The Role Of Immunoglobulin Receptors In The Pathogenesis Of Rheumatoid Arthritis

Vikki Martyne Abrahams

Thesis submitted for Doctor of Philosophy in Immunology

University College London

2000

i ProQuest Number: 10797830

All rights reserved

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a com plete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest

ProQuest 10797830

Published by ProQuest LLC(2018). Copyright of the Dissertation is held by the Author.

All rights reserved. This work is protected against unauthorized copying under Title 17, United States C ode Microform Edition © ProQuest LLC.

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106- 1346 ABSTRACT Historically, a key immunological feature of rheumatoid arthritis is the presence of small circulating immunoglobulin G (IgG) rheumatoid factor-based immune complexes. More recently, tumour necrosis factor alpha (TNFa) has been shown to be an important mediator of inflammation in rheumatoid arthritis. The aim of this thesis was to test a hypothetical model outlining an effector mechanism for inflammation in rheumatoid arthritis. This model predicts that small immune complexes comprising self-associating IgG rheumatoid factors, owing to their small size, evade complement-mediated clearance from the circulation and access tissue macrophages. These complexes may initiate inflammation within synovium by binding to one of three immunoglobulin receptors for IgG (FcyR). Unlike FcyRI and FcyRII, FcyRIIIa expression by macrophages correlates w ith the location of synovitis and extra-articular features in patients with rheumatoid arthritis. Selective FcyRIIIa crosslinking by IgG rheum atoid factor complexes may trigger macrophage activation and the subsequent production of proinflammatory mediators such as TNFa.

Murine IgG anti-FcyR monoclonal antibodies (mAb) to each FcyR were used as surrogates for small immune complexes. Of the three anti-FcyR mAb, only that directed against FcyRIIIa triggered adhered monocytes to release TNFa, IL-la and reactive oxygen species. This confirmed previous evidence that for signalling, FcyRI and FcyRII require multiple receptor aggregation. Further findings indicated anti-FcyRIII mAb- induced TNFa release to be dependent on crosslinking of either three FcyRIIIa receptors, or two receptors where one FcyRIIIa bound an IgG Fc fragment. Small, soluble human IgG anti-NIP mAb immune complexes were also able to trigger adhered monocytes to release TNFa in a subclass dependent manner. Moreover, although this TNFa response was found to be dependent upon both FcyRII and FcyRIIIa, FcyRIIIa appeared to play a dom inant role.

The findings of this thesis support the hypothesis that small IgG rheumatoid factor based immune complexes may trigger the production of TNFa and other proinflammatory mediators from macrophages in rheum atoid arthritis, and implicate an im portant role for FcyRIIIa.

2 For Mum and Dad

In memory of Nana

3 THANKS To Jo Edwards and Jo Cambridge. You've both inspired, encouraged and supported me over the past few years. I cannot thank you enough.

Thanks go to Celltech Therapeutics Ltd for their financial support.

Many thanks go to the staff at Haematology day care (PPW2) and Haematology outpatients, University College Hospital for all their help.

ACKNOWLEDGEMENTS I am very grateful to Professor Peter Lydyard for technical advice, useful discussions and providing the anti-FcyRI monoclonal antibody (32.2), Professor Michael Fanger for kindly providing the anti-FcyRIII monoclonal antibody cell line and purified anti-FcyRII monoclonal antibody, Dr Nancy Hogg for generously donating the anti-FcyRI monoclonal antibody, Professor Dennis Carson and Professor Pojen Chen for kindly giving me the human IgG rheumatoid factors monoclonal antibody clones and Professor Peter Lachmann and Dr Julia Voice for kindly providing the NIP immune complex reagents.

Thanks also go to Dr Andrew Beavil (The Randall Institute, Kings College London, London) for his help with my HPLC experiments and to Dr John Hoddersall (Department of Nephrology, University College London, London) for providing me time on the chemiluminometer.

DECLARATION The work presented in this thesis was carried out by the author unless otherwise stated.

4 CONTENTS

Title 1 Abstract 2 Dedication 3 Thanks 4 Acknowledgements 4 Declaration 4 Abbreviations 16

Chapter 1. Introduction 19 1.1. History of rheumatoid arthritis 20 1.2. Clinical and pathological features of rheumatoid arthritis 20 1.2.1. Extra-articular features of rheumatoid arthritis 21 1.3. Aetiological features of rheumatoid arthritis 22 1.4. Mediators of inflammation and tissue damage in rheumatoid arthritis 24 1.4.1. Cytokines and rheumatoid arthritis 24 1.4.2. Tumour necrosis factor 25 1.4.3. Tumour necrosis factor receptors 24 1.4.4. Interleukin-1 26 1.4.5. Interleukin-1 receptors 26 1.4.6. Tumour necrosis factor, interleukin-1 and rheumatoid arthritis 27 1.4.7. Interleukin-6 and other cytokines 29 1.4.8. Chemokines and growth factors 30 1.4.9. Reactive oxygen species 31 1.4.10. Matrix metalloproteinases 32 1.4.11. Com plem ent 33 1.5. The immunopathogenesis of rheumatoid arthritis 33 1.6. Rheumatoid factors and rheumatoid arthritis 35 1.6.1. Prevalence of rheumatoid factors and methods of detection 36 1.6.2. Rheum atoid factor isotype 38 1.6.3. Fine specificity of rheumatoid factors 38 1.6.4. IgG as an immunogen for rheumatoid factor production 41 1.6.5. Origins of rheum atoid factors 42

5 1.6.6. Rheumatoid factors and the formation of immune complexes 46 1.6.7. IgM rheumatoid factor immune complexes 46 1.6.8. IgG rheumatoid factor immune complexes 47 1.6.9. Rheumatoid factor complexes and Waldenstrom's hypergammaglobulinemic purpura 50 1.6.10. Fixation and activation of complement by rheumatoid factors 51 1.7. Human immunoglobulin receptors 53 1.7.1. FcyRI (CD64) 55 1.7.2. FcyRII (CD32) 55 1.7.3. FcyRIII (CD16) 56 1.7.4. Fey receptor subunits and signalling 57 1.7.5. Fey receptor polymorphisms 61 1.7.6. Modulation of Fey receptor expression 64 1.7.7. Fey receptor expression by tissue macrophages 65 1.8. A novel hypothesis for the pathogenesis of rheumatoid arthritis 67 1.8.1. Effector mechanism for the initiation of inflammation in rheumatoid arthritis 67 1.8.2. IgG rheum atoid factor subclass 68 1.8.3. Chronicity of the initial inflammatory event in rheumatoid arthritis 68 1.9. Hypothesis and aims of this thesis 72 1.9.1. Hypothesis and previous work by other groups 72 1.9.2. Aim s 73 1.9.3. Experimental approaches 73

Chapter 2. Materials and methods 75 2.1. Primary cells and hybridomas used in these studies 76 2.1.1 Primary cells 76 2.1.2. Mouse cell lines 76 2.1.3. H um an cell lines 76 2.2. Culturing of primary cells and continuous cell lines 76 2.3. Murine monoclonal antibodies used in these studies 76 2.3.1. Anti-FcyR! monoclonal antibodies 76

6 2.3.2. Anti-FcyRII monoclonal antibodies 78 2.3.3. Anti-FcyRIII monoclonal antibodies 78 2.3.4. Anti-TNF monoclonal antibodies 79 2.4. Human monoclonal antibodies used in these studies 79 2.4.1. IgG rheumatoid factor monoclonal antibodies 79 2.4.2. IgG anti-NIP m onoclonal antibodies 80 2.5. Purification of monoclonal antibodies by affinity chromatography 80 2.5.1. Purification of murine monoclonal antibodies 80 2.5.2. Preparation of F(ab')2 fragments of the anti-FcyRIII mAb 81 2.5.3. Preparation of Fab fragments of the anti-FcyRIII mAb 81 2.5.4. Purification of human monoclonal antibodies 82 2.6. Measurement of protein concentration 82 2.7. Measurement of immunoglobulin (IgG) concentration 83 2.7.1. Measurement of murine IgG concentration 83 2.7.2. M easurem ent of hum an IgG concentration 83 2.8. Measurement of rheumatoid factor activity 84 2.9. Immune complexes used in these studies 85 2.9.1. M easurem ent of hum an IgG rheum atoid factor mAb immune complex size by analytical high-performance liquid chromatography 85 2.9.2. Separation of human IgG rheumatoid factor mAb immune complexes by preparative high-performance liquid chromatography 85 2.9.3. Preparation of human IgG anti-NIP mAb immune complexes 86 2.10. Source of human peripheral blood monocytes 86 2.10.1. Isolation of hum an peripheral blood monocytes from whole blood 87 2.11. Immunofluorescent staining and flow cytometry 88 2.11.1. Indirect immunofluorescent staining of whole blood 88 2.11.2. Indirect immunofluorescent staining of isolated monocytes 89 2.11.3. Flow Cytometry 90 2.11.4. Calibration of the FACScan 91 2.11.5. Calculation of the number of Fey receptor binding sites per cell 91

7 2.12. Culturing of human monocytes and production of cytokinesin vitro 92 2.13. Detection and measurement of cytokines in culture supernatants 94 2.13.1. ELISA for the detection and measurement of TNFa in culture supernatants 94 2.13.2. Alternative ELISA for the detection and measurement of TNFa in culture supernatants 95 2.13.3. ELISA for the detection and m easurem ent of IL-la in culture supernatants 96 2.14. Detection and measurement of reactive oxygen species production by adhered human monocytesin vitro 97 2.15. Statistical analysis 98 2.16. Data representation 98

Chapter 3. Results 99 3.1. Fey receptor expression by human peripheral blood monocytes 100 3.1.1. Murine anti-FcyR mAb: confirmation of antibody specificity and determination of optimal binding levels 100 3.1.2. FcyR Expression by isolated and adherent hum an monocytes in vitro 108 3.1.3. The effect of TNFa on FcyR expression by adhered monocytes in vitro 115 3.2. TNFa production by human peripheral blood monocytes in vitro 119 3.2.1. The effects of peripheral blood monocyte isolation and adherence on TNFa production 119 3.2.2. The effects of culturing adhered monocytes on TNFa Production 120 3.2.3. Reproducibility and donor variability of TNFa production by adhered human monocytes 121 3.3. The effects of Fey receptor ligation on TNFa production by adhered human monocytesin vitro 122 3.3.1. TNFa release from adhered monocytes following incubation with murine anti-FcyR mAb 122

8 3.3.2. The effects of blocking FcyRI or FcyRII on the anti-FcyRIII mAb induced TNFa response from adhered human monocytes 127 3.3.3. TNFa release from adhered monocytes following Incubation with F(ab')2 or Fab fragments of the anti-FcyRIII mAb 129 3.3.4. TNFa release from adhered monocytes following incubation with more than one anti-FcyR mAb 131 3.4. The effects of Fey receptor ligation on IL-la production by adhered human monocytes in vitro 132 3.4.1. IL-la release from adhered monocytes following incubation with anti-FcyR mAb 132 3.4.2. The effect of an anti-TN Fa mAb on both the LPS and anti-FcyRIII mAb induced IL-la release from adhered human monocytes 136 3.5. The effects of Fey receptor ligation on reactive oxygen species production by adhered human monocytes in vitro 140 3.5.1. Reactive oxygen species production by adhered monocytes following incubation with anti-FcyR mAb 140 3.5.2. Reactive oxygen species production by adhered monocytes following incubation with PMA or LPS 144 3.5.3. Reactive oxygen species production by adhered monocytes following incubation with anti-FcyRIII mAb F(ab')2 fragments 149 3.6. The effects of human IgG immune complexes on TNFa production by adhered monocytes in vitro 153 3.6.1. TNFa release from adhered monocytes following incubation with immune complexes of human IgG anti-NIP mAb 153 3.6.2. The effects of IgG immune complex subclass on TNFa production by adhered monocytes 157 3.6.3. The effects of antigen epitope density on human IgG Immune complex induced TNFa production by adhered monocytes 160 3.6.4. The effects of blocking FcyR on small human IgG immune complex triggered TNFa production 160

9 3.7. Human IgG rheumatoid factor monoclonal antibodies 162 3.7.1. Determining the potential for the IgGl rheumatoid factor mAb to self-associate 162 3.7.2. Determ ining the potential for the IgG2 rheum atoid factor mAb to self-associate 167 3.7.3. Determining the effects of concentration on the elution profile of the IgGl rheumatoid factor mAb 169 3.7.4. Determining the effects of fractionation on the stability of the IgGl rheumatoid factor mAb 171 3.7.5. TNFa release from adhered monocytes following incubation with the human IgG rheumatoid factor mAb 172 3.8. Inhibitor studies 176 3.8.1. The effects of inhibitors on TNFa release from adhered Monocytes 176

Chapter 4. Discussion 181 4.1. Fey receptor ligation by murine anti-FcyR mAb 182 4.1.1 Production of cytokines and free radicals following incubation with anti-FcyR mAb 182 4.1.2. FcyRIIIa crosslinking by anti-FcyRIII mAb and T N Fa release 184 4.1.3. FcyRIIIa crosslinking by anti-FcyRIII mAb and reactive oxygen species production 187 4.1.4. Fey receptor crosslinking by anti-FcyR mAb and subsequent signalling 188 4.1.5. Effect of inhibitors on anti-FcyRIII mAb and LPS triggered TNFa release 189 4.2. Fey receptor ligation by human IgG im m une complexes 190 4.2.1. H um an IgG anti-NIP im m une complexes and TN Fa release from adhered monocytes 190 4.2.2. H um an IgG im m une complex-induced T N Fa release and the involvement of Fey receptors 192 4.2.3. Effect of inhibitors on human IgG immune complex and LPS induced TN Fa release 192 4.2.4. H um an IgG anti-NIP im m une complexes 193 4.3. Human IgG rheumatoid factors 195 4.3.1. IgG rheumatoid factor mAb self-association 195 4.3.2. H um an IgG rheum atoid factor-induced T N Fa release 196

10 4.4. TNFa and IL-la production by adhered human monocytes 196 4.5. Rheumatoid factor, Fc receptors and rheumatoid arthritis 198 4.6. B cell survival and IgG rheumatoid factor production in rheumatoid arthritis 199 4.7. A new treatment for rheumatoid arthritis 201 4.8. Future work 202

References 204

Appendices 262 Appendix 1. Relevant publications 263 Appendix 2. Calibration of the HPLC column 301 Appendix 3. Calibration of the FACScan 302 Appendix 4. TNFa ELISA 303 Appendix 5. Buffers and media 304

11 FIGURES

Introduction 1 Schematic showing the formation of IgM rheumatoid factor immune complexes 47

2 Schematic showing the proposed cyclic structure form ed during the self-association of two IgG rheumatoid factor molecules 48

3 Schematic diagram of the termination of IgG rheumatoid factor polym erisation by excess norm al IgG 50

4 Schematic diagram of surface expressed human Fey receptors and their associated subunits 60

5 Schematic diagram outlining the initiating effector mechanism for inflammation in rheumatoid arthritis 71

Results 6 Dotplot showing the FACS profile of whole blood 101

7 The binding of the three anti-FcyR mAb to leucocytes in whole blood 105

8 The effects of culturing adhered human monocytes on FcyR expression 111

9 The effects of culturing adhered human monocytes on FcyRIIIa expression 114

10 The effects of TN Fa on FcyRIIIa expression by adhered human monocytes 117

11a TNFa release from adhered monocytes following incubation with anti-FcyR mAb 125

12 lib Inhibition of LPS-induced TNFa production from adhered human monocytes by the presence of polymixin B 126

11c The effect of polymixin B on the anti-FcyRIII mAb -induced TNFa response 126

12 Barchart showing the effects of blocking FcyRI or FcyRII on the anti-FcyRIII mAb induced TNFa response 128

13 The effects of anti-FcyRIII F(ab')2 and Fab fragments on TNFa release 130

14 IL-la release from adhered monocytes following incubation with anti-FcyR mAb 135

15 The effects of a neutralising anti-TNFa mAb on TNFa detection and IL-la release 138

16 Reactive oxygen species production by adhered monocytes following incubation with anti-FcyR mAb 143

17 The effects of PMA and LPS on reactive oxygen species production 147

18 The effects of anti-FcyRIII F(ab ')2 fragments on reactive oxygen species production 152

19 The effect of IgGl anti-NIP immune complex preparation on TNFa release 155

20 The effect of small IgGl anti-NIP immune complex (prepared in a molar excess of antigen) concentration on TNFa release 156

21 The effects of immune complex subclass on TNFa release 159

22 HPLC analysis of the IgGl rheumatoid factor mAb 165

13 23 HPLC elution profiles for the IgGl and IgG2 rheumatoid factor mAb 168

24 The effect of concentration on the HPLC elution profile of the IgGl rheumatoid factor mAb 170 25 The effects of IgGl and IgG2 rheumatoid factor mAb on TNFa release 175

26 The effects of inhibitors on TNFa release 179

Discussion 27 A schematic diagram showing that an anti-FcyR mAb has the potential to ligate up to three Fey receptors 184

28 The signalling for TN Fa release through FcyRIIIa requires recruitm ent of either a) three FcyRIIIa receptors or b) two FcyRIIIa receptors where one interaction m ust involve binding through an Fc domain 187

29 Schematic diagram showing formation of a) hum an IgG anti-NIP dim er b) human IgG rheumatoid factor dimer 194

30 The stereochemistry of the IgG anti-NIP dimers could not be controlled 195

TABLES

Introduction 1 The distribution of FcyRIIIa expression by norm al tissue macrophages correlates with the pattern of systemic disease seen in patients with rheumatoid arthritis 66

14 Materials and methods 2 Primary monoclonal antibodies used in the indirect immunofluorescent staining of whole blood and/ or isolated monocytes 89

3 FACScan instrument settings 90

4 Monoclonal antibodies incubated with adhered human monocytes for measurement of cytokine production in vitro 93

5 Monoclonal antibodies incubated with adhered human for monocytes measurement of reactive oxygen species production in vitro 98

Results 6a Mean fluorescent intensities following the immunostaining of whole blood with each of the three anti-FcyR mAb 102

6b The binding of each anti-FcyR mAb at saturating binding concentrations to human leucocytes in whole blood 102

7 Mean fluorescent intensities following the immunostaining of whole blood with the whole anti-FcyRIII mAb, anti-FcyRIII F(ab)2 fragments or anti-FcyRIII Fab fragments 104

8 The effects of adherence to plastic on TNFa release by isolated human monocytes 119

9 LPS induced TNFa production from adhered monocytes isolated from three different donors 122

10 The effects of anti-FcyRIII mAb concentration on TNFa release from adhered monocytes 123

15 ABBREVIATIONS

ADCC Antibody dependent cellular cytotoxicity BCR B cell receptor BES N, N-bis [2-Hydroxyethyl]-2-aminoethanesulfonic acid BIC Bicarbonate buffer BSA Bovine serum albumin °C Degrees Celsius cDNA Complementary deoxyribonucleic acid CH Heavy chain constant domain CGM Complete growth medium CO2 Carbon dioxide cps Counts per second CR1 Complement receptor 1 CR2 Com plem ent receptor 2 DAF Decay accelerating factor EBV Epstein-Barr virus ELISA Enzyme linked immunosorbant assay F Phenylalanine Fab Monovalent antigen binding fragment of immunoglobulin F(ab')2 Divalent antigen binding fragment of immunoglobulin FACS Fluorescence Activated Cell Sorter Fc Fragment crystallisable FcaR Fc receptor for IgA FceR Fc receptor for IgE FcyR Fc receptor for IgG/Fc gamma receptor FCS Fetal calf serum FITC Fluorescein isothiocynate FPLC Fast protein liquid chromatography Y Gamma G-CSF Granulocyte colony-stimulating factor GM-CSF Granulocyte macrophage colony-stimulating factor GPI Glycophosphotidylinositol H Histidine HBSS Hank's balanced salt solution without phenol red HEPES N-2-Hydroxyethylpiperazine-N'-2-ethane sulphonic add HLA Human leucocyte antigen

16 H2C>2 Hydrogen peroxide HOC1 Hypochlorous acid HPLC High performance liquid chromatography HRP Horse radish peroxidase I Iodine IgA Immunoglobulin A IgE Immunoglobulin E IgG Immunoglobulin G IgM Immunoglobulin M IL-1 Interleukin-1 IL-4 Interleukin-4 IL-6 Interleukin-6 IL-8 Interleukin-8 IL-10 Interleukin-10 IL-1R Interleukin-1 receptor IL-IRa Interleukin-1 receptor antagonist IFNy Gamma interferon IT AM Immunoreceptor tyrosine-based activation motif ITIM Immunoreceptor tyrosine-based inhibitory motif

K Kappa X Lambda L Leucine LGL Large granular lymphocyte LPS Lipopolysaccharide mAb Monoclonal antibody MCP-1 Monocyte chemoattractant protein-1 M-CSF Macrophage colony-stimulating factor MFI Mean fluorescent intensity MgCl Magnesium chloride MHC Major histocompatibility complex MM Modified medium MMP Matrix metalloproteinase MPO Myeloperoxidase m RNA Messenger ribonucleic acid MW Molecular weight NA Neutrophil antigen NaCl Sodium chloride

17 NaOH Sodium hydroxide NIP 5-iodo-4-hydrox-3-nitrophenacetyl NK Natural killer cell O r Superoxide OH Hydroxyl radical PBS Phosphate buffered saline PBT Phosphate buffered saline with 0.1% tween p g e 2 Prostaglandin E2 PH Log of free proton concentration PMA Phorbol-12 myristate 13-acetate PMN Polymorphonuclear cell PTK Protein tyrosine kinase R Arginine RF Rheumatoid factor RPMI Roswell Park Memorial Institute essential media rsc Receptor sites per cell S.D. Standard deviation SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis SLE Systemic lupus erythematosus SPA Staphylococcal protein A SPG Streptococcal protein G TACE TNFa converting enzyme TCR T cell receptor TGFp Transforming growth factor beta TMB S^'AS'-tetramethylbenzidine (Dihydrochloride) TNF Tumour necrosis factor TNF-R Tumour necrosis factor receptor Tris Tris [hydroxymethyl] aminoethane V Valine VC AM-1 Vascular cell adhesion molecule-1 VH Variable light chain VL Variable heavy chain E Zeta

18 CHAPTER 1

INTRODUCTION

19 1.1. History of rheumatoid arthritis Indian medical literature from 2000 years ago (Sturrock et a l, 1977) and writings by Sydenham from 1676 may both have been describing patients with rheumatoid arthritis. However, while studies in human palaeopathology have shown the existence of both osteoarthritis and ankylosing spondylitis in the ancient and medieval worlds, convincing evidence for the existence of rheumatoid arthritis as a disease entity prior to the nineteenth century is lacking (Short, 1974; Thould & Thould, 1983). The first detailed account of a patient with all the clinical features of rheumatoid arthritis was not reported until the early 1800’s by the French medical student, Augustin-Jacob Landre-Beauvais (Snorrason, 1952). This student wrote a case report on a patient presenting with what was thought to be a previously undefined form of gout (Short, 1974). The term rheumatoid arthritis was first used by the London physician Sir Alfred Baring Garrod in 1859 (Garrod, 1859).

Notwithstanding the uncertainty over its antiquity, rheumatoid arthritis has been known about for at least 200 years and yet its precise aetiology and mechanism of pathogenesis remain undetermined.

1.2. Clinical and pathological features of rheumatoid arthritis Rheumatoid arthritis is a debilitating inflammatory joint disease which primarily targets the synovial membrane. Approximately 1-3% of the world's population is affected by rheumatoid arthritis (Harris, 1990), with a femalermale ratio of 3:1 (da Silva & Hall, 1992). The onset of symptomatic disease often appears between 35 and 55 but can affect individuals of any age. Rheumatoid arthritis often begins as a swelling within the joints of the hands (metacarpophalangeal and proximal interphalangeal) wrists or feet (metatassophalangeal), initially in a symmetrical pattern. It may then progress to any other synovial joint, the knee being a common target. Patients present with a raised erythrocyte sedimentation rate, indicative of a systemic process and obvious swellings of the soft tissue over joints, due to inflammation of the synovium. Around 80% of patients with rheumatoid arthritis are seropositive for rheumatoid factor (see 1.6. Rheumatoid factors and rheumatoid arthritis).

20 The normal synovium consists of an intimal lining layer of fibroblast­ like synoviocytes and bone marrow derived macrophages (Edwards et a l, 1982). Macrophages are also present within the subintima and deeper into the synovial tissue. Lymphocytes can be seen within the normal synovium, but at low levels. In contrast, the inflamed rheumatoid synovium is characterised by the infiltration of macrophages, lymphocytes, plasma cells and some polymorphonuclear (PMN) cells. Most notably, intimal macrophages increase in both number and size resulting in "lining hyperplasia" (Henderson et al., 1988). Many lymphocytes appear scattered throughout the tissue, however lymphoid aggregates often occur. In approximately 10% of patients, secondary follicles of B cells with germinal centres form (Edwards & Wilkinson, 1995).

Cartilage destruction and pannus formation follows. At this stage, joints are obviously swollen and muscle wasting may become apparent, particularly in the hands. Eventually, inflammation becomes less prominent and the joint erosion ultimately results in permanent distortion.

1.2.1. Extra-articular features of rheumatoid arthritis In addition to synovitis, about 20% of seropositive patients display extra- articular features such as rheumatoid nodules, lung and cardiac involvement (Maini & Zvaifler, 1998). Seronegative patients with rheumatoid arthritis rarely have these extra-articular features.

Rheumatoid nodules occurring at points of pressure, such as over the bony prominences around the elbow, are generally associated with more severe and erosive arthritis. They can also be found in the connective tissue of the lung or heart (Ziff, 1990). The rheumatoid nodule consists of a central necrotic core of fibrin which is surrounded by a palisade of mostly macrophages and fibroblasts (Hedfors et a l, 1983). The connective tissue layer surrounding this palisade consists of infiltrating monocytes, some T cells and immunoglobulin-secreting plasma cells (Ziff, 1990).

The most common form of lung involvement is pleurisy, however, pulmonary fibrosis and alveolitis are also found in patients with

21 rheumatoid arthritis (Anaya et al., 1995). Other systemic features include pericarditis and a rise in liver enzymes (Maini & Zvaifler, 1998).

The occurrence of systemic disease is often neglected when attempting to describe the pathogenesis of rheumatoid arthritis. In order to fully elucidate a disease mechanism, both intra- and extra-articular features need to be explained.

1.3. Aetiological features of rheumatoid arthritis The prevalence of rheumatoid arthritis has familial links. Compared with more distant relatives, first degree relatives are more likely to develop rheumatoid arthritis. Additionally, the frequency of this disease in monozygotic twins is greater than in dizygotic twins (Reveille, 1998). However, there is only a 15% concordance rate in monozygotic twins (Jarvinen & Aho, 1994), suggesting other contributing factors. Hence, rheumatoid arthritis is a multifactorial disease.

The genetics of rheumatoid arthritis is partly determined by gender, with more women that men being affected. Furthermore, hormonal changes appear to effect disease onset and severity (da Silva & Hall, 1992).

Infectious agents are an attractive explanation for the initiation of rheumatoid arthritis. A large number of patients with rheumatoid arthritis have higher antibody titres to Epstein-Barr virus (EBV) antigens compared with controls (Alspaugh et al., 1981). However, some patients with rheumatoid arthritis lack anti-EBV antibodies (Venables et al., 1988). Other viral infections such as retroviruses and parvoviruses (Krause et a l, 1996) have been associated with rheumatoid arthritis, as have a number of bacterial infections such as Mycobacterium tuberculosis and Proteus mirabilis (Holoshitz et al., 1986, Senior et al., 1995). Nonetheless, there is no evidence to suggest a causative role for any bacterial or viral agents in rheumatoid arthritis.

Rheumatoid arthritis is genetically associated with the class II major histocompatibility complex (MHC). Major histocompatibility complex molecules are encoded by human leucocyte antigen (HLA) genes which are located on chromosome 6 and are highly polymorphic. The genes for

22 Class IIMHC are found in region D which can be further subdivided into regions R, Q and P. These genes are therefore referred to as HLA-DR, -DQ or -DP (van Jaarsveld et al., 1998a). The HLA-DR p chain gene, DRB1, encodes for HLA-DR alleles 1-4, now termed 01 - 04 (Gregersen et al., 1987).

Rheumatoid arthritis is linked with specific HLA-DR allotypes. Approximately 60-70% of Caucasian patients with rheumatoid arthritis are HLA-DR4 positive and these patients are more likely to have severe erosive disease with extra-articular features (Winchester, 1994; van Jaarsveld et al., 1998b; Meyer et al., 1999). Stastny (1978) was the first to report this link and found the DR4 subtype, Dw4 (HLA-DRB1*0401), to be associated with susceptibility to rheumatoid arthritis. Since this finding, other DR4 alleles have also been associated with rheumatoid arthritis, such as Dwl4 (HLA-DRBl*0404/0408) and Dwl5 (HLA-DRB1*0405) (Nepom et al., 1989; Yelamos et al., 1993; Meyer et al., 1999). However, a number of Caucasian patients and individuals from other racial backgrounds with rheumatoid arthritis are DR4 negative. In these cases different HLA-DR types have been connected with disease susceptibility and severity. HLA-DR1 (HLA-DRB1*0102) has been found in DR4 negative Caucasians as well as in Israeli Jews with rheumatoid arthritis (de Vries et al., 1993), while possession of HLA-DR10 has a high frequency in Spanish patients (Yelamos et al., 1993).

In 1987 Gregersen et al. proposed a hypothesis to explain the link between HLA-DR and rheumatoid arthritis. It was found that certain HLA-DR alleles shared amino acid sequences around position 70 in the third hypervariable region, encoding for the first MHC molecule domain. It was suggested that this "shared epitope" promoted disease susceptibility by influencing the interaction between MHC class II and T cell receptor (TCR) at the level of antigen presentation. Additionally the "shared epitope" may influence TCR selection in the thymus resulting in a genetically predetermined immune repertoire (Gregersen et al., 1987; Kohsaka et al., 1998). However, a link between the genetics of MHC molecules and either the susceptibility to, or the severity of rheumatoid arthritis may simply imply that an adaptive immune response is involved since antigen presentation is common to all adaptive

23 responses. This leaves open the question of how the genetics of MHC molecules can contribute to rheumatoid arthritis in conjunction with other genetic and non-genetic factors.

1.4. Mediators of inflammation and tissue damage in rheumatoid arthritis

1.4.1. Cytokines and rheumatoid arthritis Cytokines are soluble proteins which mediate signals between cells, either in an autocrine or paracrine manner. Cytokines are an important component of host defence when produced locally. However, excess, sustained or systemic cytokine production can be harmful to the host and may even contribute to disease. Two cytokines, tumour necrosis factor (TNF) and interleukin-1 (IL-1) have been particularly implicated in the pathogenesis of rheumatoid arthritis.

