQRADGDCDLAAVILHVKRRRICVSPHNHT 58 [1-73] CCL27 LSQWFEHQERKLHG------TLPKLNFGMLRKMG----- 88 CCL28 VKQWMKVQAAKKNGKGNVCHRKKHHGKRNSNRAHQGKHETYGHKTPY 105 [1-81]

B CT

CT

NT NT

Figure 4.1 A: Sequence alignment of CCL27 and CCL28. Disulfide bonds are indicated by lines and the two C-terminal truncations, at position 73 in CCL27 and 81 in CCL28, discussed later in the text, are indicated in red. B: Monomeric structure of CCL27 (green) with the two disulfide bonds shown in yellow and a homology model of CCL28 (cyan) based on sequence alignment with CCL27. Two of the three predicted disulfide bonds align with CCL27 and the third pair of cysteines are indicated in yellow, with the expected disulfide bond represented by an arrow.

4.3.2: Addition of an amino-terminal Phenylalanine to CCL27 produces a super-

agonist of human CCR10

Figure 4.2a shows that wild type CCL27 induces migration of CCR10-expressing

L1.2 cells resulting in a typical chemotaxis profile with an average potency of 50 nM.

98

Because CCL27 has a Phe at the N-terminus, and CCL28 does not have an aromatic

residue in the N-terminal region, the first mutation involved the addition of a second Phe

at the N-terminal position of CCL27, in order to evaluate its contribution to the activation

of CCR10. The resulting construct, labeled F-CCL27, displayed super-agonist activity in

all three functional assays. Figure 4.2a shows a 5-fold increase in migration potency

compared to WT. Because the signaling domain of chemokines has been shown to

involve several residues within the N-terminus [6, 7, 16, 103], a second CCL27 construct

was designed with a Leu residue at the N-terminal position, labeled L-CCL27, in order to

determine the specificity of the Phe in terms of a super-agonist. Figure 4.2a

demonstrates that the addition of Leu has no effect on CCR10 migration. The calcium

flux data shows significantly increased activity for F-CCL27 compared to WT (Figure

4.2b).

99

A Chemotaxis with CCR10 transfected L1.2 cellsB Calcium flux with CCR10 transfected L1.2 cells 0.80 300

0.70 WT CCL27 250 F-CCL27 F-CCL27 0.60 L-CCL27 200 0.50

0.40 150

0.30

Fluorescence 100 0.20 Cells Migrated/Total Cells Cells 50 0.10

0.00 0 0 10-9 10-8 10-7 10-6 0 0.01 0.1 0.5 1 5 25 50 100 Concentration (M) Time (s) C D Desensitization Assay, 250 nM F-CCL27 Desensitization Assay, 250 nM WT CCL27 250 250

200 200

150 150 250 nM WT CCL27 250 nM F-CCL27 100 100 Fluorescence Fluorescence

50 50

0 0 0 100 200 300 400 500 600 0 100 200 300 400 500 600 Time (s) Time (s)

Figure 4.2 Functional analysis of F-CCL27, using L1.2 cells transfected with the chemokine receptor CCR10. A: Chemotaxis assay for WT, F-CCL27, and L-CCL27. B: Maximum fluorescence signals from calcium influx assays for WT and F-CCL27. C: Desensitization assay measuring calcium influx after an initial addition of 250 nM F- CCL27, followed by addition of 250 nM CCL37 at 300 s. D: Desensitization assay measure calcium influx of an initial addition of 250 nM WT CCL27, followed by addition of 250 nM F-CCL27 at 300 s.

The results for the desensitization assays are shown in Figures 4.2c and d.

Addition of 250 nM WT CCL27 to 250 nM F-CCL27 produces the expected result – the calcium influx is significantly decreased because the surface receptor was internalized following the initial addition of F-CCL27 (Figure 4.2c). However, the addition of 250 nM

F-CCL27 to 250 nM WT CCL27 elicits a different response (Figure 4.2d). In this case,

100

F-CCL27 causes a greater increase in calcium flux than the initial addition of WT,

suggesting that WT does not desensitize the receptors to F-CCL27. Taken together, the

calcium flux data in Figure 4.2 suggest that, in addition to increased potency to

migration, F-CCL27 has an increased affinity for CCR10 compared to WT, and is able to

compete with WT for CCR10 activation. It is interesting that the presence of Phe would

have such a significant impact on both the binding and activation of CCL27 to CCR10,

given that CCL28 does not have an N-terminal aromatic residue. This suggests a

potentially different mechanism for CCR10 activation between these two chemokines.

4.3.3: Amino-terminal modifications of CCL27 produce partial agonists

Mutation of Phe1 to an Ala (F1A) resulted in a 5-fold decrease in potency and significantly decreased efficacy. The results are shown in Figure 4.3a and this type of response was termed a partial agonist. To further evaluate the importance of Phe1, the

N-terminal truncation mutant [3-88] CCL27, which removes Phe1 and Leu2 putting a Leu at the first position, was designed and tested for activation of CCR10 induced migration.

Figure 4.3a shows that [3-88] CCL27 has the same 5-fold decrease in potency and a similar decrease in efficacy as F1A when compared with WT. These results combined with the L-CCL27 migration assay suggest that a Leu residue by itself at the N-terminus will not induce migration in CCR10 to the same extent as WT, unless the Phe is present within the first two residues of the sequence. Figure 4.3b shows the results of the desensitization assays for F1A and [3-88] CCL27 compared to WT. Both mutants show very low initial calcium flux followed by a large signal after the addition of 250 nM WT.

These results suggest that these mutations significantly compromise the ability of CCL27 to activate CCR10 in a G-protein dependent pathway.

101

A Chemotaxis Assays with CCR10 B Desensitization Assays with CCR10 0.70 250 WT CCL27 250 nM WT 0.60 F1A 200 250 nM F1A 0.50 [3-88] 250 nM [3-88]

0.40 150 250 nM 0.30 100 WT CCL27

0.20 Fluorescence

Cells Migrated/Total Cells Migrated/Total Cells 50 0.10

0.00 0 -9 -8 -6 0 10 10 10-7 10 0 100 200 300 400 500 600 Concentration (M) C D TIme (s) 0.80 250

0.70 WT 200 0.60 P4AP5A 250nM 0.50 150 WT CCL27 0.40

0.30 100 Fluorescence 250nM

Cells Migrated/Total Cells Migrated/Total Cells 0.20 50 P4AP5A CCL27 0.10

0.00 0 -9 -8 -7 -6 0 10 10 10 10 0 100 200 300 400 500 600 Concentration (M) TIme (s)

Figure 4.3 Functional analysis of N-terminal mutants, assays performed using L1.2 cells transfected with CCR10. A: Chemotaxis assay of WT, F1A, and [3-88] CCL27. B: Desensitization assays measuring calcium influx for initial additions of 250 nM WT, F1A, and [3-88] CCL27, followed by addition of 250 nM WT CCL27 at 300 s. C: Chemotaxis assay of WT and P4AP5A CCL27. D: Desensitization assays measuring calcium influx of initial addition of 250 nM P4AP5A CCL27, followed by addition of 250 nM WT CCL27 at 300 s.

Finally, we wanted to examine the role of Pro4 and Pro5 in CCR10 activation. In terms of chemokine structure, N-terminal Pro residues have been shown to play a key role in oligomerization. For example, the P8A mutant of CCL2 is a monomeric variant, and our previous analysis of CCL27 showed that the double mutant P4AP5A required higher concentrations to produce WT-like oligomerization [21, 102]. We therefore measured migration and desensitization for the P4AP5A CCL27 mutant, and the results,

102 shown in Figures 4.3c and d, are very similar to WT. This suggests that the N-terminal

Pro residues in CCL27 play more of a role in structure than in function.

4.3.4: The chemical nature of the N-terminal residues of CCL27 compared to

CCL28 is not critical for signaling

CCL28 has an Ile residue at the N-terminal position and induces a response in

CCR10 very similar to that of CCL27 (data not shown) [67, 68]. In order to determine if receptor activation is dependent on the N-terminal amino acid sequence, we generated a CCL27 mutant with Phe1 truncated and Leu2 replaced with an Ile, labeled [2-88]L2I

CCL27. As seen in Figure 4.4a, this mutant very closely resembles CCL28 at the N- terminus. Interestingly, the results from the chemotaxis assay in Figure 4.4b, show the same 5-fold decrease in potency and a similar decrease in efficacy as seen for both F1A and [3-88]L3A CCL27. Because CCL28 activates a second receptor, CCR3, we measured the activity of [2-88]L1I CCL27 with CCR3 and found that it was unable to induce a migration response (data not shown). These results suggest that chemokine specificity to receptor activation is not dependent only on the amino acid sequence of the

N-terminal signaling domain of the chemokine.

103

A CCL27 FLLPPSTACCTQ 12 CCL28 -ILPIASSCCTE 11

B

0.80

0.70 WT CCL27

0.60 [2-88]L2I

CCL28 0.50

0.40

0.30

0.20 Cells Migrated/Total Cells 0.10

0.00 0 10-9 10-8 10-7 10-6 Concentration (M)

Figure 4.4 A; Sequence alignment of the N-terminal regions of CCL27 and CCL28. B: Chemotaxis assay, measuring activity of WT, [2-88]L2I CCL27 and WT CCL28, using L1.2 cells transfected with the chemokine receptor CCR10.

4.3.5: Deletion of N-terminal residues results in mutants with antagonist

properties

By definition, a protein antagonist is inactive to receptor activation, but maintains binding capability. In order to determine exactly which residues are necessary for

CCR10 activation, and to characterize potential antagonists, two N-terminal truncation mutants of CCL27 were generated. The first truncation mutant removed Phe1, Leu2, and replaced Leu3 with an Ala, [3-88]L3A CCL27. [3-88]L3A CCL27 is functionally

104

similar to [4-88] CCL27 however because residue 4 is a Pro, it was necessary to leave one Ala at the N-terminal position in order to cleave this mutant from its Ubiquitin-fusion expression partner protein. The second truncation mutant removed the first five residues from the N-terminus, [6-88] CCL27. Figure 4.5a shows that these two NT truncation mutants are not able to activate CCR10 to induce cellular migration. Similar to the results from the migration assay, Figure 4.5b shows that [3-88] L3A and [6-88]

CCL27 do not flux calcium above baseline, even up to 250 nM. Finally, the desensitization assays shown in Figure 4.5c suggest that the truncation mutants are not able to activate CCR10 to any pathways that result in receptor internalization. These experiments demonstrate that residues within the 1-5 region of CCL27 are essential for inducing migration, cellular activation, and internalization of CCR10.

105

AB Chemotaxis Assay with CCR10 Calcium Flux Assay with CCR10 0.70 300

WT WT 0.60 250 [3-88]L3A [3-88]L3A 0.50 [6-88] [6-88] 200 0.40 150 0.30

Fluorescence 100 0.20

Cells Migrated/Total Cells 0.10 50

0.00 0 0 10-9 10-8 -7 10-6 10 0 1 5 10 25 50 250 Concentration (M) Concentration (nM) C D Desensitization Assay with CCR10 Competition Chemotaxis with CCR10 250 120 WT [3-88]L3A 100 200 [6-88 0 80 150 60 100 [3-88]L3A

Fluorescence 40 [6-88]

50 20 Percentage of Cells Migraged 0 0 0 -9 -8 -7 -6 0 100 200 300 400 500 600 10 10 10 10 Time (s) Concentration (M)

Figure 4.5 Functional analysis of N-terminal truncation mutants, [3-88] L3A and [6-88] CCL27 using L1.2 cells transfected with CCR10. A: Chemotaxis assay, B: Maximum fluorescence signals for calcium influx assays, C: Desensitization assays measuring calcium influx following an initial addition of 250 nM WT, [3-88] L3A CCL27, and [6-88] CCL27, as well as a media-only injection, followed by a second addition of 250 nM WT CCL27 at 300 s. D: Chemotaxis competition with 50 nM WT CCL27 in the lower chamber of each well, which results in maximum migration for a standard migration assay.

Once we had determined that these truncation mutants are not active, we next wanted to measure their ability to bind CCR10. We therefore first performed a radioactive binding assay, similar to binding assays performed on other chemokine systems [104]. However, the addition of 125I to WT CCL27 interfered with its ability to

106

bind CCR10 making it impossible to determine Kd values (data not shown). This situation has occurred with other chemokine systems, so as an alternative approach to determine ligand:receptor interaction, we used a chemotaxis competition assay. This assay does not measure binding constants, but we can determine whether or not mutant

CCL27 retains the ability to interact with CCR10, making it a useful alternative technique. In this competition assay, the L1.2 cells expressing CCR10 were incubated in the presence of 50 nM WT CCL27 (induces maximum migration signal) and increasing amounts of the truncation mutant. Figure 4.5d shows that both of these mutants are able to compete with 50 nM WT CCL27, resulting in a decrease in CCR10 migration, suggesting they retain the ability to interact with CCR10. However, Figure

4.5d also shows that it requires concentrations above 100 nM for [6-88] CCL27 to decrease activity to migration, suggesting the first five N-terminal residues together may also play a role in CCR10 binding. Since our previous results suggested that Pro4 and

Pro5 play a role in the overall structure of CCL27, it is possible that the two Pro residues, which did not appear to affect activity, may play a role in receptor binding. Overall, our results suggest that truncation of the first three and first five N-terminal residues of

CCL27 result in mutants with antagonist properties.

4.3.6: C-terminal truncation of CCL27 results in a partial agonist

Compared with other known chemokines, CCL27 and CCL28 have relatively long

C-terminal regions. For example, CCL27 extends 18 residues beyond the α-helix [102],

and while there is no known structure for CCL28, its primary sequence is 17 residues

longer than that of CCL27 (Figure 4.1a). In contrast, the chemokines CCL1 and CXCL8

have C-terminal regions that extend only 3 and 6 residues respectively beyond the α-

helix [82, 92]. In order to examine the functional relevance of these C-terminal regions,

107

truncation mutants were made for both CCL27 and CCL28. [1-73] CCL27 removes the

C-terminal region 3 residues beyond the α-helix and [1-81] CCL28 removes the last 24

residues, making it more similar in length to CCL27 (Figure 1a). We first performed an

15N-1H HSQC experiment using 15N-labeled [1-73] CCL27 and compared the results to

that WT in order to verify that this truncation mutant was properly folded (Figure 4.6).

Figure 4.6 15N-1H HSQC spectra acquired with a 600 MHz Bruker Avance II system with a 5 mm CryoProbe. Samples were prepared in 50 mM Acetate buffer at pH 5.6 and spectra were acquired at 302.3K. A: 15N-labeled WT CCL27, B: 15N-labeled [1-73] CCL27. Signals are broadened, suggesting increased dynamics.

The results from the chemotaxis assay for [1-73] CCL27 with CCR10 are shown in Figure 4.7a. Surprisingly, this C-terminal truncation mutant shows a significant decrease in efficacy in the induction of cellular migration. In order to verify these results further, calcium flux assays were performed and the results also show a decreased signal for [1-73] CCL27 compared to WT.

108

A Chemotaxis Assay with CCR10 B Calcium Flux Assay with CCR10 0.60 250

WT WT e

s 0.50 [1-73] 200 [1-73] 0.40 150 0.30 100 0.20 Cells Migrated/Total CellCells Migrated/Total

Maximum Fluorescenc Maximum 50 0.10

000 0 -9 -8 -7 -6 0 10 10 10 10 0 1 10 25 50 100 250 Concentration (M) Concentration (nM)

C Desensitization Assay with 250 nM WT D CCL28 Chemotaxis Assay with CCR10 300 0.80

WT 0.70 CCL28 250 [1-73] s [1-81]CCL28 0.60 200 250 nM WT 0.50 150 0.40

0.30

Fluorescence 100 0.20 Cells Migrated/Total Cell 50 0.10

0 0.00 0 100 200 300 400 500 600 0 10-9 10-8 10-7 10-6 Time (s) Concentration (M)

Figure 4.7 Functional analysis of [1-73] CCL27 using L1.2 cells transfected with CCR10. A: Chemotaxis assay, B: Maximum fluorescence signal from calcium influx assay, C: Desensitization assay measuring calcium influx following initial addition of 250 nM WT and [1-73] CCL27, with a second addition of 250 nM WT CCL27 at 300 s. D: Chemotaxis assay comparing WT CCL28 and [1-81] CCL28.

The results at 10 nM in the calcium flux assay suggest a decrease in the potency of [1-73] CCL27 as well (Figure 4.7b). Interestingly, unlike the N-terminal mutants, the desensitization assay does show a decrease in signal upon addition of 250 nM WT

(Figure 4.7c), suggesting that [1-73] CCL27 is still able to bind to CCR10 and some receptor internalization is occurring. In comparison, the results for [1-81] CCL28 show

109

that truncation of the C-terminal region has no apparent effect on CCR10-induced

cellular migration, if anything the potency is slightly increased (Figure 4.7d). The

difference in C-terminal activity for CCL27 and CCL28 is interesting because some have

speculated that the long C-terminal domain of CCL28 is proteolytically cleaved in vivo.

The fact that [1-81] CCL28 has the same ability to induce migration as WT suggests that the protein is able to functional normally without the C-terminal region.