1.4.2. Tumour necrosis factor The proinflammatory cytokine, TNF, acquired its name from the observation that its production in response to bacterial endotoxin caused the necrosis of tumours in vivo (Carswell et al., 1975; Balkwill, 1989). TNF exists as two isoforms, TNFa and lymphotoxin (formally TNFp) (Balkwill, 1989). Lymphotoxin (TNF(3) is now known as lymphotoxin-a following the identification of an additional member of the TNF family, lymphotoxin-p (Bronwing et al., 1993). TNFa and lymphotoxin-a exert similar effects on target cells. Lymphotoxin is only produced by T lymphocytes while TNFa can also be produced by macrophages and mast cells (Balkwill, 1989). The genes for TNFa and lymphotoxin-a have been mapped to the MHC region on chromosome 6 and their protein products exhibit 30% homology (Pennica et al., 1984; N edw in et ah, 1985; Spies et al., 1986). Lymphotoxin-p has also been mapped to the same region on chromosome 6 and exhibits 21% homology with TNFa and 24% homology with lymphotoxin-a (Browing et ah, 1993).

TNFa is produced intracellularly as a 26-kDa polypeptide. TNFa is also found in the plasma membrane in this form and at this site has cytotoxic activity (Kriegler et ah, 1988). Surface bound TNFa is enzymatically cleaved to release a trimer (of 17-kDa subunits) of biologically active,

24 extracellular TNFa (Wingfield et ah, 1987). Membrane-associated pro- TNFa is processed by the enzyme, TNFa converting enzyme (TACE), which is membrane-bound and a member of the a disintegrin and metalloprotease (ADAM) family (Black et ah, 1997; Moss et al., 1997) More recently, work has revealed that TACE also plays a role in the shedding of other surface molecules including TNF receptors (Peschon et al., 1998).

1.4.3. Tum our necrosis factor receptors Two receptors for TNFa and lymphotoxin-a have been identified and their genes cloned (Brockhaus et a l, 1990; Gray et al., 1990; Loetscher et al., 1990; Schall et al., 1990). The type I TNF receptor (TNF-RI) has a molecular mass of 55-kDa while the type II TNF receptor (TNF-RII) has a molecular mass of 75-kDa (Brockhaus et ah, 1990). Both receptors have similar extracellular domains containing four cysteine-rich domains and are homologous to the cell surface molecules, nerve growth factor receptor, B cell activating CD40, the human T cell activation marker CD27 and the rat T cell activation marker 0X40 which all belong to the same receptor family (Dembic et al., 1990; de Jong et al., 1991). Both receptors can bind TNFa and lymphotoxin-a (Gray et al., 1990; Tartaglia & Goeddel, 1992). Furthermore, the extracellular domains of both receptors can exist as functional soluble receptors and therefore natural inhibitors of TNF (Nophar et al., 1990). The cytoplasmic domains of TNF- RI and TNF-RII are unrelated and, as a result, trigger distinct pathways (Dembic et ah, 1990). TNF-RI mediates most TNF related effects including necrosis of tumours and cytotoxicity, while the functions of TNF-RII are more limited. Although both TNF-R can activate nuclear factor kB , the cytoplasmic domain of TNF-RI contains TNF-RI-associated death domains which allow this receptor to also trigger apoptosis (Yuan, 1997). Lymphotoxin-p cannot bind either TNF-RI or TNF-RII but instead when complexed with lymphotoxin-a to form a heterotrimer, binds the specific lymphotoxin-p receptor which is expressed by activated lymphocytes (Force et ah, 1995).

TNFa, as many cytokines, is pleiotrophic and can therefore affect many different cell types. This is made possible by the wide expression of TNF receptors. Monocytes, macrophages and natural killer cells express both TNF-RI and TNF-RII (Dembic et ah, 1990; N aum e et ah, 1991). Resting T

25 and B lymphocytes express little, if any TNF receptors. However, following stimulation both CD4 and CD8 positive T cells and B cells express predominantly the type II TNF receptor (Erikstein et a l, 1991; W are et a l , 1991), while fibroblasts and epithelial cells express mainly the type I TNF receptor (Brennan et a l, 1992a).

1.4.4. Interleukin-1 There are three members of the IL-1 gene family, located on the long arm of chromosome 2 at position 2ql3 - 2q21 (Webb et a l, 1986). These genes encode for IL-la, IL-p and IL-1 receptor antagonist (IL-IRa). All three genes have been cloned (Auron et al., 1984; Lomedico et a l, 1984; Eisenberg et a l, 1990). IL-1 is produced by a wide range of cells, most notably monocytes, macrophages and B lymphocytes, and acts on many cell types, mediating multiple biological effects including fever, anorexia, inflammation, leucocyte activation and the induction of gene expression (reviewed in Dinarello, 1991). Both IL-la and IL-lp proteins are initially synthesised as 31-kDa precursors (pro-IL-1) and proteolytic processing results in 17-kDa peptides (Hazuda et al., 1988). IL-la and IL-ip 17-kDa peptides and the phosphorylated precursor form of IL-la are all biologically active (Dinarello, 1991). IL-la tends to remain on the cell surface and exert its effects locally. IL-lp is only fully active following intracellular cleavage into the 17-kDa peptide, following which it is readily released (Lonnemann et al., 1989; Dinarello, 1996), however, pro- IL-ip may have some biological activity (Jobbling et a l, 1988).

1.4.5. Interleukin-1 receptors Two IL-1 receptors have been identified. IL-1 receptor type I (IL-1RI) is an 80-kDa protein and the type II receptor (IL-1RH) is a 68-kDa protein (Dinarello, 1991) Both receptors belong to the immunoglobulin-gene family and their genes have been mapped to chromosome 2 (Sims et al., 1995). Both receptors are transmembrane glycoproteins with an extracellular region that contains three immunoglobulin-like domains and both can exist in soluble form. However, only IL-1RI can transduce signals and this is reflected in the short cytoplasmic tail of IL-1RD. Although both receptor types bind IL-la, pro-IL-la and IL-ip, IL-la appears to bind to IL-1RI better, while IL-1RII has a higher affinity for IL- lp. Since IL-IRQ cannot transduce a signal, it acts as an IL-1 antagonist

26 (Colotta et al, 1994). Another natural antagonist for IL-1 is the IL-lRa. This protein has a molecular weight of 22-kDa and competes with IL-1 for receptor occupancy (Arend et a l, 1990). IL-1RI is expressed by monocytes, T cells, fibroblasts, keratinocytes, endothelial cells, hepatocytes and chondrocytes while IL-1RII expression is more limited being expressed by B cells, neutrophils and bone marrow cells (Dinarello, 1991).

1.4.6. Tumour necrosis factor, interleukin-1 and rheumatoid arthritis IL-la is present at high levels in rheumatoid synovial fluids (Fontana et al. , 1982) and was the first pro-inflammatory cytokine to be implicated in rheumatoid synovitis and joint destruction (Saklatvala, 1986). TNFa is, however, now thought to be the dominant mediator of inflammation and tissue injury in rheumatoid synovitis (reviewed in Brennan et a l, 1992a). TN Fa has been detected in the synovial fluid (Hopkins & Meager, 1988), synovial tissue (Chu et a l, 1991a) and serum from patients w ith rheumatoid arthritis (Saxne et a l, 1988). Immunohistochemical studies have shown the majority of TNFa-secreting cells in rheumatoid synovia to be macrophages within the synovial lining layer and deeper into the synovium (Wilkinson & Edwards, 1991).

TNFa and IL-1 may be acting alone, additively or synergistically. In vitro studies have provided evidence supporting a role for TNFa and IL-1 in mediating tissue injury and inflammation in rheumatoid arthritis. Both TNFa and IL-1 have been shown to stimulate human synovial cells, dermal fibroblasts (Mizel et a l, 1981; Dayer et a l, 1985) and synovial fibroblasts (Saklatvala et a l, 1985) to produce prostaglandin E 2 (PGE2 ) and collagenase. Human chondrocytes have also been shown to produce PGE2 following stimulation with IL-1 and TNFa (Campbell et a l, 1990). Proteoglycan is an essential component of cartilage and proteoglycan loss is seen in patients with rheumatoid arthritis. TNFa has been shown to not only induce osteoclastic bone resorption but also inhibit proteoglycan and collagen synthesis and induce proteoglycan degradation by chondrocytes (Bertolini et a l, 1986; Saklatvala, 1986; Thompson et a l, 1987). IL-1 also causes chondrocytes to degrade and inhibit the synthesis of new proteoglycan (Saklatvala et a l, 1984; Tyler, 1985; Benton & Tyler, 1988) and to bring about cartilage destruction and bone resorption (Gowan et a l, 1983; Saklatvala et a l, 1985).

27 TNFa and IL-1 have both been shown to act as chemoattractants for monocytes and polymorphonuclear cells (Luger et a l, 1983; Gamble et a l, 1985; Ming et al., 1987; Richter et al., 1989; Richter et al., 1995) and to promote leucocyte adhesion to endothelial cells (Bevilacqua et al., 1985). IL-1 and TNFa can also induce the release of chemotactic factors, such as monocyte chemotactic activating factor from fibroblasts (Larsen et al., 1989), and neutrophil chemotactic factor from endothelial cells (Strieter et a l, 1989).

Studies by Buchan et al. (1988) showed that IL-la, IL-ip, TNFa and TNFp messenger RNA (mRNA) could all be detected in stimulated joint cell cultures isolated from patients with rheumatoid arthritis. Further studies demonstrated that mononuclear cells from rheumatoid synovium or synovial fluids would spontaneously secrete TNFa and IL-1 in the absence of exogenous stimuli. Furthermore, the utilisation of an anti- TNFa monoclonal antibody (mAb) showed this IL-1 production to be dependant upon the presence of TNFa (Brennan et al., 1989). A separate study showed that spontaneous IL-1 production from peripheral blood monocytes in vitro was greater from patients with rheumatoid arthritis than controls or patients with rheumatoid arthritis receiving gold treatment (Danis et a l, 1987).

The potential for TNFa and IL-1 to contribute to the pathogenesis of rheumatoid arthritis has also been demonstrated in vivo in a number of experimental models. For example, transgenic mice expressing a 3' modified human TNFa gene developed arthritis (Keffer et a l, 1991). By backcrossing this transgenic model with DBA/1 mice, which are susceptible to dev eloping arthritis, disease onset was accelerated and more aggressive (Butler et al., 1997). Administration of a neutralising anti-TNFa/p monoclonal antibody (mAb) to mice with collagen type II- induced arthritis, either prior to the onset of disease or following disease manifestation, significantly reduced disease severity (Williams RO et a l, 1992). However, the most convincing evidence for a role for TNFa in rheumatoid arthritis has been from human clinical trials. Following treatment with a chimeric anti-human TNFa monoclonal antibody, patients with rheumatoid arthritis showed a marked improvement (Elliott et a l, 1993; Charles et a l, 1999).

28 Following injection of IL-1 into the knee joints of rabbits, both proteoglycan loss and inflammatory cell infiltration were observed (Pettipher et al., 1986). The administration of both anti-IL-la and anti-IL- lp antibodies to mice before the onset of collagen-induced arthritis completely prevented disease, while treatment after disease manifestation significantly reduced inflammation and cartilage destruction. The anti-IL-lp antibody given alone also significantly reduced disease (van den Berg et al., 1994). Local administration of murine IL-1RI into rats with antigen-induced arthritis reduced joint swelling and tissue injury (Dauer et al., 1994). Therapeutic blockade of IL- 1 with recombinant human IL-IRa in humans is effective, but less striking than for TNFa (Bresnihan et al., 1998).

The evidence discussed so far strongly supports a role for both TNFa and IL-1 in the pathogenesis of rheumatoid arthritis. The suggestion is that both cytokines, but particularly TNFa, act locally either alone or synergistically to induce an inflammatory response, the degradation of surrounding tissues and the recruitment of leucocytes to the site of inflammation.

A host of pro-inflammatory mediators and other cytokines have also been associated with rheumatoid arthritis. Although there is little evidence to support a primary role in disease pathogenesis, these additional agents may arise as a result of the inflammatory response initiated by TNFa and IL-1, and may contribute to inflammation, tissue injury and the generation of a localised immune network in the rheumatoid synovium. Some of the more important mediators of inflammation and tissue injury associated with rheumatoid arthritis will be discussed.

1.4.7. Interleukin-6 and other cytokines Interleukin -6 (IL-6 ) is a B cell activator but also induces the production of acute phase proteins, an important systemic feature of rheumatoid arthritis. Elevated levels of IL- 6 , have been detected in the synovial tissue, synovial fluid and serum from patients with rheumatoid arthritis (Houssiau et al., 1988; Waage et a l, 1989; Sack et a l, 1994). Synovial macrophages appear to be the main source of IL - 6 in the rheumatoid

29 synovium. Immunohistochemical studies have shown IL- 6 -containing macrophages to be located close to immunoglobulin-producing plasma cells. Since IL -6 stimulates antibody production this may be significant in the generation of rheumatoid factors within the synovium (Field et ah, 1991), as will be discussed later. Interestingly, the treatment of rheumatoid arthritis patients with an anti-TNFa monoclonal antibody significantly reduced IL -6 levels (Elliott et a l, 1993). This has suggested that although IL -6 might have a role to play in the perpetuation of inflammation in rheumatoid arthritis, its contribution may be secondary to the production of TNFa.

Other cytokines such as interleukin-11 (Mino et ah, 1998) and interleukin-15 (Mclnnes & Liew, 1998) have been associated with rheumatoid arthritis, however, there is little evidence to support a primary role for these cytokines in disease pathogenesis. Instead, their presence is likely to be a consequence of the inflammatory process within the rheumatoid synovium.

1.4.8. Chemokines and growth factors Chemoattractants are instrumental in the recruitment of leucocytes to sites of inflammation. The chemokines, monocyte chemoattractant protein-1 (MCP-1) and interleukin - 8 (IL-8 ) are potent chemotactic factors for both neutrophils and monocytes (Gerszten et ah, 1999). Elevated levels of IL -8 have been detected in the synovial fluids and tissues from patients with rheumatoid arthritis (Brennan et ah, 1990a; Deleuran et ah, 1994), while chondrocytes stimulated with IL-1 have been shown to release MCP-1 in vitro (Villiger et ah, 1992).

A number of growth factors have also been detected at raised levels in patients with rheumatoid arthritis. Granulocyte macrophage colony- stimulating factor (GM-CSF) has been found at elevated levels in rheumatoid synovial fluids (Xu et ah, 1989). Moreover, isolated synovial macrophages from patients with rheumatoid arthritis spontaneously secreted GM-CSF in culture and this appeared to be TNFa dependent (Harworth et ah, 1991). Additionally, articular chondrocytes have been shown to produce both GM-CSF (Alsalamen et ah, 1994) and macrophage colony-stimulating factor (M-CSF) (Campbell et ah, 1993) following

30 stimulation with IL-1 and TNFa. Transforming growth factor beta (TGFp), particularly TGFfll, as well as abundant latent TGFpi binding protein have also been detected in the synovial tissue and fluid from patients with rheumatoid arthritis (Miossec et ah, 1990; Brennan et ah, 1990b; Chu et ah, 1991b; Taketazu et ah, 1994).

1.4.9. Reactive oxygen species Possibly the most potent means of phagocytic defence against microorganisms is the generation of free radicals. Following cell stimulation, a rapid respiratory burst occurs and it is the conversion of its products into oxidising radicals that mediates cell damage (Babior, 1984a). The metabolic pathway begins with the conversion of oxygen into superoxide (C> 2‘), catalysed by the membrane bound enzyme, NADPH- oxidase (Babior, 1984b). Dismutation of C> 2_ results in the formation of hydrogen peroxide (H 2O2). From F^C^, the powerful hydroxyl radical (‘OH) is generated (Edwards, 1994). Additionally, H 2O2 is the substrate for the enzyme myeloperoxidase (MPO) (Babior, 1984a). Myeloperoxidase is found in the azurophilic granules of neutrophils and the lysosomes of monocytes and some macrophages (Edwards, 1994). Mature macrophages lack MPO activity (Fleit et ah, 1982). The oxidation of H 2Q2 in the presence of chloride ions generates the highly toxic hypochlorous acid (HOC1).

While low levels of reactive oxygen species are thought to be involved in signalling by acting as second messengers, when produced in excess these molecules can cause oxidative stress which may result in cell and tissue injury (Finkel, 1998).

Reactive oxygen species, although short-lived, have been detected in the synovium of patients with rheumatoid arthritis by electron spin resonance spectroscopy (Mapp et ah, 1995). However, studies on the effects of free radicals in vitro have provided probably the best evidence that oxidative stress may mediate tissue injury, most notably cartilage damage, in rheumatoid arthritis. Hypochlorous acid and ‘OH have been shown to facilitate the fragmentation of collagen by the degradation of proteoglycans (Halliwell, 1978; Kauanko et ah, 1989; Dayer et ah, 1993), while both HOCL and H 2 O2 can inhibit proteoglycan synthesis (Kauanko

31 et ah, 1989). Furthermore, TNFa has been shown to activate the enzyme NADPH oxidase (Dusi et ah, 1996) and to cause oxidative stress in cell lines (Zimmerman et ah, 1989). This has again suggested that TNFa is a major mediator of pathology in rheumatoid arthritis, by acting both directly and indirectly on surrounding tissue.

1.4.10. M atrix metalloproteinases Matrix metalloproteinases (MMP) are proteolytic enzymes that specifically act upon matrix components. MMPs are important in the remodelling of tissues and in facilitating cell migration. There are three major classes of MMP; stromelysins, collagenases and gelatinases (Woessner, 1991). These enzymes are initially produced in latent form and are activated following enzymatic cleavage (Kleiner & Stetler- Stevenson, 1993). Tissue inhibitors of metalloproteinases are naturally occurring MMP inhibitors (Nagase & Woessner, 1999). The stromelysins, for example MMP-3, act upon a wide range of matrices including proteoglycans, fibronectin, laminin and collagen type IV. Collagenases, like neutrophil collagenase (MMP- 8 ) and interstitial collagenase (MMP- 1), are specific for collagen types I, II and m (Woessner, 1991). The gelatinases degrade collagen type IV and denatured collagen and are produced by a wide range of cell types. Gelatinase A (MMP-2) is probably the most abundant and is produced by chondrocytes, fibroblasts, osteoblasts and monocytes, while Gelatinase B (MMP-9) is expressed by monocytes and alveolar macrophages and is found in the specific granules of neutrophils (Birkedal-Hansen et ah, 1993).

Matrix metalloproteinases have been implicated in the cartilage damage seen in rheumatoid arthritis. Increased levels of collagenases and stromelysins have been detected in the synovial fluid and synovial tissues from patients with rheumatoid arthritis (Clark et ah, 1993; Beekm an et ah, 1997). Moreover, MMP production appears to be inducible by both IL-1 and TNFa (Hanemaaijer et ah, 1997; Vincenti et ah, 1998). This has been supported by the observation that the treatment of rheumatoid arthritis patients with an anti-TNFa monoclonal antibody resulted in the reduction of serum stromelysin levels (Brennan et ah, 1997).

32 1.4.11. C om plem ent Complement is an important mediator of inflammation and there is evidence for localised complement activation in patients with rheumatoid arthritis. Complement consumption (Pekin & Zvaifler, 1964; Hedberg, 1967; Ruddy et al., 1969), as well the presence of the membrane attack complex (Morgan et al., 1988) have been detected in the rheumatoid synovium. However, as will be discussed later, it is likely that the role complement may be secondary to the initial disease mechanism in rheumatoid arthritis.

1.5. The immunopathogenesis of rheumatoid arthritis As already discussed, TNFa and IL-1 both appear to mediate local inflammation and tissue injury in the rheumatoid synovium, as well as triggering the production of other degradative compounds and pro- inflammatory mediators in rheumatoid arthritis. Histological studies indicate that the synovial macrophage is the major source of TNFa in the rheumatoid synovium. The initial trigger for macrophage activation and subsequent TNFa production in rheumatoid arthritis is unknown and has been the focus of research for many years.

In 1981 the hypothesis that T cells played a principle role in the initiation of inflammation in the rheumatoid synovium was proposed (Janossy et al., 1981). It was postulated that self antigens were presented to CD4 positive T lymphocytes by antigen presenting cells, such as dendritic cells (Thomas & Lipsky, 1996). It was thought that this interaction was the initiating process in rheumatoid arthritis. Subsequent macrophage activation would then generate proinflammatory mediators such as TNFa and IL-1 which would in turn act on synovial chondrocytes and fibroblasts to result in the cartilage and bone destruction associated with rheumatoid arthritis.

The activation of autoreactive T cells has been attributed to either molecular cross-reactivity with foreign antigen (Albani & Carson, 1996), or the availability of cryptic epitopes previously hidden or at low concentrations (Warnock & Goodacre, 1997). These views have been supported by experimental models such as adjuvant (Pearson, 1956), streptococcal cell wall (Cromartie et al., 1977) and collagen-induced

33 arthritis (Trentham et ah, 1977). However, despite the search for T cell responses to microbial or articular antigens, there still remains no solid evidence of an antigen specific-T cell response in rheumatoid arthritis (Fox, 1997; Mclnnes & Liew, 1998). Moreover, the reason for and timing of failure of T cell tolerance has not been adequately addressed.

The T cell hypothesis of Janossy et ah (1981) was also based upon the observation that the majority of lymphocytes within the rheumatoid synovium were found to be CD4 positive T cells (Duke et al., 1982). This was supported by the protection of animals against collagen type II and streptococcal cell wall-induced arthritis by pre-treatment with anti-CD4 monoclonal antibodies (Goldschmidt et al., 1992; van den Broek et al., 1992). Although, most animal models for rheumatoid arthritis are T cell- dependent, many show features more representative of the seronegative arthropathies than seropositive rheumatoid arthritis (Oliver & Brahn, 1996). Rheumatoid factor positivity in experimental models is rare. Although the M RL/lpr mouse has high serum titres of both im m unoglobulin M (IgM) and im m unoglobulin G (IgG) rheum atoid factor, this model is more characteristic of SLE rather than of rheumatoid arthritis (O'sullivan et a l, 1995). Furthermore, the extra-articular features associated with rheumatoid arthritis are absent in experimental models, for example streptococcal cell wall-induced arthritis lacks rheumatoid nodules (Oliver & Brahn, 1996).

Although T cell depleting treatments, such as anti-CD4 therapy are effective at T lymphocyte ablation, human trials have not demonstrated a significant clinical improvement (Moreland et a l, 1995; Tak et a l, 1995; van der Lubbe et ah, 1995; Bresnihan et ah, 1998), unlike therapies directed towards macrophage-derived cytokines (Elliott et ah, 1993). This implies that macrophage activation and subsequent cytokine production in rheumatoid arthritis is not a T cell driven mechanism. This is further supported by histological studies showing synovial intimal macrophage activation to be the initial event in rheumatoid arthritis, proceeding T cell accumulation (Kraan et ah, 1998). Furthermore, macrophage activation in rheumatoid arthritis appears to be spatially distinct from infiltrating T cells (Kurosaka & Ziff, 1983). This therefore raises the question of, in the absence of direct T cell involvement, what is

34 triggering macrophage activation and cytokine production in rheumatoid arthritis.

The initial trigger for macrophage activation in the rheumatoid synovium still remains unsolved. However, its identity may have already been described and all that is required in an effector mechanism. A key feature of rheumatoid arthritis is its association with an autoantibody known as the rheumatoid factor (RF). Others have suggested that rheumatoid factor-based immune complexes could evoke inflammation by interacting with monocytes (Nardella et al., 1983; Mannik & Nardella, 1995).

1.6. Rheumatoid factors and rheumatoid arthritis 60 years ago, Dr E. Waaler in Bregen, Norway, described a component in the serum from patients with rheumatoid arthritis which caused the agglutination of sheep erythrocytes (Waaler, 1939; Waaler, 1940). In fact, the experiment in which Waaler titrated sera from patients with rheumatoid arthritis against sheep erythrocytes sensitised with rabbit IgG, dates from 11th December 1937 (Milgrom, 1988). Waaler's finding was rediscovered in 1948 by Rose et al. (1948) at the Columbia-Presbyterian Medical Centre in New York and the active component was later named the rheumatoid factor (Pike et al., 1949).

In 1957 it was found that rheumatoid factors were serum gamma­ globulins with a sedimentation rate of 19S (Franklin et al . , 1957). Physical studies by analytical ultracentrifugation demonstrated that this 19S component, later found to be an IgM (Kunkel et al., 1959), complexed with gamma-globulins that sedimented at a rate of 7S, to form 22S aggregates (Franklin et al., 1957). These aggregates of an IgM rheumatoid factor complexed with autologous IgG were shown to associate and dissociate under different conditions of pH and IgG concentration (Kunkel et a l, 1959).

Rheumatoid factors are classically defined as polyclonal antibodies of the class IgM that recognise and bind the Fc (fragment crystallisable) region of IgG (Moore & Domer, 1993). However, rheumatoid factors of class IgG, immunoglobulin A (IgA) and immunoglobulin E (IgE) have also been

35 described (Mannik, 1992). Little is known about IgE rheumatoid factors and most research has been directed at rheumatoid factors of the classes IgM and IgG. Some autoantibodies with rheumatoid factor activity also have the capacity to crossreact with other antigens such as nuclear components, viral proteins (Williams, 1992a; Moore & Dorner, 1993) or P2-microglobulin (Williams et al ., 1992b). Autoantibodies to the Fab rather than the Fc portion of IgG may also occur and can contribute to rheumatoid factor activity as measured by standard assays (Milgrom, 1988). These latter autoantibodies are generally not considered to be true rheumatoid factors, although this has been a matter of controversy. Rheumatoid factors have become the basis of standard diagnostic tests for rheumatoid arthritis. However, their role in the pathogenesis of rheumatoid arthritis has remained a matter of debate.

1.6.1. Prevalence of rheumatoid factors and methods of detection Elevated levels of IgM rheumatoid factor, detected by conventional agglutination techniques, are found in the serum of 70-80% of patients with rheumatoid arthritis (Harris, 1990). However, circulating rheumatoid factors are not confined to patients with rheumatoid arthritis. Rheumatoid factors have been detected in the serum from patients with other rheumatic diseases such as systemic lupus erythematosus (SLE), systemic sclerosis and primary Sjogren's syndrome (Shmerling & Delbanco, 1991) and in patients with chronic infections such as leprosy, tuberculosis and malaria (Williams, 1974; Shmerling & Delbanco, 1991). Low levels of rheumatoid factors are also found in the serum from many normal individuals as part of a normal immune response to foreign antigen (Levine & Axelrod, 1985) and levels are seen to rise with age (Hallgren et al., 1973; van Schaardenburg et al., 1993). A small proportion of patients with rheumatoid arthritis test negative for circulating rheumatoid factor. These seronegative patients tend to develop less erosive disease and rarely have extra-articular features, but their disease is otherwise indistinguishable (Conway et al., 1994).

Many attempts have been made to use the detection of rheumatoid factors in sera as a diagnostic test for rheumatoid arthritis. The first method was based on the original techniques of Waaler and Rose and has become known as the Rose-Waaler agglutination test (Waaler, 1940;

36 Rose et al., 1948). This test detects IgM rheumatoid factor by measuring the agglutination of sheep red blood cells sensitised with rabbit IgG. In 1956 the latex fixation test was developed in which the sheep red blood cells were replaced by latex particles coated with human IgG (Singer & Plotz, 1956). This method is sensitive and often used as a screen but is limited by its low specificity (Shmerling & Delbanco, 1991). Although a number of other tests, including nephelometric techniques (Roberts- Thom pson et al., 1985) have been developed in an attempt to optimise sensitivity, specificity and reproducibility, the agglutination of IgG-coated particles remains the basis for routine screening tests. Solid phase assays have also been developed to detect rheumatoid factor of specific isotypes by determining the binding of rheumatoid factor to either human (Carson et al., 1977; Powell et al., 1985), rabbit (Swedler et al., 1997) or goat IgG Fc fragments coated onto a plastic substrate (Tuomi, 1989). The rheumatoid factors are detected using a radiolabelled or enzyme- conjugated anti-human IgG, IgA or IgM. An important consideration in relation to these types of assay is whether they measure the same populations of autoantibodies. Conformational changes may occur in the Fc fragment of IgG when bound either to a particle such as a sheep red cell, or a plastic surface. In solid-phase assays further conformational changes may be associated with the cleavage of IgG substrate to yield isolated Fc fragments, leading to significant changes in avidity. There are, therefore, technical reasons why rheumatoid factor activity measured by these assays may not give a direct measure of their potential biological properties.

Despite the development of many different tests, the specificity and sensitivity of rheumatoid factor assays for rheumatoid arthritis remain too low to be of major clinical value. The diagnosis of rheumatoid arthritis is still largely based on clinical criteria. This implies that if rheumatoid factors are pathogenic, they act through a property which overlaps with, but is not identical to, rheumatoid factor activity as identified by IgG Fc binding assays.

37 1.6.2. Rheumatoid factor isotype One possible reason for the divide between rheumatoid factor positivity and the development of rheumatoid arthritis, is that pathogenicity may be dependent on the isotype of the rheumatoid factor. The predominant class of rheumatoid factor in both normal individuals and patients with rheumatoid arthritis is IgM. However, there is no clear relationship between IgM rheumatoid factor levels and the severity of inflammation Qacoby et a l, 1973; Aho et al., 1997). Patients with high serum titres of IgG rheumatoid factor tend to have more severe disease and display extra- articular manifestations (Hay et al., 1979; Robbins et a l, 1986). More recently, a study in Finland has shown the early detection of raised serum IgG levels in normal individuals to be indicative for the development of seropositive rheumatoid arthritis later in life (Aho et al., 1997). A number of studies have also found the presence of IgA rheumatoid factor in early synovitis to be a marker for the development of more severe disease and to be associated with the manifestation of extra-articular features in patients with rheumatoid arthritis (Teitsson et al., 1984; Moore et a l , 1994; J6 nsson et al., 1995; Aho et a l, 1997). Additionally, Rudge et al. (1985) observed a fall in disease activity and IgG rheumatoid factor and IgA rheumatoid factor levels but no change in IgM rheumatoid factor levels following treatment with gold. While evidence for the importance of individual rheumatoid factor isotypes is inconclusive, there remains a suggestion that IgG and IgA rheumatoid factor may be more relevant to pathogenesis than IgM rheumatoid factor in rheumatoid arthritis. This is emphasised by the occurrence of IgM rheumatoid factors in normal individuals and as will be discussed later, the differentiation between IgM rheumatoid factor and IgG rheumatoid factor based immune complexes.

1.6.3. Fine specificity of rheumatoid factors The fine specificity of rheumatoid factors may provide important information equating to their pathogenicity in rheumatoid arthritis. Epitopes on the Fc region of IgG that are recognised by rheumatoid factors may be influential in determining the type of rheumatoid factor-based immune complex formed as well as the relationship between affinity, avidity and dissociation kinetics for interactions between the rheumatoid

38 factor and the IgG Fc fragment. These factors may differentiate between a potentially pathogenic and a non-pathogenic rheumatoid factor.