Analysis of CCL27:GAG interactions

The results from our previous structural analysis of CCL27 suggested that this chemokine exists as a mixture of oligomeric states, with the higher order oligomers stabilized by the presence of Heparin octasaccharide [102]. It is possible that this heterogeneity is involved in GAG specificity, with multiple oligomeric forms stabilized by different GAGs. Therefore, we sought to map the GAG binding region of CCL27, from both a global, and a more residue-specific standpoint, in order to better understand the possibly mechanisms by which the oligomeric forms may be stabilized.

4.3.7: CCL27 interactions with both Heparin Tetrasaccharide

In order to determine the global structural effect of GAGs, we measured 15N-1H

HSQC chemical shift perturbations of 15N-CCL27 in the presence of increasing amounts of both Heparin tetrasaccharide and Heparin hexasaccharide.

The addition of tetrasaccharide initially appeared to cause some aggregation.

However, the protein was fully in solution after a thirty minute incubation at room temperature. Figure 4.8a shows the 15N-1H HSQC spectra with increasing amounts of

110

Heparin tetrasaccharide. As is shown, some residues experienced a significant change in chemical shift.

A 110 No Tetrasaccharide 40L 1:0.5 Tetrasaccharide 61L 27L 1:1 Tetrasaccharide 1:2 Tetrasaccharide 26V

115 12Q N (ppm) 15 53C 20D 120 14Y 42A

31L 47L 125 10 987 1H (ppm) B 0.6

0.5

0.4 ppm

∆ 0.3

0.2

0.1

0 64 1F 2L 3L 4P 5P 7T 9C 6S 52I 54I 27I 8A 59P 61L 56P 65F 66E 17P 18L 22L 23L 40L 43F 45L 47L 69E 72L 11T 13L 30E 31L 33E 75T 76L 77P 79L 81F 84L 58N 60S 62S 50R 51S 53C 55H 67H 15R 16K 19S 20D 21K 24R 25K 35D 37D 38C 39H 46H 70R 71K 73H 10C 78K 80N 85R 86K 57Q 63Q 41Q 42A 44V 48A 49Q 68Q 12Q 14Y 26V 28Q 29V 32Q 34A 36G 74G 82G 83M 87M 88G

Protein Residue

Figure 4.8 Results from chemical shift perturbation analysis of CCL27 titration with Heparin Tetrasaccharide. A: 15N-1H HSQC spectrum showing changes in resonance signals with increasing molar amounts of heparin tetrasaccharide. B: The changes in chemical shift for 0 and 1:4 CCL27:tetrasaccharide plotted as a function of protein residue.

111

The histogram in Figure 4.8b shows the magnitude of the chemical shift change

as a function of protein residue. The residues most affected by the tetrasaccharide are

Asp20 – Lys25 in the N-terminal 310 α-helix, Val26, Glu30, Leu31 in the first β-strand,

Asp35 – Gln41 in the 30s loop, Leu47 in the second β-strand, Gln49 and Arg50 in the

40s loop, Ile52 – Ile54 in the third β-strand, and Ser60, Leu61, Trp64, Glu66, His67, and

Arg70 in the C-terminal α-helix.

Comparing these results to other chemokines shows that CCL2, CCL3, and

CXCL12 also have GAG binding residues in the 310 helix and 30s loop regions [13, 35,

105]. In addition, CCL2, CCL4, and CXCL12 have GAG-binding residues in the C- terminal α-helix. Although other chemokines do not have such a significant contribution from the third β-strand as CCL27, in the context of the monomeric structure, these results make sense. The third β-strand contains Cys53, which joins this region to the unstructured N-terminus that leads directly to the 310 helix. The third β-strand is also connected directly to the C-terminal α-helix by a short loop region. The chemical shift changes were the same for Heparin hexasaccharide, but solubility was more of an issue above 1:1 CCL27:hexasaccharide, as indicated by signal attenuation in the 15N-1H

HSQC spectra (data not shown).

In order to better visualize the chemical shift changes in Figure 4.8, the results were mapped onto the monomeric structure of CCL27, as shown in Figure 4.10a with the shift change of 0.1 – 0.2 ppm colored in pink, 0.2 – 0.3 ppm in burgundy, and >0.3 ppm in red. The regions interacting with both tetrasaccharide and hexasaccharide appear to involve predominantly the 310 helix, the 30s loop, and the third β-strand. As

112 indicated by Figure 4.9a, many of the residues within these regions are within loop regions and surface exposed, predominantly on one face of the monomeric structure.

A

B

+72 -72

Figure 4.9 A: Results from chemical shift perturbation analysis with heparin tetrasaccharide mapped to monomeric structure of CCL27, Changes in chemical shift are measure by color, with 0.1 – 0.2 ppm colored in pink, 0.2 – 0.3 ppm in burgundy, and >0.3 ppm in red. B: Electrostatic surface gradient with positively charged residues in blue to negatively charged residues in red.

Figure 4.9b shows the calculated electrostatic surface, indicated positively charged residues in blue and negatively charged residues in red. A comparison of

113

Figure 4.10a to 4.9b suggests that the GAG-binding residues involving the 310 helix and

third β-strand compose a positively charged surface area, which would form stable

electrostatic interactions with negatively charged GAGs. While the 30s loop region has

fewer positively charged residues, our previous structural analysis of CCL27 suggested

that this region plays a role in oligomerization, and therefore may aid in stabilizing the

complex chemokine:GAG structure.

4.3.8: Residue specific identification of the Heparin binding sites for CCL27

We next wanted to determine the effect of GAG-binding on the functional capabilities of CCL27. In order to do this, we needed to determine residues critical for

GAG binding. One caveat to chemical shift perturbation analysis is that the results do not necessarily reflect resides that bind directly, but also those experiencing changes to their local environment as a result of the GAGs. We therefore combined single Alanine point mutations with a Heparin binding assay in order to isolate mutant forms that had significantly decreased GAG affinity, to be used to test function. Although Heparin may not be the most physiologically relevant GAG with respect to surface localization of chemokines, it is a highly sulfated soluble GAG and that is readily obtainable, well- behaved, and similar to the more heterogeneous heparan sulfate, known to interact with chemokines in vivo [78, 106].

GAGs in general are negatively charged and therefore tend to interact with the positively charged basic residues on the surface of chemokines. Using the results from chemical shift perturbation analysis, our initial GAG binding assays targeted Alanine mutants involving solvent accessible, positively charged residues that experienced a significant change in chemical shift. We prepared thirty-five single alanine mutants, one

114

single glutamate mutant K25E (charge reversal), and one double mutant K21A/K25A, as well as examined the N- and C-terminal truncation mutants shown previously to have an effect on receptor activation. Each CCL27 mutant was evaluated by quantifying the concentration of NaCl, [NaCl] required for elution from a Heparin sepharose column relative to WT, ∆[NaCl] (Figure 4.8a). WT CCL27 eluted at 516.7 (±1.8) nM NaCl, and

the single alanine mutants with the highest ∆[NaCl] values were R15A, K16A, R24A,

K25A, and R50A at 470.2 ± 4.5, 425.3 ± 1.6, 392.6 ± 4.1, 445.3 ± 2.9, 387.1 ± 1.5, and

403.2 ± 2.2 nM NaCl respectively (Figure 4.10a). Reversing the charge at position 25 by

mutating the lysine to a glutamate (K25E) considerably decreased the affinity to 305.8 ±

1.9 nM. And when two of these primary Heparin binding sites were mutated to make

K21A/K25A, there was a further decrease in affinity to 283.6 ± 1.3 nM. The C-terminal

truncation mutant [1-73] exhibited a slight decrease in affinity, but truncation of the N-

terminus does not appear to disrupt Heparin association (Figure 4.10a).

115

A NaCl (mM)

B * * * * * Normalized NaCl (mM) NaCl Normalized

C Protein Residue

Figure 4.10 Results from Heparin binding assays of Alanine point mutations for CCL27. A: Concentration of NaCl required to elute mutant CCL27 from a Heparin sepharose column. B: The results in A, normalized to the [NaCl] required for elution from sepharose alone. C; The positive results from B mapped to the monomeric structure of CCL27.

To further evaluate the specificity of these mutants as opposed to electrostatic interactions, binding to a nonspecific cation exchange column (S-sepharose) was measured and compared to the results with Heparin sepharose, resulting in a value termed ∆∆[NaCl] [13]. For some of the mutants, ∆∆[NaCl] values were negative (Figure

116

4.10b), indicating a higher affinity for the nonspecific S-sepharose column than the

Heparin sepharose. However, consistent with the data in Figure 4.10a, the single alanine mutants K21A, K25A, R50A, and the charge reversal mutant K25E, as well as the double mutant K21A/K25A had highly positive ∆∆[NaCl] values, suggesting that these residues have the most significant contribution to the specific interaction of CCL27 with Heparin. Figure 4.10c shows the results from this analysis mapped to the monomeric structure of CCL27. The residues necessary for Heparin binding appear to be distinct from those important for receptor activation. The mutant K25A, which demonstrated the most significant decrease in Heparin affinity, was selected for further functional analysis.

4.3.9: The non-GAG-binding mutant K25A CCL27 has decreased activity to

transendothelial cellular migration

In order to determine if the ability to bind GAGs is directly connected to the function of CCL27, we compared the potential of WT and K25A CCL27 to activate

CCR10-induced migration across an endothelial layer. In this assay, Human Umbilical

Endothelial Cells (HUVECs) were cultured to confluency on the upper surface of each porous filter. Either WT or mutant CCL27 was added to the lower chamber and L1.2 cells stably expressing CCR10 were added to the top chamber above the layer of

HUVECs representing the model endothelium. As a control, we first ran this transendothelial migration assay for the superagonist mutant F-CCL27 and the antagonist mutant [3-88] L3A CCL27. Neither of these mutants were expected to have an impact on GAG binding, and the results in Figure 4.11 suggest that the behavior of these mutants is the same for trans-filter and transendothelial.

117

A B Transfilter Migration Transendothelial Migration

0.70 0.02

WT 0.60 0.01 WT [3-88]L3A [3-88]L3A F-CCL27 0.01 0.50 F-CCL27 0.01 0.40 0.01 0.30 0.01 0.20 0.00 Cells Migrated/Total Cells 0.10 Cells Migrated/Total Cells 0.00

0.00 0.00 0 1 5 0 1 5 10 50 10 50 0.1 0.1 100 250 500 100 250 500 1000 1000 Chemokine Concentration (nM) Chemokine Concentration (nM)

4.11 Migration assays were run using L1.2 cells stably transfected with CCR10. A: Trans-filter migration assay of WT, F-CCL27, and [3-88] L3A CCL27, B: Transendothelial migration assay for WT, F-CCL27, and [3-88] L3A CCL27.

However, the results for K25A, shown in Figure 4.12A and B, demonstrate that this GAG-deficient mutant form of CCL27 has WT-like activity in terms of trans-filter cellular migration, but significantly decreased potency and efficacy to cellular migration across an endothelial layer, compared to WT.

Transfilter Migration Transendothelial Migration

0.04 WT 0.60 WT K25A K25A

0.40

0.02

0.20 0.01

0 0 -9 -8 -7 -6 -5 010-9 10-8 10-7 10-6 10-5 01010 10 10 10

Figure 4.12 Migration assays with L1.2 cells stably expressing CCR10, comparing WT to K25A CCL27. A: Transfilter assay, B: Transendothelial assay with HUVECs cultured on the surface of the filter to form a representative endothelium.

118

4.4 Discussion

The chemokine network is highly complex in terms of receptor specificity and biological function. Although the monomeric forms of chemokines are highly homologous, the oligomeric forms are more varied. To date, CCL27 binds and activates only one receptor, CCR10. This receptor is also activated by CCL28, which in turn binds and activates CCR3, a receptor for nine other chemokine ligands. Our previous analysis of the structural properties of CCL27 revealed a dynamic system with a potentially unique oligomeric interface. CCL27 appears to exist as a mixture of oligomeric states, with the higher order forms stabilized by the presence of octasaccharide. We therefore sought to elucidate interactions between this ligand and its receptor CCR10, compared to CCL28, as well as several different GAGs, in order to better understand the details driving specificity.

Looking at the first five residues of CCL27, Phe1 plays a crucial role in CCR10 activation and addition of a second N-terminal phenylalanine residue increases function by an order of magnitude. Similar studies of other chemokines have shown that N- terminal additions to chemokines such as CCL5 (“AOP-RANTES”) and CCL1 (Ser-

CCL1) actually decreased activity, presumably by hindering access of the N-terminal residue to the receptor activation site [40, 107]. In this case, CCL27 activity is enhanced

by the additional aromatic phenylalanine. A sequence comparison shows that although

CCL28 does not have an aromatic residue at its N-terminus, it does have a nonpolar

isoleucine. However, when the N-terminus of CCL27 is replaced with an isoleucine, the

results show a 5-fold decrease in receptor activation. Overall, these results suggest that

the mechanism of CCR10 activation goes beyond hydrophobic interactions Further

119 truncation of the first three and first five N-terminal residues of CCL27 shows that this region is absolutely necessary for CCR10 activation to both migration and calcium flux.

At the same time, these mutants, [3-88] L3A and [6-88] CCL27, are still able to compete with WT to decrease cellular migration. This indicates that CCR10 binding is partially maintained without these N-terminal residues, suggesting these mutants have antagonist properties. These results provide additional support for the 2-stage model of chemokine:receptor interaction and activation described previously [7, 14, 17, 94, 108].

This is also reminiscent of CCL2/MCP-1 and CCL5 where N-terminal truncation mutants, specifically [9-68] CCL5 and [9-76] CCL2 were shown to be antagonists to cell migration and calcium flux [14, 16, 40]. However, in the case of CCL2, the truncation mutants were still able to desensitize CCL2-induced calcium flux, while in this case, [3-88] L3A and [6-88] CCL27 were not [16]. Similarly, truncation of the first three N-terminal residues of CXCL11/ITAC resulted in a potent antagonist to its receptor CXCR3, while a similar mutation in a second ligand for CXCR3 (CXCL8/IP-10), led to a loss of binding capacity [109]. The fact that the N-terminal regions of some chemokines are used in both receptor binding and activation, while others, such as CCL27, appear to be predominantly involved in activation, provides additional information in terms of receptor specificity within the chemokine network.

Furthermore, and perhaps the most surprising result in this study, the C-terminal region of CCL27 plays a role in CCR10 activation. Traditionally the C-terminal regions of chemokines are not considered to be involved in receptor interaction or activation and our previous NMR studies of CCL27 show this region to be relatively unstructured [102].

It has been shown previously that C-terminal truncation of certain chemokines can result in decreased activity. For example, the C-terminal truncation of CCL2 results in a

120

decreased potency to cell migration [110] and CXCL8/IL-8 showed a 10-20 fold reduction in activity for C-terminal truncation variants [111]. It is also interesting that the

C-terminal region of CCL28, which is considerably longer than most other chemokines,

including CCL27, has no significant effect on activity. In fact, [1-81] CCL28 actually

shows a slight increase in potency but is more or less the same as WT.

It was postulated in the case of CXCL8 that the C-terminal α-helix played a role

in GAG binding and that was why the truncation variants were less active [111].

However, in our GAG binding studies, the C-terminus does not appear to be involved, as

[1-73] CCL27 does not show a significant change in either the ∆[NaCl] or the ∆∆[NaCl] value and chemical shift perturbations for the entire region are relatively low. Our previous structural analysis of CCL27 suggested that certain residues in the C-terminal

α-helix appear to be affected by oligomerization, but only the first few residues beyond the helix showed any significant changes [102]. On the other hand, CCL27 does appear to bind Heparin at Lys21, Lys25, and Arg50, which cluster together along with the residues shown to be involved in tetrasaccharide interaction, Arg24, Gln49, and Ile52, to form a predominantly positively charged region on the protein surface. Similar results were seen in studies with CXCL12/SDF1-α, where the Heparin binding site was a basic surface region composed of Lys24, Lys27, Arg41, and Lys43 [80]. Because the dimer structure for CXCL12 is known, they were also able to show that this Heparin binding site was located near to the dimer interface for CXCL12 [80].

In order to determine if these results suggest that like other chemokine systems, the GAG binding region is potentially occurring at the oligomeric interface of CCL27, the resides from the chemical shift perturbation analysis that experienced a change >0.2

121 ppm were mapped to CC and CXC dimer models generated previously for CCL27

(Figure 4.13) [13, 20, 80, 102].

A

C

0 ppm > 0.4 ppm

Figure 4.13 A: CC dimer model for CCL27 shown with an octasaccharide derived from the structure of dodecasaccharide (pdb #1hpn), C: CXC dimer model of CCL27 with a hexasaccharide model derived from dodecasaccharide (PDB code1hpn) and drawn to scale with the dimer models.

122

Previous structural analysis of CCL27 suggests that the oligomeric form has

characteristics of both a CC and a CXC chemokine dimer [102]. Interestingly, Figure

4.12a and b suggests that these residues appear to span the interface of both dimer structures. The regions of a dodecasaccharide, corresponding to the equivalent of two tetrasaccharides in the CC dimer (Figure 4.12a) and one tetrasaccharide in the CXC dimer (Figure 4.12b) were included, in order to illustrate how different oligosaccharides may be stabilizing the different dimeric structures. These results suggest that the multiple oligomeric states of CCL27 may play a role in GAG specificity, with different oligosaccharides able to interact with different interfaces.

It has also been demonstrated in this study that the non-GAG-binding variant,

K25A CCL27, was not capable of inducing in vitro transendothelial migration, suggesting that the ability to bind GAGs is directly related to CCL27 function. Ultimately, this analysis further supports the hypothesis we put forth previously, that the plasticity of the oligomerization state of CCL27 allows it to adopt different oligomeric structures, depending on its GAG (or other) binding partners, thereby increasing the functional diversity of this chemokine [102].

123

4.5 Conclusion

Overall, our results demonstrate that the interaction between chemokines and their chemokine receptors, ultimately resulting in G-protein activation, is highly complex.

Like many other chemokines, CCL27 appears to activate CCR10 through its N-terminus and deletion of the first three and first five residues result in mutants with antagonist properties. However, the C-terminal region, while previously demonstrated to be unstructured, was shown here to be important for activity [102]. Finally, similar to previous results for other chemokine systems, the GAG binding sites consist primarily of positively charged solvent exposed regions of CCL27. However, other residues affected by both tetrasaccharide and hexasaccharide are consistent with regions we believe to be involved with oligomerization. In addition, these results suggest that GAG interaction is involved in the biological diversity of CCL27, with different oligomeric forms potentially stabilized through interaction with different GAGs. Because CCL27 has been implicated in several skin inflammatory diseases such as psoriasis, and is believed to play a role along with CCR10 in melanoma metastasis to secondary sites within skin tissue, this study represents a major step in designing antagonists with potential therapeutic applications [41, 62, 112].

124

4.6 Materials and Methods

4.6.1: Expression and purification of WT and mutant CCL27 and CCL28

Both WT and mutant CCL27 and CCL28 were expressed as a His-ubiquitin fusion using the pHUE vector and purified from inclusion bodies [69]. Unlabeled protein was generated at 37˚C in LB at 200 µg/mL Carbenicillin. All growths, from initial inoculation to harvest were carried out within one day to avoid potential toxicity issues and induced with 1.0 mM IPTG at A600 of approximately 0.700. The purification procedure used for both proteins was described previously [102]. Purification was confirmed by MALDI mass spectrometry for unlabeled protein.

4.6.2: Cell culture

The mouse lymphoma L1.2 cells, both untransfected and stably expressing human CCR10 were obtained from Eugene Baker at Stanford University and maintained in RPMI 1640 media + 10% fetal bovine serum, 1.0% MEM non-essential amino acids,

1.0% sodium pyruvate, 0.1% BME, and 700 µg/mL Geneticin was used for selection.

4.6.3: Chemotaxis assays

Twelve hours prior to the assay, cells were split to 1.0 x 106 cells/mL and induced

with culture media containing 5 mM sodium butyrate. Immediately prior to the assay, the

cells were counted, centrifuged and resuspended at a concentration of 2.5 x 106 cells/mL in RPMI 1640 + 10% fetal bovine serum. The assay was performed as described previously [14], with several modifications. Briefly, 100 µL aliquots of cells were placed in the upper chamber of a transwell (Corning-Costar) with 5 µm pore size. Varying concentrations of CCL27 or CCL28 were in the lower chamber. Following a 2 hour

125

incubation at 37ºC/5% CO2,/60% relative humidity, the filters were removed and the cells which had migrated to the lower chamber were counted using a FACS BD flow cytometer (number of cells counted in 30 s). All experiments were performed in triplicate and the results shown here were repeated on two separate occasions. Migrated cells were normalized to the number of cells counted in 30 seconds that had been added to a well without a filter present.

4.6.4: Calcium flux and desensitization assays

Intracellular concentrations of calcium were measured using the FLIPR Assay Kit calcium indicator dye, part # R7447 (Molecular Devices). Assay buffer consisted of 1x

Hank’s Balances Salt Solution (HBSS, Invitrogen), with 20 mM Hepes, pH 7.2. One bottle of dye was prepared in 10 mL Assay buffer. One hundred µL L1.2 cells expressing CCR10 in RPMI media, at a concentration of 2.0 x 106 cells/mL, were added

to each well of a clear-bottom assay plate, followed by 100 µL calcium indicator dye.

The cells were then incubated for 45 min at 37ºC/5% CO2/60% humidity. Chemokine

samples were prepared in 50 mM Hepes buffer, pH 7.2 and added at appropriate

concentrations to a second V-bottom assay plate. The assays were performed at 37ºC,

using a FlexStation 3 (Molecular Devices). Ligand was added to the cell plate at time 0

s, for calcium flux assays, and at time 0 and 300 s, for receptor desensitization assays.

4.6.5: In vitro Heparin binding assays

The amount of salt required to elute both WT and mutant CCL27 from a Heparin-

sepharose column was determined as previously described [58], with several minor

adjustments. Using an AKTA FPLC system, 0.5 mL of 25 µM protein in 50 mM Hepes,

pH 7.2 was injected onto a 1 mL Heparin sepharose column (HiTrap Heparin HP, GE

126

Healthcare) or a non-specific sepharose column (HiTrap SP HP, GE Healthcare) equilibrated in the same buffer. The column was eluted at a flow rate of 1 mL/min in 50 mM Hepes, pH 7.2, using a NaCl gradient from 0 to 1 M over 60 minutes. This assay was performed at least three times for each mutant. The NaCl concentration at peak elution was calculated from the conductivity at peak elution normalized against conductivities at baseline and 1 M NaCl. The concentrations of NaCl required to elute the proteins from the Heparin or SP-sepharose columns ([NaCl]H and [NaCl]S, respectively) were used to calculate a specificity index ∆∆[NaCl] for each mutant,

∆∆[NaCl] = ∆[NaCl]H – ∆[NaCl]S

where ∆[NaCl]H = [NaCl]H,WT – [NaCl]H,mutant and ∆[NaCl]S = [NaCl]S,WT – [NaCl]S,mutant

[13]. A positive specificity index for a mutant means that the difference between the

amount of NaCl required to elute the mutant from the Heparin sepharose column as

compared to WT was greater than the same difference for the SP-sepharose column. In

other words, electrostatic interactions between a mutant with a positive specificity index

and the heparin-sepharose column are weakened to a greater extent than those

between the same mutant and the nonspecific SP-sepharose column.

4.6.6: NMR analysis, 15N-1H HSQC and chemical shift perturbation

Heparin tetrasaccharide and Heparin hexasaccharide were purchased commercially by Neoparin, inc. and prepared in water, as described previously [13]. 15N-

labeled protein was purified as described previously but the growths were carried out at

37˚C in MOPS minimal media, 10x MOPS was prepared with 0.1 mM FeSO4, 2.76 mM

K2SO4, 5.0 mM CaCl2, 5.28 mM MgCl2, 0.5 M NaCl, 10 mL Micronutrients per liter and

200 µg/mL Carbenicillin, in the presence of 98% 15N ammonium sulfate. Purity and

percentage of label incorporation were assessed by ESI mass spectrometry. NMR

127 experiments were conducted using a Bruker Avance II NMR spectrometer with a 5 mm

CryoProbe. Protein samples were prepared in 50 mM Acetate buffer, pH 5.6 and experiments were acquired at 302.3 K.

128

4.7 Acknowledgements

Chapter 4 is currently being prepared for the submission of publication of the material. The following are the contributing authors: Jansma, A.; Hsu, A.; Handel, T.

CHAPTER 5

Preliminary Functional Analysis of CCL28 Suggests a

Role for the C-terminal Region Independent of

Leukocyte Migration

5.1 Summary

CCL28, also known as the Mucosal Epithelial Chemokine (MEC), binds and activates the chemokine receptor CCR10 and CCR3. Although it is constitutively expressed in mucosal tissues, CCL28 is upregulated in response to inflammation and has been shown to have antimicrobial properties. Initial functional analysis of this chemokine focuses on a C-terminal truncation mutant, [1-81] CCL28. Using a combination of limited proteolysis and migration assays, this study demonstrates that the

C-terminal regions easily undergoes proteolysis, and is not essential for the activation of

CCR10 to cellular migration.

129 130

5.2 Introduction

As part of their role in the inflammatory response, several chemokines have been shown to have antimicrobial activity [113]. Of the ~50 human chemokines, CCL28 is unique because it has six Cysteine residues which form three disulfide bonds, in addition to an extended C-terminal domain. As discussed in Chapter 4, it is sequentially homologous to the chemokine CCL27 and is the second ligand for the chemokine receptor CCR10. However, CCL27 has no antimicrobial activity and is primarily found in skin cells, while CCL28 is constitutively expressed in virtually every type of mucosal tissue [114, 115]. In addition, CCL27 and CCL28 are most homologous in their N-

terminal regions, and much less similar when comparing the C-terminal domain.

CCL28 also attracts eosinophils through interaction with a second receptor,

CCR3, which in turn has nine other chemokine ligands [114, 116, 117]. Recent studies

have implicated interaction between CCL28 and CCR3 in pathological responses to

airway microbial insult, such as asthma and chronic obstructive pulmonary disease

(COPD) [68, 116]. The antimicrobial properties of CCL28 were first identified due to the homology of its C-terminal region to the known antimicrobial peptide, histatin 5 [114].

Although several other constitutively expressed epithelial chemokines, such as CCL6,

CCL14, and CCL15 have also been shown to have antimicrobial properties, very little is

known about the structural requirements needed to facilitate destruction of bacteria [113,

118].

Mutagenesis studies of CCL28 were originally conducted as part of the functional analysis of CCL27. However, these initial results have led to the hypothesis that the C-

131 terminal region of CCL28 is proteolytically cleaved in vivo and this C-terminal region adopts antimicrobial activity while the remainder of the chemokine functions in leukocyte migration. A recent study by Liu, et. al. supports this hypothesis by demonstrating that the antimicrobial activity of CCL28 is dependent on the positively charged amino acid residues in the C-terminus [113].

132

5.3 Results

5.3.1: The C-terminal region of CCL28 is not necessary for cellular migration

Figure 4.1 shows a sequence alignment of CCL27 and CCL28. As part of the

functional analysis of CCL27, C-terminal truncation mutants were made of both proteins,

[1-73] CCL27 and [1-81] CCL28. While [1-73] CCL27 exhibited significantly decreased

potency to migration, [1-81] CCL28 had virtually no change when compared to WT

(Figure 5.1).

CCL28 Chemotaxis Assay with CCR10 0.80

0.70 CCL28

s [1-81]CCL28 0.60

0.50

0.40

0.30

0.20 Cells Migrated/Total Cell

0.10

0.00 0 10-9 10-8 10-7 10-6

Concentration (M)

Figure 5.1 Adapted from Chapter 4, migration assay with L1.2 cells stably transfected with CCR10, comparing WT CCL28 to the C-terminal truncation mutant [1-81] CCL28. The C-terminus does not appear to be necessary for activation of CCR10 to migration. 5.3.2: Limited Proteolysis of CCL28

As part of initial crystallography trials, limited proteolysis of CCL28 was performed in order to determine if certain regions are more susceptible to proteolytic

133 cleavage, and thus potentially unstructured, making crystallization more difficult.

Proteolysis with a 1:100 molar ratio of Trypsin to CCL28 showed that a cleavage product is generated within the first 0.5 h and this truncated protein remains stable up to 5 h

(Figure 5.2a). Because the C-terminal region is predicted to be unstructured, and the mutant [1-81] CCL28 had no effect on cellular migration, this assay was repeated with

[1-81] CCL28. The results, shown in Figure 5.2b, demonstrate that this truncated variant remains stable in the presence of Trypsin for up to 5 hours, identical to the cleavage product of WT CCL28. These results suggest that it is the C-terminal region being cleaved by Trypsin in the WT sample. While time did not permit mass spectrometry analysis of this sample, the next step of this project is to isolate the cleavage product from the WT incubation and use MALDI in order to confirm that it is the C-terminus which has been cleaved.

A WT CCL28 B [1-81] CCL28

19 19 15 15 CCL28 Cleavage Product [1-81] CCL28 6 1 hour 2 hours 5 hours 30 min. 30 1 hour 2 hours 5 hours

6 min. 30

Figure 5.2 Limited Proteolysis, with Trypsin comparing WT CCL28 to [1-81] CCL28.

134

5.3.2: Chemical cross-linking experiments suggest the C-terminal region of CCL28

is involved with GAG binding

In order to examine the effect of the C-terminal region of CCL28 on GAG binding, chemical cross-linking experiments were performed with Sulfo-EGS in the presence of increasing amounts of the GAG Heparin decasaccharide. Figure 5.3A shows that WT

CCL28 has a very strong propensity to oligomerize on GAGs.

A: WT CCL28 with Heparin Decasaccharide B: [1-81] CCL28 with Heparin Decasaccharide Ratio: 0 0 8:1 4:1 2:1 1:1 1:2 1:4 0 0 8:1 4:1 2:1 1:1 1:2 1:4

EGS --+ +++ +++ +++++++

Figure 5.3 Chemical cross linking of CCL28 using Sulfo-EGS with increasing molar amounts of the GAG Heparin decasaccharide. A: The molar ratios of CCL28:Heparin decasaccharide are indicated above the gel and presence of Sulfo-EGS indicated below. B: The mutant [1-81] CCL28, with CCL28:Heparin decasaccharide ratio indicated above the gel and the presence of Sulfo-EGS indicated below.

The decreased signal at 1:1 CCL28:Heparin decasaccharide suggests that much

of the protein is no longer in solution. As the GAG concentration increases, competing

with CCL28 oligomerization, discrete bands corresponding to dimer, tetramer, and

higher order oligomeric forms become apparent. Comparing these results to the [1-81]

CCL28 mutant shown in Figure 5.3B, shows a stronger signal for the bands

corresponding to the monomeric form of the protein. While a 1:1 [1-81] CCL28:Heparin

135 decasaccharide ratio does show the presence of oligomerization, the size of the oligomeric forms, as well as the intensity of the bands, is less than what is seen for WT.

Similar to the results for CCL27, shown in Chapter 3, excess GAG stabilizes the dimeric form and the presence of higher order oligomers decreases. Taken together, these results suggest that the C-terminal region of CCL28 is not significantly involved in activation of CCR10, but instead plays a significant role in GAG binding.

136

5.4 Discussion and Future Directions

Previous studies of CCL28 have shown that in addition to its role in leukocyte migration, the C-terminal region of this chemokine has antimicrobial properties. While these results are very preliminary, they do suggest that the C-terminal region of CCL28 is not necessary for leukocyte migration and it easily cleaved by Trypsin, suggesting proteolytic cleavage may occur in vivo.

There is currently no known structure for CCL28. The C-terminal region has been predicted to be intrinsically unstructured. However, in the case of the antimicrobial peptide histatin 5, it has been suggested that this protein adopts an α-helical structure

necessary for its role in bacterial destruction [119]. Future analysis of CCL28 should

therefore involve a structural characterization, comparing WT to [1-81] CCL28 under

various conditions to determine formation of secondary structural elements in the C-

terminal region. Additional functional assays, such as calcium flux and receptor

desensitization should also be measured for [1-81] CCL28 in order to determine if the C-

terminal region plays any role in cellular function. It would also be interesting to evaluate

the effect of the C-terminus on interaction with glycosaminoglycans (GAGs) through

chemical shift perturbation analysis. Finally, Pulsed Field Gradient (PFG) diffusion

analysis of WT and [1-81] CCL28 would help to determine if the C-terminus plays a role

in oligomerization. The hypothesis is that this C-terminal region is not necessary for the

chemokine-like activity of CCL28 and instead acts independently as an antimicrobial

peptide.

CHAPTER 6

Investigation of the Effect of Glycosaminoglycans on

CXCL11 Oligomerization

6.1 Summary

Interactions between chemokines and glycosaminoglycans (GAGs) on cell

surfaces are required to localize chemokines to specific anatomical spaces. Most

chemokines are basic proteins that utilize Arg/Lys/His residues in clusters, such as

“BBXB” and “BBXXB”, as recognition epitopes for GAGs. The chemokine CXCL11/I-

TAC is conformationally heterogeneous, as indicated by the number of peaks in the 15N-

1H HSQC spectra. While this study focused on characterizing the GAG-binding epitopes

of CXCL11, my involvement in this project was to determine the oligomerization profile

by Pulsed Field Gradient (PFG) diffusion NMR analysis. In addition, a series of 15N-1H

HSQC experiments were acquired in the presence and absence of the GAGs Heparin tetrasaccharide and Arixtra. The results suggest that GAGs do not stabilize one form of

CXCL11, but appear to stabilize the oligomeric state, as indicated by signal attenuation in the spectra. This hypothesis was further confirmed by solubility assays with Heparin octasaccharide.

137 138

6.2 Introduction

Glycosaminoglycans (GAGs) are negatively charged linear polysaccharides that

have an exceptional range of size and sequence variability including patterns of sulfation

and acetylation. All chemokines interact with heparin, which serves as a model

compound for Heparan sulfate, the most ubiquitous class of GAG that is expressed on

virtually every cell. Beyond their role in localizing chemokines, the interaction with GAGs

may contribute to the selectivity and fine-tuning of the chemokine system, which appears

redundant in terms of ligands and receptors. For example, the expression pattern and

composition of GAGs on cell surfaces changes during diseases such as inflammation

and interactions with GAGs may play an important role in disease progression by

accumulating specific chemokines that recruit and activate specific cell types and

downstream signaling pathways [120].