Rheumatoid factors from patients with rheumatoid arthritis are polyclonal and may react with a variety of epitopes expressed by the Fc fragment of IgG (Kunkel & Tan, 1964). A number of studies have focused upon the characterisation of the fine specificity of rheumatoid factors in an attempt to reveal the function of these autoantibodies. Most have studied IgM rheumatoid factors from patients with rheumatoid arthritis. Owing to the polyclonality of these antibodies, a number of groups have chosen to utilise monoclonal IgM rheumatoid factors found in patients with Waldenstrom's macroglobulinemia and essential mixed cryogloubulinemia (Sasso et a l, 1988). However, it has been found that significant differences exist between these paraproteins and rheumatoid factors derived from patients with rheumatoid arthritis, in both their specificity and avidity for IgG (Bonagura et a l, 1993; Robbins et a l, 1993). A direct comparison with paraproteins from patients with Waldenstrom's macroglobulinemia was achieved by using monoclonal IgM rheumatoid factors derived from patients with rheumatoid arthritis. These monoclonal rheumatoid factors were generated by the EBV transformation of either synovial or peripheral blood B cells from patients with rheumatoid arthritis. Furthermore, this process did not affect the specificity of the rheumatoid factors produced for genetically engineered IgG (Bonagura et a l, 1999). In one study it was shown that monoclonal IgM rheumatoid factors derived from patients with rheumatoid arthritis reacted with epitopes within the second heavy chain constant domain (CH 2, also referred to as Cy 2 ) of IgG, that were identical to those recognised by the majority of polyclonal rheumatoid factors examined (Williams & Malone, 1994).

The dominant binding site on the Fc fragment of IgG for both polyclonal (IgG and IgM) and m onoclonal (IgM) rheum atoid factors, has been located to the CH 2-CH3 interface (Nardella et a l, 1985; Sasso et a l, 1988). This was determined by measuring the levels of binding of rheumatoid factors to proteolytically cleaved fragments of IgG. The binding site on IgG for Staphylococcal Protein A (SPA) is also located at the CH 2-CH3 interface. Both IgG and IgM rheumatoid factors and this microbial

39 protein recognise some of the same amino add residues (Nardella et al., 1985; Sasso et al., 1988). In one study IgG and IgM rheumatoid factors were found to carry the internal image of SPA (Oppligner et al., 1987). Streptococcal Protein G (SPG) was also found to bind IgG at the CH 2-CH3 interface (Nardella et al., 1981) and to inhibit the binding of both polyclonal and monoclonal IgM rheumatoid factors to IgG (Stone et al., 1989). Although not identical, the binding sites for SPA and SPG are similar. By genetically engineering IgG using site directed mutagenesis and exon exchange, it was shown that at least three regions of the CH 2 domain and one region of the CH 3 domain contributed to the epitope recognised by a monoclonal IgM rheumatoid factor derived from a patient with Waldenstrom's macroglobulinemia (Artandi et al., 1992). This work supported previous observations for the specificity of IgM rheumatoid factors from patients with rheumatoid arthritis for IgG myeloma proteins (Natvig et ah, 1972).

The CH 2-CH3 interface also includes the antigen, designated Ga, which is recognised by both polyclonal and monoclonal IgM rheumatoid factors. Ga is found on IgGl, IgG2 and IgG4, but not IgG3 (Allen & Kunkel, 1966). The isotype distribution of the Ga antigen and the binding site for SPA follows the same pattern, while SPG binds all four IgG subclasses (Stone et al., 1989). In keeping with the Ga antigen distribution, serum derived IgM rheumatoid factors from patients with rheumatoid arthritis were found to bind IgGl, IgG2 and IgG4, but not IgG3. However, rheum atoid factors derived from rheumatoid synovium exhibited greater specificity and affinity for IgG3 relative to the binding of other subclasses by serum rheumatoid factors (Robbins & Wistar, 1985, Robbins et al., 1987, Robbins et al., 1993). A rtandi et al. (1992) showed the non-binding to IgG3 to be dependent upon a single amino acid residue. At amino acid position 435, IgGl, IgG2, and IgG4 express a Histidine, while IgG3 expresses an Arginine. These findings have suggested that rheumatoid factors derived from the synovium may have different fine specificities compared with circulating rheumatoid factors. This may be relevant to the type of immune complex formed in the circulation and the rheumatoid synovium and is an area that will be discussed later in this chapter.

40 1.6.4. IgG as an immunogen for rheumatoid factor production In terms of traditional immunological theory, the existence of rheumatoid factors is a puzzle. It is difficult to understand how an immune response to a plasma protein such as IgG, present in high concentration, can arise. Lang et a l (1999) have recently reported circulating IgG Fab fragment specific T cells clones in patients with rheumatoid arthritis, supporting previous work by Radoiu et a l (1982). These findings raise the possibility of a true failure of T cell tolerance to IgG. However, the consensus view has been that T cell responses to IgG are insignificant in rheumatoid arthritis.

Another group has reported high levels of a protein designated p205 in rheumatoid synovial fluid. This antigen was found to have amino acid sequences that matched those in the IgG CH 2 domain, the area where rheumatoid factors bind (Blass et a l, 1999). This may indicate a cross­ reactivity between anti-P205 T cells and IgG Fc peptides. However, the question of how tolerance to such a ubiquitous protein can occur remains the same.

An alternative explanation for the occurrence of rheumatoid factor production, is that the immunogen may be IgG which has been altered in some way, resulting in a structural change. Stimulation of an immune response to altered IgG might generate rheumatoid factors which recognise altered IgG specifically, but may also extend to the production of rheumatoid factors which bind native IgG. Some support for this idea comes from studies showing that rheumatoid factors from patient sera react with IgG from species other than human (Jones et a l, 1990). Initially, Robbins & Wistar (1985) found that serum IgM rheumatoid factor had a greater specificity for rabbit, than for human IgG. However, in a subsequent study serum rheumatoid factors from patients with rheumatoid arthritis were found to react equally with rabbit and human IgG, while rheumatoid factor produced by cells extracted from rheumatoid synovium had a higher avidity for human IgG (Robbins et a l, 1987).

It has been suggested that the glycosylation status of IgG is somehow related to the production of rheumatoid factor. The glycosylation state of

41 IgG varies w ith age (Parekh et al., 1988), however, it has been shown that patients with rheumatoid arthritis have an increased expression of agalactosyl IgG glycoforms compared with age-matched controls (Parekh et al., 1985). Nevertheless, this loss of galactose on the Fc region of IgG may not be specific for rheumatoid arthritis (Rademacher et al., 1988; Tomana et al., 1988). Moreover, it has been shown that the binding of IgM rheumatoid factors to IgG is not dependent upon the nature of the carbohydrates, but instead the amino acid sequence of the IgG Fc region (Newkirk et a l, 1990). However, the binding of a monoclonal IgG rheumatoid factor derived from a patient with rheumatoid arthritis to IgG Fc did appear to be influenced by carbohydrate (Newkirk & Rauch, 1993).

As will be discussed later, recent analysis of the possible mechanisms of rheumatoid factor production makes the need for an altered IgG immunogen less relevant. However, the detailed steric and glycosylation states of IgG may still be highly relevant in the context of rheumatoid factor immune complex formation.

1.6.5. Origins of rheumatoid factors Rheumatoid factors, as all immunoglobulins, are generated by cells from the B lymphocyte lineage (Male & Roitt, 1993). B lymphocytes exist as three distinct populations termed B-la, B-lb and B-2 (Kantor, 1991). Cells of the B-la population express the surface antigen CD5 and were originally referred to as Leu-1 B cells (Kipps, 1989). The progenitors of B- la cells are B-lb cells which are surface CD5" but CD5 mRNA+ (Mantovani et al., 1993). In the adult, CD5+ B cells constitute up to 30% of circulating and splenic B lymphocytes, while in the fetal spleen and newborn cord blood, CD5+ B cells are the dominant subpopulation. B-2 or ''conventional" B cells, arising from the bone marrow, constitute the majority of B lymphocytes in the adult. Unlike B-2 lymphocytes, CD5+ B cells maintain their numbers by self-repopulation (Casali & Notkins, 1989a; Kipps, 1989). Previous work by Casali and colleagues demonstrated that the majority of B cells generating polyreactive IgM autoantibodies with rheumatoid factor activity were B-la cells (reviewed in Casali & Notkins, 1989b). In contrast to B-2 cells which produce high affinity, monospecific antibodies following an antigen-driven immune response,

42 B-la derived autoantibodies arise from an antigen-independent process and remain close to germline configuration (Bona, 1988; Casali & Notkins, 1989a). Rheumatoid factors such as these can be found in normal individuals following infection or immunisation (Carson et al., 1987) and these so called "natural" autoantibodies may have some physiological function relating to their low affinity and polyreactivity (Cohen & Cooke, 1986). Natural autoantibodies may play a protective role against autoimmune disease by binding to self antigens which present epitopes also expressed by pathogenic antigens (Cohen & Cooke, 1986). Alternatively, the low affinity of natural autoantibodies may allow them to act as opsonins for the silent clearance of breakdown products which may be potentially immunogenic (Avrameas, 1991). Furthermore, natural autoantibodies may enhance normal immune responses by promoting complement fixation and immune complex clearance, increasing the avidity of low affinity IgG antibodies or enhancing antigen presentation (Carson et al., 1987; Kipps, 1989).

Increased circulating levels of CD5+ B cells have been reported in patients with rheumatoid arthritis (Plater-Zyberk et al., 1985; Hardy et al., 1987). However, a study by Jones et al. (1993) found patients with rheumatoid arthritis and IgA nephropathy to have decreased levels of circulating CD5+ B cells and increased levels of CD5- B cells, compared with normal individuals and patients with Graves’ disease where circulating rheumatoid factors are thought to be non-pathogenic. This study would therefore support a protective role for rheumatoid factors derived from CD5+ B cells and may suggest a distinct origin for potentially pathogenic rheumatoid factors. This concept is supported by the observation that in patients with rheumatoid arthritis, IgG rheumatoid factor secreting B cells were predominately CD5" whereas in controls and Graves disease they were mainly CD5+ (Jones et al., 1993). The contribution made by B-la cells to rheumatoid factor levels in vivo is far from clear but may provide some clue as to the origin of rheumatoid factors from patients with rheumatoid arthritis and from normal individuals.

Studies on IgM rheumatoid factors from normal individuals and patients with rheumatoid arthritis suggest that potentially pathogenic rheumatoid factors from rheumatoid synovium and those from normal

43 subjects differ significantly in a way that reflects their origin. Thompson et al. (1994) found that following immunisation with foreign antigen, the majority of IgM rheumatoid factors in normal individuals had their variable light chain (Vl) derived from the Vk3 gene segments, while the variable heavy chain (Vh) gene segments utilised families VhI/ Vh3 and V h4. Rheumatoid synovium derived rheumatoid factors predominantly used the heavy chain gene segments of the Vh3 family. Interestingly, similar studies by other groups have shown an over-representation of Vh4 gene segments by rheumatoid synovium derived rheumatoid factors (Brown et al., 1992; Pascual & Capra, 1992; Williams et al., 1999). Thom pson et al. (1995a) showed that rheum atoid factors from norm al individuals had undergone some somatic hypermutation, but little, if any affinity maturation. In contrast, the synovial rheumatoid factors from patients with rheumatoid arthritis exhibited a pattern of hypermutation distinct from that in normals, as well as both affinity maturation and class switching. Peripheral blood rheumatoid factors from patients with rheumatoid arthritis displayed a pattern of V-gene usage intermediate of that seen in rheumatoid synovium and normal individuals (Thompson et al., 1995b). These findings suggested that rheumatoid factors in patients with rheumatoid arthritis may arise, not only from the subset of B cells responsible for natural rheumatoid factor production but also from conventional B cells, as a result of a gene mutation in the variable region coding for immunoglobulins of unrelated specificities.

Thom pson et al. (1995a) demonstrated that rheumatoid factors from both normal immunised individuals and patients with rheumatoid arthritis bear evidence of somatic hypermutation. It is believed that in order to undergo somatic hypermutation, B cells must receive T cell help. B cells with rheumatoid factor specificity should be unable to obtain this help, since T cells specific for IgG Fc are deleted or made anergic (Bikoff, 1983). However, B cells with rheumatoid factor specificity may obtain T cell help through one of two mechanisms, both of which bypass the need for Fc specific T cells. They may obtain T cell help by presenting peptides derived from an irrelevant foreign antigen taken up in the form of IgG containing immune complexes via the B cell surface receptor specific for IgG Fc (Roosnek & Lanzavecchia, 1991). Alternatively, they may present

44 peptides from their own immunoglobulin V regions which might be shared by a common foreign antigen to which T cells were responsive (Thompson et al., 1994). The first mechanism requires the presence of foreign antigen and might be expected to operate during an immune response to infection or vaccination (Roosnek & Lanzavecchia, 1991). The second mechanism does not require the presence of foreign antigen and might be more important in the long term production of rheumatoid factors seen in association with rheumatic disease. Although in theory, the sharing of a peptide between an rheumatoid factor V region and a foreign antigen, such as SPA, could allow the second mechanism to operate, little or no information is available about the help received by rheumatoid factor specific clones.

Rheumatoid factors from patients with rheumatoid arthritis exhibit both affinity maturation and class switching to IgG and IgA, as expected from an antigen-driven response (Ermel et al., 1993). Rheumatoid factors from normal individuals, however, do not display affinity maturation or class switching, although Mantovani et al. (1993) have shown that B-la cells from one patient with rheumatoid arthritis generated high affinity IgM rheumatoid factors which had undergone somatic hypermutation, without class switching. Nevertheless, in most cases rheumatoid factors from normal individuals may be prevented from undergoing affinity maturation and therefore remain non-pathogenic, while disease- associated rheumatoid factors may escape this control mechanism (Thompson et al., 1994; Thom pson et al., 1995a). IgG rheum atoid factors are commonly found in rheumatoid arthritis patients with a high IgM rheumatoid factor titre, suggesting that class switching may be determined by the level of rheumatoid factor response. It may therefore be that IgG rheumatoid factors are derived from B cells previously producing IgM rheumatoid factor as a result of class switching. Therefore, a proportion of rheumatoid factor in rheumatoid arthritis may arise, as in normal individuals, as a physiological "spillover" from an antigen specific immune response. Alternatively, IgG rheumatoid factor may arise from B cell clones of unrelated specificities. In support of this concept Williams et al. (1999) have shown rheumatoid synovium derived IgA and IgM rheumatoid factors to be clonally related, whilst no clonal relationship was observed between rheumatoid synovial IgG and

45 IgM rheumatoid factors. However, there is growing evidence to suggest that a proportion of rheumatoid factor in patients with rheumatoid arthritis arise and persist through quite different mechanisms, which may relate in some way to their potential pathogenicity (Edwards et al., 1999).

1.6.6. Rheumatoid factors and the formation of immune complexes A rheumatoid factor of any class will bind to the Fc region of IgG and since IgG is present in most body fluids, including normal synovial fluids (Hrncfr et al., 1972), rheumatoid factor based immune complexes exist in vivo (Mannik, 1992). Immune complexes of rheumatoid factors may play a primary role in the initiation of inflammation and the pathogenesis of rheumatoid arthritis. Rheumatoid factor isotype may strongly influence the size, stereochemistry and stability of the immune complex formed, and therefore may be important in determining which complexes may have pathogenic properties.

1.6.7. IgM rheumatoid factor immune complexes IgM exists as a pentameric structure. Therefore, one IgM rheumatoid factor molecule has the potential to bind up to five IgG molecules (Figure la). Steric hindrance prevents the binding of more IgG molecules (Mannik, 1992). This structure is, however, unstable and may be in a constant state of association and dissociation (Mannik & Nardella, 1988). The avidity of an IgM rheumatoid factor molecule for aggregated IgG is far greater, forming a more stable complex (Mannik, 1992). An IgM molecule will only activate the complement component Clq once its structure has been altered to form a "five legged, table-like" configuration (Feinstein & Munn, 1969; Perkins et ah, 1991). This conformation will not occur following the binding of monomeric IgG as shown in Figure la. Recently, crystal structure studies have shown the "table-like" structure to form if either side of the Fc region of monomeric IgG is symmetrically bound by two IgM rheumatoid factor Fab fragments (Corper et ah, 1997) (Figure lb). The formation of circulating IgM rheumatoid factor complexes are unlikely to play a pathogenic role in rheumatoid arthritis since those stable enough to form may be rapidly cleared following activation of Clq.

46 Figure 1. Schematic showing the formation of IgM rheumatoid factor immune complexes with a) monomeric IgG as described by Mannik (1992) and b) m onom eric IgG as described by Corper et al. (1997).

a)

IgM-RF IgG

<^>

b)

1.6.8. IgG rheumatoid factor immune complexes Kunkel et al. (1961) were the first to describe "intermediate" immune complexes in the serum of patients with rheumatoid arthritis that had a sedimentation rate lying between 9S and 17S. These complexes were found to consist of IgG (Kunkel et al., 1961; Schrohenloher, 1966) and possess rheumatoid factor activity (Schrohenloher, 1966). Similar complexes were also detected in the synovial fluids of patients with rheumatoid arthritis (Hannestad, 1967; Winchester et al., 1970) and rheumatoid synovium (Munthe & Natvig, 1971; Munthe & Natvig,

47 1972). Characterisation by Pope et al. (1974) revealed these intermediate complexes to consist of two IgG rheumatoid factors that had self­ associated to form a dimer. A model was proposed by Mannik and colleagues (Pope et al., 1974) wherein the Fab portion of one IgG rheumatoid factor was bound to the Fc fragment of the other and visa versa, so that each IgG rheumatoid factor molecule was simultaneously acting as both antibody and antigen. Hence, a cyclic ring structure was formed, stabilised by the two antigen-antibody bonds (Figure 2).

Figure 2 . Schematic showing the proposed cyclic structure formed during the self-association of two IgG rheumatoid factor molecules. Taken from Pope et al. (1974).

Fab Fab

Fab Fab

IgG rheumatoid factor self-association enables the formation of immune complexes in the absence of any other antigen. From the estimation of dissociation constants, it was predicted that IgG rheumatoid factors would preferentially self-associate in the presence of normal monomeric IgG. The interaction of an IgG rheum atoid factor molecule with monomeric IgG will have an dissociation constant of approximately 10-5 L/mol (Pope et al., 1974). However, when two IgG rheumatoid factors self-associate the stabilising effect of two interactions may reduce the dissociation constant by an order of magnitude or more, thus resulting in an increased avidity. Although this interaction may have a longer on- rate than the binding of an IgG rheumatoid factor with a normal IgG molecule, the off-rate will also be slower making the self-associating IgG rheumatoid factor interaction more stable.

48 It was proposed that once self-association of the IgG rheumatoid factors had occurred, these complexes then had the potential to undergo concentration-dependent aggregation to form tetramers, octamers and 1 2 - mers (Pope et al., 1974), since one antigen binding site would still be available on the Fc fragment of each IgG rheumatoid factor molecule (Pope et al., 1975). Full exposure of these antigenic determinants would arise during the conformation changes that occurred with ring closure (Mannik, 1992). However, it was predicted that when a normal monomeric IgG molecule bound to an IgG rheumatoid factor, this second antigenic determinant became unavailable, owing to steric hindrance (Nardella et ah, 1981). Normal IgG molecules may be constantly associating and dissociating with these available sites at a rate such that in excess normal IgG, further polymerisation of the IgG rheumatoid factor dimer may be inhibited (Figure 3). The binding of two normal monomeric IgG molecules to an IgG rheumatoid factor dimer may explain why circulating complexes are small while the synovium houses larger polymers (Hannestad, 1967; Winchester et ah, 1970; N ardella et ah, 1981). As will be discussed later, this may be important with respect to rheumatoid arthritis as these small complexes containing self-associating IgG rheumatoid factor may be crucial to the initiating disease mechanism and the development of systemic manifestations.

Not all IgG rheumatoid factor may have the potential to self-associate. Self- association may only occur if the two IgG rheumatoid factors bind to the same or complementary epitopes (Pope et ah, 1975). Differences in epitope pairing may result in differences in pathogenic potentiality. Additionally, IgG rheumatoid factor dimers may differ in pathogenic potential when compared to IgG rheumatoid factors complexed with non-rheumatoid factor IgG. These considerations further emphasise the concept that only a proportion of IgG rheumatoid factors may generate pathogenic immune complexes and may explain why rheumatoid factor levels as measured at present, do not correlate precisely with the presence of disease.

49 Figure 3. Schematic diagram of the termination of IgG rheumatoid factor polymerisation by excess normal IgG. Taken from: Nardella et al. (1981).

IgG-RF dimer IgG

Fab Fab

Fab Fab

Fab Fab

Fab Fab

1.6.9. Rheumatoid factor complexes and Waldenstrom's hypergammaglobulinemic purpura In 1943, Waldenstrom (1943) described the syndrome of hypergammaglobulinemic purpura, characterised by a recurrent purpura of the lower extremities, mild anaemia and hypergammaglobulinemia. These features may occur in patients with rheumatoid arthritis and especially Felty's syndrome, however, they may also occur in the absence of rheumatoid arthritis (Clark et al, 1974). Moreover, intermediate complexes of IgG rheumatoid factor have been detected in the sera of patients with Waldenstrom's hypergammaglobulinemic purpura in the absence of synovitis (Kunkel et al., 1961; Capra et al, 1971; Kyle et al., 1971). Clark et al. (1974) performed a direct comparison between the rheumatoid factor complexes found in the sera of patients with

50 rheumatoid arthritis and the complexes in the sera of patients with only Waldenstrom’s hypergammaglobulinemic purpura. It was found that sera from patients with Waldenstrom’s hypergammaglobulinemic purpura contained a higher proportion of IgG and IgA rheumatoid factor complexes than from patients with rheumatoid arthritis. Furthermore, the IgG rheumatoid factor complexes from patients with Waldenstrom's hypergammaglobulinemic purpura were predominantly intermediate complexes, while the sera from patients with rheumatoid arthritis contained mostly large, IgM rheumatoid factor based complexes. Analysis of these intermediate IgG rheumatoid factor complexes by Capra et al. (1971) indicated the prevalence of the subclass IgGl. However, the radial immunodiffusion technique used to identify IgG subclass was relatively unsophisticated. Two separate studies further analysed the self associating IgG rheumatoid factor dimers from patients with Waldenstrom's hypergammaglobulinemic purpura and found the predominant subclass in these complexes to be IgG3 (Bignold et a t, 1980; Serino et a t, 1983). In comparison, a study on rheumatoid factor subclass in the serum from patients with rheumatoid arthritis revealed IgGl and IgG4 as the dominant subclasses (Cohen et a l, 1987). These observations further suggest that only a subgroup of IgG rheumatoid factors may be pathogenically involved in the generation of synovitis in rheumatoid arthritis.

1.6.10. Fixation and activation of complement by rheumatoid factors A number of studies have shown complement consumption in the synovial fluids of patients with rheumatoid arthritis (Pekin & Zvaifler, 1964; Hedberg, 1967; Ruddy et a l, 1969) relative to complement levels within the serum (Winchester et a l, 1970). Complement consumption occurring within the synovium was thought to be a result of a localised immune response. It was therefore thought the rheumatoid factor complexes found in the synovium and synovial fluids of patients with rheumatoid arthritis were behaving pathogenically by activating the complement system. Work by Winchester and his colleagues (Winchester et a t, 1970) demonstrated the ability of large IgG rheumatoid factor complexes, isolated from the synovial fluids of patients with rheumatoid arthritis, to fix and activate components of human complement. Complement fixation was further enhanced by the addition

51 of IgM rheumatoid factor to the IgG rheumatoid factor complexes. Smaller complexes, isolated from the synovial fluids of patients with rheumatoid arthritis, failed to precipitate with Clq. Similarly, Bianco et al. (1974) demonstrated complement fixation by IgM rheumatoid factor. As found by Bianco et al. (1974) and other groups, IgG rheumatoid factors from patients with rheumatoid arthritis did not fix complement (Tanimoto et a l, 1976; Sabharwal et al., 1982). This was probably because in these studies, the polymerisation of the IgG rheumatoid factors was blocked by the presence of normal IgG and these complexes were too small to fix complement. Brown et al. (1982) demonstrated the ability of polymerised, self-associated IgG rheumatoid factor complexes from patients with rheumatoid arthritis to fix and activate the human complement system. They suggested that within the joint it is more likely to find larger IgG rheumatoid factor polymers and therefore complement activation, explaining the observations by Winchester et al. (1970) who found that high synovial fluid rheumatoid factor titres correlated with depressed complement levels.

In rheumatoid arthritis, rheumatoid factors may be generated both in the synovium and in lymphoid organs. From either source they may gain access to the circulation either in isolation or in the form of complexes. There is a large body of evidence indicating the activation of complement in the joint in rheumatoid arthritis, but relatively little evidence of systemic complement activation. In approximately 2-5% of patients with high rheumatoid factor titres, micro vascular injury may occur. Vasculitis in rheumatoid arthritis usually manifests as nail fold infarcts and ulcers of the lower leg (Maini & Zvaifler, 1998). SLE patients with glomerulonephritis and other forms of microvascular damage exhibit immune complex deposition in basement membranes and systemic complement activation (Schur, 1993). However, in rheumatoid arthritis there is evidence of neither (Conn et a l, 1993), suggesting that complement activation is not the primary mechanism of rheumatoid vasculitis. Instead, vasculitic lesions in rheumatoid arthritis may occur as a result of platelet activity within the microvasculature (Cunningham et al., 1986). High circulating rheumatoid factor levels may create a sludging effect at the sites of venular injury common to rheumatoid arthritis, Waldenstrom's macroglobulinaemia and Waldenstrom’s

52 hypergammaglobulinemic purpura. Platelet aggregation and activity in response to aggregated immunoglobulin at these sites (Fink et al., 1979) may therefore be the cause of vasculitis in these disorders, rather that immune complex deposition and complement activation.

Complement activation may contribute to synovitis once the tissue has become populated with rheumatoid factor secreting plasma cells. When an immune complex interacts with complement it may either generate inflammation if this occurs within tissues or it may be cleared if this occurs within the circulation. It may be that the rheumatoid factor containing complexes responsible for disseminating the disease to joints are those which do not interact with complement i.e. small self­ associating IgG rheumatoid factor complexes. These small IgG rheumatoid factor immune complexes, (unlike IgM rheumatoid factor complexes or IgG rheumatoid factor multimers) may escape clearance in the circulation, but may still be able to generate inflammation in the extravascular space through a mechanism independent of complement. Such a mechanism may involve the direct activation of synovial macrophages through the binding of small IgG rheumatoid factor complexes to surface immunoglobulin receptors.

1.7. Human immunoglobulin receptors Human immunoglobulin receptors have been identified and characterised for IgG (FcyR), IgA (FcaR) and IgE (FceR) (reviewed in: Ravetch & Kinet, 1991; van de Winkel & Anderson, 1991; Morton et al., 1996). Fc receptors function as a link between the antibody response and effector cell function. The three human FcyR, which will be discussed in detail, are glycoproteins belonging to a subgroup of the immunoglobulin supergene family (Williams & Barclay, 1988). FcyR have the ability to mediate the phagocytosis of particles, such as microbes, the internalisation of immune complexes, antibody dependent cellular cytotoxicity (ADCC) and the release of pro-inflammatory factors, such as cytokines or free radicals (reviewed in: Fanger et al., 1989; van de Winkel & Capel, 1993; Gergely & Sarmay, 1994). Fc receptors for IgG were first described in the late 1960's (Berken & Benacerraf, 1967). They were identified as receptors on human monocytes and macrophages that bound and phagocytosed red blood cells coated with human IgG, in the

53 absence of complement (Huber & Fudenberg, 1968a; Huber et al., 1968b; Huber et al., 1969). Three murine FcyR have also been identified and characterised. Although some homology exists, murine FcyR differ from human FcyR in their signalling capacities, structure and cell distribution (reviewed in: Mellman et al., 1988; Unkeless et a l, 1988).

Fey receptors may play a dominant role in the pathogenesis of rheumatoid arthritis by binding immune complexes of IgG rheumatoid factor and mediating macrophage activation. Several studies in vivo have elegantly demonstrated IgG immune complex-induced inflammation and tissue injury to be FcyR-mediated (Sylvestre & Ravetch, 1994; Sylvestre et al., 1996; Clynes et al., 1998; Clynes et al., 1999). However, equating the role of individual Fey receptors in these experimental models to human disease is dangerous since murine and human FcyR differ, particularly in terms of signalling capacities.

The possibility of Fc receptor involvement in the inflammatory process in rheumatoid arthritis emphasises the pathogenic potential of IgG rheumatoid factor and the possible non-pathogenicity of IgM rheumatoid factor. There is no substantial evidence for the existence of an IgM- specific Fc receptor. Although IgM rheumatoid factor may bind and activate macrophages through complement receptors (Walport, 1993), as will be discussed later, it seems unlikely that circulating IgM rheumatoid factor would be able to access tissue macrophages in order to initiate inflammation in rheumatoid arthritis. In contrast, small immune complexes of self-associating IgG rheumatoid factor may be the subpopulation of rheumatoid factor based immune complexes that can selectively initiate inflammation in rheumatoid arthritis.

As discussed previously, the presence of circulating IgA rheumatoid factor in patients with rheumatoid arthritis also appears to correlate with disease activity. FcaR (CD89) is expressed by granulocytes and monocytes and can bind both monomeric and polymeric IgAl and IgA2 (Monteiro et a l, 1990; M onteiro et al., 1992). The binding of IgA rheumatoid factor complexes to FcaR may also activate macrophages in rheumatoid arthritis and this will be expanded upon in the discussion (see: 4.5. Rheumatoid factor, Fey receptors and rheumatoid arthritis).

54 1.7.1. FcyRI (CD64) FcyRI is a heavily glycosylated 72-kDa molecule that was originally isolated from human monocytes and U937 cells by affinity chromatography (Anderson, 1982). FcyRI is the only Fc receptor w ith a high affinity (Ka = 10 8 - 109 M"l) for monomeric IgG (Anderson & Abraham, 1980). FcyRI is constitutively expressed by monocytes and macrophages (Looney et ah, 1986a), and its expression can be increased 5- to 10-fold by treatment with gamma interferon (IFN-y) (Guyre et ah, 1983). It is also expressed by dendritic cells and CD34+ myeloid progenitors (Deo et ah, 1997). FcyRI expression by neutrophils can also be induced following treatment with IFN-y or granulocyte colony stim ulating factor (G-CSF) (Perussia et ah, 1983; Kerst et ah, 1993). FcyRI binds human IgG subclasses in the order of magnitude: IgGl=IgG3>IgG4»>IgG2 (Anderson & Abraham, 1980), while the receptor subclass specificity for murine IgG is: 2a=3»l, 2b (Jones et ah, 1985).

The three FcyRI genes; FcyRI A, FcyRIB and FcyRIC have been m apped to the long arm of chromosome 1 (lq21.1) (Ernst et ah, 1992). FcyRIA encodes for a receptor (FcyRIa) that consists of a 292 amino acid extracellular region which contains three immunoglobulin-like domains, a 2 1 amino acid transmembrane region and a highly charged cytoplasmic domain of 61 amino acids (Allen & Seed, 1989). FcyRIB encodes for two transcripts, FcyRIbl and FcyRIb2, while FcyRIC encodes for the receptor FcyRIc (van de W inkel et ah, 1991). All three transcripts have an extracellular region containing two immunoglobulin-like domains (Ernst et ah, 1992). Both FcyRIbl and FcyRIc are soluble receptors which lack membrane anchorage, as a result of a translation stop codon in exon EC3. Cells expressing FcyRI all contain transcripts for FcyRIa, w hich binds m onom eric IgG w ith high affinity, and FcyRIb2, which only binds complexed IgG (Porges et ah, 1992).

1.7.2. FcyRU (CD32) FcyRII is a 40-kDa protein which was first purified from hu m an peripheral blood monocytes and U937 cells by affinity chromatography (Anderson & Abraham, 1980). It is a low affinity FcyR and cannot efficiently bind monomeric IgG (Ka < 10 7M_1). It can, however, bind

55 complexed human IgG and murine IgGl and IgG2b Qones et ah, 1985; Looney et ah, 1986a). Of all Fey receptors, Fc/RII has the widest distribution. It is expressed by monocytes, macrophages, neutrophils, eosinophils (Looney et al., 1986b), basophils, dendritic cells (Deo et ah, 1997), platelets (Rosenfeld et ah, 1985), B cells (Cohen et ah, 1983; Looney et ah, 1986a), Langerhans cells (Schmitt et ah, 1990) and placental endothelial cells (Sedmak et ah, 1991).

The cloning and sequencing of complementary DNA (cDNA) has shown that FcyRII has an extracellular region of 180 amino acids which contains two immunoglobulin-like domains, an transmembrane region of 27-29 amino acids and a cytoplasmic domain that varies from 44 to 76 amino acids (Stuart et ah, 1987; Hibbs et ah, 1988; Stengelin et ah, 1988). The three genes for FcyRII (FcyRIIA, FcyRIIB, FcyRIIC) have been m apped to lq23-24 of chrom osom e 1 and encode the six isoforms: FcyRIIal, FcyRIIa2, FcyRIIbl, FcyRIB, FcyRIIb3 and FcyRIIc (Qui et ah, 1990). All transcripts have identical extracellular domains but are heterogeneous in their cytoplasmic dom ains. FcyRIIa2 is a soluble isoform as a result of alternate RNA splicing (Warderdam et ah, 1990). FcyRIIal and FcyRIIc are expressed by monocytes, macrophages and neutrophils, while the FcyRIIb isoform is expressed by monocytes, macrophages and B cells (Brooks et ah, 1989). FcyRIIb binds hum an IgG subclasses in the order of 3>1>4»2 and murine IgG in the order of 2b>2a>l»3 (Ravetch & Kinet, 1991). The subclass specificity for FcyRIIa varies w ith allotype (see 1.7.5. FcyR polymorphisms).

1.7.3. FcyRIII (CD16) FcyRIII is an extensively glycosylated protein with a molecular weight ranging from 50 to 80 kDa. It was first identified as a receptor on neutrophils that differed from the previously identified monocyte Fey receptors in the binding of both human IgGl and a murine monoclonal antibody (3G8) (Fliet et ah, 1982). Two FcyRIII genes have been identified and mapped to lq23-24 on the long arm of chromosome 1 (Ravetch & Perussia, 1989). FcyRIIIA encodes the transcript, FcyRIIIa, and FcyRIIIB encodes for FcyRIIIb. Both receptors possess an extracellular region of approximately 190 amino acids which contains two immunoglobulin­ like domains. FcyRIIIa has a transm em brane region and a 25 am ino acid

56 cytoplasmic tail (Simmons & Seed, 1988), whereas FcyRIIIb is attached to the outer side of the plasma membrane by a glycophosphatidylinositol (GPI) anchor (Huizinga et al., 1988; Selvaraj et ah, 1988). This variance between FcyRIIIa and FcyRIIIb is the result of a single am ino acid substitution at position 203. At this amino acid position, FcyRIIIA expresses a phenylalanine which results in a transmembrane and cytoplasmic region, while a serine at position 203 results in the GPI moiety (Lanier et al., 1989a; Ravetch & Kurosaki, 1989). Both FcyRIIIa and FcyRIIIb can exist as soluble receptors, generated by proteolytic cleavage (Huizinga et ah, 1990a; H arrison et al., 1991).

FcyRIIIa is expressed by macrophages (Fleit et ah, 1982), a subpopulation of monocytes (Ziegler-Heitbrock et al ., 1988), natural killer (NK) cells and large granular lymphocytes (LGL) (Lanier et ah, 1983; Perussia & Trinchier, 1984). Modulation of FcyRIIIa expression by monocytes will be discussed later in this section. FcyRIIIa is a moderate-affinity receptor since in addition to binding complexed IgG, it can also bind monomeric IgG, albeit with a lower affinity than FcyRI (Ka ~ 3xl07 M_1) (van de Winkel & Anderson, 1991; Vance et ah, 1993). FcyRIIIb is expressed constitutively by neutrophils (Fleit et ah, 1982) and on IFN-y treated eosinophils (Hartnell et ah, 1992) and has a low affinity for human IgG (Ka < 107M_1). Both receptor isoforms bind human IgG subclasses in the order of: 1=3»>2,4 and murine IgG subclasses in the order of: 3>2a>2b»l (van de Winkel & Capel, 1993). The IgG binding site for both FcyRIIIa and FcyRIIIb lies w ithin the second im m unoglobulin-like (membrane-proximal) domain (Hibbs et ah, 1994; Tam m et ah, 1996; Tamm & Schmidt, 1996).

1.7.4. Fey receptor subunits and signalling Unlike FcyRII, FcyRI and FcyRIII lack a signalling m otif and therefore rely on associated subunits for signal transduction (Figure 4). FcyRIIIa (but not FcyRIIIb) is associated with disulphide linked subunits. Macrophage FcyRIIIa is associated w ith a gamm a chain hom odim er (y-y) which was first described for the high affinity mast cell receptor, F ceR I (Hibbs et ah, 1989; Ra et ah, 1989), and is also associated with FcaR (Pfefferkorn & Yeaman, 1994). FcyRIIIa expressed by NK cells may be associated w ith this y-chain dimer, however, it can also be found associated with either a

57 homodimer of the zeta chain (£-£) found in the TCR-CD3 complex (Lanier et al., 1989b; A nderson P. et ah, 1990) or a heterodim er (£-y) (Masuda & Roos, 1993). Expression of these subunits w ith FcyRIIIa is essential for receptor expression and the prevention of its degradation (Kurosaki & Ravetch, 1989; Kurosaki et ah, 1991). Furthermore, work by Withmuellar et ah (1992) has shown that the y-chain is essential for FcyRIIIa signal transduction.

FcyRI and FcyRII have both been show n to associate w ith the y-chain dimer (Ernst et ah, 1993; Masuda & Roos, 1993). As for FcyRIIIa, this y- chain subunit may be necessary for FcyRI receptor expression (van Vugt et ah, 1996), however, others have demonstrated that not all membrane bound FcyRI is complexed with the y chain (Allen & Seed, 1989; Masuda & Roos, 1993). Although FcyRI utilises the y-chain for signalling (Scholl & Geha, 1993; Melendez et ah, 1998), in some circumstances it may recruit FcyRII, which contains a signalling motif within its cytoplasmic domain (Melendez et ah, 1998). Moreover, work by Miller et ah (1996) has suggested that the y-chain may increase FcyR affinity for IgG and this may explain the high affinity of FcyRI and interm ediate affinity of FcyRIIIa for m onom eric IgG.

The cytoplasmic domains of both the y and £ chains contain a structural signalling motif, essential for their activating properties and common to all multichain immune recognition receptors. This structure was first described by Reth (1989) and later termed the Immunoreceptor Tyrosine- based Activation Motif (ITAM) (Cambier et ah, 1995). It has the sequence YXXL-7x-YXXL where 7x denotes seven variable amino acids (Cambier et ah, 1994). FcyRII also contains an ITAM within its cytoplasmic tail, however, this ITAM is unique in that the two YXXL sequences border 12 am ino acids (Reth, 1989). Therefore, FcyRII can signal independently, however, recruitment of the y-chain homodimer can modulate the signalling behaviour of this receptor (Van den Herik-Oudijk et ah, 1995a). Fey receptor crosslinking or aggregation results in the phosphorylation of tyrosine residues within the ITAM by sre-family protein tyrosine kinases (PTK). This is followed by the binding of syk-family PTK to the phosphorylated ITAM. Subsequent activation of these syk PTK results in a cascade of intracellular physiological responses.

58 Within the cytoplasmic tail of FcyRIIbl lies an Immunoreceptor Tyrosine-based Inhibitory Motif (ITIM) (Amigorena et al ., 1992) w hich consists of a single YXXL sequence (Daeron, 1997) and this gives this receptor its inhibitory activity. For example, the simultaneous crosslinking of FcyRIIbl and the B cell receptor (BCR) downregulates antibody secretion (Amigorena et a l, 1992).

59 Figure 4. Schematic diagram of surface expressed human Fey receptors and their associated subunits. Adapted from Deo et ah (1997).

FcyRI FcyRII la

Ib2 He lib

Y-Y Y-Y

ITAM

ITIM

FcyRIII

Ilia mb

X>< £ “ Y Y-Y GPI l-s-s- s-s- m o t if

60 1.7.5. Fey receptor polymorphisms If Fc/ receptors play a role in the pathogenesis of rheumatoid arthritis, then receptor polymorphisms may influence either an individual's susceptibility to development of disease, or may have an affect upon disease severity. However, to date, there are no reports linking any of the following FcyR polymorphisms with rheumatoid arthritis.

A rare expression polymorphism for FcyRI was identified by Ceuppens et al. (1988). They found that four healthy sisters from a Belgium family lacked the expression of FcyRI by their peripheral blood monocytes. Monocytes from 3000 normal individuals were tested from the same area and all expressed FcyRI (Van de W inkel et a l, 1995). Monocytes from the FcyRI deficient individuals were unable to bind the Fc region of mouse IgG2a or monomeric human IgGl and were unreactive with two anti-FcyRI monoclonal antibodies. Furthermore, monocytes from these individuals failed to support mouse IgG2a anti-CD3 monoclonal antibody-induced T cell proliferation (Ceuppens et a l, 1985a; Ceuppens et a l, 1985b). Further analyses showed this expression polymorphism to be a result of a single nucleotide substitution in the open reading frame of the FcyRIA gene, causing an early stop codon in the first im m unoglobulin-like extracellular dom ain of FcyRIa. Messenger RNA instability was observed and reverse transcription polymerase chain reaction analysis showed the FcyRIa transcript to be 15-20 fold lower than in controls. Interestingly, FcyRIb2 transcripts were elevated in the FcyRIa deficient individuals and it was postulated that this may account for their normal phagocyte function (van de Winkel et al., 1995).

FcyRIIa expressed by monocytes, macrophages, neutrophils and B cells displays a well described genetic polymorphism. This was first identified by the ability of human monocytes to bind a murine IgGl anti-CD3 monoclonal antibody and induce T cell proliferation (Tax et al., 1983). Monocytes from 70% of Caucasian individuals possessed this ability and were termed high responders, while the remaining 30% were termed non responders. This distribution was found to be reversed within the Asian population, with 15% high responders and 85% low responders (Abo et al ., 1984).

61 Molecular studies found that the polymorphic forms of FcyRIIa differed in a single amino acid at two amino acid positions. At position 27, within the first immunoglobulin-like domain (membrane-distal), FcyRIIa from high responders displayed a glutamine while low responders expressed a tryptophan (Clark et al., 1991). However, the FcyRIIa polym orphism was found to be due to the amino acid difference that occurred at position 131, within the second immunoglobulin-like domain (Warderdam et al., 1990). This membrane-proximal domain is the location of the IgG binding site (Hulett et al., 1995). At this position, FcyRIIa from high responders expressed an arginine (R) and low responders expressed a histidine (H). The allotypes were therefore termed FcyRIIa-131R (high responder) and FcyRII-131H (low responder). While FcyRIIa-131R binds murine IgGl well, it was found that this allotype had only a weak affinity for hum an IgG2 dimers. Conversely, FcyRIIa-131H, which has a low affinity for murine IgGl, was found to bind human IgG2 dimers strongly. Both allotypes bound dimers of human IgGl and IgG3 (Warderdam et al., 1991). Therefore the subclass specificity for hum an IgG is; FcyRIIa-131R: 1=3>2>4 and FcyRIIa-131H: 1=3=2>4 (Salmon et al., 1992), and for murine IgG; FcyRIIa-131R: 2a=2b=l and FcyRIIa-131H: 2 a= 2 b » l (van de W inkel & Capel, 1993). Furthermore, the murine anti-FcyRII monoclonal antibody, 41H16, binds to an epitope only expressed by the FcyIIa-131R allotype (Gosselin et al., 1990).

The FcyRIIa polymorphism has also been found to influence neutrophil function. Neutrophils expressing FcyRIIa-131H were found to have a greater capacity for the phagocytosis of human IgG2 or IgG3-coated erythrocytes than neutrophils expressing FcyRIIa-131R (Salmon et al., 1992; Bredius et al., 1994). FcyRIIa polym orphism has also been associated with increased susceptibility for infections (Fijen et al., 1993; Sanders et al., 1994; Platonov et al., 1998) and the autoimmune disease SLE (Duits et a l, 1995).

FcyRIIIb expresses a polym orphism designated neutrophil antigen 1 (NA1) and neutrophil antigen 2 (NA2). 37% of Caucasians express the FcyRIIIb-NAl allotype, while the remaining 63% express FcyRIIIb-NA2 (van de Winkel & Capel, 1993). The two isoforms display differing levels of glycosylation and deglycosylated NA1 and NA2 have distinct

62 electrophoretic mobility (Ory et al., 1989a). Ravetch & Perussia (1989) found 5 nucleotide differences between NA1 and NA2. However, the polymorphism was found to be due to a two amino acid difference occurring at positions 65 and 82. This difference was found to result in two extra N-linked glcosylation sites within the membrane-distal domain of FcyRIIIb-NA2, giving it a total of six sites (Ravetch & Perussia, 1989; Ory et a l, 1989b, Huizinga et a l, 1990b). Furthermore, the monoclonal antibody, CBL Gran 11 only bound to FcyRIIIb-NAl, while the monoclonal antibody, GRM1 only reacted with FcyRIIIb-NA2 (van de Winkel & Capel, 1993). Functionally, neutrophils expressing the FcyRIIIb- NAl allotype were found to have a greater capacity for the phagocytosis of both IgGl and IgG3 coated particles than neutrophils expressing the FcyRIIIb-NA2, irrespective of the FcyRIIa polym orphism (Salmon et a l, 1990; Bredius et a l, 1994). However, there was no difference between the two NA allotypes in the levels of binding of opsonised erythrocytes (Salmon et a l, 1992).

FcyRIDb polymorphism has been associated with severe renal disease in systemic vasculitis (Wainstein et a l, 1996) and with susceptibly to bacterial infections in combination with FcyRIIa polymorphism (Fijen et a l, 1993). Recent data has suggested a link between FcyRIIIb polymorphism and multiple sclerosis (Myhr et a l, 1999). FcyRIIIb deficiencies have also been observed. A patient with SLE was found to lack the expression of FcyRIIIb as a result of gene disorganisation (Clark et a l, 1990), while neutrophils from two normal individuals were found by Huizinga et a l (1990c) to lack the gene, FcyRIIIB.

A num ber of polym orphism s have been identified for FcyRIIIa. Vance et a l (1993) have described an expression polymorphism for FcyRIIIa on LGL and NK cells. By measuring the levels of human monomeric IgGl binding to these cells it was found that some individuals expressed twice the num ber of FcyRIIIa than others. This was also reflected functionally since LGL/NK cells from individuals with high levels of FcyRIIIa showed a greater capacity for the ADCC of erythrocytes. The first structural polymorphism for FcyRIIIa was identified by Ravetch & Perussia (1989). They found that a nucleotide substitution at position 559 resulted in either a Valine (V) or a phenylalanine (F) at amino acid position 158 in

63 the m em brane-proxim al dom ain of FcyRIIIa, expressed by natural killer cells. A second structural polymorphism was described by de Haas et al (1996). They found at am ino acid position 48, NK cell FcyRIIIa possessed either a leucine (L), a histidine or an arginine, resulting in a triallelic polym orphism . N atural killer cells therefore expressed either FcyRIIIa- 48L, FcyRIIIa-48H or FcyRIIIa-48R. Characterisation of this polym orphism showed that NK cells expressing FcyRIIIa-48R or FcyRIIIa-48H had a greater capacity for binding human IgGl and IgG3 than cells expressing FcyRIIIa-48L. Furtherm ore, it was found that 86% of Caucasian individuals expressed the FcyRIIIa-48L allotype, while 8% expressed FcyRIIIa-48H and 6% expressed FcyRIIIa-48R. In contrast Japanese individuals were only found to express the FcyRIIIa-48L allotype. Further studies of both the -158V/F and -48L/H/R polymorphisms demonstrated that the two were linked. Individuals expressing the -158F genotype also expressed FcyRIIIa-48L, while those positive for the -158V genotype expressed either FcyRIIIa-48H or FcyRIIIa-48R (Koene et a l, 1997). However, NK cells expressing FcyRIIIa-158V bound higher levels of hum an IgGl and IgG3 than NK cells expressing FcyRIIIa-158F and this was independent of the -48L/H/R polymorphism (Koene et a l, 1997; W u et a l, 1997). Moreover, the selective binding of the anti-FcyRIII m onoclonal antibody, MEM154, to the FcyRIII-158V allotype was found to be dependent upon this amino acid (Koene et a l, 1997) and not the linked polymorphism at amino acid position 48 as originally thought (de Haas et a l, 1996). However, the amino acid expressed at position 48 influenced the binding of the monoclonal antibody, B73.1 (Koene et a l, 1997). The FcyRIIIa-158F genotype may be associated w ith an increased risk of developing systemic lupus erythematosus (Keone et a l, 1998).

1.7.6. M odulation o f Fey receptor expression The levels of FcyRI expression can be enhanced by IFN-y or G-CSF (Guyre et a l, 1983; Perussia et a l, 1983; Kerst et a l, 1993). FcyRI and FcyRII expression can also be downregulated by interleukin-4 (IL-4) (te Velde et a l, 1990). Previous studies have shown that a number of cytokines and growth factors can m odulate monocyte FcyRIIIa expression. Liao and colleagues found that the treatment of monocyte-derived macrophages w ith TN Fa or IL-ip resulted in the rapid dow nregulation of FcyRIIIa expression (Liao & Simone, 1994; Liao et a l, 1994). FcyRIIIa levels

64 returned to at least that of the controls within 24 hours following initial cytokine exposure. It has also been demonstrated that the culturing of freshly isolated monocytes with either TGF-pl or TGF-p2 induced FcyRIIIa expression (Welch et al., 1990; W ong et ah, 1991). However, work generated from our laboratory suggests that TGFp may in fact act as an inhibitor of FcyRIIIa expression by hum an monocytes in vitro (Bhatia et al., in preparation). More recently, it has been shown that human monocytes cultured with interleukin-10 (IL-10) induced high levels of FcyRIIIa expression (Calzada-Wach et al., 1996). FcyRIIIa expression has also been shown to be downregulated by the presence of IL-4 (te Velde et al., 1990; W ong et al., 1991).

1.7.7. Fey receptor expression by tissue macrophages As discussed previously, synovial macrophages are activated in the rheumatoid synovium. It is possible that macrophage activation and the subsequent production of damaging cytokines may be occurring in response to a small IgG rheumatoid factor complex following its binding to a surface immunoglobulin receptor.

Tissue macrophages constitutively express FcyRII (van de Winkel & Capel, 1993) and some tissue macrophages express FcyRI at levels equal to, or greater than blood monocytes (Anderson CL et al., 1990). However, FcyRIIIa expression is highly restricted. Both alveolar and peritoneal macrophages from normal individuals have been shown to express FcyRIII (Anderson CL et a l, 1990; Kindt et al., 1991), later confirmed to be the transm em brane isoform, FcyRIIIa (Levy et a l, 1991). Broker et al. (1990) investigated FcyR distribution in the synovium from patients with either rheumatoid arthritis or osteoarthritis. In both cases FcyRIIIa was found to be strongly expressed by synovial intimal macrophages. However, it was not clear whether this was constitutive or secondary to a local inflammatory response. More recently, studies on the expression of FcyR in norm al tissues have dem onstrated that high levels of FcyRIIIa are expressed by synovial intimal macrophages (Edwards et al., 1997a; Bhatia et al., 1998). Interestingly, in fetal limbs, FcyRIIIa was only expressed in the synovial intima (Edwards et a l, 1997a). Furtherm ore, expression of FcyRIIIa by macrophages in other norm al tissues (Bhatia et a l, 1998) appears to correlate closely with the localisation of extra-

65 articular features (Maini & Zvaifler, 1998) in patients with rheumatoid arthritis (Table 1).

Table 1. The distribution of FcyRIIIa expression by norm al tissue macrophages correlates with the pattern of systemic disease seen in patients with rheumatoid arthritis. Taken from Bhatia et a l, 1998.

Sites of FcyRIIIa expression Systemic disease in RA

Synovium Synovitis (Intimal macrophages) Lung Alveolitis (Alveolar macrophages) Pericardium Pericarditis

Liver Hepatic dysfunction (Kupffer cells) Acute phase response Dermis under mechanical stress Subcutaneous nodules (Palisading macrophages) Bone marrow A naem ia (Suppressed haematopoiesis)

These findings provide strong evidence for a role for FcyRIIIa in macrophage activation in rheumatoid arthritis. If either FcyRI or FcyRII were to be the dominant receptor involved in the binding of pathogenic IgG rheumatoid factor immune complexes in rheumatoid arthritis, then one would expect to see evidence of pathology in a less restrictive pattern. In theory, a small IgG rheumatoid factor complex would be able to activate any blood monocyte or tissue macrophage at any site. Instead, only a few specific tissues are affected in patients with rheumatoid arthritis. Furthermore, FcyRI is expressed very weakly in normal synovium, being limited to the subintima (Edwards et al., 1997a; Bhatia et a l, 1998). FcyRI has a high affinity for IgG and its saturation with monomeric IgG at concentrations less than physiological levels can inhibit the binding of small IgG oligomers (Ceuppens et a l, 1988; Tax & van de Winkel, 1990). Only a large immune complex would have the

66 ability to displace any monomeric IgG bound by FcyRI (Tax & van de Winkel, 1990). FcyRII is functionally associated with large immune complexes (Huizinga et al., 1989a). FcyRIIIa has higher affinity than FcyRII for aggregated IgG and unlike FcyRII, FcyRIIIa can bind m onom eric IgG (Vance et al., 1993). Furthermore, neutrophil studies have shown FcyRIII to be the dominant receptor for the binding of dimeric immune complexes (Klaassen et al., 1988; Huizinga et al., 1989b). Therefore, it appears that FcyRIIIa may be the likely candidate for the selective binding of small IgG rheumatoid factor complexes, and owing to its restricted distribution, may explain tissue targeting in rheumatoid arthritis.

1.8. A novel hypothesis for the pathogenesis of rheumatoid arthritis From the evidence discussed so far and a number of other observations which will also be discussed, a hypothetical model for the pathogenesis of rheumatoid arthritis was constructed (Edwards & Cambridge, 1998).

1.8.1. Effector mechanism for the initiation of inflammation in rheumatoid arthritis The initiation of an inflammatory response within the synovium may occur as follows (Figure 5). Intermediate IgG rheumatoid factor complexes, owing to their small size, may evade clearance through complement receptor 1 (CR1) on erythrocytes (Tausk & Gigli, 1990; Wong, 1990), and therefore may persist within the circulation. In contrast, IgM-rheumatoid factor immune complexes may be rapidly cleared from the circulation by anti-phlogistic complement fixation (W alport et al., 1997). The small IgG rheumatoid factor complexes may then, again owing to their size, leave the circulation by crossing endothelium to enter the extravascular space, where they may access tissue macrophages. In the case of the synovium, intimal macrophages express the im m unoglobulin receptor, FcyRIIIa. The small IgG rheum atoid factor complexes may selectively ligate FcyRIIIa. Crosslinking of this receptor may trigger macrophage activation, resulting in the production of pro-inflammatory cytokines and the generation of an intense respiratory burst. The release of both TNFa and reactive oxygen species could then cause local inflammation and tissue injury (reviewed by M oulton, 1996).

67 1.8.2. IgG rheum atoid factor subclass The subclass of the IgG rheumatoid factor may be a major influence on the pathogenicity of these immune complexes, in addition to their size. Recent work has highlighted the importance of both IgG subclass and size on interactions between immune complexes and Fey receptors and complement receptors (Voice & Lachmann, 1997). Complexes of the subclasses IgG2 and IgG4 have been shown not to bind to neutrophils or monocytes, while IgGl and IgG3 dimers bind neutrophils preferentially through FcyRIII (Klaassen et al., 1988; Huizinga et ah, 1989b; Voice & Lachmann, 1997). However, IgGl and IgG3 may differ in the way they interact with this receptor (Morgan et ah, 1995). Furtherm ore, the Fc binding sites on an IgG3 dimer are approximately 200A apart (with the hinges extended), while the Fc binding sites on an IgGl dimer are around 75 A apart (Edwards & Sutton, 1998). Therefore, IgG3 may only dimerise w ith another IgG3 molecule and although IgG3 binds FcyRIII very well, these steric properties may prevent FcyRIIIa signalling as effectively as with IgGl dimers. This may explain the high levels of IgG3 rheumatoid factor dimers in patients with Waldenstrom's hypergammaglobulinemic purpura, with no evidence of any synovitis (Bignold et ah, 1980; Serino et ah, 1983). Therefore the proposed mechanism outlined above may be triggered by small rheumatoid factor complexes of the subclass IgGl.

1.8.3. Chronicity of the initial inflammatory event in rheumatoid arthritis The mechanism outlined in section 1.8.1. describes an effector mechanism for the initiation of inflammation within the synovium. However, for clinical synovitis and eventually joint degradation to manifest symptomatically, this inflammatory process must be a persistent one. In addition to causing local inflammation, TNFa has been shown to act upon synovial stromal cells. Unlike fibroblasts within most other tissues, synovial fibroblasts are particularly responsive to TNFa. As for bone marrow stromal cells, TNFa induces synovial fibroblasts to express vascular cell adhesion molecule-1 (VCAM-1), decay accelerating factor (DAF) and complement receptor 2 (CR2) (Marlor et ah, 1992; Edwards et ah, 1997b). Expression of these molecules is necessary for the survival and differentiation of B lymphocytes. Within germinal centres in normal lymphoid tissues, VCAM-1, DAF and CR2 are co-expressed by

68 follicular dendritic cells (Lampert et a l, 1993; Koopm an et al., 1994; Lui et al., 1997; Fang et al., 1998). The expression of these molecules by fibroblasts within the rheumatoid synovium (Wilkinson et al., 1993; Edwards et al., 1997b) could create a microenviroment capable of supporting the accumulation of B lymphocytes and allow their differentiation into plasma cells. This is seen in most rheumatoid synovia (Dechanet et al., 1995), however, true ectopic germinal centres form in the synovium of only 10 - 20% of patients with rheumatoid arthritis (Gardner, 1992). Some of these accumulating plasma cells within the synovium may produce IgM or IgG rheumatoid factors (Youinou et al., 1984; Carson, 1993). IgG rheumatoid factors generated locally may form large polymers since non-rheumatoid factor IgG levels are low compared with those in the circulation. These IgG rheumatoid factor multimers as well as IgM-rheumatoid factor complexes will have the ability to fix and activate complement in the environment of the synovium, and this may amplify the inflammatory process, originally initiated by the small circulating IgG rheumatoid factor complexes. The formation of IgG rheumatoid factors within the synovium may also augment macrophage activation and the subsequent release of proinflammatory factors by crosslinking FcyRI and FcyRII. Additionally, FcyRIIIa crosslinking results in the release of the chem okine, MCP-1 (Marsh et al., 1997) which may recruit monocytes from the peripheral blood into the synovium.

The question of why there should be a continuous production of IgG rheumatoid factor has not yet been addressed. A recent extension of the hypothesis proposes that IgG rheumatoid factors may have the potential to perpetuate their own existence by bypassing control mechanisms at both the T cell help and germinal centre levels (Edwards et al., 1999). This aspect of the hypothesis is largely outside the context of this dissertation but will be returned to briefly in the discussion.

69 Legend for Figure 5. Schematic diagram outlining the initiating effector mechanism for inflammation in rheumatoid arthritis.

1. The mechanism begins when small IgG rheumatoid factor complexes leave the circulation by crossing endothelium and entering tissues such as the synovium.

2. Here the small IgG rheumatoid factor complexes may access tissue macrophages which are expressing FcyRIIIa in addition to FcyRII and low levels of FcyRI. These complexes may selectively ligate FcyRIIIa.

3. Crosslinking of FcyRIIIa by the small IgG rheumatoid factor complex may activate the macrophage to produce proinflammatory cytokines like TNFa and to generate an intense respiratory burst, causing local tissue injury and inflammation.

4. Additionally, the TNFa may upregulate the expression of certain surface molecules by synovial stromal cells. TNFa induced-expression of VCAM-1, CR2 and DAF by synovial fibroblasts could create a microenviroment that can support B lymphocyte survival.

5. B lymphocytes may accumulate within the synovium and differentiate into plasma cells. Some of these plasma cells may then generate local rheumatoid factor production.

6. Production of both IgM and IgG rheumatoid factors within the synovium may generate large complement fixing polymers which may amplify the inflammatory response. Large IgG rheumatoid factor complexes may also crosslink FcyRI and FcyRII and perpetuate macrophage activation.

70 *

71 1.9. Hypothesis and aims of this thesis

1.9.1. Hypothesis and previous work by other groups The hypothesis of this study is that FcyRIIIa may preferentially ligate small IgG rheumatoid factor immune complexes and mediate the activation of macrophages to produce proinflammatory mediators such as TNFa in patients with rheumatoid arthritis.

The aim of this study was to test this hypothesis in vitro by investigating the ability of human monocyte-derived macrophages to produce pro- inflammatory mediators, such as TNFa, following ligation of each of the three hum an Fc receptors for IgG (FcyR) by small IgG im m une complexes.

FcyR mediated TNFa production by human monocytes in vitro has been previously investigated (Debets et al., 1988; Polat et ah, 1993). However, few studies have focused upon the roles of individual Fey receptors. Work by Debets et al. (1990) studied the ability of FcyRI and FcyRII crosslinking to induce TNFa release from human peripheral blood monocytes. TNFa production through FcyRI was only observed in response to murine IgG2a in solid phase, while FcyRII only evoked a response to murine IgGl in solid phase, following receptor treatment w ith proteolytic enzymes. H um an FcyRIIIa has also been show n to mediate TNFa production following receptor crosslinking with either immune complexes or murine monoclonal antibodies. However, data is only available for natural killer cells (Anegdn et ah, 1988; H endrich et ah, 1991).