GAG binding sites have been determined for several chemokines, and when

mapped onto their surfaces, show considerable topological diversity [13, 14, 35, 78]. A

BBXB motif has been identified in the 40s loop for CCL3/MIP-1α, CCL4/MIP-1β, and

CCL5/RANTES, whereas CXCL12/SDF-1α mediates GAG binding through a BBXB motif in the 20s loop. CCL2/MCP-1 binds GAGs through the 20s loop as well, but also has contributions from the C-terminal α-helix through Lys58 and His66. Similarly, CXCL8/IL-

8 binds GAGs via the C-terminal helix, through Arg68.

CXCL11 was found to bind Heparin sepharose more tightly than most chemokines. CXCL11 signals through only one known receptor, CXCR3, which it shares with two additional chemokines, CXCL10/IP-10 and CXCL9/MIG. CXCR3 is

139

expressed on activated T-lymphocytes, mainly of the Th1 subtype, but also on a subset

of B-cells, monocytes, and NK cells. In order to determine the oligomeric profile for

CXCL11, Pulsed Field Gradient (PFG) NMR diffusion analysis was performed and 15N-

1H HSQC spectra were monitored under several different solution conditions. 15N-1H

HSQC analysis was then performed in the presence of both Heparin tetrasaccharide and

Arixtra in order to determine if GAGs stabilized one form of CXCL11. Acquisition of

NMR data in the presence of GAGs was complicated by signal attenuation, which was attributed to oligomerization based on the results of solubility assays with Heparin octasaccharide.

140

6.3 Results

6.3.1: 1H-15N HSQC spectra reveal conformational heterogeneity in CXCL11

To corroborate the above experiments to define the GAG binding site on

CXCL11, we intended to use chemical shift perturbation experiments with Heparin fragments of defined size and composition. To this end, 15N-labeled CXCL11 samples were prepared with a Trp residue added to the C-terminus. Trp has a distinct proton chemical shift, at approximately 10 ppm, and therefore is a good indicator of the number of species present in solution. The HSQC spectra were recorded at various temperatures and pH conditions. However, unlike most other chemokines, CXCL11 showed evidence of multiple species in slow exchange, by the presence of more crosspeaks than one would expect, with most peaks at least doubled, and in some cases tripled, as evidenced by the presence of two Trp signals (Figure 6.1).

141

115

120 N (ppm) 15

125

10 9 8 7 1H (ppm)

Figure 6.1 15N-1H HSQC of 1.0 mM CXCL11, acquired at 600 MHz with an Avance II NMR spectrometer, equipped with a 5 mm TCI CryoProbe, at 296 K in 50 mM acetate buffer at pH 5.6. The Trp residues are seen at ~10 ppm and suggest the presence of multiple forms.

6.3.2: PFG diffusion analysis indicates CXCL11 is dimeric

Determination of the diffusion coefficients of CXCL11 over a concentration range of 0.25 mM – 1.0 mM indicates that this chemokine is a dimer under the conditions used for HSQC data collection (Figure 6.2).

142

Figure 6.2 Pulsed Field Gradient (PFG) diffusion profile of WT CXCL11, performed on a Bruker Avance II at 600 MHz with a TCI CryoProbe at 302.3K, over a range of concentrations from 0.25 mM to 2.0 mM. Assuming the diffusion coefficient at 0.25 mM corresponds to the monomeric form, the theoretical Ds value for the dimer was estimated using the Stoke’s-Einstein equation as 1.09 x 10-10 m2/s.

These results, combined with the uniform size of the crosspeaks in the HSQC, suggest that the extra peaks are not due to the presence of both monomer and dimer, but rather alternate conformations of the chemokine. This is consistent with a previous report describing an NMR structure of CXCL11 (corresponding to a canonical chemokine fold) which was determined at low pH (4.5) and high temperature (40oC), since these were the only conditions that gave a single species [109].

143

6.3.3: GAG binding analysis by HSQC chemical shift perturbation

In order to determine if the presence of multiple species contributes to GAG interactions, HSQC spectra were recorded after the addition of phosphate, Heparin tetrasaccharide, and Arixtra, in order to determine if one species is stabilized over the others. Figures 6.3 shows the results of 100 mM phosphate added to a 0.5 mM sample at pH 5.6. A visual comparison of the HSQC spectra suggests that the presence of phosphate does not stabilize one form over the others.

AB 0.25 mM, pH 5.6 0.25 mM, pH 5.6, 100 mM Phosphate

116 116

120 N (ppm) N (ppm) 120 15 15

124 124

10 9 8 7 10 9 8 7 1H (ppm) 1H (ppm)

Figure 6.3 15N-1H HSQC spectra for CXCL11 acquired at 296K using a Bruker Avance II 600 MHz spectrometer equipped with a 5 mm CryoProbe. A: Sample prepared in 50 mM acetate buffer at pH 5.6. B: Same conditions as (A) with 100 mM phosphate. The spectra show no significant changes in the presence of phosphate.

We next monitored the effect of GAGs on CXCL11 by measuring the HSQC spectra in the presence of increasing amounts of Heparin tetrasaccharide. Figure 6.4 demonstrates that the addition of Heparin tetrasaccharide results in significant signal attentuation and by a ratio of 1:1 CXCL11:tetrasaccharide (Figure 6.4d), the signal is almost completely gone. However, these results suggest that the different forms of

CXCL11 are equally affected by GAGs, as opposed to one stabilized over the others, which is evident by the presence of both Trp resides in all four spectra.

144

Figure 6.4 Tetrasaccharide titration with 0.5 mM 15N-CXCL11 acquired at 296K using a Bruker Avance II 600 MHz spectrometer equipped with a 5 mm CryoProbe. Samples were prepared in 50 mM Acetate buffer, pH 5.6. The ratios of CXCL11:Heparin tetrasaccharide are indicated above each panel.

In order to determine the effect of a more homogeneous GAG, this experiment was repeated with increasing amounts of Arixtra. Arixtra is a homogeneous highly sulfated Heparin pentasaccharide. It is longer than many other commercially available homogeneous GAGs and therefore more closely mimics physiological interactions, but because it is homogeneous, the chemokine tends to stay in solution at NMR concentrations. The HSQC results for the titration are shown in Figure 6.5 and despite the homogeneity of Arixtra, the signal attenuation was more severe in this experiment than in the presence of tetrasaccharide, with almost no visible signal at 2:1

CXCL11:Arixtra (Figure 6.5c). Also similar to the previous analysis, both Trp signals are

145

present in all four spectra, suggesting the multiple species of CXCL11 respond similarly

to the presence of GAGs.

Figure 6.5 Arixtra titration with 0.5 mM CXCL11 acquired at 296K using a Bruker Avance II 600 MHz spectrometer equipped with a 5 mm CryoProbe. Samples were prepared in 50 mM Acetate buffer, pH 5.6. The ratios of CXCL11:Arixtra are indicated above each panel.

It is possible that the signal attenuation seen with both tetrasaccharide and

Arixtra is a result of CXCL11 oligomerization. The linewidths suggest that the species in solution are monomeric, however this would make sense if the chemokine interacting with the GAG is no longer in solution. Ultimately, because of the complexity of the spectra under most conditions, and the signal attenuation in the presence of GAGs, the

146

chemical shift perturbation analysis, which would require challenging spectral assignments at best, was abandoned. Nevertheless, the results are intriguing and suggest some unique conformational plasticity for CXCL11.

6.3.4: Solubility analysis of CXCL11 with Heparin octasaccharide

In order to determine if the GAG-induced signal attenuation in the HSQC spectra is due to loss of protein in solution, which would indicate possible oligomerization, a solubility assay was performed in the presence of increasing amounts of Heparin octasaccharide. The results suggest that at 2:1 CXCL11:octasaccharide (0.1 mM

CXCL11), the majority is insoluble, with only approximately 20% remaining in solution.

The results at 25 µM were identical (data not shown) suggesting that CXCL11 oligomerizes in the presence of GAGs. This behavior may help to account for the signal attenuation in the HSQC spectra.

0.10 mM CXCL11 + Octasaccharide 1.2

1 280nm ∆ 0.8

0.6

0.4

0.2 Normalized Absorbance, Absorbance, Normalized

0 01:1 0.10.20.30.40.50.60.70.80.91:21:3 1:4 1:5 1:6 1:7 1:8 CXCL11:Heparin Octasaccharide

Figure 6.6 Solubility analysis of 0.10 mM CXCL11 in the presence of increasing amounts of Heparin octasaccharide. Solubility was determine by measuring absorbance at 280 nm and normalizing to the value at 0 mM octasaccharide.

147

6.4 Discussion

The interactions of chemokines with cell surface GAGs, in addition to high affinity interactions with their receptors on leukocytes, play an essential role in cell recruitment.

This has been demonstrated in vivo using chemokine mutants that have a compromised ability to bind GAGs in vitro, but intact receptor binding; although such GAG-binding deficient mutants are generally competent to recruit cells in vitro in simple trans-well assays, they are incapable of inducing migration in vivo [20, 121] or in vitro across endothelial monolayers [121]. The requirement for the GAG interaction is thought to reflect, in part, the need for the sequesteration of chemokines on cell surfaces where they provide directional cues for migrating cells. Indeed, a mutant of RANTES,

[44AANA47]-RANTES, when injected intraperitoneally, rapidly accumulates in serum, in contrast to the WT protein [33]. Notably, this mutant was able to block the in vivo cell recruitment to the WT protein, and in three murine models of inflammation (thioglycolate and ovalbumin induced cell recruitment and MOG-induced experimental autoimmune encephalomyelitis). Similarly, a GAG mutant of MCP-3 (CCL7) antagonizes chemokine mediated cell recruitment in an air puoch in vivo model and in vitro to the synovial fluid from patients with Rheumatoid Arthritis (RA) [121]. Finally, a GAG mutant of MCP-1

(CCL2) inhibits recruitment of cells to WT MCP-1 in an in vivo peritoneal recruitment assay (unpublished data, Proudfoot, A.). Since there are approximately 50 ligands in the human chemokine system, it remains to be seen if GAG binding is important for the vast majority of the ligands, and for fully understanding the effects of GAG-binding on chemokine function. The chemokine family also provides an extraordinary opportunity for investigating the extent to which GAG-interactions add to the specificity of chemokine

148 interactions. Finally, the ability of GAG mutants to antagonize chemokine function, offers the potential for the development of novel therapeutics.

While oligomeric structures of CXCL10 have been solved by crystallography, crystallizing CXCL11 has not been feasible. The spectra shown in Figure 8, however, provide an obvious explanation for why crystallization has been challenging: CXCL11 is conformationally heterogeneous. This heterogeneous behavior is likely to be relevant to its biological function involving different receptor and/or GAG interactions. While most chemokines studied to date appear to adopt a single conformation based on their HSQC spectra, another example of structural promiscuity has been reported for the human chemokine lymphotactin/XCL1. Similar to CXCL11, the HSQC spectra of XCL1 are indicative of two species. And in fact, XCL1 exits in equilibrium between two entirely distinct folds, one that binds GAGs but does not bind or activate the receptor, and one that has the complementary receptor binding ability but does not bind GAGs [122].

CXCL11 may also be subject to dynamic structural interconversion, although it is not likely to be as extreme as lymphotactin since the CXCL11 structure is locked down by two disulfides whereas XCL1 has only a single disulfide. Neverthless, it is likely to be biologically relevant, and we are currently investigating the structural nature and functional significance of this behavior in CXCL11.

149

6.5 Materials and Methods

6.5.1: NMR spectroscopy

Several conditions of pH and temperature were screened for spectral quality, but the protein consistently showed conformational heterogeneity. For the representative spectra shown in Figure 6.3.1, and PFG diffusion profile shown in Figure 6.3.2, the sample was prepared at a concentration of 1 mM in 50 mM Acetate buffer pH 5.3, and the data was collected at 23°C on a Bruker Avance II spectrometer equipped with a TCI

CryoProbe. The PFG diffusion experiment is detailed in Appendix I. Values for d20 and p30 were 150 ms and 2.0 ms respectively. ArixtraTM (fondaparinux sodium) was

purchased from Organon Sanofi-Synthelabo LLC (West Orange, NJ) and was desalted

and quantified for experiments as described previously [123].

6.5.2: Solubility assays

Samples were prepared in 50 mM acetate buffer at pH 5.6. The protein

absorbance values were measured at 280 nm using a NanoDrop 2000c

spectrophotometer (Thermo Scientific). The data was normalized to the absorbance

values measured in the absence of Heparin octasaccharide.

150

6.6 Acknowledgements

Chapter 6 is, in part, has been submitted for publication of the material to the

Journal of Biological Chemistry. This chapter reflects my contributions to this work. The following are the contributing authors: Sielaff, I.; Gaudry, J. P.; Johnson, Z.; Kungl, A.;

Gesslbaur, B.; Mulloy, B.;Power, C.; Proudfoot, A. E.; Handel, T.

CHAPTER 7

Expression and Purification of Silent Chemokine

Receptors D6 and DARC

7.1 Summary

Chemokine receptors are a specific class of G-protein coupled receptors

(GPCRs) which represent the largest family of membrane protein in the human genome

[25]. GPCRs are highly versatile proteins, with ligands ranging in size from photons to peptides and proteins. Chemokine receptors are implicated in a broad range of inflammatory diseases, as well as cancer, AIDS, and malaria [3, 7, 124, 125]. They are therefore excellent targets for drug discovery. Yet despite all this, many of the mechanistic details of chemokine receptor function are currently unknown. In addition, there is no high resolution structural information for chemokine receptors, and only a few high resolution GPCR structures – Rhodopsin, the β1- and β2-adrenergic receptors, and the A2A adenosine receptor [126-129]. As with other GPCRs, a major obstacle in biophysical analysis and structural studies of chemokine receptors is the difficulty involved with obtaining high levels of pure and functional protein. Difficulties involving proper folding of seven trans-membrane helices, post-translational modifications necessary for optimal function, and cellular toxicity, to name a few, limit many options in terms of protein expression systems. This study involves the application and preliminary optimization of a tetracycline-inducible expression system using Human Embryonic

Kidney (HEK) 293s cells, applied to the “silent” chemokine receptors D6 and DARC

151 152

[130]. These results represent crucial first steps toward the long-term goals of this project, which involve evaluating the effects of different ligands on the functional mechanisms of these receptors and ultimately to obtain high resolution crystal structures.

153

7.2 Introduction to Silent Chemokine Receptors

The chemokine network includes three known “silent” receptors, D6, the Duffy

Antigen Receptor for Chemokines (DARC), and CCXCKR [131]. All three of these receptors bind their chemokine ligands with high affinity and different specificities.

However, unlike standard chemokine receptors, they do not signal through G-proteins, and have therefore been termed “silent.” They are believed to act as chemokine decoys and thus regulators of the inflammatory response.

7.2.1: The D6 chemokine receptor

D6 is a seven-transmembrane helical receptor that, in contrast to standard chemokine receptors, does not signal through G-protein following ligand binding [132-

134]. There is, in fact, an alteration in the canonical DRYLAIV receptor signaling motif located in the second intracellular loop. This region consists of the sequence DKYLEIV in D6 [135]. While D6 is a highly promiscuous receptor, in that it binds eleven known chemokine ligands, it shows specificity for inflammatory CC chemokines and is predominantly expressed in lymphatic endothelial cells [135]. As described in Chapter 1, following chemokine binding, and subsequent G-protein activation, standard chemokine receptors are phosphorylated and internalized by a β-arrestin dependent mechanism. In contrast, it has been shown that D6 is constitutively associated with β-arrestin and C- terminal phosphorylation does not appear to be dependent on chemokine binding [132,

134]. In addition, the C-terminal region of D6 contains a stretch of acidic residues not present in signaling chemokine receptors and this is believed to play a role in both β- arrestin association and receptor recycling [132, 134]. As a result, D6 is believed to be a constitutively recycling receptor that actively internalizes inflammatory CC chemokines

154

and targets them for intracellular degradation [132]. This “decoy receptor” is thus

believed to play a role in regulating the inflammatory response [133, 136].

7.2.2: The Duffy Antigen Receptor for Chemokines (DARC)

DARC, located on the surface of erythrocyte cells, as well as venule endothelial cells, is a seven transmembrane helical receptor for chemokines as well as malaria parasites (Figure 7.1) [125, 137].

Figure 7.1 Schematic illustration of the silent chemokine receptor DARC [125].

While DARC is specific to inflammatory chemokines, its sixteen known ligands are from both the CC and CXC families of chemokines [138]. Like D6, DARC does not signal through G-protein and also lacks the DRY motif in the second intracellular loop

[139]. It is believed that DARC plays a role in modulating leukocyte recruitment to sites of inflammation by facilitating the movement of chemokines across the endothelium

[140]. In addition, it is believed that DARC prevents excessive activation of the

155 inflammatory response by acting as a “sink,” or reservoir, for circulating chemokines, binding soluble chemokines in the blood and transporting them to distil sites [141].

Because of its direct connection to erythrocyte invasion by the malarial parasite

Plasmodium vivax, DARC represents a potential target in the treatment of malaria [142-

144]. There is therefore, a significant interest in understanding both the structural and functional mechanisms for this silent chemokine receptor.