72 1.9.2. A im s 1. TNFa production by human monocytes or macrophages following FcyRIIIa ligation had not been previously demonstrated. Therefore, the initial aim of this thesis was to determine whether specific ligation of each of the three FcyR could mediate TNFa production by adherent human monocytes in vitro. This was achieved using murine monoclonal antibodies to each FcyR.

2. To demonstrate the production of other pro-inflammatory mediators by adherent human monocytes following ligation of each of the three Fey receptors using murine monoclonal antibodies. To examine any differential effects following ligation of each Fey receptor type.

3. To determine whether small IgG rheumatoid factor immune complexes could induce TNFa production by adherent human monocytes in vitro and to establish which FcyR were involved. For this purpose, IgG immune complexes of various size and subclass composition, as well as human IgG rheumatoid factor monoclonal antibodies were also used to ligate FcyR.

1.9.3. Experimental approaches Peripheral blood monocytes were used in this project instead of synovial derived macrophages. Synovial macrophages from patients with rheumatoid arthritis spontaneously produce TNFa in culture due to their activation in vivo (Brennan et al., 1989) and normal synovial macrophages were not readily available. To create an in vitro system that was representative of the environment within the synovium, it was decided to use adherent monocytes. Many previous in vitro studies have used non-adherent monocytes and it was felt that this was not an accurate representation of in vivo cell conditions.

Within the synovium and other tissues targeted in patients with rheumatoid arthritis, tissue macrophages express all three FcyR, although within synovium FcyRI expression is often low. Therefore, the in vitro system used in these studies required expression of all three Fey receptors by the adherent human monocytes. Peripheral blood monocytes constitutively express FcyRI and FcyRII, however, FcyRIIIa expression is

73 low if at all evident (Fleit et al. , 1982). A number of studies report that approximately 10% of peripheral blood monocytes express FcyRIIIa (Ziegler-Heitbrock et a l, 1988; Passlick et al., 1989). Consequently, there have been studies investigating the upregulation of FcyRIIIa expression by peripheral blood monocytes in vitro. Fleit et al. (1982) found that after 7 days of culture in Teflon beakers, 15% of peripheral blood monocytes were strongly FcyRIIIa positive. Clarkson et al. (1987) confirmed these observations and showed that by day 10 of culturing in Teflon plates, 100% of monocytes expressed FcyRIIIa. W ork by Klaassen et al. (1990) demonstrated that when peripheral blood monocytes were adhered to plastic and cultured, FcyRIIIa expression was also upregulated with time and that levels of expression plateaued after only 3 days in culture. This later system was chosen for the studies in this thesis since FcyRIIIa expression could be induced quickly on adherent peripheral blood monocytes, without the need to use exogenous cytokines.

74 CHAPTER 2

MATERIALS AND METHODS

75 2.1. Primary cells and hybridomas used in these studies

2.1.1 Primary cells The only primary cell cultures used in these studies were of human peripheral blood monocytes, isolated from whole blood.

2.1.2. Mouse cell lines The mouse cell line 3G8 which was a kind gift from Professor MW Fanger (Department of Immunology, Dartmouth College, Lebanon, NH) produced a m urine IgGl anti-hum an FcyRIII mAb.

2.1.3. Human cell lines The two hybridomas, Cl and C2 were generous gifts from Professor PP Chen (Department of Medicine, University College of Los Angeles, Los Angeles, CA) and Professor DA Carson (Department of Medicine, University of California, San Diego, CA). Both cell lines produced human IgG rheumatoid factor mAb. The Cl clone produced a mAb of the isotype IgG2, while C2 produced an IgGl rheum atoid factor mAb.

2.2. Culturing of primary cells and continuous cell lines All primary cells and hybridomas were cultured in complete growth medium (CGM) which constituted modified medium (MM) (RPMI-1640 with 25mM HEPES; Gibco BRL, Paisley, Scotland) supplemented with 2(Vg/ml gentamycin, 2mM L-glutamine, lOOU/ml penicillin, lOO^ig/ml streptomycin, 2% non-essential amino acids, ImM sodium pyruvate (all from Gibco) and 10% heat inactivated (incubated at 56°C for 30 minutes) fetal calf serum (FCS) (Sigma Chemical Co, Poole, Dorset). All cells were cultured in a humidified incubator set at 37°C in an atmosphere of 5% carbon dioxide in air.

2.3. Murine monoclonal antibodies used in these studies

2.3.1. Anti-FcyRI monoclonal antibodies The anti-FcyRI mAb, designated 10.1, was a kind gift from Dr N Hogg (Imperial Cancer Research Fund, London). This murine mAb was originally derived from a fusion of Sp2/0-Agl4 cells with spleen cells from a BALB/ c mouse previously immunised with rheumatoid synovial

76 fluid cells and purified human monocytes (Dougherty et al.r 1987). The mAb 10.1, was found to be a murine IgGl by double immunodiffusion in agar using murine class and subclass specific antisera.

The specificity of this monoclonal antibody for the high affinity receptor FcyRI, was determined following the immunoprecipitation of a 71-kDa protein by 10.1 from 125Iodine (I) labelled U937 cells (a hum an monocytic cell line). This was supported by indirect immunofluorescent staining and flow cytometry of human leucocytes and cell lines. It was found that 10.1 reacted with a surface molecule expressed by human monocytes, U937 cells and HL60 cells (a human promyelocte cell line) and a small number of human neutrophil samples.

Studies of the inhibition of rosette formation between U937 cells and opsonised human red blood cells by 10.1 suggested that this monoclonal antibody bound to an epitope located near to, or within, the ligand binding site of FcyRI. Monoclonal antibody, 10.1 also inhibited murine IgG2a anti-CD3 induced T cell proliferation. However, the binding of monomeric human IgG to U937 cells was not inhibited by 10.1.

The mAb 32.2 was a kind gift from Professor PM Lydyard (Department of Immunology, University College London, London). This murine anti- FcyRI mAb was derived from a fusion between the NS-1 myeloma cell line and splenocytes from a mouse previously immunised with partially purified FcR from U937 cells (Anderson et al.r 1986). The mAb 32.2 is a murine IgGl as determined by immunoblot assays using isotype-specific antisera. The specificity of this antibody for FcyRI was confirmed by the ability of intact 32.2 and also its Fab fragments to immunoprecipitate a 72- kDa molecule from radioiodinated U937 cells. Indirect immunofluorescent staining and flow cytometry of human leucocytes and cells lines showed the mAb 32.2 to only bind FcyRI bearing cells (human monocytes, U937s, HL60s and some neutrophil samples). The mAb 32.2 did not react w ith m urine FcRI or FcRII.

77 Blocking experiments using human IgG demonstrated the mAb 32.2 to bind an epitope distinct from the ligand binding site of FcyRI. Therefore,, the mAb 32.2 did not interfere with ligand binding by the receptor, nor did ligand binding interfere with mAb 32.2 binding.

2.3.2. Anti-FcyRII monoclonal antibodies The murine IgG2b anti-FcyRII mAb named IV.3 was a kind gift from Professor MW Fanger. This murine mAb was generated from a fusion of a BALB/ c m yelom a cell line (P3. x 63.Ag8.653) w ith spleen cells from a CAF1 mouse previously immunised with the human erythroblastic cell line K562 (Looney et ah, 1986b).

The specificity of this monoclonal antibody for FcyRII was determined by the immunoprecipitation of a 40-kDa surface molecule by IV.3 from both K562 and U937 cells. The complete inhibition of FcR mediated rosette formation between human 0+ erythrocytes opsonised with human IgG anti-Rh and U937 cells by IV.3 suggested that the mAb was binding to an epitope within the ligand binding site of FcyRII. This was confirmed by the inhibitory effect of IV.3 on FcyR mediated superoxide production by human neutrophils (Looney et al ., 1986b).

2.3.3. Anti-FcyRUI monoclonal antibodies The anti-FcyRIII mAb (3G8) producing cell line was cultured to generate supernatants from which the antibody could be purified for F(ab ')2 and Fab production. Purified 3G8 was purchased from Immunotech (Marseilles, France) to be used for all whole antibody experiments. Purified 3G8 was also purchased from PharMingen (San Diego, CA) for quality control. Endotoxin concentrations were <0.01ng/pg protein, as determined by Lim ulus amebocyte lysate assay.

The mAb, 3G8 was originally raised following the fusion of the P3U1 myeloma cell line with spleen cells from CD 2 F mice previously immunised with human PMN (Fleit et ah, 1982). The mAb, 3G8 was typed as a murine IgGl by Ouchterlony analysis and did not react with m ouse FcR.

78 Using the immunofluorescent staining of human leucocytes and cell lines, 3G8 was found to react with all PMN, all eosinophils and a small population of peroxidase-positive lymphocytes (—16%). All monocytes, U937 cells and HL60 cells and peroxidase-negative lymphocytes were negative for 3G8 binding. The specificity of 3G8 for FcyRIII was confirmed by ability of the mAb to immunoprecipitate a molecule ranging from 53 to 66-kDa from 125I labelled PMN.

2.3.4. Anti-TNF monoclonal antibodies The murine IgGl anti-human TNFa mAb (CB006) was a kind gift from Celltech Therapeutics Ltd (Slough, Berkshire). This mAb blocks the biological activity of human TNFa (Charpentier et al.f 1992) by binding an epitope involved in TNF receptor binding (Dr S Stephens, personal communication).

2.4. Human monoclonal antibodies used in these studies

2.4.1. IgG rheumatoid factor monoclonal antibodies The cell lines Cl and C2 were cultured to produce supernatants from which human IgGl (C2) or IgG2 (Cl) rheumatoid factor mAb could be purified. These IgG rheumatoid factor secreting hybridomas were generated following the fusion of synovial fluid mononuclear cells from the knee of a patient with seropositive rheumatoid arthritis (less that 1 year duration) with the heterohybridoma cell line, K6H6/B5 (Lu et al., 1993). The products of the cell lines Cl and C2 were typed as IgG2 and IgGl respectively by enzyme linked im m unosorbant assay (ELISA). Characterisation by ELISA also showed the products of both cell lines to have rheumatoid factor activity and were monospecific since there was no significant binding to human collagen IV, bovine serum albumin, chicken ovalbumin, key-holelimpet hemocyanin, tetanus toxoid or calf thymus single stranded DNA. Fast protein liquid chromatography (FPLC) demonstrated the ability of these human IgG rheumatoid factor mAb to self-associate (Lu et al., 1992). Molecular characterisation of these two mAb showed that both possess X light chains. The IgGl mAb (C2) used the heavy chain variable gene family Vh4.11 and the IgG2 mAb (Cl) used a light chain variable gene from a new VX family designated VX9.

79 2.4.2. IgG anti-NIP monoclonal antibodies Purified chimeric anti-5-iodo-4-hydroxy-3-nitrophenacetyl (NIP) monoclonal antibodies were a generous gift from Professor PJ Lachmann (Medical Research Council Centre, Cambridge University, Cambridge) and Dr J Voice (Department of Immunology, Scripps Research Institute, San Diego, CA). These chimeric mAb (containing mouse variable regions and human constant regions) of each IgG subclass were raised against the hapten, 4-hydroxy-3-nitrophenacetyl (NP) (Briiggemann et al., 1987) and were produced by the following cell lines: THG1-24 (IgGl), JW183/5/1 (IgG2), TH3-MP-2-19-3-8 (IgG3) and JW 184/2/1 (IgG4). Each of these monoclonal antibodies were purified by affinity chromatography using a NIP-caproate-o-succinimide Sepharose column. Characterisation of these mAb with respect to their serology, mobility in sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and ability to bind Protein A showed their behaviour to be identical to their human counterparts (Briiggemann et al., 1987). Each of the IgG anti-NIP mAb were checked by ELISA against the four human IgG subclasses and mAb purity were confirmed by SDS-PAGE (Voice & Lachmann, 1997).

2.5. Purification of monoclonal antibodies by affinity chromatography

2.5.1. Purification of murine monoclonal antibodies Cell-free culture supernatants containing the anti-FcyRIII mAb (3G8) were pooled, adjusted to pH8 with 1M sodium hydroxide (NaOH) and filtered (pore size 0.2fim) (Gelman, Portsmouth). The supernatant was then concentrated under pressure (Nitrogen at 15 pounds per square inch/1 kilogram per centimetre squared) with continuous stirring, in an Amicon concentrator over a Diaflow ultrafiltration membrane with a molecular weight (MW) cut-off at 30-kDa (Millipore, Watford, Hertfordshire). Both the concentrated supernatant and the filtrate were collected and tested for protein concentration and total murine IgG content according to the protocols described later in this chapter.

The anti-FcyRIII mAb was then purified from the concentrated supernatant using affinity chromatography. Briefly, the supernatant was loaded onto a 2 ml Protein G-Sepharose 4B column (Sigma) previously equilibrated with phosphate buffered solution (PBS, pH8) at 4°C.

80 Following washing with PBS, the mAb was eluted with one column volume of 3M magnesium chloride (MgCl) and dialysed into PBS. The purified mAb was then concentrated using a centricon 30 with a MW cut­ off at 30-kDa (Millipore). The concentrated monoclonal antibody was then filtered, aliquoted and stored at -70° C. The protein and murine IgG concentrations were then tested.

2.5.2. Preparation ofF(ab ')2 fragments of the anti-FcyRIII mA b Preparation of anti-FcyRIII F(ab ’)2 fragments was performed by Cybos Ltd (Southampton). Briefly, the anti-FcyRIII mAb was concentrated to 5mg/ml using a Vivascience Vivapore concentrator and then dialysed into 20mM sodium acetate buffer (pH4.2) overnight at 4°C. Immobilised pepsin (Peirce, Rockford IL) was washed four times in sodium acetate buffer (pH4.2) and 125pl of this was added to the antibody in a glass bottle. The mixture was then incubated at 37°C on an orbital mixer for 12 hours, after which the proteolytic reaction was stopped by microcentrifuging the mixture and removing the supernatant. The digest was purified from the supernatant using a Protein G-Sepharose 4B column. The F(ab’)2 fragments were collected as flow through, dialysed into PBS and concentrated. Purity of the F(ab ’)2 digest was confirmed by SDS-PAGE.

2.5.3. Preparation of Fab fragm en ts of th e anti-FcyRIII mAb Preparation of anti-FcyRIII Fab fragments was performed by Cybos Ltd. The anti-FcyRIII mAb was concentrated to 5mg/ml using a Vivascience Vivapore concentrator and then dialysed into 20mM sodium acetate buffer (pH4.2) overnight at 4°C. Immobilised papain (Peirce) was washed four times in sodium acetate buffer (pH4.2) and 125pl of this was added to the antibody in a glass bottle. The mixture was then incubated at 37°C on an orbital mixer for 12 hours, after which the proteolytic reaction was stopped by microcentrifuging the mixture and removing the supernatant. The digest was purified from the supernatant using a Protein G-Speharose 4B column. The Fab fragments were collected as flow through, dialysed into PBS and concentrated. Purity of the Fab digest was confirmed by SDS-PAGE.

81 2.5.4. Purification of human monoclonal antibodies Cell-free culture supernatants containing either the IgGl (C2) or the IgG2 (Cl) rheumatoid factor mAb were pooled and adjusted to pH 7.4 by adding one volume of 1M phosphate buffer to nine volumes of supernatant. The supernatant was concentrated as described previously for the murine anti-FcyRIII mAb. Both IgG-RF mAb were purified by affinity chromatography. Briefly, the concentrated supernatant was loaded onto a 2 ml Protein G-Sepharose 4B column previously equilibrated with PBS (pH 7.4) at 4°C. Following washing with PBS, the mAb was eluted with one column volume of 0.1M glycine (pH 2.8) and rapidly concentrated using polyethylene glycol 20,000 at 4°C as a hydroscopic agent. The purified mAb was filtered, aliquoted and then snap frozen after which it was stored at -70°C. Prior to use, the mAb was adjusted to pH 7.4 by adding one volume of 1M phosphate buffer (pH 7.4) to nine volumes of the purified monoclonal antibody. Protein concentration, human IgG concentration and rheumatoid factor activity were tested according to the protocols described below.

2.6. Measurement of protein concentration A modified Bradford assay (Bradford, 1976; Locke, 1994) was used to determine protein concentration. Serial doubling dilutions of bovine serum album in (BSA; Sigma) dissolved in PBS starting from a concentration of 2mg/ ml were used to construct a standard curve. 10^1 of the test solution (in PBS) or BSA standards were pipetted into a 96 well Immulon II plate (Dynex Technologies, Chantilly, VA). All samples were assayed in duplicate. 19CV1 of the modified dye reagent was then added to all wells and the plate left to stand for 5 minutes before absorption was measured at 595nm. A standard curve was constructed by plotting the averaged optical densities of the standards on the y-axis, against concentration on the x-axis. The protein concentration of the test samples was then calculated from the standard curve.

82 2.7. M easurem ent of im m unoglobulin (IgG) concentration Murine and human IgG concentrations were determined by ELISA.

2.7.2. Measurement of murine IgG concentration Immunol II ELISA plates (Dynatech) were coated with F(ab')2 goat anti­ mouse IgG antibody (Fab specific; Sigma) at 1 ng/ml in bicarbonate buffer (BIC) for 1 hour at 37°C. After washing 3 times with PBS/0.1% tween (PBT) the plates were blocked with 200^1 of 2%BSA/PBT at 37°C for 1 hour. Following washing as before, lOOpl of test samples at various dilutions or an isotype-matched murine IgG (Southern Biotechnology Associates, Birmingham, AL) in doubling dilutions from lpg/ml as standards were loaded onto the plates. All dilutions were made in 1% BSA/PBT and all samples were assayed in duplicate. The plates were incubated at 37°C for 1 hour, washed three times with PBT, then 100^1 of alkaline phosphatase conjugated goat anti-mouse IgG (whole molecule specific, Sigma) at l^g/ ml was added to each well and the plates incubated as before. Following washing three times with PBT and three times with BIC, lOOjxl of phosphatase substrate (p-Nitrophenyl phosphate disodium (Sigma) at lmg/ml in BIC with lOmM MgCl) was added to each well. After a 10 minute incubation at 37°C, absorbance was measured at 405nm with a reference of 490nm. A standard curve was constructed by plotting the averaged optical densities of the standards along the y-axis and In concentration on the x-axis. Averaged optical densities of the test samples were read off the standard curve and the IgG concentration calculated.

2.7.2. Measurement of human IgG concentration Immunol II ELISA plates (Dynex) were coated with goat F(ab')2 anti­ hum an IgG (Fc specific; Sigma) at lu g/m l in BIC for 1 hour at 37°C. After washing 3 times with PBT the plates were blocked with 200jil of 2%BSA/PBT at 37°C for 1 hour. Following washing as before, lOOpl of test samples at various dilutions or human IgG (Sigma) in doubling dilutions from lpg/ml as standards were loaded onto the plates. All dilutions were made in 1% BSA/PBT and all samples were assayed in duplicate. The plates were incubated at 37°C for 1 hour, washed three times with PBT, then lOOpl of alkaline phosphatase conjugated goat anti­ hum an IgG (y chain specific; Sigma) at l\xg/ml was added to each well and the plates incubated as before. Following washing three times with PBT

83 and three times with BIC, lOOpl of phosphatase substrate was added to each well. After a 10 minute incubation at 37°C, absorbance was measured at 405nm with a reference of 490nm. Averaged optical densities of the standards were plotted along the y-axis and In concentration plotted on the x-axis. Averaged optical densities of the test samples were read off the standard curve and the IgG concentration calculated.

2.8. Measurement of rheumatoid factor activity To determine rheumatoid factor activity of the purified human IgG rheumatoid factor mAb (Cl and C2), the ability of these lambda (X) chain containing monoclonal antibodies to bind human IgG Fc was determined by a modification of an ELISA described previously (Lu et al., 1992). Briefly, Im m unol II plates were coated w ith Fc fragments of hum an IgG (Jackson Immunoresearch Laboratories inc., Westgrove, PA) at 50pg/ml in BIC. The plate was incubated at 37°C for 1 hour and then washed three times with PBT. The plates were then blocked by incubating at 37°C for 1 hour with 200pl of 4% BSA/PBT. Following washing as before, 100^1 of the purified IgG-RF mAb samples diluted to known concentrations were loaded onto the plate. A X chain containing human IgGl or IgG2 myeloma protein (The Binding Site, Birmingham) was used as a negative control and a X chain containing rheumatoid factor (Cambridge Life Science, Cambridge) was used as a positive control. All controls were assayed at concentrations equal to the test monoclonal antibody. All dilutions were made in 4%BSA/PBT and all samples were assayed in duplicate. The plates were incubated at 37°C for 1 hour and then washed as before. lOOpl of an alkaline phosphatase conjugated goat anti-human X light chain antibody (bound and free; Sigma) was added to each well at lpg/ml and the plates incubated as before. Following washing three times with PBT and three times with BIC, lOOpl of phosphatase substrate was added to each well. After a 1 hour incubation at 37° C, absorbance was measured at 405nm with a reference of 490nm. Rheumatoid factor activity of the IgG-RF mAb was expressed as a percentage of the positive control.

84 2.9. Immune complexes used in these studies

2.9.2. Measurement of human IgG rheumatoid factor mAh immune complex size by analytical high-performance liquid chromatography High-performance liquid chromatography (HPLC) was used to measure the molecular weights of immune complexes consisting of the IgGl (C2) or the IgG2 (Cl) rheumatoid factor monoclonal antibodies. The purified IgGl or IgG2 rheumatoid factor mAb was adjusted to a concentration of lmg/ml in PBS and was then passed through a Biosep size exclusion column S3000 (Phenomenex inc, Torrance, CA) which had been pre­ equilibrated with 0.5M Tris/0.25M NaCl (pH 7.2). A X chain containing human IgGl or IgG2 myeloma protein (The Binding Site) was also analysed as a control. The column was set to run at 1ml/min and protein was detected at an absorbance of 280nm. Previously the column was calibrated using markers of known molecular weights (Sigma). The data from these calibration markers was used to construct a standard curve from which elution times of the IgG-RF mAb could be read to obtain molecular weights of the IgG-RF containing immune complexes (Appendix 2).

2.9.2. Separation of human IgG rheumatoid factor mAb immune complexes by preparative high performance liquid chromatography To obtain purified IgG-RF mAb containing immune complexes of a known size, HPLC was performed as described above. However, as the IgG-RF mAb complexes passed through the column, fractions were collected every 20 seconds. The time at which each fraction was collected was read off the calibration curve and the size of the sample could be determined. Protein concentration of each fraction was determined by measuring absorbance at 280nm. Approximate immunoglobulin concentrations were determined by dividing this absorbance reading by 1.4 for an immunoglobulin concentration expressed as mg /ml. All fractions were filtered and where necessary, concentrated using a centricon 30 with a MW cut-off at 30-kDa (Millipore). Unless stated otherwise, the fractions were placed on ice and used in further experiments on the same day as preparation. Prior to the incubation of complexes with cell cultures, the fractions were dialysed into PBS.

85 2.9.3. Preparation of human IgG anti-NIP mAb immune complexes Immune complexes were formed between human IgG anti-NIP mAb of each IgG subclass and NIP conjugated to BSA. Both were generous gifts from Professor PJ Lachmann and Dr J Voice. The NIP conjugated BSA used in these studies had been previously prepared (Voice & Lachmann, 1997) and stored at -70°C. Briefly, after coupling NlP-caproate-O- succinimide to the BSA at various concentrations, any free NIP was removed and the average molar ratios of NIPrBSA determined. The NIP:BSA molar ratios used in these experiments were 2.4NIP:1BSA, 8.9NIP:1BSA and 21.5NIP:1BSA where the number of NIP molecules per BSA will be referred to as the epitope density of the antigen.

Immune complexes were prepared as described previously (Crockett- Torabi & Fantone, 1990; Voice & Lachmann, 1997). Briefly, the IgG anti- NIP mAb of a selected subclass was diluted to 200^g/ml in Hank's balanced salt solution without phenol red (HBSS; Gibco). The antibody was then mixed with the antigen (NIP/BSA) at a selected epitope density and concentration such that complexes were prepared in either a molar excess of antigen, a molar excess of antibody or in an antigen:antibody molar ratio where the epitope density was equivalent to the number of antibody molecules (referred to as equivalence). The antibody/antigen preparations were incubated at 37°C for 1 hour and then cooled on ice for 20 minutes. The immune complexes were used in further experiments on the same day as preparation.

2.10. Source of human peripheral blood monocytes Peripheral blood monocytes constitute only 10% of circulating leucocytes (Ziegler-Heitbrock et al., 1988). Therefore, to consistently obtain a highly purified monocyte preparation, large quantities of whole blood were required on a regular basis. Whole blood was generously supplied by both Haematology Day Care (PPW 2) and Haematology Outpatient Departments at University College Hospital (London). Venous blood (300-500ml) was collected from patients with either polycythaemia or haemachromatosis undergoing therapeutic bloodletting, into sterile wet packs (Baxter Healthcare Ltd, Thetford, Norfolk) containing 63ml citrate phosphate dextrose adenine 1 anti-coagulant solution. The majority of blood used in these studies was collected from patients with

86 haemachromatosis. Blood was stored at 4°C and used within 24 hours of collection.

Although monocyte function has not been examined in patients with haemachromatosis, there have been reports that both neutrophils and monocytes from patients with polycythaemia display abnormally low reactive oxygen species production in response to formyl-methionyl- leucyl-phenylalanine. However, both resting and phorbol-12-myristate 13-acetate (PMA) stimulated neutrophils and monocytes displayed normal free radical production (Samuelsson et ah, 1994). Therefore, for all chemiluminescence studies, monocytes were isolated from patients with haemachromatosis.

Whole blood and isolated monocytes from normal donors were also used in a small number of experiments including FACS analysis of whole blood, chemiluminescent studies and some titrations of reagents used to study the production of cytokines in vitro. These experiments were used to corroborate results from similar experiments performed on whole blood and monocytes from patients with polycythaemia and haemachromatosis.

2.10.1. Isolation of human peripheral blood monocytes from whole blood Nine volumes of whole blood was mixed with one volume of 0.9% sodium chloride/ 6 % dextran to sedimentate erythrocytes. One volume of buffy coat was then layered onto 2 volumes of Histopaque (density 1.077 g/m l; Sigma) and centrifuged at 400g for 30 m inutes at room temperature. The mononuclear cell layer lying at the plasma/Histopaque interface was removed and diluted 1:1 with MM. Following mixing by inversion, the cell suspension was centrifuged at 250g for 10 minutes at room temperature. The supernatant was discarded, the pellet resuspended in MM and the suspension centrifuged at 100# for 10 minutes at room temperature. This was repeated another five times to remove contaminating platelets. The mononuclear cell pellet was then resuspended in MM supplemented with 10% heat inactivated FCS to a concentration of approximately 5 x 10 6 cells/ml. The cell suspension was transferred into 100mm plastic petri dishes (Bibby Sterilin Ltd, Staffordshire) and incubated at 37°C for 2 hours. Nonadherent cells were

87 removed by vigorously washing six times with warmed HBSS. Adhered cells were gently harvested using a rubber policeman, resuspended in CGM and the concentration adjusted to 5 x 10 5 cells/ml (unless stated otherwise).

Monocyte purity was determined by indirect immunofluorescent staining and flow cytometry for CD14 and CD3 expression and was shown to be greater than 95%. CD14 and CD3 expression were determ ined using the murine mAb TuK4 and UCHT1 respectively (DAKO, Ely, Cambridge). Trypan blue (Sigma) exclusion showed monocyte viability to be greater than 98%. Monocyte purity and viability were checked routinely throughout the study to ensure high levels of both were maintained.

2.11. Immunofluorescent staining and flow cytometry

2.11.1. Indirect immunofluorescent staining of whole blood Indirect immunofluorescent staining of whole blood was performed as described previously (Maeda et a l, 1996) with some modifications. Briefly, 2ml of whole blood was diluted 1:5 with heparinised (10 units of heparin/ml) staining buffer (1% BSA and 0.1% azide in PBS) and centrifuged at 250g for 5 minutes at 4°C. The supernatant was discarded and the wash repeated. The pellet (plasma-free blood) was then made up to a total volume of 2 ml with staining buffer.

25pl of the test antibody or isotype-matched control (Table 2) at final concentrations as indicated by each experiment, was mixed with lOOpl of plasma-free blood and incubated on ice for 45 minutes. 150pl of ice cold staining buffer was added to each sample and mixed by gentle pipetting. The samples were then centrifuged in a bench top microfuge at 1850# for 5 minutes and the supernatants discarded. This wash was repeated and the pellets resuspended with 25^1 of fluorescein isothiocynate (FITC) conjugated goat F(ab ')2 anti-mouse Ig ( 1 :2 0 dilution; DAKO). Following incubation on ice for 45 minutes in the dark and then washing twice as before, the cell pellets were resuspended in 2ml of lysing buffer (Becton Dickenson, Oxford). After mixing by inversion the samples were incubated at room temperature for 15 minutes in the dark, washed twice w ith 2 ml of cold staining buffer and the erythrocyte-free pellets then

88 resuspended in 250^1 of 2% paraformaldehyde in PBS to fix the cells. The samples were stored at 4°C in the dark before being analysed by flow cytometry.

2.11.2. Indirect immunfluore scent staining of isolated monocytes lm l of freshly isolated human monocytes (lxl0 6 /ml) were seeded into 24 well plastic tissue culture plates (Becton Dickenson) and allowed to adhere at 37°C for 24 hours (unless stated otherwise). Prior to staining, the adhered cells were harvested by cooling on ice and gentle pipetting. The monocyte suspension was washed twice with lml of cold staining buffer. The monocyte pellets were then resuspended with 25pl of the test antibody or isotype-matched control (Table 2) at final concentrations as indicated by each experiment, incubated on ice for 45 minutes and then 150pl of ice cold staining buffer added to each sample and mixed by gentle pipetting. The samples were centrifuged in a bench top micofuge at 1850# for 5 minutes and the supernatants discarded. This wash was repeated and the pellets resuspended with 25^1 of FITC conjugated goat F(ab ')2 anti-mouse Ig (1:20, DAKO). The samples were incubated on ice for 45 minutes in the dark, washed twice as before, and the cells were then fixed with 250^,1 of 2% paraformaldehyde in PBS. Samples were kept at 4°C in the dark until analysed.

Table 2 . Primary monoclonal antibodies used in the indirect immunofluorescent staining of whole blood and/or isolated monocytes.

M onoclonal Species Isotype Specificity for antibody human surface antigen UCHT1 m ouse IgGlK CD3 o & CM TiiK4 m ouse £ CD14

1 0 .1 m ouse IgGl FcyRI (CD64) IV.3 m ouse IgG2bK FcyRII (CD32) 3G8 m ouse IgGlK FcyRin (CD16)

The mouse monoclonal antibodies IgGlK (DAK-G01), IgG2ax (DAK-G05) and IgG2bx (DAK-G09) which are all specific for Aspergillus niger glucose oxidase (all from DAKO) were used as isotype-matched controls.

89 Aspergillus niger glucose oxidase is an enzyme which is neither present nor induced in mammalian tissues.

2.11.3. Flow Cytometry Flow cytometry was performed using a Fluorescence Activated Cell Sorter (FACScan; Becton Dickenson). For each sample, 10,000 events were acquired. The instrument settings used for the analysis of whole blood and purified, cultured monocytes are shown in Table 3. The forward light scatter detected cell size while the side 90° light scatter detected cell granularity. The FL-1 channel detected the light emission of the excited fluorochrome, FITC, as fluorescent intensity. For the analysis of whole blood, each cell population was gated on, according to cell size and 90° light scatter. No such gating was necessary for the analysis of isolated monocytes.

Table 3. FACScan instrument settings. a) FACScan settings for analysis of whole blood.

Volts Gains Threshold

Forward scatter 0 1 .0 0 (linear) 50-200 Side scatter 400 1 .0 0 (linear) off FL1 548 log off b) FACScan settings for analysis of isolated monocytes.

Volts Gains Threshold Forward scatter 4 4.76 (linear) 25-200 Side scatter 400 1 .0 0 (linear) off FL1 548 log off

The data was analysed using Win MDI software (Microsoft).

90 2.11.4. Calibration o f the FACScan The FACScan was calibrated using standard fluorescent beads Fluorospheres; DAKO) as described previously (Le-Bouteiller et a l, 1983). These beads are a mixture of 5 populations of 3.2jim microspheres, each with a different fluorescent intensity. Each bead population is covered with a known amount of fluorochrome that can be excited at any wavelength between 364 and 650nm. The laser beam of the FACScan excited the fluorochrome at 488nm and the light detection set at 548nm to calibrate the FL-1 channel for FITC. The mean fluorescent intensities (MFI) for each bead population were measured and using the assigned values for molecules of FITC per bead, a calibration curve was constructed (Appendix 3).

2.11.5. Calculation of the number of Fcyreceptor binding sites per cell The mean fluorescent intensity was directly proportional to the degree of anti-FcyR mAb cellular staining and therefore the levels of Fey receptor expression. Using the MFI's previously obtained from the standard fluorescent calibration beads, the equation below for the number of binding sites per cell could then be solved.

*Receptor binding sites per cell = Test FITC molecules - Control FITC molecules f / p ratio f / p ratio W here : FITC molecules = exp {((In MFI - c)/m)} f/p ratio: E 495 nm/E278 nm = 0-61 ± 0.05 corresponding to a m olar FITC/protein ratio of 2.3

* Receptor binding sites per cell is equivalent to the number of secondary antibody molecules (F(ab ')2 goat anti-mouse Ig) bound per cell (Fanger et a l, 1989), however, by accounting for the f/p ratio, the actual number of FcyR binding sites were calculated (Maeda et a l, 1996).

91 2 .1 2 . Culturing of human monocytes and production of cytokines in vitro 200pl of freshly isolated monocytes were seeded into 96 well plastic tissue culture plates (Becton Dickenson) and incubated at 37°C for 24 hours (unless stated otherwise). The culture supernatants were removed and replaced with fresh CGM.

The adherent human monocytes were then incubated with various monoclonal antibodies (Table 4) at final concentrations as indicated by each experiment, in a total volume of 200pl CGM. As negative controls, cultures were incubated with either isotype-matched controls at concentrations equal to that of the test mAb, or medium alone. As positive controls, cultures were incubated with lipopolysaccharide (LPS) from Escherichia coli 026:B6 (Sigma) in the absence of polymixin B or PMA (Sigma), both at a final concentration of 0.5pg/ml. Possible LPS contamination of reagents was excluded by performing all incubations in the presence of polymixin B (10 pg/ml; Gibco), which consistently abrogated the LPS-induced TNFa response. Since supernatants were to be collected at various time intervals, each culture was prepared in multiples so that each culture corresponded with an individual time point. All cultures were incubated at 37°C and cultures supernatants collected at various time points between 0 and 24 hours (unless stated otherwise). Once collected, the supernatants were centrifuged in a bench top microfuge at 7000g for 5 minutes. Cell-free supernatants were recovered and stored at -70°C.

92 Table 4. Monoclonal antibodies incubated with adhered human monocytes for measurement of cytokine production in vitro.

Antibody Species Isotype Specificity 10 .1 Mouse IgGl H um an FcyRI IV.3 Mouse IgG2bK H um an FcyRII 3G8 Mouse IgGlK H um an FcyRIII F(ab ')2 3G8 Mouse IgGlK H um an FcyRIII Fab 3G8 Mouse IgGlK H um an FcyRIII Cl H um an IgG2k H um an IgG Fc

C2 H um an IgGlX Human IgG Fc THG1-24 Human IgGlX NIP JW183/5/1 H um an IgG2X NIP TH3-MP-2-19-2-8 H um an IgG3k NIP JW184/2/1 Human IgG4X NIP

The mouse monoclonal antibodies IgGlK (MOPC-21; Sigma), IgG2aK (DAK-G05) and IgG2bK (DAK-G09) and the hum an IgGl (k) and IgG2 (X) myeloma proteins (The Binding Site) were used as isotype-matched controls. Where cultures were incubated with preformed complexes of human IgG anti-NIP mAb with NIP/BSA, control cultures were incubated with either the IgG anti-NIP mAb alone or NIP/BSA alone. Where necessary, antibodies were predialysed into PBS to remove any traces of azide.

For the inhibitor studies, cultures were incubated with monoclonal antibodies, LPS or medium (all in the absence of polymixin B) in the presence of cycloheximide (50^g/ml), actinomycin D (2pg/ml) or colchicine (ljig/ml) (all from Sigma). All three inhibitors prevented LPS- induced TNFa release in a dose-dependent manner and were used at concentrations according to the manufacturers specification. A gelatinase-specific MMP inhibitor (1746) was a kind gift from Celltech. This MMP inhibitor prevented LPS-induced TNFa release in a dose dependent manner and had an IC 50 ~15pM. Maximum inhibition was achieved at 50|jM. The presence of the above inhibitors had no effect upon monocyte viability.

93 2.13. Detection and measurement of cytokines in culture supernatants The presence and concentrations of TNFa and IL-la in culture supernatants were determined by ELISA. All assays showed high levels of reproducibility. Additionally, all antibodies and reagents used in determining the production of cytokines were assayed by ELISA in the absence of cells. This was to ensure positive results were not due to any cross-reactivity or interference of reagents with the assay.

2.13.1. ELISA for the detection and measurement of TNFa in culture supernatants A modification of the ELISA described by Charpentier et al. (1992) was used to detect and measure human T N F a in culture supernatants. This ELISA had an upper limit of 2000pg/ml and detected free, soluble TNFa. 96 well maxisorb plates (Gibco) were coated with 100^1 of a murine IgGl anti-human T N F a monoclonal antibody (CB006, Celltech) at 8 [xg/ml in 0.02M N, N-bis [2-Hydroxyethyl]-2-aminoethanesulfonic acid (BES at pH7; Sigma). The plates were incubated at room temperature for 16-20 hours, after which the wells were aspirated and replaced with 200^ 1 of blocking buffer (1% BSA in 0.02M BES at pH7). The plates were incubated at room temperature for 1 hour and the wells were aspirated as before. Unless stated otherwise, all samples were diluted in 2 % normal mouse serum (Seralab, Loughbrough, Leicstershire)/1%BSA/PBS. Prior to loading the samples, 50pl of this diluent was added to all wells. Recombinant human TNFa (National Institute for Biological Standards and Control (NIBSC), Potters Bar, Hertfordshire) standards in doubling dilutions from 2000pg/ml were used to calibrate the assay. Three concentrations of hum an T N F a (600, 200 and 60pg/ml) diluted in CGM were also used as controls. 50pl of the standards, controls or the undiluted culture supernatants were then loaded onto the plates in duplicate. The plates were incubated for 1 hour at room temperature with continuous agitation using a microplate shaker (Camlab, Cambridge) set at 800rpm. Following aspiration of the wells, the captured T N F a was detected by incubating with 100nl of a sheep polyclonal anti-human TNF (Celltech) at a 1:10,000 dilution and the plates incubated as before. Following aspiration, lOOpl of horseradish peroxidase (HRP) conjugated anti-sheep IgG (H & L) (Jackson Immunoresearch Laboratories) was added to each well at 0.4pg/ml. The plates were incubated for 30 minutes at room

94 temperature with continuous agitation. After four washes with PBS, 100^1 of peroxidase substrate (100^ig/ml 3,3',5,5'-tetramethylbenzidine (Sigma) in citrate buffer with 0.05% H 2O2) was added to all wells. The plates were incubated at room temperature with agitation for 15-30 minutes. The assay was then stopped by adding 50^1 of 2M sulphuric acid to all wells and the absorbance was measured at 450nm with a reference of 570nm. A standard curve was constructed by plotting the averaged optical densities of the TNFa standards on the y-axis and the concentration along the x-axis. The equation of the standard curve was then used to calculate the TNFa concentration in the culture supernatants from averaged optical densities.

2.13.2. Alternative ELISA for the detection and measurement of TNFa in culture supernatants. When TNFa concentrations were tested in supernatants from cultures incubated with either the IgGl or IgG2 rheumatoid factor mAb, high optical densities were observed. Similarly, when the IgG rheumatoid factor mAb were cultured in the absence of monocytes and the supernatants then tested for TNFa using the above ELISA, comparable optical densities were recorded (Appendix 4). These false positive readings suggested that the IgG rheumatoid factor mAb was interfering or cross-reacting w ith the T N Fa ELISA. Therefore, for experiments where monocytes were incubated with either the IgGl or IgG2 rheumatoid factor mAb, an alternative ELISA with an upper limit of 2000pg/ml was used to detect TNFa in culture supernatants.

Briefly, 96 well maxisorb plates (Gibco) were coated with murine IgGl anti-human T N F a mAb (MAbl; PharMingen) at 2pg/ml in BIC and incubated overnight at 4°C. Following aspiration of the wells, the plates were blocked with 200pl of 1%BSA in PBS. The plates were incubated at room temperature for 1 hour and then washed three times with PBT. Unless stated otherwise, all samples were diluted in 1%BSA/PBT. Recombinant human T N F a (NIBSC) standards in doubling dilutions from 2000pg/ml were used to calibrate the assay. Three concentrations of hum an T N F a (600, 200 and 60pg/ml) diluted in CGM were also used as controls. 95^1 of the standards, controls, or the undiluted culture supernatants were then loaded onto the plates in duplicate. The plates

95 were incubated overnight at 4°C and then washed as before. The captured TNFa was detected by incubating all wells with lOOpl of biotin conjugated mouse anti-human TNFa mAb (MAB 11; PharMingen) at l^g/m lfor 1 hour at room temperature. The plates were washed 4 times with PBT and lOOpl of HRP-conjugated streptavidin (1:500 dilution; DAKO) was then added to all wells. The plates were incubated for 30 minutes at room temperature and then washed 5 times with PBT, after which lOOpl of peroxidase substrate was added to each well. Following incubation at room temperature for 1 hour, the reaction was stopped by adding 50pl of 2M sulphuric acid to all wells. Absorbance was measured at 450nm with a reference of 570nm. Averaged optical densities of the TNFa standards were plotted on the y-axis and concentration plotted along the x-axis. The equation of this standard curve was used to calculate the TNFa concentration in the culture supernatants from averaged optical densities.

2.13.3. ELISA for the detection and measurement of IL-la in culture supernatants The Quantikine™ IL-la ELISA (R&D Systems, Oxfordshire) with an upper limit of 250pg/ml and a lower limit of less than 0.5pg/ml was carried out according to the manufacturers protocol. Briefly, 95pl of undiluted culture supernatants were loaded onto 96 well maxisorb plates precoated with a murine anti-human IL-la mAb. 95pl of recombinant human IL-la in serial doubling dilutions from 250pg/ml were also assayed to construct a standard curve. All samples were assayed in duplicate. The plates were incubated at room temperature for 2 hours and then washed 3 times with PBT. 200pl of HRP-conjugated anti-IL-la polyclonal antibody was added to all wells and the plates incubated at room temperature for 1 hour. After washing as before, 200 (la.1 of peroxidase substrate was added to each well and the plates incubated at room temperature for 20 minutes. The reaction was stopped by adding 50|xl of 2M sulphuric acid to all wells. Absorbance was measured at 450nm with a wavelength correction of 570nm. Averaged optical densities of the IL-la standards were plotted on the y-axis and concentration plotted on the x-axis. The equation of this standard curve was used to calculate the IL-la concentration in the culture supernatants from averaged optical densities.

96 2.14. Detection and measurement of reactive oxygen species production by adhered human monocytes in vitro Reactive oxygen species production was measured by chemiluminescence as described previously (Assreuy et al ., 1994; Epperlein et a l, 1998). The chemiluminescent probe lucigenin (bis-JV- methylacridinium nitrate) could be used for the detection and measurement of reactive oxygen species. However, lucigenin detects products of the NADPH-oxidase system such as superoxide and also possibly other oxidants such as nitric oxide (Davies & Edwards, 1992). The chemiluminescent probe luminol (3-aminophthalhydrazide), although largely MPO dependent (McNally & Bell, 1996) detects products from both the MPO system and NADPH-oxidase activation (Johansson & Dahlgren, 1989). Therefore, in these experiments reactive oxygen species production was measured by luminol-dependent chemiluminescence (Allen et al ., 1972; Allen & Loose, 1976).

Chemiluminescence was measured with a luminometer (Dr A. Nurouho-Dutra, Department of Nephrology, University College London, London) which used a gallium-arseride photomultiplier tube cooled to - 20°C with a linear counting efficiency curve above 10% (most methods have an efficiency of < 0.01%) in the w avelength range of 200 - 900nm. The signal emitted from the tube was fed directly into a frequency counter unit (20 MHz) and the data were presented as average counts per second (cps). 1000 cps represented 0.2 nM of reactive oxygen species per second. lml of freshly isolated monocytes were seeded into 35mm plastic petri dishes (Corning, Buckinghamshire) and incubated at 37°C for 24 hours. The cells were then carefully washed with warm HBSS to ensure all traces of serum had been removed. 900pl of HBSS / lOmM HEPES (pH 7.2) was added to the adherent monocytes followed by 20pl of 1M luminol (Molecular Probes Europe BV, Leiden, Holland). The petri dishes were then covered with cling film and placed into one of four 37°C light tight mirrored counting chambers. The dishes were left in the chambers for 2 minutes, allowing the temperature of the cells to equilibrate and background light emission to be measured. The dishes were then removed from the chambers and lOOjid of the anti-FcyR mAb (Table 5) or

97 isotype-matched control added at final concentrations as indicated by each experiment. As a positive control, cultures were incubated with lOOpl of PM A (0.5pg/ml). All dishes were then quickly returned to the counting chambers and chemiluminescence measured in real time. Polymixin B was not used in these experiments since there were uncertainties of whether this chemical would somehow interfere with the chemiluminescent probe, luminol.

Table 5. Monoclonal antibodies incubated with adhered human monocytes for the measurement of reactive oxygen species production in vitro.

Antibody Species Isotype Specificity 1 0 .1 Mouse IgGl H um an FcyRI 32.2 Mouse IgGl H um an FcyRI IV.3 Mouse IgG2bK H um an FcyRII 3G8 Mouse IgGlK H um an FcyRIII

3G8 F(ab ')2 Mouse . IgGlK H um an FcyRIII

The mouse monoclonal antibodies, IgGlK (DAK-G01) and IgG2bx (DAK- G09) were used as isotype-matched controls.

Where necessary, antibodies were predialysed into PBS to remove any traces of azide which is an inhibitor of MPO activity (Johansson & Dahlgren, 1989).

2.15. Statistical analysis Significance was determined using unpaired T-test with Fisher's correction for small sample size.

2.16. Data representation Results have been presented as either the mean ± standard deviation (S.D.) of three experiments, or as data from one representative experiment.

98 CHAPTER 3

RESULTS

99 3.1. Fey receptor expression by human peripheral blood monocytes

3.1.1. Murine anti-FcyR mAb: confirmation o f antibody specificity and determination of optimal binding levels

Experimental design Before the murine anti-FcyR monoclonal antibodies could be used for the detection of surface FcyR on human leucocytes, the specificity of each antibody needed to be verified and optimal binding concentrations determined. This was achieved by the indirect immunofluorescent staining and flow cytometry of leucocytes in whole blood, according to the protocol described in materials and methods (2.11.1.). Plasma-free blood was stained with either the anti-FcyR! mAb (10.1), anti-FcyRII mAb (IV .3), anti-FcyRIII mAb (3G8) or murine isotype-matched controls at a final concentration of 50, 2 0 , 1 0 ,5 , 2 or lp g / ml.

Results Figure 6 shows the FACS profile of whole blood following erythrocyte-lysis, with individual gates around the lymphocyte (L), monocyte (M) and granulocyte (G) cell populations. Similar FACS profiles were observed in 5 separate repeat experiments with whole blood from different donors.

The levels of cellular staining were obtained for each of the three anti-FcyR mAb at each concentration. Mean fluorescent intensities for the anti-FcyRI and anti-FcyRII mAb were obtained by gating on the monocyte population. Since monocytes express little if any FcyRIfl (Fleit et al., 1982; Ziegler- Heitbrock et al, 1988), levels of cell staining for the anti-FcyRIII mAb were acquired by gating on the granulocyte population. All MFI were corrected for non-specific binding by subtracting the fluorescent intensity from cells stained with the isotype-matched controls. As shown in Table 6 , the optimal binding concentration of both the anti-FcyRI (10.1) and the anti-FcyRIII (3G8) mAb was 20pg/ml, while that of the anti-FcyRII mAb (IV.3) was lOpg/ml. Matching optimal concentrations were acquired in 2 repeat experiments with whole blood from different donors.

100 0 1023 FSC-H

Figure 6 . Dotplot showing the FACS profile of whole blood. Forward scatter (FSC-H) has been plotted along the x axis and side scatter (SSC-H) along the y axis and the three leucocyte populations, lymphocyte (L), monocyte (M) and granulocyte (G), individually gated. Data are from one representative experiment.

101 Table 6 a. Mean fluorescent intensities following the immunostaining of whole blood with each of the three anti-FcyR mAb. Data are from one representative experiment.

Monoclonal Anti-FcyRI (10.1) Anti-FcyRII (IV.3) Anti-FcyRIII antibody binding to binding to (3G8) binding to concentration monocytes monocytes granulocytes 0 *8 /m l) (MFI) (MFI) (MFI) 50 128.73 168.42 1598.88

2 0 148.12 173.92 1821.48

1 0 80.14 188.87 1482.48 5 73.42 160.11 1260.41

2 46.97 146.60 939.28 1 32.22 122.70 553.22

The samples displaying saturating binding for each of the anti-FcyR mAb were then further analysed for their specificity by comparing the levels of receptor binding sites per cell in each leucocyte population with known FcyR distribution (Fanger et al, 1989). The binding of each of the three anti-FcyR mAb to each of the three cell populations is shown in Figure 7. The percentage of cells stained positive for each anti-FcyR mAb and the average number of receptor sites per cell (rsc) are summarised in Table 7. These data are from one representative experiment and similar results were observed in 2 repeat experiments with whole blood from different donors.

Table 6 b. The binding of each anti-FcyR mAb at saturating binding concentrations to human leucocytes in whole blood. The table shows the percentage of cells stained positive (%+) for the anti-FcyR mAb within each gated cell population and the mean number of Fey receptor sites per cell (rsc).

Anti-FcyRI mAb Anti-FcyRII mAb Anti-FcyRIII Ab % + rsc % + rsc % + rsc

Lymphocyte --- - 30 161 Monocyte 95 12904 99 19480 58 4784

Granulocyte 50 1458 1 0 0 20617 95 154623

102 Figure 7a shows the levels of anti-FcyRI mAb binding to each of the three gated cell populations. 95% of the gated monocyte population expressed high levels of FcyRI (12904 rsc), while 50% of the granulocyte population beared low FcyRI levels (1458 rsc). However, the histogram for this later result actally appears to show about 95% of granulocytes to stain positive. This discrepency is due to the placement of the marker which gated from the far right-hand side of the isotype control. The anti-FcyRI mAb did not show significant binding to the gated lymphocyte population, although a slight shift of the anti-FcyRI profile to the right of the isotype control was observed.

Figure 7b shows the levels of anti-FcyRII mAb binding to each cell population. 99% of gated monocyte population and 100% of the granulocyte population expressed high FcyRII levels (19480 and 20617 rsc respectively). The anti-FcyRII mAb did not show significant binding to the gated lymphocyte population although, as for the anti-FcyRI mAb, there was a slight shift of the profile to the right of the isotype control.

Figure 7c shows the levels of anti-FcyRIII mAb binding to each cell population. 30% of the lymphocyte population expressed FcyRHI. Most were expressing low levels, however, approximately 1 0 % of this subpopulation were expressing high FcyRIII levels. 95% of granulocyte population expressed high levels of FcyRIII (154623 rsc). Interestingly, 58% of the gated monocyte population stained positive for the anti-FcyRIII mAb, expressing a mean of 4784 rsc. The anti-FcyRIII mAb binding to monocytes exhibited bimodal staining, suggesting that most of the FcyRIH positive cells expressed low to moderate receptor levels, while a smaller subpopulation (~ 2 0 %) expressed high receptor levels.

Following enzymatic digestion, both the F(ab ’)2 and Fab fragments of the anti-FcyRIII mAb bound to the leucocyte populations as observed for the whole antibody by flow cytometry using FITC conjugated F(ab)2 goat anti­ mouse Ig as the detecting antibody. This confirmed that following preparation, the receptor binding function of both antibody fragments was maintained. The saturating binding concentrations of the F(ab ')2 and Fab fragments were 2 0 jig/ml and 1 0 |ig/ml respectively as determined by flow cytometry (Table 7).

103 Table 7. Mean fluorescent intensities following the immunostaining of whole blood with the whole anti-FcyRIII mAb, anti-FcyRIII F(ab )2 fragments or anti-FcyRIII Fab fragments. Data are from one representative experiment.

Monoclonal Whole F(ab )2 Fab antibody anti-FcyRIII anti-FcyRIII anti-FcyRIII concentration binding to binding to binding to (|Ag/ml) granulocytes granulocytes granulocytes (MFI) (MFI) (MFI)

50 326.64 450.32 - 2 0 861.83 536.73 651.06

1 0 757.89 443.53 705.46 5 651.83 273.78 648.25

2 502.21 142.50 333.27

Legend for Figure 7. The binding of the three anti-FcyR mAb to leucocytes in whole blood. Histograms show the fluorescent staining of whole blood by the a) anti-FcyRI mAb (red) or murine IgGl (black), b) anti-FcyRII mAb (red) or murine IgG2b (black) or c) anti-FcyRIII mAb (red) or murine IgGl (black) at saturating binding concentrations. Anti-FcyR mAb staining to each of the three gated cell populations are represented by a separate histogram. Fluorescent intensity (FL1-H) has been plotted along the x axis and the number of events plotted along the y axis. Data shown are from one representative experiment.

104 Figure 7a)

Lymphocytes

(/) c CD > LU

O 10° 102 FL1-H

Monocytes

95% M1 1 I

102 10: 10' FL1-H

o Granulocytes

LU

O 102 FL1-H

105 Figure 7b)

CO Lymphocytes

c >d) LU

O 10° 102 FL1-H

Monocytes

99% M1 C CD > UJ

O 102 FL1-H

Granulocytes (N 100% M1

c >0) UJ

o 102 FL1-H

106 Figure 7c)

Lymphocytes

(/> c M1 LU 30%

o 10° 102 FL1-H

Monocytes

c >

O 10° 101 102 FL1-H

O)o Granulocytes

95% mi C0) > LU

O 10° FL1-H

107 3.1.2. FcyR Expression by isolated and adherent human monocytes in vitro

Experimental design In the previous experiment, the binding capacities and specificities for each of the three murine anti-FcyR monoclonal antibodies were confirmed. The anti-FcyR mAb were then used to establish a system suitable for the investigation of FcyR-dependent cytokine production by human macrophages in vitro. Such a system required isolated peripheral blood monocytes to express all three Fey receptors. Human peripheral blood monocytes constitutively express FcyRI and FcyRII while approximately 10% have been reported to express FcyRIIIa (Ziegler-Heitbrock et al, 1988). However, the results presented in the previous section suggest that this is an oversimplification. Approximately 58% of monocytes expressed FcyRHIa at low levels, while a subpopulation of approximately 2 0 % of the total, expressed high FcyRIIIa levels. This observation has also been reported by other groups (Anderson CL et al, 1990; Maeda et al, 1996). Since the mean level of FcyRIIIa expression was low in comparison to the other two FcyR, steps were taken to induce more comparable levels by a process of maturation. Klaassen et al (1990) found the adherence of monocytes to plastic upregulated FcyRIIIa expression. A modification of this system was chosen to be used for the studies of FcyR-mediated macrophage activation. However, it was necessary to confirm the levels of FcyR expression by both isolated and adhered monocytes so that optimum conditions could be established for future experiments.

Methods The levels of FcyR expression by freshly isolated and cultured monocytes were determined by indirect immunofluorescent staining and flow cytometry. The anti-FcyR monoclonal antibodies were used at the predetermined saturating binding concentrations. Isolated monocytes were adhered to plastic in CGM and cultured at 37°C for 24 and 48 hours. The presence of serum was essential for the survival of monocytes in culture, as well as the induction of FcyRIIIa expression and maintenance of CD14 expression (Andreesen et al, 1990). Following each culture time, the cells were then harvested and stained for each FcyR according to the protocol described in materials and methods (2.11.2.). Monocytes were also stained for FcyR expression immediately following isolation (time 0).

108 Results As shown in Figure 8 c, when isolated human monocytes were allowed to adhere to plastic, an increase in FcyRIIIa expression was observed with time. In this example, at time 0 (freshly isolated monocytes) 22% of the monocytes were stained with a high fluorescent intensity, suggesting high levels of FcyRIIIa expression by this subpopulation. Following adherence to plastic for 24 hours, 42% of monocytes were expressing FcyRIIIa and after 48 hours this rose to 51%. FcyRI and FcyRII expression levels did not significantly change with time. Following adherence to plastic for 0, 24 and 48 hours, 35%, 37% and 30% of monocytes respectively expressed FcyRI (Figure 8a). Similarly, at time 0, 60% of monocytes were expressing FcyRII. Following adherence to plastic for 24 and 48 hours, 67% and 60% of adhered monocytes respectively expressed FcyRII (Figure 8b). Similar results were obtained in two repeat experiments with monocytes from different donors.

In a separate experiment, monocytes were adhered to plastic and cultured for 3 days and 4 days, after which the cells were immunostained for FcyRIIIa expression as before. Figure 9 shows that following adherence to plastic for 3 days 64% of monocytes expressed FcyRIIIa. Following adherence to plastic for 4 days, 67% of monocytes expressed FcyRIIIa. Similar results were observed when this experiment was repeated. This suggested that by 3 days, FcyRIIIa expression by adhered human monocytes had reached a plateau, confirming observations previously reported by Klaassen et al (1990). Therefore, at time points from 24 hours to 4 days the levels of expression of all three Fey receptors were reasonably comparable, providing a workable system for subsequent studies.

109 Legend for Figure 8 . The effects of culturing adhered human monocytes on FcyR expression. Following adherence to plastic for 0,24 or 48 hours at 37°C, human monocytes were then immunostained with the a) anti-FcyRI mAb (red) or murine IgGl (black) b) anti-FcyRII mAb (red) or murine IgG2b (black) or c) anti-FcyRIII mAb (red) or murine IgGl (black). Fluorescent intensity (FL1-H) has been plotted along the x-axis and the number of events plotted along the y-axis. Data shown are from one representative experiment.

110 Figure 8a) O 00 Time 0

c (D > LU 35% M1

o

FL1-H

aoo 24 Hours

c d) > LU 37% M1

o 10° 102 FL1-H

o 00 48 Hours

C LU 30% M1

o

111 Figure 8b)

GO Time 0

w> c >

o T+rHf rWir-f 10° 102 FL1-H

24 Hours

UJ 67% mi

FL1-H

o r^ 48 Hours

(/) c >CD LU

O 10° 102 FL1-H

112 Figure 8c)

Time 0

>ID UJ

o 10° 102 FL1-H

24 Hours

c >d) UJ 42%

o 102 FL1-H

48 Hours

51% mi

102 FL1-H

113 3 Days

LU 64% M1

FL1-H

4 Days

67% M1

LU

O , 10° 102 FL1-H

Figure 9. The effects of culturing adhered human monocytes on FcyRIIIa expression. Following adherence to plastic for 3 and 4 days at 37°C, human monocytes were then stained with the anti-FcyRIII mAb (blue) or murine IgGl (green). Fluorescent intensity (FL1-H) has been plotted along the x-axis and the number of events plotted along the y-axis. Data shown are from one representative experiment.

114 3.1.3. The effect of TNFa on FcyR expression by adhered monocytes in vitro

Experimental design Several studies have demonstrated the modulation of monocyte FcyRIIIa expression by cytokines such as TGFp (Welch et al, 1990), IL-4 (Wong et al, 1991; te Velde et al, 1990) or IL-10 (Calzada-Wack et al, 1996). Two studies have reported the downregulation of macrophage-derived monocyte FcyRIIIa expression by TNFa (Liao & Simon, 1994; Liao et al, 1994), while FcyRI and FcyRII expression remained unaffected. These studies were both performed on monocytes that had been cultured in suspension. Therefore, the effects of TNFa on FcyRIIIa expression by adhered human monocytes was examined.

Methods Human monocytes were adhered to plastic and cultured in CGM at 37°C for 3 days. After replacing the media with fresh CGM, the cultures were incubated at 37°C with either recombinant human TNFa (2ng/ml) or medium alone for 1, 2 and 24 hours. Following each incubation time, the cells were harvested and FcyRIIIa expression measured by indirect immunofluorescent staining and flow cytometry as described previously. Adhered monocytes cultured for 3 days were also stained for FcyRIIIa expression prior to incubation with TNFa or medium (Time 0).

Results As shown in Figure 10, the incubation of adhered monocytes with TNFa had no notable effect on FcyRIIIa expression. Following 3 days in culture, 61% of adhered monocytes expressed FcyRIIIa at an average of 1840 rsc (Time 0). Following a 1 hour incubation with TNFa, 62% of adhered monocytes expressed FcyRIIIa (1996 rsc) compared with 69% of untreated monocytes expressing FcyRIIIa (2294 rsc). Following a 2 hour incubation, 40% of TNFa treated monocytes expressed an average of 1238 FcyRIIIa sites per cell, while 39% of monocytes in the absence of TNFa expressed an average of 1543 FcyRIIIa sites per cell. Following an incubation of 24 hours with TNFa, 65% of monocytes expressed an average of 1994 FcyRIIIa rsc, while 53% of untreated cells expressed an average of 1687 rsc. Similar results were observed when this experiment was repeated.