7.2.3: Tetracycline-inducible mammalian cell expression system

GPCRs represent approximately 30% of drug targets and despite the fact that there are at least 850 known GPCRs in the human genome, only a few high resolution structures are available [130, 145, 146]. One reason for this discrepancy is the difficulties associated with high-level expression of these 7-transmembrane helical proteins. In addition to toxicity effects often associated with expression, many GPCRs require post-translational modifications for optimal function, which ultimately makes expression in bacterial systems exceedingly difficult. In this study, the expression of D6 and DARC was accomplished using a tetracycline-inducible mammalian cell system, originally developed by the Khorana lab to overcome toxicity associated with several specific mutants of Rhodopsin [130]. This method, illustrated in Figure 7.2, is similar to bacterial expression in that mammalian HEK293s cells, stably transfected with the desired gene, grow to confluency before they are induced, with tetracycline, to express the protein of interest. The cells with the highest levels expression are then selected, by either flow cytometry, or other surface expression tests. In addition to Rhodopsin, this system has been successfully applied to the β-adrenergic receptor as well as the chemokine receptor CCR1 [130, 145, 146].

156

D D D D D D D D D Cell Sorting – top 5% D

Transfect into HEK cells Grow to confluency D Anti-”DARC”-Phycoerythrin Tetracycline Inducible Induce Cell recovery

Isolate Single Cells

D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D

Grow to confluency Induce D Anti-”DARC”-Phycoerythrin

Figure 7.2 Schematic representation of the process of tetracycline-inducible mammalian cell expression system, applied to DARC.

157

7.3 Results

7.3.1: Expression and purification of D6

This expression system was first applied to D6. Gerard Graham had shown

previously that functional D6 was able to be expressed in milligram quantities using

mouse lymphoma (L1.2) cells [147]. As part of a collaboration with this lab, we obtained

D6 in L1.2 cells in order to compare with the tetracycline-inducible HEK293s cell system.

The first step was to subclone D6, with an N-terminal HA tag and a C-terminal His6 tag, into the tetracycline-inducible CMV vector (pCMV-Tet-O), followed by stable transfection into HEK293 cells.

Cell lines from the HEK293s system were grown from single cells, by dilution into a 96-well plate. Each cell line was expanded into 2 x 10 cm plates for surface expression tests. One plate for each cell line was induced with tetracycline 24 h prior to harvest. Surface expression was tested using an anti-D6 antibody fused to phycoerythrin. Surface fluorescence was determined using a SpectraMax plate reader

(Moledular Devices). The cell lines in Figure 7.3 show the results from some surface expression tests, comparing tetracycline-induced to uninduced cells. Cell line #20 displayed sufficiently high surface expression in this group and was therefore selected for increased expression and purification.

158

Figure 7.3 Surface expression of D6 on HEK293s cells. The induced cells are shown in pink and the un-induced shown in blue. Cell line #20+ represents sufficiently high expression and was therefore selected for purification studies.

Figure 7.4 shows the results from fluorescence microscopy for induced cell line

#20 compared to constitutive expression of D6 in L1.2 cells, using the same anit-D6 antibody fused to phycoerythrin. Although both cell types display high levels of D6 expression, the HEK293s cells are much larger and have more membrane surface area to accommodate additional protein. The cells were permeabilized that Figure 7.4a suggests that there is a significant amount of D6 present inside the cell. This is consistent with the idea that D6 is constitutively phosphorylated and rapidly internalized and recycled back to the surface.

159

A B

Figure 7.4 Fluorescence microscopy using an Anti-D6 antibody fused to phycoethryrin (red). A: D6 expression in HEK cells following induction with tetracycline. B: D6 expression in L1.2 cells. Cells were permeabilized and nuclei were stained with X (blue).

The cell line #20 was split into 5 x T-150 cm2 plates and grown to confluency prior to induction with tetracycline. After a 24 h induction period, D6 was harvested and purified from the HEK293s cells using a modified version of the method previously described by Blackburn, et. al. for the purification of D6 from L1.2 cells [147]. Briefly, the

cells were lysed using an emulsiflux apparatus and the protein purified by centrifugation

followed by solubilization with 2% (w/v) Dodecylmaltoside (DDM) and 0.05% (w/v)

cholesterol hemisuccinate. Purification involved double affinity chromatography, first by

a Ni-NTA column, followed by an HA affinity column, at 4ºC. The details of this

purification scheme are outlined in Appendix I.F. Figure 7.5 shows the Western blot

analysis for the fractions eluted from the HA column. WT D6 is glycosylated, as

indicated by the two bands. The protein was detected using an Anti-HA antibody.

160

Figure 7.5 Western blot analysis of the fractions of D6 eluted from the HA column, indicating both glycosylated and non-glycosylated D6.

7.3.2: Expression and purification of DARC

Since one of the ultimate goals of this analysis is to obtain sufficient levels of purified DARC for X-ray crystallography trials, protein homogeneity is essential. The results shown previously for D6 suggest that glycosylation is a potential hindrance to a homogenous sample. Therefore, based on N-linked glycosylation patterns of other chemokine receptors, the Asn residue at position 16 was mutated to Gln. There are two other potential N-linked glycosylation sites at positions 27 and 33 within the N-terminus.

However, our analysis began with Q16N in order to determine if one point mutation was sufficient.

Q16N DARC was then cloned with an N-terminal HA tag and C-terminal His6 tag and subsequently transfected into the tetracycline-inducible HEK293s cell expression system by the same method as D6. However, instead of growing cell lines from single cells and testing individual receptor surface expression, we used flow cytometry to sort for the cells displaying the highest fluorescence signal in the presence of an Anti-DARC-

161

phycoerythrin antibody (Scheme outlined in Figure 7.2). The majority of these high-

expressing cells were isolated, grown to confluency on T-150 cm2 plates, and frozen for

future test expressions.

Cell line #35 was selected for purification and solubilization testing. The cells

were grown to confluency in 5 x T150 cm2 plates and induced with tetracycline. DARC

was purified by the same method used for D6, outlined in Appendix I.F. Figure 7.6

shows the results from Western blot analysis of the fractions eluted from the HA affinity

column, using an anti-His antibody for detection.

B

FT Fraction: 12 3 4

50 DARC

30

15

Figure 7.6 Fractions collected following DARC purification by HA affinity chromatography. DARC is indicated by the black arrow and appears to be non- glycosylated. An unknown cleavage product is indicated by the red arrow.

The results shown in Figure 7.6 indicate that DARC is pure and appears to be non-glycosylated. This suggests that the Q16N mutation was sufficient to generate a non-glycosylated, homogeneous sample. However, Figure 7.6 also suggests the presence of an unknown cleavage product, predominantly in the HA column flow-

162 through. Based on the intensity of the band, this appears to be a significant portion of the total protein yield and as a result, it is necessary to determine the site of cleavage in order to develop techniques to minimize protein loss.

Because the majority of this cleavage product did not adhere to the HA affinity column, the assumption was that cleavage is occurring at the N-terminal region of the protein and this product therefore lacks the N-terminal HA tag. In order to confirm this theory, a second aliquot of these fractions was run on an SDS-PAGE gel and Western blot analysis was performed using and anti-HA antibody. The results in Figure 7.7 show no indication of a cleavage product, thus confirming the theory that this cleavage occurs toward the N-terminus of DARC.

FT Fraction: 1234

DARC

Figure 7.7 Results from HA affinity purification of DARC, probed using an anti-HA antibody.

The next step for this system is to isolate the cleavage product and use FT-mass spectrometry in order to obtain residue-specific information regarding the exact location

163 where the cleavage occurs and substitute that residue with something less susceptible to cleavage.

164

7.4 Discussion and Future Directions

Both D6 and DARC are considered enigmatic members of the chemokine receptor family. While they bind multiple chemokine ligands, they are not coupled to the signaling pathways commonly used by standard chemokine receptors, mainly the activation of Gαi. However, one defining property of these “silent” receptors is their

ability to internalize multiple chemokine ligands without signaling through G-protein

[133]. Whether the ligand is targeted for cellular degradation or brought back to the

surface and released, appears to be dependent on the receptor, the ligand, and the cell

type [124, 133]. While there is some information regarding this mechanism, such as the

idea that D6 is constitutively phosphorylated in order to be rapidly internalized along with

its chemokine ligands, many details remain unknown.

Effect of receptor interactions on the fate of the chemokine ligand

Both D6 and DARC are highly promiscuous receptors [133, 139]. D6 currently

has eleven known chemokine ligands, all from the inflammatory CC family. In addition,

DARC not only binds sixteen chemokine ligands, as well as the malaria parasite P.

vivax, it is also the only known chemokine receptor with ligands that cross over between

both the CC and CXC families [133, 139, 143]. As mentioned previously, more is known

regarding the mechanisms of D6. For example, it is currently believed that ligand

interaction is not necessary for receptor internalization, but instead the intracellular C-

terminal domain is constitutively phosphorylated, allowing for constant interaction with β-

arrestin, resulting in rapid receptor turnover [134]. However, very little is known

regarding the structural mechanism by which different chemokine ligands bind to D6 and

what impact this may have on receptor function in terms of conformational changes,

165

activation of additional pathways, etc. At this time, there is even less information

available regarding the mechanism by which DARC internalizes its chemokine ligands

and recycles them back to the surface. It is believed that this process transports

chemokines from one location to another, enabling them then to bind and activate

signaling receptors [125].

Because the unifying theme for these two receptors involves ligand

internalization, the first long-term aim of this project is to determine the residue-specific

mechanisms by which the different ligands interact with D6 compared to DARC. This

type of analysis would address potential differences between CC and CXC chemokine

interactions with DARC, and how this differs from CC interactions with D6. In addition,

because DARC also binds the duffy binding-like domain of P. vivax, insight into the

differences between parasite interaction and chemokine interaction may lead to the

design of novel therapeutics in the treatment of malaria [143]. There are several

methods by which to study ligand:receptor interactions. One such method involves

using H/D exchange coupled to mass spectrometry, in order to determine residue-

specific information of both the ligand and the receptor [148].

Structural analysis of D6 and DARC

As discussed previously, there are currently no high resolution structures available for chemokine receptors. The closest structures for comparative studies and ligand:receptor complex modeling are that of Rhodopsin and the β2-adrenergic receptor

[127, 128]. Therefore, in order to fully understand the functional mechanisms of these receptors, it is crucial to have structural information at the atomic level. Despite the many challenges associated with crystallization of membrane proteins, several

166

techniques, including nanodiscs and bicelles, have proven to be successful. If sufficient

quantities of functional protein are obtained through this expression system, the ultimate

aim of this project is to obtain high resolution structures for both D6 and DARC and in

the presence of their different ligands.

Directions for the immediate future

In order to accomplish the ligand binding studies and structural characterization

described above, the expression of both proteins must first be scaled up in order to

generate sufficient quantities of protein for analysis. This should begin with D6 since the

preliminary data suggests the purification of full length is feasible, as opposed to DARC,

which is proteolytically cleaved during expression. One method by which to increase D6

expression is to grow the HEK293s cells in suspension, which would enable significantly

more cells per liter of media [145]. The next step is to screen detergents in order to

determine the optimal solubilization conditions for protein function [145]. Once these conditions have been optimized, the final step is to verify protein function. This is more straightforward with standard signaling receptors, where it is possible to measure cellular responses to activation, such as calcium mobilization. However, in the case of D6, which does not have any known signaling pathways, ligand binding assays are the most efficient way to confirm protein function [147]. One advantage in the analysis of silent receptors, is that they do not require the presence of G-protein for optimal ligand interaction, which has been a hindrance in the functional reconstitution of other signaling chemokine receptors. Although there is much yet to be accomplished toward the final aim of this project, the analysis presented here represents a critical first step along the way to understanding the structural and functional mechanisms of these silent chemokine receptors.

CHAPTER 8

Conclusions

CCL27, CCL28, and CXCL11 are members of the chemokine family. These proteins are best known for their role in leukocyte migration, both as part of standard immune system maintenance as well as in response to inflammation. As such, chemokines and their receptors have been implicated in a variety of pathophysicological processes. The first aim of this project was to investigate the monomeric structure and dynamics, as well as oligomeric properties of CCL27 using NMR spectroscopy. The first step involved developing a method, described in Chapter 2, to express, purify, and refold milligram quantities of both unlabeled and isotopically labeled functional protein. The oligomeric properties of CCL27 were then examined using pulsed field gradient (PFG)

NMR diffusion experiments, and the results indicated that CCL27 transitions between a monomer, dimer and tetramer species over a relatively narrow concentration range.

This data, in conjunction with 15N-1H HSQC chemical shift perturbations, filtered

(HC)NH-NOEs, and site-directed mutagenesis, suggest that several equilibria involving

different interfaces seem to be simultaneously at work. The overall results of this study

indicate that while the monomeric form of CCL27 has the standard chemokine motif, it

oligomerizes up to a tetramer by itself in solution, and the presence of the GAGs, shifts

the equilibrium to the oligomeric state, suggesting an important role for the oligomeric

form in GAG binding. It is possible that CCL27 is able to adopt different oligomeric

structures, depending on its GAG (or other) binding partners, thereby increasing the

functional diversity of this chemokine.

167 168

The interaction of chemokines with their receptors represents a very complex

field, with some chemokines specific to one receptor, while others activate multiple

receptors, expressed on diverse cells types. Because CCL27 is highly specific to the

receptor CCR10, the second aim of this project was to determine the structural

properties of CCL27 that induce activation of CCR10 and to compare these results to

studies involving CCL28, the other ligand for CCR10. Through targeted mutagenesis,

coupled to cellular migration assays, along with calcium influx, and receptor

desensitization assays, several mutants were isolated with both partial agonist and

antagonist properties. In addition, the results indicate that the N-terminal phenylalanine

residue is necessary for optimal function, with the addition of a second phenylalanine at

the N-terminus resulting in a super-agonist, with a 10-fold increase in activity. The

results also indicate that the C-terminal region of CCL27 plays a significant role in

receptor activation. The latter is interesting since traditionally, it is the N-terminal region

of chemokines that is responsible for receptor activation. Comparing these results to

CCL28, suggests that chemokine receptor specificity cannot be determined by the

primary sequence alone.

Interaction with GAGs has been shown to be crucial for chemokines to

successfully direct in vivo leukocyte migration, and GAG binding stabilizes the oligomeric

form. Given the heterogeneous expression and structural diversity of GAGs, these

interactions are believed to contribute to chemokine tissue specificity and biological

diversity. As such, the third aim of this project involved analysis of the GAG binding

regions of CCL27. GAG binding regions were determined by 15N-1H HSQC chemical shift perturbation analysis, combined with titration of the GAGs Heparin tetrasaccharide and Heparin hexasaccharide. Mapping the results from these titrations onto the surface

169 of CCL27 dimer models suggests that multiple oligomeric structures may be stabilized by interaction with different GAGs. Finally, previous studies using a binding assay, with

Heparin as the representative GAG, revealed that single alanine mutations to several surface exposed, positively charged residues, particularly K25A, significantly compromised Heparin affinity. Therefore, using the K25A mutant, it was shown that this

GAG-deficient mutant has decreased activity in terms of transendothelial cellular migration, suggesting GAG binding is directly related to CCL27 function. These results lend credence to the proposed hypothesis that CCL27 may adopt different oligomeric forms depending on its GAG binding partner. The methods developed for this project were used as part of a GAG-binding study of the chemokine CXCL11, which has been shown to have abnormally high GAG affinity.

The final aim of this project involved the expression and purification of the silent chemokine receptors D6 and the Duffy Antigen Receptor for Chemokines (DARC).

These receptors are termed silent receptors because they bind many chemokines with high affinity and specificity, but they do not signal through G-proteins. Instead, it is believed that they act as regulators, either by targeting their chemokine ligands for degradation, or by shuttling them from one location to another. In this study, protein expression was optimized by transfecting both D6 and DARC into a tetracycline- inducible mammalian cell expression system. This technique is similar to bacterial expression in that the cells grow to confluency prior to the induction of protein expression, thus helping to minimize cellular toxicity normally associated with the expression of these large seven trans-membrane helical proteins. Both D6 and DARC were purified from these mammalian cells and subsequently solubilized, demonstrating several advantages of this method over other expression systems.

APPENDIX I

PROTOCOLS

A. PFG diffusion NMR by Diffusion Ordered Spectroscopy

(DOSY)

A.1: Background

Pulsed Field Gradient (PFG) diffusion effectively measure the average distance traveled by solute molecules in an NMR sample during a fixed period of time [75]. The

method relies on the use of pulse field gradients to spatially encode the NMR signals

such that their effective position in the NMR tube can be monitored [81]. Measuring the

signal attenuation in successive proton spectra, as a function of increasing gradient field

strength, enables calculation of the self-diffusion coefficient, Ds, which is directly related

to the hydrodynamic radius of the protein in solution via the Stokes Einstein relationship

[44, 77, 81]. Briefly, a relatively slowly decaying signal corresponds to a smaller value of

Ds, which equates to a larger apparent molecular weight and vice versa. By accurately calibrating gradient strengths with solvents of known Ds, one can obtain an accurate

measurement of the Ds of the protein of interest. Table I.1 lists the pulse sequences that

I have optimized, as well as a brief description, and their experimental applications.

170 171

Table A.1 Description of optimized PFG diffusion pulse sequences, set up to run with Bruker DOSY command.

A.2: Calibrations

For accurate calculation of a diffusion coefficient, the maximum gradient strength

must be measured and stored in the Bruker Topspin software. The file is accessed

using “gradpar.”