115 Legend for Figure 10. The effects of TNFa on FcyRIIIa expression by adhered human monocytes. Following adherence to plastic for 3 days (Time 0) monocytes were incubated with and without recombinant human TNFa for 1 , 2 or 24 hours at 37°C. Untreated cells were stained with the anti- FcyRIII mAb (blue) or murine IgGl (green). Similarly, TNFa treated monocytes were stained with the anti-FcyRIII mAb (red) or murine IgGl (black). Fluorescent intensity (FL1-H) has been plotted along the x-axis and the number of events plotted along the y-axis. Histograms are from a representative experiment.

116 Figure 10

Time 0

m C >

o

FL1-H

tn 1 Hour

Cin >Q) UJ

O 10° FL1-H

m 2 Hours

in C >

o TTTT^ 10° 102 103 FL1-H

24 Hours

mC >a> LU

o 102 103 FL1-H

117 Summary The three murine anti-FcyR monoclonal antibodies each reacted with human leucocytes in whole blood according to the cellular distribution of each Fey receptor (Fanger et al, 1989), thereby confirming the monospecificity of each monoclonal antibody. Saturating binding concentrations for each anti-FcyR mAb were determined for use in subsequent experiments.

The adherence of isolated human monocytes to tissue culture plastic induced the upregulation of FcyRIIIa expression reaching a plateau after 3 days in culture, although a significant increase in FcyRIIIa expression was observed after only 24 hours in culture. Furthermore, FcyRIIIa expression by monocytes adhered to plastic was unaffected by the presence of exogenous TNFa, in agreement with observations by Gessl et al (1994). FcyRI and FcyRII expression by adhered monocytes remained constant with time.

Adhered monocytes were therefore cultured for 24 hours prior to experimentation. This culture time was selected on the basis of both receptor expression and optimum TNFa production, which will be discussed in the following section (3.2.1. & 3.2.2.). Thus an in vitro system was established where the effects ligating each Fey receptor on macrophage activation could be investigated.

118 3.2. TNFa production by human peripheral blood monocytesin vitro

3.2.1. The effects of peripheral blood monocyte isolation and adherence on TNFa production

Experimental design The adherence of monocytes to plastic tissue culture plates has been shown to induce cell activity such as the generation of superoxide, the release of 13- glucuronidase (Krause et al, 1996-97; Kelley et ah, 1987) and TNF gene activation (Darville et al, 1992). Therefore, following the adherence of freshly isolated monocytes to plastic at 37°C, supernatants were collected at various time points over 24 hours and tested for TNFa as described in materials and methods (2.13.1.).

Results As shown in Table 8, TNFa was rapidly released into the culture supernatants following the adherence of isolated monocytes to plastic.

Table 8 . The effect of adherence to plastic on TNFa release by isolated human monocytes. Values are expressed as the mean ± S.D. from three separate experiments.

Time after adherence (hours) TNFa (ng/ml)

0 0.027 ±0.05

1 0.034 ±0.06

2 0.244 ± 0.06 4 0.366 ±0.06

6 0.477 ± 0.10 18 0.437 ±0.08

24 0 .6 8 6 ± 0.28

119 3.2.2. The effects of culturing adhered monocytes on TNFa production

Experimental design The adherence of monocytes to plastic, compared with monocytes that have been matured and subsequently stimulated in suspension, may affect the nature of cell activation. For example, compared with monocytes in suspension, adhered monocytes may exhibit enhanced TNFa and IL-ip production, while the ability to release IL - 6 is diminished (Krause et al, 1987). There have been reports that the long-term culturing of monocytes in suspension increases phagocytic activity, while peroxidase activity decreases (Fleit et al, 1982). One group reported that following 24 hours in culture, non-adherent monocytes were able to be stimulated to produce maximal levels of TNFa and this ability for cytokine production remained constant with time (Young et al, 1990). Therefore, the effects of culturing adhered monocytes on their capacity for TNFa production was investigated. Adhered human monocytes were incubated with LPS (a known stimulator of TN Fa production) or the murine IgGl anti-FcyRIII mAb (see 3.3. for further details) to determine the optimal conditions for TNFa production.

Methods Adhered human monocytes were cultured for 1 or 4 days at 37°C. Following each culture time, the medium was removed and the cultures then incubated with LPS (0.5pg/ml; in the absence of polymixin B) or the anti-FcyRm mAb, 3G8 (50jig/ml). Cell-free supernatants were collected after 4 hours and tested for TNFa as described previously.

Results When adhered monocytes were cultured for 1 or 4 days and then incubated with LPS, all cultures released TNFa into the supernatants. However, the levels of LPS-induced TNFa released from the 1 day cultures (1.162 ± 0.44 ng/m l) were greater than those released from the 4 day cultures (0.144 ± 0.13 ng/ml). Similarly, levels of TNFa released from the 1 day cultures incubated with the anti-FcyRIII mAb (0.711 ± 0.12 ng/ ml) were greater than those from the 4 day cultures (0.269 ± 0.16 ng/m l).

120 Summary The adherence of human peripheral blood monocytes to plastic was found to trigger the release of TNFa into culture supernatants. Following 24 hours in culture, adherent human monocytes released TNFa upon stimulation. However, longer culture times resulted in a diminished capacity for TNFa release following stimulation. Therefore, in order achieve a system whereby adherent monocytes were expressing all three Fey receptors, were able to produce a strong TNFa response upon stimulation and the baseline TNFa production at time 0 tested negative, the system was standardised as follows: Freshly isolated peripheral blood monocytes were adhered to plastic at 37°C for 24 hours in CGM. Following this incubation, the culture media was removed and replaced with fresh CGM. Appropriate stimulation/incubations could then be performed (see materials and methods 2 .1 2 . for full details).

3.2.3. Reproducibility and donor variability of TNFa production by adhered human monocytes

Experimental design Monocytes were separated from three different donors on the same day. Following adherence of the monocytes to plastic for 24 hours, the culture media was replaced with fresh and the cells incubated with LPS (0.5 pg/ ml; in the absence of polymixin B) in triplicate for 4 hours. Following collection, the supernatants were tested for TNFa as described previously. All samples were assayed on the same ELISA plate.

Results As shown in Table 9, when monocytes from the three donors were stimulated with LPS, differing levels of TNFa were produced. This diversity in monocyte responsiveness to stimulation reflected the donor variability. However, the levels TNF a released into the supernatants from monocytes isolated from one individual, were highly consistent, indicating reliable reproducibility of both the in vitro system and the cytokine assay.

121 Table 9. LPS induced TNFa production from adhered monocytes isolated from three different donors.

LPS induced TNFa (ng/m l after a 4 hour incubation Donor Sample 1 Sample 2 Sample 3 Mean ± S.D.

1 2.487 2.563 2.654 2.668 ±0.19 2 0.267 0.287 0.299 0.284 ± 0.02 3 1.093 1.272 1.005 1.123 ±0.14 Mean ± S.D. 1.282 ± 1 .1 2 1.374 ± 1.14 1.319 ± 1.21

3.3. The effects of Fey receptor ligation on TNFa production by adhered hum an monocytes in vitro

3.3.1. TNFa release from adhered monocytes following incubation with murine anti-FcyR mAh

Experimental design The ability of each Fey receptor to mediate TNFa release from adhered human monocytes was investigated by specifically ligating each FcyR using a murine anti-FcyR monoclonal antibody in the absence of secondary crosslinking, according to the protocol described in materials and methods (2.12.).

Results When adhered human monocytes were incubated with the anti-FcyRI, anti- FcyRII or anti-FcyRIII mAb (all at 50(ig/ml), TNFa was detected only in those cultures incubated with the anti-FcyRIII mAb (Figure 11a). The anti- FcyRIII mAb response was concentration dependent (Table 10) and polymixin B resistant (Figure 11c). Although flow cytometry showed anti- FcyRIII saturating binding levels to granulocytes to be obtained at 20jig/ml, optimum TNFa production from adhered monocytes was achieved with the anti-FcyRIII mAb at a concnetration of 50 pg/ml. This inconsistency may be due to the ability of the anti-FcyRIII mAb to bind the granulocyte isoform, FcyRIdb with a greater affinity than the macrophage isoform, FcyRIIIa (Anderson CL et al, 1990). The levels of TNFa produced in response to the anti-FcyRIII mAb rapidly rose with time, peaking at 4 hours, after which the

122 response declined. TNFa was undetectable in the supernatants from cultures incubated with medium alone or the murine IgG2b (50pg/ ml). In this representative experiment minimal levels of TNFa were detected in cultures incubated with the murine IgGl (50pg/ml) at 2 and 6 hours only. In other experiments TNFa was undetectable. TNFa release from adhered monocytes was induced by both LPS (in the absence of polymixin B) and PMA. The LPS-induced TNFa response rose rapidly with time and peaked after 6 hours, as observed by Matic & Simon (1991), after which the levels of TNFa declined with time. The PMA-induced response displayed a time course which rose continuously over 24 hours. Similar time courses for all the above incubations were observed in 5 repeat experiments with adhered monocytes from different donors. Only the LPS response was polymixin B sensitive (Figure lib & 11c). Furthermore, the inhibition of LPS-induced TNFa release by polymixin B was consistent in all subsequent experiments.

Table 10. The effects of anti-FcyRIII mAb concentration on TNFa release from adhered monocytes. Culture supernatants were collected for measurement of TNFa 4 hours after incubation with anti-FcyRIII mAb (3G8) at various concentrations. Data from three separate experiments are shown.

Anti-FcyRIII 1 2 3 mAb TNFa (ng/ml) TNFa (ng/ml) TNFa (ng/ml) (pg/ml) 2 0 0 0.399 0.876 0.910 1 0 0 0.274 1.641 0.950 50 0.150 1.776 0.510 25 0.032 1.036 0.169 1 0 0 .0 0 0 0.431 0.026 5 0 .0 0 0 0.079 0 .0 0 0 1 0.000 0.000 0.000

123 Legend for Figure 11a. TNFa release from adhered monocytes following incubation with anti-FcyR mAb. Figure shows the time courses for TNFa production by adhered monocytes following incubation with the anti-FcyRI mAb (pink open triangle), the anti-FcyRII mAb (light blue open circle), or the anti-FcyRIII mAb (red solid circle). The negative controls included medium alone (black open square), murine IgGl (pink open diamond) and murine IgG2b (blue solid diamond). The positive controls were LPS in the absence of polymixin B (green solid triangle) and PMA (blue solid square). The data are from one representative experiment.

Legend for Figure lib. Inhibition of LPS-induced TNFa production from adhered human monocytes by the presence of polymixin B. Adhered human monocytes were cultured with LPS (ljig/ml) in the presence of various concentrations of polymixin B for 4 hours after which supernatants were collected and tested for TNFa.

Legend for Figure 11c. The effect of polymixin B on the anti-FcyRIII mAb- induced TNFa response. Adhered human monocytes were cultured with either LPS (ljig/m l) or the anti-FcyRIII mAb (50^g/m l) in the presence (10 [rg/ ml) or absence of polymixin B. Supernatants were collected after 4 hours and assayed for TNFa.

124 •fl M P e n M CTQ ho o

1 T • • i 3 3 3 r> r> r> 73 p3 > > *Ti sr. sr. W cji •n 73 r> 00 OJ o £ N> M N> O O CT 3 a c 125 s; ? a s. C/5 C/5 2 CTQ

Time (Hours) TNF (ng/ml) TNF(ng/ml) 0 0.6 0 0.0 0 0.4- 0 0 0.4- 0 1 1 1 Figurelib ...... Figure 11cFigure 6 2 6 8 2 8-1 2 0 0 ------' - 0 2 - 4 LPS Anti-Fc Anti-Fc LPS 6 Polymixin BPolymixin (ug/ml) 8 126 10 12 41 820 18 16 14 0 + Polymixin B Polymixin + 0 B Polymixin - □ y RHI mAb

22 3.3.2. The effects of blocking FcyRI or FcyRII on the anti-FcyRIII mAh induced TNFa response from adhered human monocytes

Experimental design The three anti-FcyR mAb can be viewed as surrogates for the interactions between small IgG immune complexes and each individual Fey receptor in isolation. Each Fab region of an anti-FcyR mAb can bind one FcyR, thereby allowing the crosslinking of two Fey receptors of the same type. A third FcyR of any type may be non-specifically recruited through the Fc fragment of the monoclonal antibody. The three murine anti-FcyR mAb block IgG Fc binding by recognising epitopes in or dose to their receptor's binding site (Dougherty et al, 1987; Looney et al, 1986b; Fleit et al., 1982). Therefore, the anti-FcyRI and anti-FcyRII mAb, which were non-stimulatory for TNFa, were used to block binding of the anti-FcyRIII mAb Fc fragment to either FcyRI or FcyRII. This was to determine whether TNFa release induced by the anti-FcyRIII mAb was dependent upon any additional Fey receptor recruitment through the mAb Fc region.

Methods Adhered human monocytes were preincubated at 37°C for 30 minutes with either the anti-FcyRI mAb or the anti-FcyRII mAb (both at the saturating binding concentration of 20pg/ ml), or medium alone. The anti-FcyRIII mAb (50pg/ml) was then added to all cultures. Supernatants were collected after a 4 hour incubation with the anti-FcyRIII mAb and tested for TNFa as described previously.

Results Adhered monocytes incubated with LPS (in the absence of polymixin B) released TNFa into the supernatant (1.308 ± 0.53 ng/ml).As observed previously, no TNFa was detected in cultures incubated with either medium, the anti-FcyRI mAb or the anti-FcyRII mAb alone. TNFa was detected in the supernatants from cultures that had been preincubated with medium alone and then incubated for 4 hours with the anti-FcyRIII mAb (1.279 ± 0.12 ng/ml). Cultures incubated with the anti-FcyRI mAb prior to incubation with the anti-FcyRIII mAb also released similar amounts of TNFa (1.635 ± 0.44 ng/m l), as did those cultures preincubated with the anti-FcyRII mAb (1.424 ± 0.19 ng/ml) (Figure 12). Therefore, the blocking of either FcyRI

127 or FcyRII with their respective monoclonal antibodies had no statistical significant effect upon the levels of TNFa released in response to the anti- FcyRIII mAb.

Figure 12. Barchart showing the effects of blocking FcyRI or FcyRII on the anti-FcyRIII mAb-induced TNFa response. Adhered human monocytes were preincubated with either medium alone, the anti-FcyRI mAb (10.1) or the anti-FcyRII mAb (IV.3). All cultures were then incubated for 4 hours with the anti-FcyRIII mAb (3G8). Values are expressed as the mean ± S.D. from three separate experiments.

0.6 -

0.4-

0 . 2 -

0.0 Anti-Fc YRI Anti-Fc YRIIMedium

128 3.3.3. TNFa release from adhered monocytes following incubation with F(ab')2 or Fab fragments of the anti-FcyRIII mAb

Experimental design The previous experiment had not eliminated the possibility of the anti- FcyRIII mAb recruiting another FcyRIIIa receptor through its Fc region. Therefore, to determine whether TNFa release in response to the anti-FcyRIII mAb was dependent upon the Fc region of the monoclonal antibody interacting with an Fey receptor, adhered monocytes were incubated with either F(ab ')2 or Fab fragments of the anti-FcyRlH mAb. To compensate for the differing levels of antigen binding sites in the three antibody preparations, the concentrations of the F(ab ')2 and Fab fragments were adjusted. Therefore, the anti-FcyRIII F(ab ')2 fragments were adjusted to 2/3 the concentration of the whole antibody and the anti-FcyRQI Fab fragments were adjusted to 1/3 of the whole anti-FcyRIII mAb concentration.

Results Figure 13 shows that following a 4 hour incubation with the anti-FcyRIII mAb (50|xg/ml), adhered monocytes released TNFa into the supernatant (0.499 ± 0.12 ng/ml). Low levels of TNFa were detected in the supernatants from cultures incubated with either the anti-FcyRIII mAb F(ab ')2 fragments at 33pg/ml (0.025 ± 0.04 ng/ml) or the anti-FcyRIII Fab fragments at 17pg/ml (0.036 ± 0.06 ng/ml). However, similarly low levels of TNFa were also detected in the supernatants from cultures incubated with medium alone (0.026 ± 0.05 ng/ml). LPS (in the absence of polymixin B) also induced adhered monocytes to release TNFa as observed previously (1.076 ± 0.57 ng/ml). Similar results were observed when cultures were incubated with the anti-FcyRIII mAb F(ab ')2 or Fab fragments at a final concentration equal to the whole monoclonal antibody (50pg/ml) (data not shown).

Moreover, when adhered monocytes were preincubated with F(ab ’)2 fragments of the anti-FcyRIH mAb at binding saturating levels (20pg/ml), the whole anti-FcyRIII induced TNFa response was inhibited by 69.0 ± 6.2%, confirming the functional ability of the anti-FcyRIII F(ab ')2 fragment to bind

FcyRHI. The lack of total inhibition by the F(ab )2 fragments may be because a slightly higher F(ab ) 2 concnetration was required, again highlighting the differing binding affnity of the anti-FcyRQI mAb for FcyRIQa and FcyRIDb.

129 Figure 13. The effects of anti-FcyRIII F(ab ')2 and Fab fragments on TNFa release. Barchart showing the production of TNFa after adhered monocytes were incubated for 4 hours with either the anti-FcyRIII (3G8) mAb, anti- FcyRIII F(ab ')2 fragments, anti-FcyRIII Fab fragments or medium alone. Values are expressed as the mean ± S.D. from three separate experiments.

0.7 1

to I

0 . 1 -

0.0 Anti-Fc yRHI F(ab)2 of Fab of Medium anti-FcyRIII anti-FcyRIII

130 3.3.4. TNFa release from adhered monocytes following incubation with more than one anti-FcyR mAb

Experimental design Looney et al. (1986b) reported that the anti-FcyRII mAb (IV.3) alone had no inhibitory effect upon rosette formation between granulocytes and rabbit IgG-sensitised erythrocytes, whereas rosette formation was partially inhibited by the anti-Fc/RIH mAb, 3G8. However, in combination, IV.3 and 3G8 completely inhibited this rosette formation. Although neither the anti- FcyRI nor the anti-FcyRII mAb induced TNFa release from adhered monocytes, an additive effect may be possible if both are used together or in combination with the stimulatory anti-FcyRIII mAb. Therefore, adhered monocytes were incubated for 4 hours with various combinations of the three anti-FcyR mAb (all at a final concentration of 50pg/ml) and the levels of TNFa production were compared.

Results The incubation of adhered human monocytes with more than one anti-FcyR mAb had no statistically significant effect upon the levels of TNFa released into the culture supernatant. Following a 4 hour incubation, cultures incubated with the anti-FcyRIII mAb alone (0.747 ± 0.57ng/ml) released TNFa as observed previously. Similar levels of TNFa were released from cultures incubated with either the anti-FcyRI and the anti-FcyRIII mAb (0.840 ± 0.95ng/ml), the anti-FcyRII and the anti-FcyRIII mAb (0.877 ± 0.60ng/ml) or all three anti-FcyR mAb (0.958 ± 0.82ng/ml). TNFa levels were low if at all detectable in the supernatants from cultures incubated with either the anti-FcyRI mAb alone (0.000 ± O.OOng/ml) or the anti-FcyRII mAb alone (0.023 ± 0.03ng/ml). TNFa was also undetectable in the supernatants from cultures incubated with both the anti-FcyRI and the anti-FcyRII mAb or medium alone. As observed previously, LPS in the absence of polymixin B induced TNFa release from adhered monocytes (0.779 ± 0.45 ng/m l).

Summary The experiments presented in this section have demonstrated the ability of the murine anti-FcyRIII mAb to induce TNFa release from adhered human monocytes in the absence of secondary crosslinking: a novel finding. The anti-FcyRI and anti-FcyRII mAb were both found to be non-stimulatory for

131 TNFa release by the adhered human monocytes, supporting previous findings (Debets et al, 1990). Furthermore, neither the F(ab ')2 nor the Fab fragments of the anti-FcyRIII mAb were able to induce TNFa release from the adhered monocytes. Using the anti-FcyRI and anti-FcyRII mAb to block receptor binding, it was found that there was no additional recruitment of either FcyRI or FcyRII by the Fc region of the anti-FcyRIII mAb during the induction of a TNFa response.

3.4. The effects of Fey receptor ligation on IL-la production by adhered human monocytes in vitro

3.4.1. IL-la release from adhered monocytes following incubation with anti-FcyR mAb

Experimental design The experiments presented in the previous section (3.3.) demonstrated the ability of FcyRIIIa to mediate TNFa release from adhered human monocytes, following receptor ligation with a murine anti-FcyR monoclonal antibody. It therefore seemed appropriate to investigate the ability of each Fey receptor to mediate the release of other cytokines following receptor ligation by these anti-FcyR monoclonal antibodies. Work by Chantry et al. (1989) has demonstrated the ability of immune complexes derived from patients with type II mixed cryoglobulin to induce IL-la and IL-p production from human monocytes. The crosslinking of FcyR by IgG immune complexes had already been shown to induce the release of IL-1 p from human monocytes (Polat et al, 1993). Although IL-1 a tends to be membrane associated it was decided to see if FcyR ligation could also induce IL-la release from adhered human monocytes. This could have implications in the local environment of the synovium since this proinflammatory cytokine may play a similar role to TNFa in mediating pathology in rheumatoid arthritis. This is supported by the observation of spontaneous and prolonged IL-la production from isolated rheumatoid synovial cells (Buchan et al, 1988). Therefore, the ability of each Fey receptor to mediate IL-la release by adhered human monocytes in vitro was investigated by ligating each FcyR using the murine anti-FcyR monoclonal antibodies in the absence of secondary crosslinking as previously described. Culture supernatants were collected at various time points over 72 hours and tested for IL-la, according to the protocol

132 described in materials and methods (2.13.3.). Supernatants were also collected after a 4 hour incubation and assayed for TNFa as described previously.

Results When adhered human monocytes were incubated with the anti-FcyRI^ anti- FcyRII or anti-FcyRIII mAb, only those cultures incubated with the anti- FcyRIII mAb released IL-la into the supernatant (Figure 14). The time course for IL-1 a release in response to the anti-FcyRIII mAb was very different form the anti-FcyRIII-induced TNFa response (see section 3.3.1.). IL-la was not detectable until after an incubation of 11 hours. The IL-la levels then gradually rose with time and appeared to plateau after 33 hours. IL-la was not detected in supernatants from cultures incubated with either medium alone or the murine isotype-matched controls. LPS (in the absence of polymixin B) also induced adhered monocytes to release IL-la and the time course was again, different form the LPS-induced TNFa response since IL- la was only detected in supernatants following a 10 hour incubation. Unlike the anti-FcyRIII mAb-induced IL-1 a response, the IL-la response induced by LPS rose steeply and continuously over the 72 hour time course. Interestingly, it was found that although higher IL-la levels were measured in response to LPS, compared with the IL-la levels induced by the anti- FcyRIII mAb, the levels of TNFa released into supernatants were similar following a 4 hour incubation with either LPS (1.279ng/ml) or the anti- FcyRQI mAb (1.347ng/ml). Similar results were observed in 2 repeat experiments with monocytes from different donors.

133 Legend for Figure 14. IL-la release from adhered monocytes following incubation with anti-FcyR mAb. Figure shows the times course for IL-la production by adhered monocytes following incubation with the anti-FcyRI mAb (pink solid square), the anti-FcyRII mAb (light blue open square), or the anti-FcyRIII mAb (red solid circle). The negative controls included medium alone (blue open triangle), murine IgGl (black open circle) and murine IgG2b (black solid diamond). LPS in the absence of polymixin B (green solid triangle) was used as a positive control. The data shown are from one representative experiment.

134 >TI £ M •-t rD 4^ CTQ ti M * O

^ 00 oj hi 00 s s ti —i

m > > Ch Ch h n 3 3 a* n o n 2 5 S J? J? CfQ (N oS-. S £ o £ £ T- n O n a 135 IL-1 (pg/ml)IL-1 ro 1 o U o OJ Ul CJ1 VI o U l o U l U l Time (H ours) 3.4.2. The effect of an anti-TNFa mAb on both the LPS and anti-FcyRIII mAb induced IL-la release from adhered human monocytes

Experimental design Dinarello et al. (1986) first described the capacity of human recombinant TNFa to induce the production of IL-1 from human mononuclear cells in vitro. IL-1 production by rheumatoid synovial mononuclear cells in culture has also been shown to be dependent on the presence of TNFa (Brennan et al., 1989). Furthermore, immune complexes may work in synergy with TNFa to enhance IL-1 production from human monocytes (Chantry et al, 1989). The potential dependency upon TNFa of the FcyRIIIa-mediated and LPS- induced IL-la production by adhered human monocytes shown in the previous experiment (3.4.1.), was investigated using a neutralising anti- TNFa mAb (CB006). The anti-TNF mAb was specific for the active site of TNFa and therefore prevented both its detection by the ELISA used in these studies and its interaction with TNF receptors (Dr S Stephens, personal communication).

Methods Adhered human monocytes were incubated with either LPS (in the absence of polymixin B) or the anti-FcyRIII mAb (50pg/ml) in the presence of the murine monoclonal anti-TNFa mAb at 100,10, 1 or Opg/ml. Following a 4 hour incubation, supernatants were collected and tested for TNFa and after a 48 hour incubation supernatants were collected and tested for IL-la as described previously.

Results As shown in Figure 15a, the anti-TNFa mAb efficiently bound soluble TNFa in the culture supernatants in a dose dependent manner, preventing its detection by the TNFa ELISA. This was possible since the neutralising anti- TNFa mAb was the same as the coating antibody in the TNFa ELISA. There was no cross-reactivity between the anti-TNFa mAb and the ELISA which may have given false positive results. However, to have fully ruled out the possibility of the anti-TNF a mAb in solution interferring with the detection of TNFa, recombinant TNFa should have been titrated to show that the ELISA could still detect excess, unbound cytokine. Additionally, to have fully established neutalisation by the anti-TNFa mAb, a TNFa bioassay

136 should have been performed. Nevertheless as shown in figure 15a, at the anti-TNFa mAb concentration of lpg/m l, the detection of TNFa release in response to LPS was inhibited by 50% while the detection of the anti-FcyRIII mAb TNFa response was inhibited by 38%. At the anti-TNFa mAb concentration of lOpg/ ml, the detection of the TNFa responses following incubation with either LPS or the anti-FcyRIII mAb were both inhibited by 94%. At lOOpg/ml, the anti-TNF mAb inhibited TNFa detection in response to LPS and the anti-FcyRIII mAb both by 100%. Similar levels of inhibition were observed when this experiment was repeated.

As shown in Figure 15b, IL-la release from adhered monocytes following incubation with the anti-FcyRIII mAb was inhibited by the anti-TNFa mAb in a dose dependent manner. At the anti-TNFa mAb concentration of Ijxg/ml the anti-FcyRIII mAb-induced IL-la response was inhibited by 6 6 %. The anti-FcyRIII mAb-induced IL -la response was inhibited by 82% when cultures were incubated in the presence of the anti-TNF mAb at both 10jig/ml and lOOpg/ml. No such inhibition was seen when cultures were incubated with LPS in the presence of the anti-TNFa mAb (Figure 15c). This was not due to the anti-TNFa mAb interfering with the IL-la ELISA Additionally, IL-la was not detected in the supernatants from cultures incubated with recombinant human TNFa alone at concentrations ranging between 0 and 5ng/ ml. Similar results were observed when this experiment was repeated.

137 Figure 15. The effect of a neutralising anti-TNFa mAb on TNFa detection and IL-la release. Barchart shows the levels of a) TNFa detection in cultures following a 4 hour incubation with LPS (in the absence of polymixin B) or the anti-FcyRIII mAb; b) IL-la release from cultures following a 48 hour incubation with the anti-FcyRIII mAb; c) IL-la release from cultures following a 48 hour incubation with LPS (in the absence of polymixin B) all in the presence of an anti-TNFa mAb at various concentrations. Data are from one representative experiment.

a)

3.01 □ 3G8 □ LPS 2.5-

2.0- I 1.5- i.o-

0.5-

0.0 0 1 10 100 Anti-TNF mAb g/ml)

138 b)

1. 0 -

0.8 -

0 . 6 “ g cJ 0.4

0.2 J

0.0 1 10 100 Anti-TNF mAb (V g/ml)

c)

8

6

2

0 0 1 10 100 Anti-TNF mAb (ng/ml)

139 Summary The experiments presented in this section have again demonstrated the ability of FcyRHIa to mediate cytokine production by adhered monocytes following receptor ligation with the anti-FcyRIII mAb in the absence of secondary crosslinking. Of the three anti-FcyR monoclonal antibodies, only the anti-FcyRIII mAb was able to induce IL-la release from adhered human monocytes in vitro. The time course for IL-la release from adhered monocytes in response to either the anti-FcyRQI mAb or LPS was different from the profiles observed for TNFa release. Furthermore, the anti-FcyRIQ mAb-induced IL-la production from adhered monocytes appeared to be dependent upon the presence of TNFa, while the LPS induced IL-la production occurred independently of TNFa.

3.5. The effects of Fey receptor ligation on reactive oxygen species production by adhered human monocytes in vitro.

3.5.1. Reactive oxygen species production by adhered monocytes following incubation with anti-FcyR mAb

Experimental design Phagocytic leucocytes respond to certain stimuli by generating an intense respiratory burst (Babior, 1984b). After establishing the ability of FcyRIQa to mediate both TNFa and IL-la production from adhered human monocytes (3.3.1. & 3.4.1.), these studies were further extended to Fey receptor-mediated production of reactive oxygen species. Trezinni et al (1990) had previously demonstrated the production of reactive oxygen species by non-adhered monocytes through an FcyRIQa pathway, following receptor ligation with the murine anti-FcyRQI mAb (3G8). As with the production of TNFa and IL- la described previously in this thesis, Trezzini's group found that both the anti-FcyRI (32.2) and the anti-FcyRII mAb (IV.3) failed to induce free radical production by monocytes in suspension. Therefore, the ability of each FcyR to mediate a respiratory burst by adhered human monocytes in vitro was investigated. Each Fey receptor was ligated using the murine anti-FcyR monoclonal antibodies in the absence of secondary crosslinking and reactive oxygen species production measured by luminol-dependent chemiluminescence according to the protocol described in materials and methods (2.14.).

140 Results When adhered human monocytes were incubated with either the anti-FcyRI (32.2), anti-FcyRII (IV.3), anti-FcYRHI (3G8) mAb or the isotype control murine IgGl (all at 15pg/ml), only those cultures incubated with the anti- FcyRIII mAb generated a reactive oxygen species response (Figure 16). Cultures incubated with the anti-FcyRI mAb, the anti-FcyRII mAb or the murine IgGl showed levels of chemiluminescence that fluctuated between 7 and 20 cps over the whole time course. These levels of chemiluminescence were similar to those recorded from the adhered monocytes prior to the addition of the monoclonal antibodies (background light emission ~ 1 2 cps). Following the addition of the anti-FcyRIII mAb to the adhered monocytes, the chemiluminescence response rapidly peaked within 2 minutes (126 cps). The light emission remained high for another 8 minutes and then gradually decayed with time, returning to near background levels (2 1 cps) 60 minutes after the anti-FcyRQI mAb was first added.