Figure A.1 Gradient calibration window, accessed using the command “gradpar.”

For these experiments, the gradient constant for the spectrometer was calibrated

using a doped water sample. The literature value for the diffusion coefficient of D2O is

1.872 x 10-9 m2/s at 298 K. Setting the Gradient calibration constant to 5.7839158

-9 2 G/mm gave a diffusion coefficient of 1.92 x 10 m /s, for 0.1 M CuSO4 in 99% D2O,

acquired at 298 K.

172

A.3: Parameter optimization

The diffusion time (d20), also known as ∆, as well as the length of the gradient

pulse (p30), need to be optimized for each sample. This is accomplished using a

reference experiment and a final experiment, 1-dimensional versions of the diffusion

pulse sequence (for example, ledbpgppr1d2s), with the gradient set to minimum and

maximum values. Times for d20 are ~ 50 – 200 ms and p30 < 2.5 ms. Good starting

point values for d20 and p30 are 100 ms and 1.0 ms respectively. In the reference

experiment, the gradient amplitude (gpz6) is set to 2%, representing the minimum value.

The experiment is then acquired using the command zg. In the final experiment, gpz6 is

set to 95%, representing the maximum gradient amplitude. The intensity of the peaks in

the final experiment should be ~2% that of the reference experiment, and just above the

noise. If not, adjust d20 and p30 accordingly.

A.4: Data acquisition

The Bruker au program “dosy” is used to acquire diffusion experiments. This

program creates a “difframp” file used for acquisition at different gradient amplitudes, as

well as a “difflist” used for fitting the final results. In order to acquire, type dosy and use

the following settings:

 1st grad amp – 2

 final grad amp – 95

 number of points – 16

 ramp type (l/e/q) – l

 start acq? – y

This will result in 16 separate 1-dimensional proton experiments, acquired with a linearly increasing gradient amplitude from 2% to 95%. Figure A.2 illustrates the resulting FIDs.

173

Increasing gradient strength

Figure A.2 Series of FIDs acquired with an increasing gradient amplitude of 2 – 95% [GET BETTER FIGURE FOR THIS]

A.5: Data processing, pseudo-2D using DOSY plot

The experiment is initially processed using xf2. Under the processing (ProcPars) tab, selecting D will display the DOSY parameters, shown in Figure A.3.

Figure A.3 This is an example of the DOSY parameters.

Clicking on the icons labeled 1 and 2 allows the DOSY acquisition parameters to

be automatically uploaded. The icon designated as 3 will plot the data as a Pseudo 2D

174

experiment, with the proton chemical shift values in the x-axis and the negative log of the

diffusion coefficient in the y-axis (Figure A.4).

Figure A.4 Example of a Pseudo-2D spectrum processed using DOSY plot

A.6: Data processing, calculation of the self-diffusion coefficient, Ds

The self-diffusion coefficient, Ds is a quantitative measure of how quickly the sample diffuses in the solution. The value is calculated based on the decrease in signal intensity as a function of gradient amplitude, using the Bruker T1/T2 application. Using the ProcPars tab, Figure A.5 illustrates which icon re-processes the data for quantitate analysis (the command xf2 may also be used). The resulting spectrum is shown in

Figure A.6. From the Analysis window, selecting T1/T1 Relaxation creates the screen illustrated in Figure A.7.

175

Figure A.5 DOSY processing window. The red circle indicates the icon that re- processes the data for quantitate analysis of the diffusion coefficient. The same can be accomplished using the command xf2.

Figure A.6 Spectrum resulting from xf2 processing, used for calculation of Ds

176

Figure A.7 Display for calculating the diffusion coefficient.

The icon Extract slice processes the first slice of the experiment as a 1- dimensional proton spectrum. The peaks are then integrated using the Peaks/Ranges icon and selecting manual integration. Selecting Relaxation Window brings up the following decay curve, shown in Figure A.8, for each peak integrated.

Figure A.8 Example of fit decay curve using the equation described in the spectral window. Using the function highlighted by the red arrow lists the calculated Ds values for each of the integrated proton resonances.

177

The resulting decay curves are fit using the command key highlighted by the red

circle, and the Ds values are calculated using the equation:

I(g) = I(0) exp [-(g)2 D(∆ - (/3))] where I(0) is 4.167 x 10-1,  is 4.258 x 103 Hz/G, g was calibrated at 5.784 G/mm,  was set to 4.0 ms, and ∆ was set to 150 ms. These values are listed below the command buttons on the screen (Figure A.8). The theoretical change in Ds from the monomeric to dimeric form of a protein may be estimated by approximating the monomer-monomer interaction as a hard-sphere molecular contact. Using the expected change in the frictional coefficient in conjunction with the Stokes-Einstein equation, the theoretical ratio

Ds,dimer/Ds,monomer is 0.75 [81].

A.7: Applications

The Pseudo-2D spectra have several applications. One example is the

application of DOSY Plot involves polymer assisted diffusion experiments of a mixture of

species with similar molecular weights and chemical compositions. This type of

experiment is useful in natural products chemistry to extract a compound of interested

from an isolated mixture of species. The enhancement of diffusion differences is

achieved using a soluble polymer, polyvinylpyrrolidone (PVP) as the “stationary phase”

in the NMR tube. The resulting “separation” of analytes, based on their relative

interaction with the polymer, resembles normal-phase chromatography [149]. In Figure

A10, these molecules on their own have virtually identical diffusion coefficients, but can

be resolved in the presence of PVP.

178

Figure A.9 Results from polymer-assisted DOSY analysis of a mixture of the illustrated This experiment was optimized using the dstebpgp3s_cpmg_aj pulse sequence which has an added CPMG sequence in order to minimize the signal contributions from the large polymer.

Figure A.10 illustrates a second application, protein oligomerization, which has been used in several chapters presented throughout this document. For example, the results for CCL27 at concentrations ranging from 0.25 mM to 3,0 mM are shown in the

Pseudo-2d spectrum in Figure A.10 and indicate that the protein is oligomerizing as the concentration increases.

179

Figure A.10 Overlaid pseudo-2D spectra for CCL27 at several concentrations. The change in diffusion indicates that CCL27 is becoming larger as the concentration increases.

Finally, the data in Figure A.10 may be quantified using the T1/T2 relaxation method in order to calculate a diffusion coefficient at each concentration. The resulting

Ds values are plotted as a function of concentration and serve to generate a diffusion profile, such as the example in Figure A.11.

B Oligomerization Profile of CCL27 1.50 m2/s

-10 Monomer 1.30

1.10 Dimer

0.90

Diffusion Coefficient x 10 Tetramer 0.70 00.511.522.5335 . Concentration (mM)

Figure A.11 Oligomerization profile for CCL27 (from Chapter 3), generated using the T1/T2 relaxation method, described above, to calculate the diffusion coefficient at each concentration.

180

B: Migration assay, CCL27 with L1.2 cells expressing CCR10

1. Re-suspend a bit of lyophilized CCL27 in 1 ml of 50 mM hepes pH 7.2. Determine the concentration by absorbance at 280 nm 2. Adjust the concentration to 100 uM to use that as your stock solution. 3. A good range to test the chemokine for migration would be from 0 nM to 1000 nM. 4. Prepare the dilution in the bottom of the 24 transwell filter plates by the following dilutions for a standard assay (Tables prepared in Excel):

Concentration No filter No filter No filter Stock Vol. Protein Vol. Media Concentration 0 0.115 10 50 Stock 0 0.01 0.11 1 10 Vol. Protein 06636 3 Vol. Media 600 594 594 597 594 597 Concentration 0 0.115 10 50 Stock 0 0.01 0.11 1 10 Vol. Protein 06636 3 Vol. Media 600 594 594 597 594 597 Concentration 0 0.115 10 50 Stock 0 0.01 0.11 1 10 Vol. Protein 06636 3 Vol. Media 600 594 594 597 594 597

50 Concentration (L1.20 50 (L1.20 50 (L1.20 Stock 10 10 10 Vol. Protein 333 Vol. Media 597 597 597 Concentration 100 250 500 1000 Stock 10 25 25 25 Vol. Protein 66 12 24 Vol. Media 594 594 588 576 Concentration 100 250 500 1000 Stock 10 25 25 25 Vol. Protein 66 12 24 Vol. Media 594 594 588 576 Concentration 100 250 500 1000 Stock 10 25 25 25 Vol. Protein 66 12 24 Vol. Media 594 594 588 576

181

5. Move the filters to the well containing media + chemokine. 6. Do a cell count and prepare a cell suspension of 2.5 X 106 cells per mL. 7. Add cell suspension to each filter well (100 ul of 2.5 X 106 cells/ml suspension). Your positive control will be the addition of that same amount of cell directly in the bottom of the wells without filter/without chemokine.

8. Incubate at 37C/ 5% CO2 for 2 hours. 9. Pipet up all the 600 ul in the bottom of each well and transfer them to a small flow cytometry tube. Do flow cytometry (count the number of events happening in a given time (30 sec).

182

C: Transendothelial migration assay

C.1 Human Umbilical Endothelial Cell (HUVEC) Maintenance

Media: 53% M199 (Cell-gro, 10-060-CV) 25% Endothelial Growth Media, EGM (Cell Applications, 211-500) 20% FBS 1% Sodium pyruvate 1% Glutamax PenStrep, 1x

NOTE: Always prep T75 plates for HUVECs – add 10 mL of 1% gelatin in PBS, incubate at 37ºC for ~15 minutes, remove gelatin and add media

Defrosting HUVEC Cells: 1 vial defrosted into 1 x T75

1. Thaw at 37ºC for several minutes 2. Add cells to 10 mL Media 3. Centrifuge 4. Remove media and resuspend in 12 mL media 5. Add to T75

6. Incubate at 37ºC, 5% CO2

Change media 24 hours later to remove any residual DMSO

Splitting HUVEC Cells (split 1:5 every 5-6 days):

1. Coat new T75 with 1% Gelatin in PBS 2. Remove media from old cell plate 3. Add 4 mL Trypsin-EDTA 4. Incubate at 37ºC for ~3 min. a. Check constantly under microscope b. Quench as soon as cells come off the bottom of the plate

183

5. Quench trypsin with 5 mL media 6. Spin cells 7. Remove Gelatin from new T75 8. Resuspend cells in 5 mL media 9. Add 1 mL to new T75 10. Maximum of 10 passages total for cells

Freezing stocks

1. Add Trypsin-EDTA and incubate at 37 for 3 minutes 2. Quench with media and spin cells 3. Resuspend in FBS with 10% DMSO 4. Freeze 1 T25 per cyrovial

C.2 Trans Endothelial Migration Assay

1. Assay plates, Transwell Filter plates (5μm filter) 2. Coat transwell inserts with fibronectin (1:40 dilution of 1mg/mL stock in PBS, final: 25 ug/mL, use 60 uL per well). Allow the wells to sit for several hours in the incubator a. Remove extra liquid. b. Extra plates/wells can be wrapped and kept in the refrigerator.

3. Add media to the wells. Allow them to equilibrate overnight 4. Seed HUVECs (1 x 105 cells/well) onto 24-well transwell inserts. Make sure that there is 100uL of media on the top of the transwell. 5. Let the cells grow for 48 hours. Check electrical resistance.

Assay Media, for L1.2 cells: RPMI 10% FBS

184

Controls: 3 wells with no filter (maximum possible migration) 3 wells with filters only (no endothelial cells) 6. Before experiment, remove HUVEC media and carefully wash cells 3x with Assay Media. 7. Add Assay Media + chemokine (total volume, 600 uL) to outside chamber 8. Add 100 μL, containing 250,000 cells in Assay Media to the top chamber

9. Incubate in TC hood, 2 hours, 37ºC, 5% CO2 10. Remove Filters 11. Count the cells that migrated to the lower chamber by flow cytometry a. Count for 30 seconds b. Normalize to wells with no filter

Measuring filter-only wells (control): Evan’s Blue - BSA: Evan’s blue – BSA in media Final concentration: 0.67 mg/mL Evan’s blue, 4% BSA  Dilute 0.8 g BSA and 13.4 mg Evan’s blue in 20 mL media 1. Add 100 uL of Evan’s blue – BSA solution to the top chamber (time = 0) 2. At 30, 60, 90, and 120 minutes, transfer filter to new wells 3. Measure the absorbance of 150uL aliquots of the samples on the plate reader at 630 nm (Figure C.1)

185

Evan's Blue Assay

0.35

0.3

0.25

0.2 Transfilter Transendothelial 0.15

0.1 Absorbance, 630nm

0.05

0 02 Hours

Figure C.1 Results from a standard Evan’s blue assay

186

D: Calcium flux and receptor desensitzation assays

CCR10/L1.2 cells Volumes used: 100ul cells (200,000 cells) + 100ul calcium dye + 50ul ligand.

Materials: Calcium 4 assay kit: Molecular Devices #R8142 Assay buffer: 1x Hanks Balanced salt solution, 20mM Hepes, pH7.4, 0.1% BSA Assay plates: BD Biocoat 96-well #356640 Compound/ligand plates: Costar 0.3ml v-bottom 96-well #3357

Ligand plate 1. Ligand stocks – 50 uM 2. Ligand will be diluted 1:5 (50 uL into 200 uL, total of 250 uL) 3. Make solutions at 5x desired concentrations of ligand using assay buffer a. 200 uL total in the ligand plate

Receptor plate 1. Harvest cells, wash twice in PBS/BSA, resuspend in assay buffer, count and dilute to 2.0x106 cells/mL using assay buffer a. ~12 mL of cells is require per 96-well plate 2. Add 100ul of cells to each well in a 96-well biocoat plate 3. Make up calcium dye in 10.5 mL assay buffer – make sure all goes into solution by pipetting up and down (it will foam, be careful). One vial of dye is sufficient for 1 96-well plate 4. Carefully add 100ul of dye in cell plate

Only do step #5 if using adherent cells (for L1.2 cells, just put them straight into the incubator) 1. Spin plate down for 2-3 minutes at 200g (remember to turn down rates of acceleration/decceleration). 2. Check homogeneity of cells by microscopy and place in 37C incubator for 1 hour before measuring

Flex Station Use calcium flux protocol and adjust accordingly. NOTE: be careful to make sure tips pull from the appropriate wells in the ligand plate and load into appropriate wells in cell plate

187

E: Ubiquitinase purification

E.1: Culture

1. Pick a colony and inoculate 2.5 mL of sterile LB 2. 2-3 hours later, transfer to 25mL

3. Grow to ~A600 = 0.6 - 0.7 4. Innoculate 4 x 1 L LB with 2.5 mL of initial culture

5. Grow to ~A600 = 0.35 at 37 C 6. Prior to induction collect 1 mL sample to run on gel 7. Induce cells with IPTG (final concentration = 0.5 mM); in this case add 500 μL of 1 M IPTG

8. Grow for 3-4 hours after induction (collect 1 mL sample to run on gel); record A600

E.2: Harvest

9. Pour cells into Beckman buckets; balance 10. Spin down in JLA 8.1 rotor for 15 min, 5.5 krpm, 4C 11. Pour off supernatant 12. Resuspend pellet in ~50 mL of Buffer A (see table below) 13. Pour into conical tube 14. Store in -80 C overnight E.3: Purification

Reagents 50 mM K2HPO4/KH2PO4, pH 8.0 500 mM NaCl 20 mM Imidazole 5 mM MgCl2 1 mM BME, 10% Glycerol

Lysis: 15. Add ~30 mg of lysozyme to thawed cells 16. Incubate on ice for ~20 min 17. Lyse cells by sonication: 30 s bursts, 7 X, setting 8 18. Centrifuge lysed cells at 4 C, 20 min, 15,000 rpm with JA25.50 rotor

188

19. Keep Supernatant in vial (protein is soluble, so it is in the supe) 20. Filter supernatant with Durapore filter, keep on ice until ready to FPLC

NOTE: This makes quite a bit of crude protein so the Durapore filter will clog. The best thing seems to be dividing into several aliquots during centrifugation (use buffer A to dilute). Pool all supe and filter (plan to filter in stages using several different filters).

FPLC – Ni-NTA and S75 21. Make buffers for FPLC:

Buffer A Buffer B 50 mM K2HPO4/KH2PO4, pH 8.0 50 mM K2HPO4/KH2PO4, pH 8.0 500 mM NaCl 500 mM NaCl 1 mM BME, 10% Glycerol 1 mM BME, 10% Glycerol 20 mM Imidazole 500 mM Imidazole pH buffers to 8 and steri-filter

22. Set up FPLC for use with Ni-NTA column - Pumpwash with water - Column wash two column volumes (40 mL) at 1 mL/min with water - Pumpwash with your buffers - Equilibrate column by column washing with your buffers (40 mL) at 2 mL/min 23. Divide protein supe into 2 aliquots and run separate on Ni-NTA column - Keep 1 aliquot at 4ºC while the other runs on Ni-NTA

For each of the 2 Aliquots: 24. Add protein to superloop 25. Attach superloop to FPLC 26. Before running sample through the Ni-NTA column, be sure there are no leaks at the connections—do a bypass column injection and look for any leaks 27. When ready, set method and run protein through FPLC 28. Pool fractions containing protein 29. Concentrate protein to 5 mL (be careful because it crashes out at high concentrations)

189

30. Record A280 before and after concentrating

Gel Filtration ** Note: S-75 requires ~6 h for equilibration prior to using column—plan accordingly 31. Prepare Buffer in advance

Buffer A 50 mM K2HPO4/KH2PO4, pH 8.0 300 mM NaCl 1 mM BME, 10% Glycerol

32. Run 5 mL through Sephadex-75 column (no gradient, just Buffer A) 33. Add 1:1 glycerol to the pooled fractions 34. Concentrate to 80 – 100uM 35. store at -20ºC

190

F: D6 and DARC purification from HEK Cells, N-terminal HA tag

and C-terminal His6 tag

1. Spin down cells and wash with PBS.

2. Add protease inhibitor tablets (2 per 50ml buffer) and DNAse I to resuspension buffer and resuspend.

Resuspension buffer 20mM Na phosphate, pH 8.0 0.3M NaCl 10% v/v glycerol

3. Disrupt cells carefully using the Emulsiflex run at ~1kpsi/6555 kPa.

4. Remove cell debris by centrifugation (10,000g, 20 min)

5. Centrifuge the supernatant at 125,000g for 1 hour (41 krpm using the 70Ti rotor) and resuspend the membrane pellet in solubilization buffer.

Solubilization buffer 20mM Na phosphate, pH 8.0 0.3M NaCl 10% v/v glycerol 2% (w/v) DDM 0.05% (w/v) cholesterol hemisuccinate PI tablets (no EDTA) 20mM imidazole

6. Stir gently at 4°C for 2 hours, to allow the receptors to solubilize.

7. Spin again at 125,000g for 1 hour, to remove insoluble material.

8. Apply to a Ni-NTA column and elute with 250mM imidazole.

Ni-Wash buffer 20mM Na phosphate, pH 8.0 0.3M NaCl 10% v/v glycerol 0.2% (w/v) DDM 0.005% (w/v) cholesterol hemisuccinate 20mM imidazole PI cocktail (no EDTA).

191

Ni Elution buffer 20mM Na phosphate, pH 8.0 0.3M NaCl 10% v/v glycerol 0.2% (w/v) DDM 0.005% (w/v) cholesterol hemisuccinate 250mM imidazole PI cocktail (no EDTA).

9. Apply to HA affinity column

HA Equilibration Buffer 20 mM Tris, pH 7.5 0.1 M NaCl 0.1 mM EDTA 10% (v/v) glycerol 0.2% (w/v) DDM 0.005% (w/v) CHS 1 PI Tablet

HA Wash Buffer 20 mM Tris, pH 7.5 0.1 M NaCl 0.1 mM EDTA 10% (v/v) glycerol 0.2% (w/v) DDM 0.005% (w/v) CHS 1 PI Tablet 0.05% Tween-20

HA Elution HA Peptide at 1 mg/mL in Equilibration buffer

10. Wash HA column with 5 mL Equilibration buffer

11. Load Sample and equilibrate at room temperature for 5 minutes

12. Elute with HA peptide, aliquot into 10 x 1 mL tubes

13. Concentrate samples, run on SDS-PAGE gels and do western analysis to detect presence of the protein

APPENDIX II

Preliminary Transendothelial Migration Assay

This preliminary assay was performed using the chemokine CCL7/MCP-3 with

THP-1 cells natively expressing one of the receptors for CCL7, CCR2. The protocol outlined in Appendix I.C was used with HUVEC cells grown in collaboration with the

Esko Lab. The transendothelial migration assay was run in parallel with a standard trans-filter migration assay (protocol outlined in Appendix I.B). Figure II.1 shows the results for both the transfilter and transendothelial assays.

0.18

0.16 Transfilter Assay Transendothelial 0.14

0.12

0.10

0.08

0.06

Cells Migrated/Total Cells 0.04

0.02

0.00 0 1 10 100 1000 5000 Chemokine Concentration (nM)

Figure II.1 Results from Transfilter and Transendothelial migration assays using CCL27 and TCP-1 cells natively expressing CCR2.

192 193

These results suggest that the THP-1 cells are not expressing CCR2 sufficiently to induce migration. It is possible that the endothelial cells are attracting additional proteins on the surface of the THP-1 cells, causing the migration for the transendothelial assay to be slightly higher than the transfilter. This did confirm that it will be possible to use the flow cytometer to count the migrated cells. The assay is currently being developed for L1.2 cells stably transfected to express CCR10.

APPENDIX III

Transendothelial Migration Assay Control Experiments

It is necessary to measure resistance across the membrane for each assay in order to determine that the endothelial layer has been properly cultured on the filter. We performed these experiments using the set-up in the Esko lab. Using a small ohmmeter,

700 µL PBS is added to a small chamber and the transwell filter is added on top. The media is removed from the filter and 100 µL PBS is added over the HUVEC cells. The top of the ohmmeter is placed in the filter and resistance is measured. For each assay, several filters were chosen at random to be tested and compared to filters that were treated with fibronectin, but not HUVEC cells. Figure III.1 shows the results for resistance measurements for each of the assays performed in this analysis. Each of the filters seeded with HUVECs had significantly higher resistance compared to the filters without cells.

194 195

K25A CCL27 Experiment [3-88] L3A CCL27 Experiment 80 70 75 Filters 65 70 HUV ECs 60 filters 65 HUV ECs 60 55 55 50 50

Resistance 45 Resistance 45 40 40 35 35 30 30 1234 12 Days Days

80 F-CCL27 Experiment 90 Heparin Competition Experiment

75 80 Filter 70 70 HUVECs 65 Filters HUV ECs 60 60 50 55 40 50 Resistance 30 45 Resistance (Ohm) 40 20 35 10

30 0 12 HUVECs Days

Figure III.1 Results for resistance measurements comparing filters seeded with HUVECs to filters treated with fibronectin and media. For each assay performed, the resistance across the filters was significantly higher for the ones treated with HUVECs.

REFERENCES

1. Baggiolini, M., Chemokines and leukocyte traffic. Nature, 1998. 392(6676): p. 565-8.

2. Baggiolini, M., B. Dewald, and B. Moser, Human chemokines: an update. Annu Rev Immunol, 1997. 15: p. 675-705.

3. Charo, I.F. and R.M. Ransohoff, The many roles of chemokines and chemokine receptors in inflammation. N Engl J Med, 2006. 354(6): p. 610-21.

4. Hartley, O., et al., Human immunodeficiency virus type 1 entry inhibitors selected on living cells from a library of phage chemokines. J Virol, 2003. 77(12): p. 6637- 44.

5. Kunkel, E.J. and E.C. Butcher, Chemokines and the tissue-specific migration of lymphocytes. Immunity, 2002. 16(1): p. 1-4.

6. Loetscher, P. and I. Clark-Lewis, Agonistic and antagonistic activities of chemokines. J Leukoc Biol, 2001. 69(6): p. 881-4.

7. Allen, S.J., S.E. Crown, and T.M. Handel, Chemokine: receptor structure, interactions, and antagonism. Annu Rev Immunol, 2007. 25: p. 787-820.

8. Blain, K.Y., et al., Structural and functional characterization of CC chemokine CCL14. Biochemistry, 2007. 46(35): p. 10008-15.

9. Bondue, A., et al., Characterization of the role of the N-loop of MIP-1 beta in CCR5 binding. Biochemistry, 2002. 41(46): p. 13548-55.

10. Baysal, C. and A.R. Atilgan, Elucidating the structural mechanisms for biological activity of the chemokine family. Proteins, 2001. 43(2): p. 150-60.

11. Colobran, R., et al., The chemokine network. I. How the genomic organization of chemokines contains clues for deciphering their functional complexity. Clin Exp Immunol, 2007. 148(2): p. 208-17.

12. Lau, E.K., et al., Chemokine-receptor interactions: GPCRs, glycosaminoglycans and viral chemokine binding proteins. Adv Protein Chem, 2004. 68: p. 351-91.

13. Lau, E.K., et al., Identification of the glycosaminoglycan binding site of the CC chemokine, MCP-1: implications for structure and function in vivo. J Biol Chem, 2004. 279(21): p. 22294-305.

14. Jarnagin, K., et al., Identification of surface residues of the monocyte chemotactic protein 1 that affect signaling through the receptor CCR2. Biochemistry, 1999. 38(49): p. 16167-77.

196 197

15. Wells, T.N., et al., Selectivity and antagonism of chemokine receptors. J Leukoc Biol, 1996. 59(1): p. 53-60.

16. Gong, J.H. and I. Clark-Lewis, Antagonists of monocyte chemoattractant protein 1 identified by modification of functionally critical NH2-terminal residues. J Exp Med, 1995. 181(2): p. 631-40.

17. Petit, S.J., N.E. Chayen, and J.E. Pease, Site-directed mutagenesis of the chemokine receptor CXCR6 suggests a novel paradigm for interactions with the ligand CXCL16. Eur J Immunol, 2008. 38(8): p. 2337-50.

18. Handel, T.M., et al., An engineered monomer of CCL2 has anti-inflammatory properties emphasizing the importance of oligomerization for chemokine activity in vivo. J Leukoc Biol, 2008. 84(4): p. 1101-8.

19. Johnson, Z., A.E. Proudfoot, and T.M. Handel, Interaction of chemokines and glycosaminoglycans: a new twist in the regulation of chemokine function with opportunities for therapeutic intervention. Cytokine Growth Factor Rev, 2005. 16(6): p. 625-36.

20. Proudfoot, A.E., et al., Glycosaminoglycan binding and oligomerization are essential for the in vivo activity of certain chemokines. Proc Natl Acad Sci U S A, 2003. 100(4): p. 1885-90.

21. Czaplewski, L.G., et al., Identification of amino acid residues critical for aggregation of human CC chemokines macrophage inflammatory protein (MIP)- 1alpha, MIP-1beta, and RANTES. Characterization of active disaggregated chemokine variants. J Biol Chem, 1999. 274(23): p. 16077-84.

22. Fernandez, E.J. and E. Lolis, Structure, function, and inhibition of chemokines. Annu Rev Pharmacol Toxicol, 2002. 42: p. 469-99.

23. Jin, H., et al., Investigation of CC and CXC chemokine quaternary state mutants. Biochem Biophys Res Commun, 2005. 338(2): p. 987-99.

24. Tuinstra, R.L., et al., Interconversion between two unrelated protein folds in the lymphotactin native state. Proc Natl Acad Sci U S A, 2008. 105(13): p. 5057-62.

25. Swaminathan, G.J., et al., Crystal structures of oligomeric forms of the IP- 10/CXCL10 chemokine. Structure, 2003. 11(5): p. 521-32.

26. Paoletti, S., et al., A rich chemokine environment strongly enhances leukocyte migration and activities. Blood, 2005. 105(9): p. 3405-12.

27. von Hundelshausen, P., et al., Heterophilic interactions of platelet factor 4 and RANTES promote monocyte arrest on endothelium. Blood, 2005. 105(3): p. 924- 30.

28. Crown, S.E., et al., Heterodimerization of CCR2 chemokines and regulation by glycosaminoglycan binding. J Biol Chem, 2006. 281(35): p. 25438-46.

198

29. Lowman, H.B., et al., Monomeric variants of IL-8: effects of side chain substitutions and solution conditions upon dimer formation. Protein Sci, 1997. 6(3): p. 598-608.

30. Paavola, C.D., et al., Monomeric monocyte chemoattractant protein-1 (MCP-1) binds and activates the MCP-1 receptor CCR2B. J Biol Chem, 1998. 273(50): p. 33157-65.

31. Campanella, G.S., et al., Oligomerization of CXCL10 is necessary for endothelial cell presentation and in vivo activity. J Immunol, 2006. 177(10): p. 6991-8.

32. Hoogewerf, A.J., et al., Glycosaminoglycans mediate cell surface oligomerization of chemokines. Biochemistry, 1997. 36(44): p. 13570-8.

33. Johnson, Z., et al., Interference with heparin binding and oligomerization creates a novel anti-inflammatory strategy targeting the chemokine system. J Immunol, 2004. 173(9): p. 5776-85.

34. Fermas, S., et al., Sulfated oligosaccharides (heparin and fucoidan) binding and dimerization of stromal cell-derived factor-1 (SDF-1/CXCL 12) are coupled as evidenced by affinity CE-MS analysis. Glycobiology, 2008.

35. Handel, T.M., et al., Regulation of protein function by glycosaminoglycans--as exemplified by chemokines. Annu Rev Biochem, 2005. 74: p. 385-410.

36. Imberty, A., H. Lortat-Jacob, and S. Perez, Structural view of glycosaminoglycan- protein interactions. Carbohydr Res, 2007. 342(3-4): p. 430-9.

37. Kuschert, G.S., et al., Glycosaminoglycans interact selectively with chemokines and modulate receptor binding and cellular responses. Biochemistry, 1999. 38(39): p. 12959-68.

38. Witt, D.P. and A.D. Lander, Differential binding of chemokines to glycosaminoglycan subpopulations. Curr Biol, 1994. 4(5): p. 394-400.

39. Yu, Y., et al., Chemokine-glycosaminoglycan binding: specificity for CCR2 ligand binding to highly sulfated oligosaccharides using FTICR mass spectrometry. J Biol Chem, 2005. 280(37): p. 32200-8.

40. Fox, J.M., et al., Structure/function relationships of CCR8 agonists and antagonists. Amino-terminal extension of CCL1 by a single amino acid generates a partial agonist. J Biol Chem, 2006. 281(48): p. 36652-61.

41. Homey, B., et al., CCL27-CCR10 interactions regulate T cell-mediated skin inflammation. Nat Med, 2002. 8(2): p. 157-65.

42. Ellyard, J.I., et al., Eotaxin selectively binds heparin. An interaction that protects eotaxin from proteolysis and potentiates chemotactic activity in vivo. J Biol Chem, 2007. 282(20): p. 15238-47.

199

43. Vives, R.R., et al., A kinetics and modeling study of RANTES(9-68) binding to heparin reveals a mechanism of cooperative oligomerization. Biochemistry, 2002. 41(50): p. 14779-89.

44. Veldkamp, C.T., et al., Structural basis of CXCR4 sulfotyrosine recognition by the chemokine SDF-1/CXCL12. Sci Signal, 2008. 1(37): p. ra4.

45. Dudek, A.Z., et al., Platelet factor 4 promotes adhesion of hematopoietic progenitor cells and binds IL-8: novel mechanisms for modulation of hematopoiesis. Blood, 2003. 101(12): p. 4687-94.

46. Nesmelova, I.V., et al., CXC and CC chemokines form mixed heterodimers: association free energies from molecular dynamics simulations and experimental correlations. J Biol Chem, 2008. 283(35): p. 24155-66.

47. Shahrara, S., et al., Inhibition of monocyte chemoattractant protein-1 ameliorates rat adjuvant-induced arthritis. J Immunol, 2008. 180(5): p. 3447-56.

48. Handel, T.M., et al., An engineered monomer of CCL2 has anti-inflammatory properties emphasizing the importance of oligomerization for chemokine activity in vivo Inhibition of monocyte chemoattractant protein-1 ameliorates rat adjuvant- induced arthritis. J Leukoc Biol, 2008. 84(4): p. 1101-8. 49. Azenshtein, E., et al., The CC chemokine RANTES in breast carcinoma progression: regulation of expression and potential mechanisms of promalignant activity. Cancer Res, 2002. 62(4): p. 1093-102.

50. Deng, H., et al., Identification of a major co-receptor for primary isolates of HIV-1. Nature, 1996. 381(6584): p. 661-6.

51. Godessart, N. and S.L. Kunkel, Chemokines in autoimmune disease. Curr Opin Immunol, 2001. 13(6): p. 670-5.

52. Murakami, T., A.R. Cardones, and S.T. Hwang, Chemokine receptors and melanoma metastasis. J Dermatol Sci, 2004. 36(2): p. 71-8.

53. Suzuki, K., et al., In vitro and in vivo characterization of a novel CCR3 antagonist, YM-344031. Biochem Biophys Res Commun, 2006. 339(4): p. 1217-23.

54. Sorensen, T.L., et al., Expression of specific chemokines and chemokine receptors in the central nervous system of multiple sclerosis patients. J Clin Invest, 1999. 103(6): p. 807-15.

55. Muller, A., et al., Involvement of chemokine receptors in breast cancer metastasis. Nature, 2001. 410(6824): p. 50-6.

56. O'Hayre, M., et al., Chemokines and cancer: migration, intracellular signalling and intercellular communication in the microenvironment. Biochem J, 2008. 409(3): p. 635-49.

200

57. Jabeen, T., et al., Structure of mouse IP-10, a chemokine. Acta Crystallogr D Biol Crystallogr, 2008. 64(Pt 6): p. 611-9.

58. Kuschert, G.S., et al., Identification of a glycosaminoglycan binding surface on human interleukin-8. Biochemistry, 1998. 37(32): p. 11193-201.

59. Martin, L., et al., Structural and functional analysis of the RANTES- glycosaminoglycans interactions. Biochemistry, 2001. 40(21): p. 6303-18.

60. Reiss, Y., et al., CC chemokine receptor (CCR)4 and the CCR10 ligand cutaneous T cell-attracting chemokine (CTACK) in lymphocyte trafficking to inflamed skin. J Exp Med, 2001. 194(10): p. 1541-7.

61. Amara, A., et al., Stromal cell-derived factor-1alpha associates with heparan sulfates through the first beta-strand of the chemokine. J Biol Chem, 1999. 274(34): p. 23916-25.

62. Simonetti, O., et al., Potential role of CCL27 and CCR10 expression in melanoma progression and immune escape. Eur J Cancer, 2006. 42(8): p. 1181- 7.

63. Homey, B., et al., Cutting edge: the orphan chemokine receptor G protein- coupled receptor-2 (GPR-2, CCR10) binds the skin-associated chemokine CCL27 (CTACK/ALP/ILC). J Immunol, 2000. 164(7): p. 3465-70.

64. Morales, J., et al., CTACK, a skin-associated chemokine that preferentially attracts skin-homing memory T cells. Proc Natl Acad Sci U S A, 1999. 96(25): p. 14470-5.

65. Hoeller, C., et al., In vivo imaging of cutaneous T-cell lymphoma migration to the skin. Cancer Res, 2009. 69(7): p. 2704-8.

66. Castelletti, E., et al., The mucosae-associated epithelial chemokine (MEC/CCL28) modulates immunity in HIV infection. PLoS ONE, 2007. 2(10): p. e969.

67. Pan, J., et al., A novel chemokine ligand for CCR10 and CCR3 expressed by epithelial cells in mucosal tissues. J Immunol, 2000. 165(6): p. 2943-9.

68. Wang, W., et al., Identification of a novel chemokine (CCL28), which binds CCR10 (GPR2). J Biol Chem, 2000. 275(29): p. 22313-23.

69. Catanzariti, A.M., et al., An efficient system for high-level expression and easy purification of authentic recombinant proteins. Protein Sci, 2004. 13(5): p. 1331- 9.

70. Subileau, E.A., et al., Expression of chemokines and their receptors by human brain endothelium: implications for multiple sclerosis. J Neuropathol Exp Neurol, 2009. 68(3): p. 227-40.

201

71. Szekanecz, Z., et al., Chemokines and angiogenesis in rheumatoid arthritis. Front Biosci (Elite Ed), 2009. 1: p. 44-51.

72. English, K., et al., Inflammation of the respiratory tract is associated with CCL28 and CCR10 expression in a murine model of allergic asthma. Immunol Lett, 2006. 103(2): p. 92-100.

73. Balkwill, F., Cancer and the chemokine network. Nat Rev Cancer, 2004. 4(7): p. 540-50.

74. Lusso, P., HIV and the chemokine system: 10 years later. Embo J, 2006. 25(3): p. 447-56.

75. Jansma, A., T.M. Handel, and D.J. Hamel, Chapter 2. Homo- and hetero- oligomerization of chemokines. Methods Enzymol, 2009. 461: p. 31-50.

76. Lubkowski, J., et al., The structure of MCP-1 in two crystal forms provides a rare example of variable quaternary interactions. Nat Struct Biol, 1997. 4(1): p. 64-9.

77. Veldkamp, C.T., et al., The monomer-dimer equilibrium of stromal cell-derived factor-1 (CXCL 12) is altered by pH, phosphate, sulfate, and heparin. Protein Sci, 2005. 14(4): p. 1071-81.

78. Koopmann, W. and M.S. Krangel, Identification of a glycosaminoglycan-binding site in chemokine macrophage inflammatory protein-1alpha. J Biol Chem, 1997. 272(15): p. 10103-9.

79. Rek, A., et al., A biophysical insight into the RANTES-glycosaminoglycan interaction. Biochim Biophys Acta, 2009.

80. Sadir, R., et al., Characterization of the stromal cell-derived factor-1alpha-heparin complex. J Biol Chem, 2001. 276(11): p. 8288-96.

81. Altieri, A.S., Association of Biomolecular Systems via Pulsed Field Gradient NMR Self-Diffusion Measurements. Journal of the American Chemical Society, 1995. 7: p. 740-743.

82. Handel, T.M. and P.J. Domaille, Heteronuclear (1H, 13C, 15N) NMR assignments and solution structure of the monocyte chemoattractant protein-1 (MCP-1) dimer. Biochemistry, 1996. 35(21): p. 6569-84.

83. Holmes, W.D., et al., Solution studies of recombinant human stromal-cell-derived factor-1. Protein Expr Purif, 2001. 21(3): p. 367-77.

84. Garcia de la Torre, J., M.L. Huertas, and B. Carrasco, HYDRONMR: prediction of NMR relaxation of globular proteins from atomic-level structures and hydrodynamic calculations. J Magn Reson, 2000. 147(1): p. 138-46.

202

85. Garcia De La Torre, J., M.L. Huertas, and B. Carrasco, Calculation of hydrodynamic properties of globular proteins from their atomic-level structure. Biophys J, 2000. 78(2): p. 719-30.

86. Dosset, P., et al., Efficient analysis of macromolecular rotational diffusion from heteronuclear relaxation data. J Biomol NMR, 2000. 16(1): p. 23-8.

87. Kroenke, C.D., Longitudinal and Transverse H-N Dipolar/N Chemical Shift Anisotropy Relaxation Ingerference: Unambiguous Determination of Rotational Diffusion Tensors and Chemical Exchange Effects in Biological Macromolecules. Journal of the American Chemical Society, 1998. 120(31): p. 7905-7915.

88. Cornilescu, G., F. Delaglio, and A. Bax, Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J Biomol NMR, 1999. 13(3): p. 289-302.

89. Linge, J.P., S.I. O'Donoghue, and M. Nilges, Automated assignment of ambiguous nuclear overhauser effects with ARIA. Methods Enzymol, 2001. 339: p. 71-90.

90. Nilges, M. and S.I. O'Donoghue, Ambiguous NOEs and automated NOE assignment. Progress in Nuclear Magnetic Resonance Spectroscopy, 1998. 32: p. 107-139.

91. Clore, G.M., Deviations from the simple two-parameter model-free approach to the interpretation of nitrogen-15 nuclear magnetic relaxation of proteins. Journal of the American Chemical Society, 1990. 112(12): p. 4989-4991.

92. Clore, G.M., et al., Three-dimensional structure of interleukin 8 in solution. Biochemistry, 1990. 29(7): p. 1689-96.

93. Wishart, D.S. and B.D. Sykes, The 13C chemical-shift index: a simple method for the identification of protein secondary structure using 13C chemical-shift data. J Biomol NMR, 1994. 4(2): p. 171-80.

94. Brunger, A.T., et al., Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr, 1998. 54(Pt 5): p. 905-21.

95. Hwang, T.L., P.C.M. Van Ziil, and S. Mori, Accurate quantitation of water-amide proton exchange rates using the Phase-Modulated CLEAN chemical Exchange (CLEANEX-PM) approach with a Fast-HSQC (FHSQC) detection scheme. Journal of Biomolecular NMR, 1998. 11(2): p. 221-226.

96. Laskowski, R.A., et. al., PROCHECK: a program to check the stereochemical quality of chemical structures. journal of applied crystallography, 1993. 26: p. 283-291.

203

97. Laskowski, R.A., et. al., AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. Journal of Biomolecular NMR, 1996. 8: p. 477-486.

98. Ghose, R. and J.H. Prestegard, Improved estimation of CSA-dipolar coupling cross-correlation rates from laboratory-frame relaxation experiments. J Magn Reson, 1998. 134(2): p. 308-14.

99. Bernado, P., J. Garcia de la Torre, and M. Pons, Interpretation of 15N NMR relaxation data of globular proteins using hydrodynamic calculations with HYDRONMR. J Biomol NMR, 2002. 23(2): p. 139-50.

100. Carrasco, B. and J. Garcia de la Torre, Hydrodynamic properties of rigid particles: comparison of different modeling and computational procedures. Biophysical Journal, 1999. 76: p. 3044-3057.

101. Loria, J.P., M. Rance, and A.G. Palmer, A relaxation-compensated Carr-Purcell- Meiboom-Gill sequence for characterizing chemical exchange by NMR spectroscopy. Journal of the American Chemical Society, 1999. 121(10): p. 2331-2332.

102. Jansma, A.K., J.; Hsu, A.; Handel, T.; Nietlespach, D., Manuscript in Preparation. 2009.

103. Gong, J.H., et al., RANTES and MCP-3 antagonists bind multiple chemokine receptors. J Biol Chem, 1996. 271(18): p. 10521-7.

104. Jin, H., et al., The human CC chemokine MIP-1beta dimer is not competent to bind to the CCR5 receptor. J Biol Chem, 2007. 282(38): p. 27976-83.

105. Murphy, J.W., et al., Structural and functional basis of CXCL12 (stromal cell- derived factor-1 alpha) binding to heparin. J Biol Chem, 2007. 282(13): p. 10018- 27.

106. Stringer, S.E., et al., Heparin is a unique marker of progenitors in the glial cell lineage. J Biol Chem, 1999. 274(36): p. 25455-60.

107. Proudfoot, A.E., et al., Amino-terminally modified RANTES analogues demonstrate differential effects on RANTES receptors. J Biol Chem, 1999. 274(45): p. 32478-85.

108. Wells, T.N.C., et al., The Molecular Basis of the Chemokine/Chemokine Receptor Interaction-Scope for Design of Chemokine Antagonists. Methods, 1996. 10(1): p. 126-34.

109. Booth, V., I. Clark-Lewis, and B.D. Sykes, NMR structure of CXCR3 binding chemokine CXCL11 (ITAC). Protein Sci, 2004. 13(8): p. 2022-8.

110. Zhang, Y.J., B.J. Rutledge, and B.J. Rollins, Structure/activity analysis of human monocyte chemoattractant protein-1 (MCP-1) by mutagenesis. Identification of a

204

mutated protein that inhibits MCP-1-mediated monocyte chemotaxis. J Biol Chem, 1994. 269(22): p. 15918-24.

111. Clark-Lewis, I., et al., Structure-activity relationships of interleukin-8 determined using chemically synthesized analogs. Critical role of NH2-terminal residues and evidence for uncoupling of neutrophil chemotaxis, exocytosis, and receptor binding activities. J Biol Chem, 1991. 266(34): p. 23128-34.

112. Huang, V., et al., Cutting edge: rapid accumulation of epidermal CCL27 in skin- draining lymph nodes following topical application of a contact sensitizer recruits CCR10-expressing T cells. J Immunol, 2008. 180(10): p. 6462-6.

113. Liu, B. and E. Wilson, The antimicrobial activity of CCL28 is dependent on C- terminal positively charged amino acids. Eur J Immunol, 2009.

114. Hieshima, K., et al., CCL28 has dual roles in mucosal immunity as a chemokine with broad-spectrum antimicrobial activity. J Immunol, 2003. 170(3): p. 1452-61.

115. Kagami, S., et al., CCL28 production in HaCaT cells was mediated by different signal pathways from CCL27. Exp Dermatol, 2006. 15(2): p. 95-100.

116. John, A.E., et al., Temporal production of CCL28 corresponds to eosinophil accumulation and airway hyperreactivity in allergic airway inflammation. Am J Pathol, 2005. 166(2): p. 345-53.

117. Ogawa, H., et al., Regulated production of the chemokine CCL28 in human colon epithelium. Am J Physiol Gastrointest Liver Physiol, 2004. 287(5): p. G1062-9.

118. Kotarsky, K., et al., A novel role for constitutively expressed epithelial-derived chemokines as antibacterial peptides in the intestinal mucosa. Mucosal Immunol, 2009.

119. Raj, P.A., M. Edgerton, and M.J. Levine, Salivary histatin 5: dependence of sequence, chain length, and helical conformation for candidacidal activity. J Biol Chem, 1990. 265(7): p. 3898-905.

120. Vlodavsky, I., et al., Mammalian heparanase: involvement in cancer metastasis, angiogenesis and normal development. Semin Cancer Biol, 2002. 12(2): p. 121- 9.

121. Ali, S., et al., A non-glycosaminoglycan-binding variant of CC chemokine ligand 7 (monocyte chemoattractant protein-3) antagonizes chemokine-mediated inflammation. J Immunol, 2005. 175(2): p. 1257-66.

122. Kuloglu, E.S., et al., Structural rearrangement of human lymphotactin, a C chemokine, under physiological solution conditions. J Biol Chem, 2002. 277(20): p. 17863-70.

205

123. Yu, Y., et al., Potential inhibitors of chemokine function: analysis of noncovalent complexes of CC chemokine and small polyanionic molecules by ESI FT-ICR mass spectrometry. J Am Soc Mass Spectrom, 2006. 17(4): p. 524-35.

124. Graham, G.J. and C.S. McKimmie, Chemokine scavenging by D6: a movable feast? Trends Immunol, 2006. 27(8): p. 381-6.

125. Hadley, T.J. and S.C. Peiper, From malaria to chemokine receptor: the emerging physiologic role of the Duffy blood group antigen. Blood, 1997. 89(9): p. 3077-91.

126. Jaakola, V.P., et al., The 2.6 angstrom crystal structure of a human A2A adenosine receptor bound to an antagonist. Science, 2008. 322(5905): p. 1211- 7.

127. Palczewski, K., et al., Crystal structure of rhodopsin: A G protein-coupled receptor. Science, 2000. 289(5480): p. 739-45.

128. Rasmussen, S.G., et al., Crystal structure of the human beta2 adrenergic G- protein-coupled receptor. Nature, 2007. 450(7168): p. 383-7.

129. Serrano-Vega, M.J., et al., Conformational thermostabilization of the beta1- adrenergic receptor in a detergent-resistant form. Proc Natl Acad Sci U S A, 2008. 105(3): p. 877-82.

130. Reeves, P.J., J.M. Kim, and H.G. Khorana, Structure and function in rhodopsin: a tetracycline-inducible system in stable mammalian cell lines for high-level expression of opsin mutants. Proc Natl Acad Sci U S A, 2002. 99(21): p. 13413- 8.

131. Locati, M., et al., Silent chemoattractant receptors: D6 as a decoy and scavenger receptor for inflammatory CC chemokines. Cytokine Growth Factor Rev, 2005. 16(6): p. 679-86.

132. Galliera, E., et al., beta-Arrestin-dependent constitutive internalization of the human chemokine decoy receptor D6. J Biol Chem, 2004. 279(24): p. 25590-7.

133. Nibbs, R., G. Graham, and A. Rot, Chemokines on the move: control by the chemokine "interceptors" Duffy blood group antigen and D6. Semin Immunol, 2003. 15(5): p. 287-94.

134. Weber, M., et al., The chemokine receptor D6 constitutively traffics to and from the cell surface to internalize and degrade chemokines. Mol Biol Cell, 2004. 15(5): p. 2492-508.

135. Jamieson, T., et al., The chemokine receptor D6 limits the inflammatory response in vivo. Nat Immunol, 2005. 6(4): p. 403-11.

136. Mantovani, A., et al., Extracellular and intracellular decoys in the tuning of inflammatory cytokines and Toll-like receptors: the new entry TIR8/SIGIRR. J Leukoc Biol, 2004. 75(5): p. 738-42.

206

137. de Brevern, A.G., et al., A structural model of a seven-transmembrane helix receptor: the Duffy antigen/receptor for chemokine (DARC). Biochim Biophys Acta, 2005. 1724(3): p. 288-306.

138. Czerwinski, M., et al., Mutational analysis of the N-glycosylation sites of Duffy antigen/receptor for chemokines. Biochem Biophys Res Commun, 2007. 356(3): p. 816-21.

139. Lu, Z.H., et al., The promiscuous chemokine binding profile of the Duffy antigen/receptor for chemokines is primarily localized to sequences in the amino- terminal domain. J Biol Chem, 1995. 270(44): p. 26239-45.

140. Lee, J.S., et al., Enhanced expression of Duffy antigen in the lungs during suppurative pneumonia. J Histochem Cytochem, 2003. 51(2): p. 159-66.

141. Lee, J.S., et al., The Duffy antigen modifies systemic and local tissue chemokine responses following lipopolysaccharide stimulation. J Immunol, 2006. 177(11): p. 8086-94.

142. Mayor, A., et al., Receptor-binding residues lie in central regions of Duffy- binding-like domains involved in red cell invasion and cytoadherence by malaria parasites. Blood, 2005. 105(6): p. 2557-63.

143. Singh, S.K., et al., Structural basis for Duffy recognition by the malaria parasite Duffy-binding-like domain. Nature, 2006. 439(7077): p. 741-4.

144. Yazdani, S.S., et al., Evaluation of immune responses elicited in mice against a recombinant malaria vaccine based on Plasmodium vivax Duffy binding protein. Vaccine, 2004. 22(27-28): p. 3727-37.

145. Allen, S.J., et al., Expression, purification and in vitro functional reconstitution of the chemokine receptor CCR1. Protein Expr Purif, 2009. 66(1): p. 73-81.

146. Chelikani, P., et al., The synthesis and high-level expression of a beta2- adrenergic receptor gene in a tetracycline-inducible stable mammalian cell line. Protein Sci, 2006. 15(6): p. 1433-40.

147. Blackburn, P.E., et al., Purification and biochemical characterization of the D6 chemokine receptor. Biochem J, 2004. 379(Pt 2): p. 263-72.

148. Joh, N.H., et al., Modest stabilization by most hydrogen-bonded side-chain interactions in membrane proteins. Nature, 2008. 453(7199): p. 1266-70.

149. Kavakka, J.S., I. Kilpelainen, and S. Heikkinen, General chromatographic NMR method in liquid state for synthetic chemistry: polyvinylpyrrolidone assisted DOSY experiments. Org Lett, 2009. 11(6): p. 1349-52.