In 2 repeat experiments, similar chemiluminescent responses were observed. In these repeated experiments, the anti-FcyRI mAb, 32.2 was substituted by the other anti-FcyRI mAb, 10.1. As with mAb 32.2, the anti-FcyRI mAb 10.1, induced levels of chemiluminescence equal to background light emission. In one experiment the murine IgGl isotype control was substituted by the murine IgG2b and levels of chemiluminescence were again similar to background light emission.

141 Legend for Figure 16. Reactive oxygen spedes production by adhered monocytes following incubation with anti-FcyR mAb. Figure shows the times course for reactive oxygen spedes production by adhered monocytes following incubation with the anti-FcyRI mAb (green solid square), anti- FcyRII mAb (blue solid triangle), anti-FcyRIII mAb (red solid cirde) or murine IgGl (pink solid diamond). Data are from one representative experiment.

142 \ 0 ro o Ul o o 00 o 143 O 4^ Reactive oiygen species production (cps) o 4^ o o 4^ o Time (minutes) 3.5.2. Reactive oxygen species production by adhered monocytes following incubation with PMA or LPS

Experimental design PMA and LPS are potent stimulators of phagocyte activity including activation of the respiratory burst (Dahlgren, 1987). PMA activates protein kinase C (Niedel et al, 1983). LPS stimulates monocytes through the surface marker, CD14 by first complexing with the LPS binding protein (LBP) which is ubiquitously present in serum (Wright et al, 1990; Ziegler-Heitbrock & Ulevitch, 1993). Serum has been shown to inhibit luminol-dependent PMN chemiluminescence, mainly due to the high levels of albumin in serum acting as a scavenger of oxygen radicals (Hastings et al, 1982; Holt et al, 1984). Since the chemiluminescent experiments outlined in this thesis were performed in the absence of polymixin B, the potential for LPS to stimulate a respiratory burst by adhered monocytes in the absence of serum was determined.

Results Figure 17a shows the time course for reactive oxygen species production following the incubation of adhered monocytes with the anti-FcyRIII mAb (15ng/ml), murine IgGl (15|rg/ml) or PMA (0.5pg/ml). As observed in the previous experiments (3.5.1.), the anti-FcyRIII mAb induced adhered monocytes to generate a reactive oxygen species response, while chemiluminescence levels induced by the murine IgGl were similar to those observed for background light emission. Cultures incubated with PMA generated a strong chemiluminescent response that gradually rose with time, peaking after an incubation of 25 minutes (7750cps), after which the light emission gradually decayed with time. At its peak, the PMA induced response was 6 times greater than the anti-Fc/RIH mAb induced response at its peak (126cps). Similar time courses were observed in 4 repeat experiments, with the PMA induced response ranging from 9 to 11 times greater than the anti-FcyRIII mAb induced response.

Figure 17b shows the time course for reactive oxygen species production from adhered monocytes following incubation with the anti-FcyRIII mAb, murine IgGl or LPS (0.5pg/ml). As observed previously, the anti-FcyRIII mAb induced adhered monocytes to generate a reactive oxygen species

144 response. The levels of chemiluminescence recorded from cultures incubated with LPS were comparable to the levels of chemiluminescence induced by the murine IgGl and background light emission. This result was reproduced in 2 repeat experiments.

145 Legend for Figure 17. The effects of PMA and LPS on reactive oxygen species production. Time course for reactive oxygen species production by adhered monocytes following incubation with a) the anti-FcyRIII mAb (red solid circle), murine IgGl (green solid triangle) or PMA (blue solid square) or b) the anti-FcyRIII mAb (red solid circle), murine IgGl (green solid triangle) or LPS (blue solid square). Data are from one representative experiment.

146 VJ o o o CTQ n 00 O O o O > o Ul o o o 4^ o o o 147 o o o OJ ho o o o Reactive oxygen spedes production (cps) o o o o 4^ O o o Ul o Time (Minutes) hi vi £ >-« ro o r CTQ 00 O O \D O o 00 o VI 148 n O O Ul o Reactive oxygen spedes production (cps) o o OJ N> O 00 00 o rv

Time (Minutes) 3.5.3. Reactive oxygen species production by adhered monocytes following incubation with anti-FcyRIII mAb F(ab')2 fragments

Experimental design To determine whether the generation of the oxidative respiratory burst in response to the anti-FcyRIII mAb was dependent upon an Fc-FcyR interaction, adhered monocytes were incubated with F(ab ')2 fragments of the anti-FcyRIII mAb.

Results As shown in Figure 18, adhered monocytes incubated with the anti-FcyRIII mAb (15pg/ml) generated a reactive oxygen species response as observed previously. Incubation with the murine IgGl (15pg/ml) induced levels of chemiluminescence similar to those observed for background light emission. The culture incubated with F(ab ')2 fragments of the anti-FcyRIII mAb (15pg/ml) also generated a reactive oxygen species response. Unlike the chemiluminescent time course induced by the whole anti-FcyRIII mAb, incubation of adhered monocytes with the F(ab ')2 fragments induced a weaker response. The levels of chemiluminescence rose gradually with time and peaked after an incubation time of 30 minutes (lOOcps). The response then continued at this level while the whole anti-FcyRIII mAb-induced chemiluminescent response declined with time to a similar level. These observations were confirmed in 2 repeat experiments with monocytes from different donors. In one of the repeat experiments a culture was also incubated with anti-FcyRIII mAb Fab fragments (15 pg/ml) and no chemiluminescent response was observed.

Summary The findings presented in this results section have demonstrated the ability of FcyRIHa to mediate a respiratory burst from adhered human monocytes following receptor ligation with a monoclonal antibody in the absence of secondary crosslinking. The monoclonal antibodies to FcyRI and FcyRII did not trigger reactive oxygen species production from adhered monocytes. These observations support previous experiments performed on non­ adherent human monocytes (Trezinni et al., 1990). F(ab')2 fragments of the anti-FcyRIII mAb also triggered the adhered monocytes to generate a respiratory burst. However, contrary to previous reports (Trezinni et al,

149 1990) the anti-FcyRin F(ab ')2 fragments triggered a weaker chemiluminescent response, with a profile distinct from the time course triggered by the whole monoclonal antibody.

150 Legend for Figure 18. The effects of anti-FcyRIII F(ab ')2 fragments on reactive oxygen species production. Time course for reactive oxygen species production by adhered monocytes following incubation with the anti- FcyRIII mAb (red solid circle), anti-FcyRIII F(ab ')2 fragments (blue solid square) or murine IgGl (green solid triangle). Data are from one representative experiment.

151 00 era 00 152 i-* I—i hO N> OJ CO 4^- Reactive oxygen species production (cps) CjlOCnOUlOCnO ooooooooo o 00 Ul o o o Time (minutes) 3.6. The effects of human IgG immune complexes on TNFa production by adhered monocytes in vitro The work presented in this chapter so far has been focused upon the ability of Fey receptor ligation to trigger the release of pro-inflammatory mediators such as cytokines and reactive oxygen species from adhered human monocytes in vitro. This was achieved using murine anti-FcyR monoclonal antibodies which may be viewed as surrogates for small IgG immune complexes. It was necessary to examine the effects of physiological ligands on FcyR-mediated monocyte activation in order to investigate the ability of small IgG immune complexes to induce TNFa production. Complexes of human IgG anti-NIP mAb were used to investigate the effects of IgG immune complex size and subclass on TNFa production by adhered human monocytes in vitro.

3.6.1. TNFa release from adhered monocytes following incubation with immune complexes of human IgG anti-NIP mAb

Experimental design As discussed previously, the type of immune complex implicated in the pathogenesis of rheumatoid arthritis are small, soluble IgGl complexes. The ability of small complexes consisting of the human IgGl anti-NIP mAb to trigger TNFa release from adhered human monocytes was investigated. The NIP/BSA antigen with an average epitope density of 2.4 was selected to create a heteropopulation of small complexes, theoretically comprising IgG monomers, dimers and trimers. Immune complexes were prepared in a molar excess of antigen, a molar excess of antibody (both soluble complexes) and at molar equivalence (insoluble complexes) to establish the effects of IgGl anti-NIP immune complex preparation on TNFa release.

M ethods Immune complexes of the human IgGl anti-NIP mAb coupled with 2.4NIP/BSA were prepared in either a molar excess of antigen, with the molar ratio of 1:5, a molar excess of antibody, with the molar ratio of 1:5, or in an antigen:antibody ratio of 1:2 (equivalence) so that the mean number of NIP molecules per BSA was approximately equivalent to the number of antibody molecules. The immune complexes were prepared according to the protocol described in the materials and methods (2.9.3.). Adhered human

153 monocytes were then incubated with the preformed immune complexes (at a final IgGl concentration of 50pg/ml) or appropriate controls (see materials and methods 2.12.) and following a 4 hour incubation supernatants were collected and tested for TNFa as described previously.

Results As shown in Figure 19, when adhered human monocytes were incubated with complexes of human IgGl anti-NIP mAb coupled to 2.4NIP/BSA, significant levels of TNFa were only detected in supernatants from those cultures incubated with the immune complexes prepared in a molar excess of antigen (1.121 ± 0.46ng/ml). This response was concentration dependent (Figure 20). The incubation of cultures with LPS (in the absence of polymixin B) induced similar levels of TNFa (1.427 ± 0.72ng/ml). Minimal levels of TNFa were detectable in the supernatants from cultures incubated with the immune complexes prepared either in a molar excess of antibody (0.018 ± 0.03ng/ml) or at equivalence (0.014 ± 0.02ng/ml). TNFa was undetectable in the supernatants from cultures incubated with either the 2.4NIP/BSA, the IgGl anti-NIP mAb or medium alone.

In a separate series of experiments, IgGl anti-NIP immune complexes were prepared using NIP/BSA with average epitope densities of 8.9 or 21.5. Again, TNFa was only detected in the supernatants from cultures incubated with complexes preformed in a molar excess of antigen.

154 Figure 19. The effect of IgGl anti-NIP immune complex preparation on TNFa release. Barchart showing the levels of TNFa released by adhered monocytes following a 4 hour incubation with LPS (in the absence of polymixin B) or small IgGl anti-NIP immune complexes prepared in either a molar excess of antigen, a molar excess of antibody or at molar equivalence. Values are expressed as the mean ± S.D. from three separate experiments.

2.2 -

2.0 -

1. 8 -

1 .6 -

1.4- I 1.2 - 1. 0 -

0. 8 -

0.6 -

0.4-

0 .2 -

o.o -*- Antigen excess Antigen = Antibody Antibody excess LPS

155 Figure 20. The effect of small IgGl anti-NIP immune complex (prepared in a molar excess of antigen) concentration on TNFa release by adhered monocytes following a 4 hour incubation. Data are from one representative experiment. This result was reproduced when the experiment was repeated.

l.6 i

l.o-

0.8 -

0.6 -

0.4-

0.0 1 10 100

Immune complex concentration l|i g/ml)

156 3.6.2. The effects of IgG immune complex subclass on TNFa production by adhered monocytes

Experimental design IgG of each subclass bind Fey receptors with differing affinities (van de Winkel & Capel, 1993) and therefore complexes of each IgG subclass may differ in their ability to trigger TNFa production from adhered human monocytes. This was investigated by incubating adhered monocytes with small anti-NIP immune complexes of each IgG subclass. The human IgGl, IgG2, IgG3 or IgG4 small anti-NIP immune complexes were preformed in a molar excess of antigen (2.4NTP/BSA) as described in the previous experiment (3.6.1.). Adhered human monocytes were then incubated with the preformed immune complexes (at a final IgG concentration of 50pg/ml) or appropriate controls (see materials and methods 2.12.). Supernatants were collected at various time points over 21 hours and tested for TNFa as described previously.

Results As shown in Figure 21, all cultures incubated with the small soluble IgG anti-NIP immune complexes released TNFa into the supernatant. The complexes of each IgG subclass induced a TNFa response that rapidly rose with time, peaking after an incubation of 4-6 hours and then gradually declined with time. Cultures incubated with LPS (in the absence of polymixin B) showed a similar time course for TNFa release. The IgG subclass specific complexes induced levels of TNFa production in the order of magnitude: IgGl>IgG3>IgG2>IgG4. TNFa was undetectable in the supernatants from cultures incubated with medium alone (data not shown). TNFa was also undetectable following incubation with either the 2.4NIP/BSA or the IgG anti-NIP mAb of each subclass alone. Similar time courses and subclass specific levels of response were observed in 2 repeat experiments with monocytes from different donors.

157 Legend for Figure 2 1 . The effects of immune complex subclass on TNFa release. Time course for TNFa production by adhered monocytes following incubation with small IgG anti-NIP complexes of the subclass IgGl (red solid circle), IgG2 (blue solid square), IgG3 (green solid triangle) or IgG4 (pink solid diamond), all preformed in a molar excess of antigen. Cultures were also incubated with LPS (black open circle) or medium alone (pale blue open square). Data are from one representative experiment.

158 CTQ U (O U M (O i i i i hU ^ 4* hU it* {s> N) N> N) N> N) {s> QTQ 0Q CTC CTQ CTC 0Q QTQ I—‘ i—‘ >1 i—ik ->* h 159 TNF(ng/ml) oro^a' bookj^O' bo O O O O O O O O O O n o o oo ro o 00 O Time (Hours) 3.6.3. The effects of antigen epitope density on human IgG immune complex induced TNFa production by adhered monocytes

Experimental design The effects of immune complex size on the ability of adhered human monocytes to release TNFa was investigated using soluble IgGl anti-NIP mAb complexes formed with NIP/BSA of different epitope densities.

M ethods IgGl anti-NIP mAb complexes were prepared in a molar excess (1:5) of either 2.4NIP/BSA (small complexes), 8.9NIP/BSA (moderate complexes) or 21.5NIP/BSA (large complexes) as described previously. Adhered human monocytes were then incubated with the preformed immune complexes (at a final IgGl concentration of 50jxg/ ml) or appropriate controls (see materials and methods 2.12.). Supernatants were collected after a 4 hour incubation and tested for TNFa.

Results Following a 4 hour incubation, TNFa was detected in supernatants from cultures incubated with the small soluble IgGl complexes (0.968 ± 0.28ng/ml), the moderate soluble IgGl complexes (1.475 ± 0.91ng/ml) and also the large soluble IgGl complexes (1.011 ± 0.52ng/ml). However, there appeared to be no significant relationship between the epitope density (complex size) and the levels of TNFa released.

3.6.4. The effects o f blocking FcyR on small human IgG immune complex triggered TNFa production

Experimental design To determine and identify Fey receptor involvement in the production of TNFa induced by small soluble IgGl anti-NIP mAb immune complexes, anti-FcyRI, anti-FcyRII and F(ab ')2 fragments of the anti-FcyRIII mAb were used to block FcyRI, FcyRQ and FcyRHIa receptor binding respectively.

Methods Adhered human monocytes were preincubated at 37°C for 30 minutes with either the anti-FcyRI mAb, the anti-FcyRII mAb, F(ab ')2 fragments of the anti-

160 FcyRHI mAb (all at 2 0 fig/ ml) or medium alone. Small IgGl anti-NIP complexes were preformed in a molar excess of antigen (1:5) as described previously and were then added to all cultures (at a final IgGl concentration of 50fig/ml). LPS (in the absence of polymixin B) and the anti-FcyRIII mAb were used as positive controls as described in previous experiments. Culture supernatants were collected 4 hours after the complexes were added and tested for TNFa.

Results As observed previously, cultures incubated with either LPS (in the absence of polymixin B) or the anti-FcyRIII mAb released TNFa into the supernatant (0.570 ng/ml and 0.478 ng/ml respectively). TNFa was not detected in the supernatants from cultures incubated with either the anti-FcyRI mAb, anti- FcyRII mAb, F(ab ')2 fragments of the anti-FcyRIII mAb or medium alone. When either FcyRH or FcyRHIa were blocked, the TNFa response induced by the small soluble IgGl complexes was partially inhibited by 21% and 37% respectively. However, the blocking of Fc/RI had little effect (1% inhibition) upon the TNFa response following incubation with the small IgGl immune complexes. Similar levels of inhibition were observed when this experiment was repeated.

Summary The results presented in this section have demonstrated the ability of small, soluble human IgG anti-NIP immune complexes to induce the release of TNFa from adhered human monocytes in vitro. Those complexes formed at equivalence (insoluble) or in a molar excess of antibody failed to trigger a significant TNFa response. The levels of TNFa released following the incubation of adhered monocytes with small immune complexes appeared to be subclass dependent, however, complex size did not influence the magnitude of the TNFa response. The blocking of FcyRII or FcyRIHa resulted in a partial inhibition of the release of TNF a from adhered monocytes in response to the small IgGl anti-NIP complexes. Furthermore, FcyRIIIa appeared to play a more dominant role than FcyRII in the induction of TNFa release form cultures in response to these small IgGl immune complexes.

161 3.7. Human IgG rheumatoid factor mAb This study has therefore established that small soluble human IgG immune complexes could induce the release of TNFa from adhered human monocytes under the conditions of the in vitro system used in this project. However, the actual complex relevant to the hypothesis of this thesis, was the self-associating IgG rheumatoid factor. Two IgG rheumatoid factor monoclonal antibodies (an IgGl and an IgG2) produced by Professor PP Chen and colleagues (Lu et ah, 1993) were used to investigate the ability of small IgG rheumatoid factor immune complexes to induce TNFa production by adhered human monocytes in vitro.

3.7.1. Determining the potential for the IgGl rheumatoid factor mAb to self-associate

Experimental design In previous studies using these monoclonal antibodies, Lu et al. (1992; 1993) analysed the ability of both the IgGl and the IgG2 rheumatoid factor mAb to self-assodate by FPLC. Briefly, following a 30 hour incubation at 37°C, a shift towards the higher molecular weight of 300-kDa was observed which corresponded to IgG rheumatoid factor homodimers. Prior to any functional experiments, it was essential to confirm the ability of the IgGl and the IgG2 rheumatoid factor monoclonal antibodies to self-assodate. HPLC analysis of the IgGl rheumatoid factor monodonal antibody and, as a control the IgGl myeloma protein, was performed according to the protocol described in materials and methods (2.9.1.).

Results At time 0 the elution profile of the IgGl-RF mAb showed a large peak with an estimated molecular mass of 150-kDa, corresponding to IgGl-RF monomers. There was also a smaller peak with a MW of 470-kDa, corresponding to IgGl-RF trimers and a shoulder ranging from 580-kDa to a MW of greater than 660-kDa corresponding to IgGl-RF aggregates (Figure 2 2 a).

As shown in Figure 22b, following a 15 hour incubation at 37°C, the elution profile of the IgGl-RF mAb exhibited a large peak of IgGl-RF monomers, a

162 smaller peak of IgGl-RF trimers and a slightly larger peak occurring at a MW of more than 660-kDa (IgGl-RF aggregates).

As shown in Figure 22c, after a 40 hour incubation at 37°C, the IgGl-RF mAb exhibited a similar profile to the 15 hour incubation, however, the peak at the MW of greater than 660-kDa was equal to the peak of IgGl-RF trimers.

The elution profiles of the IgGl myeloma protein following an incubation at 37°C for 0, 15 and 40 hours all exhibited a peak at the molecular weight of 150-kDa corresponding to IgG monomers (Figure 22d-f).

163 Legend for Figure 2 2 . HPLC analysis of the IgGl rheumatoid factor mAb. HPLC elution profiles for the IgGl-RF mAb following incubation at 37°C for a) 0 hours, b) 15 hours or c) 40 hours and the human myeloma IgGl protein after an incubation at 37°C for d) 0 hours, e) 15 hours or f) 40 hours. Data are from one representative experiment.

164 280 50 C1C.7500 00- 0 .0 0 0 5 1 isoe.oc 500.00 186.00 100.00 1.66 1.66 5.86 .07.56 5.00 Time 7.50 165 16.96 10.00 13.61 13.61 13.61 280 100 50C - 1500.CO 500 - 1500.00 e e . e e i -

8 96 588 00 .013.61 1.60 1.66 .07. 8 .5 7 5.00 ee 5 Time 166 Time 7

5(1 7.56 0 00 10 0 00 10 j 6irj 13.61 3.7.2. Determining the potential for the IgG2 rheumatoid factor mAb to self-associate

Experimental design In the previous experiment (3.7.1.), the IgGl rheumatoid factor monoclonal antibody appeared to self-associate spontaneously. The potential for the IgG2 rheumatoid factor monoclonal antibody was therefore analysed by HPLC without preincubation at 37°C and the elution profile was compared with that of the IgGl rheumatoid factor monoclonal antibody as described previously.

Results Figure 23a shows the elution profile of the IgGl-RF mAb. As observed previously, there was a large peak of IgGl-RF monomers, a smaller peak of trimers and a shoulder ranging from 580-kDa to a MW greater than 660-kDa (IgGl-RF aggregates).

Figure 23b shows the elution profile for the IgG2-RF mAb. Interestingly, the profile was very different when compared with the elution profile for the IgGl-RF mAb. The IgG2-RF mAb exhibited a large peak at a MW of 270- kDa, corresponding with IgG2-RF dimers. Two larger peaks were observed at the MW of 610-kDa and greater than 660-kDa, both corresponding to IgG2-RF aggregates.

The IgG2 myeloma protein exhibited an elution profile identical to that obtained for the IgGl myeloma protein in the previous experiment (data not shown).

167 Figure 23. HPLC elution profiles for the a) IgGl-RF mAb and b) IgG2-RF mAb. Data are from one representative experiment. 1 0 0 C .G 0

800.00-

8.00

nin - 100.00 © 1.60 5.00 10.00 15.00 13.61 00 f N

lOOO.GS

225O.G0-

1500.00

750.00-

- 100.00 1.60 5.00 10.00 13.61 Time

168 3.7.3. Determining the effects of concentration on the elution profile of the IgGl rheumatoid factor mAh When self-associating IgG rheumatoid factors from patients with rheumatoid arthritis were first characterised, it was found that their aggregation appeared to be concentration dependent {Pope et al, 1975). Therefore, the effect of immunoglobulin concentration on the elution profile of the IgGl rheumatoid factor mAb was analysed.

Methods The IgGl rheumatoid factor monoclonal antibody was diluted in PBS to a concentration of lmg/ml and 1 0 0 (jig/ml and the two samples were then analysed by HPLC according to the protocol described in materials and methods -(2.9.1.-).

Results As shown by Figure 24, the elution profiles for the IgGl rheumatoid factor monoclonal antibody at lmg/ml and 1 0 0 pig/ml both displayed a large peak of IgGl-RF monomers. The elution profile of the IgGl-RF at the higher concentration exhibited a peak of IgGl-RF trimers and a shoulder corresponding to IgGl-RF aggregates. However, the IgGl-RF at the lower concentration exhibited a shift to less trimers and more aggregates.

169 IgGl rheumatoid factor mAh. HPLC elution profiles for the IgGl-RF mAb at at mAb the of IgGl-RF the profile for profiles elution elution HPLC HPLC the mAh. factor on rheumatoid concentration of IgGl effect The 24. Figure lm g/m l and lOOpg/ml. Data are from one representative experiment. representative one from are Data lOOpg/ml. and l g/m lm ^ 280 2250.00 1000 lGO.OO . 0 0 - 1.60 Time 7.50 170 mgml g/m lm 12.50 15.08 3.7.4. Determining the effects of fractionation on the stability of the IgGl rheumatoid factor mAh

Experimental design Preparative HPLC was performed on the IgGl rheumatoid factor monoclonal antibody as described in materials and methods (2.9.2.), to separate IgGl-RF complexes of various sizes. Once separated, the stability of these complexes was determined by re-analysing each fraction by analytical HPLC, as described previously. Following collection, each fraction was analysed immediately. Additionally, each fraction was stored at 4°C in 0.5% azide for 4 days and then re-analysed by analytical HPLC.

Results Fractions of IgGl-RF complexes were collected and HPLC analysis demonstrated certain fractions to contain only monomers, trimers or aggregates. Following a 4 day incubation at 4°C, HPLC analysis of these fractions demonstrated no significant shifts in the elution profiles, indicating the stability of the complexes once isolated from the heteropopulation.

Summary The HPLC analyses described in these experiments have shown that the IgGl rheumatoid factor monoclonal antibody spontaneously self-associated forming a mixed population of monomers, trimers and large aggregates. The IgG2 rheumatoid factor monoclonal antibody spontaneously self-associated forming a mixture of dimers and larger aggregates. When the heteropopulation of the IgGl rheumatoid factor monoclonal antibody was separated into fractions of known size and stored at 4°C for 4 days, there was no significant difference in the complex size within each fraction compared with those analysed on the day of preparation. This suggested the IgGl-RF mAb at each complex size to be stable when separated from the heteropopulation. The only dependency on concentration found was a slight shift of the large aggregates to form larger complexes as the IgGl-RF concentration decreased.

171 3.7.5. TNFa release from adhered monocytes following incubation with human IgG rheumatoid factor mAb The HPLC analyses have shown that both the IgGl and IgG2 rheumatoid factor monoclonal antibodies spontaneously self-associated. The IgGl rheumatoid factor mAb formed a heteropopulation of monomers, trimers and aggregates, while the IgG2 rheumatoid factor mAb formed a mixed population of dimers and aggregates. The ability of these rheumatoid factor preparations to induce TNFa release from adhered human monocytes was then examined.

Methods Adhered human monocytes were incubated with either the IgGl-RF mAb, the IgG2-RF mAb (both at 250 ^g/ml) or the appropriate controls as described in materials and methods (2.12.). Culture supernatants were collected at various time points over 24 hours and tested for TNFa according to the protocol described in materials and methods (2.13.2.).

Results Following the incubation of adhered human monocytes with either the IgGl-RF mAb or the IgG2-RF mAb, significant levels of TNFa were only detected in the supernatants from cultures incubated with the IgGl-RF mAb (Figure 25). The levels of TNFa released from cultures in response to the IgGl-RF mAb rapidly rose with time and peaked at 4 hours. The TNFa response then decreased with time. The LPS (in the absence of polymixin B) induced TNFa response rose rapidly with time, peaking after 6 hours, as observed previously. Low levels of TNFa were detectable in the supernatants from cultures incubated with the IgG2-RF mAb. However, these were comparable to TNFa levels detected in the supernatant from cultures incubated with medium alone.

This experiment was repeated six times by incubating adhered monocytes with the IgGl-RF or the IgG2-RF mAb at a range of concentrations (from lmg/m l to lpg/ml). In all subsequent experiments, TNFa was not detected in any of the supernatants. Cells did however, secrete TNFa following incubation with either LPS (in the absence of polymixin B) or the anti-FcyRIII mAb (3G8) as observed previously.

172 Further investigation was carried out by incubating adhered monocytes with the separated IgGl-RF complexes of various sizes, prepared by HPLC as described previously. Adhered human monocytes were incubated with either IgGl-RF monomers, trimers or aggregates, at a final concentration of 50jig/ml. Again, TNFa was undetectable in the supernatants from cultures that had been incubated with IgGl-RF complexes of any size. Cells were however viable and able to release TNFa following incubation with either LPS (in the absence of polymixin B) or the anti-FcyRIII mAb. Similar results were obtained when this experiment was repeated.

Summary Although the initial result showing the IgGl rheumatoid factor complexes to trigger TNFa release from adhered human monocytes was promising, the subsequent experiments probably reflected the inherent problems of trying to study stable IgG rheumatoid factor oligomers. This will be referred to in more detail in the discussion (4.3.2.).

173 Legend for Figure 25. The effects of IgGl and IgG2 rheumatoid factor mAb on TNFa release. The time course for TNFa production by adhered monocytes following incubation with the IgGl-RF mAb (red solid circle), the IgG2-RF mAb (pink closed square), LPS in the absence of polymixin B (blue closed triangle) or medium alone (green closed diamond). Data are from one representative experiment.

174 1 H- i-t rt> N> N> o nj 00 n O ^ 4 ro 175 TNF (ng/ml) o 00 O O' o k> o o o o o

Time (Hours) 3.8. Inhibitor studies As shown previously in this chapter, the anti-FcyRIII mAb (3G8), LPS (in the absence of polymixin B) and the small IgGl anti-NIP mAb complexes were all able to elicit a TNFa response from adhered human monocytes, similar in both kinetics and magnitude. Therefore, the nature of the TNFa response from adhered human monocytes in vitro was further analysed using various inhibitors.

All of the following inhibitor studies were performed in the absence of polymixin B, as the effects of this chemical upon inhibitor function were unknown. Therefore, prior to these inhibitor studies, adhered monocytes were incubated with the anti-FcyRIII mAb or the small IgGl anti-NIP immune complexes in both the presence and absence of polymixin B to ensure there would be no LPS induced TNFa response occurring as a result of reagent contamination. Cultures incubated with the anti-FcyRIII mAb or the small IgGl anti-NIP complexes in the absence of polymixin B induced levels of TNFa equivalent to those released from similar cultures incubated in the presence of polymixin B.

3.8.1. The effects o f inhibitors on TNFa release from adhered monocytes

Experimental design The inhibitors cycloheximide, actinomycin D and colchicine were used to determine whether the TNFa response from adhered monocytes following incubation with the anti-FcyRIII mAb, LPS or the small IgGl anti-NIP complexes was due to the release of preformed or newly synthesised TNFa, and also to assess the role of microtubules and secretory vesicles during the release of this cytokine into culture supernatants. Matrix metalloproteinases such as the gelatinases, in addition to matrix degradation, also process surface bound pro-TNFa into the mature, soluble cytokine (Gearing et at, 1994). Therefore the effects of a gelatinase-spedfic MMP inhibitor on the release of soluble TNFa was also investigated.

M ethods Adhered human monocytes were incubated with the anti-FcyRIII mAb (50jig/ml), LPS (0.5pg/ml) or the small IgGl anti-NIP complexes (50|xg/ml) as described previously. All cultures were incubated in the presence or

176 absence of either cycloheximide (50jig/ml), actinomycin D (l|xg/ml), colchicine (2 jug/ml) or the MMP inhibitor (1746; 50jiM). Supernatants were collected following an incubation of 4 hours and tested for TNFa as described previously.

Results As seen in Figure 26a, following a 4 hour incubation, cycloheximide significantly inhibited TNFa release in response to the anti-FcyRIII mAb (100.0 ± 0.1%; p<0.001) and LPS (99.9 ± 0.3%; p<0.005). Treatment of adhered monocytes with actinomycin D also significantly inhibited both the anti- FcyRIII mAb-induced TNFa response (93.8 ± 5.2%; p<0.001) and the LPS- induced TNFa response (74.2 ± 10.4%; p<0.01). The presence of colchicine did not significantly inhibit the anti-FcyRIII mAb response (30.1 ± 22.3%) or the LPS response (32.2 ±11.1%). Presence of the MMP inhibitor significantly inhibited TNFa release in response to both the anti-FcyRIII mAb (84.2 ± 15.2%; p<0.005) and LPS (76.8 ± 13.3%; p<0.005).

In a separate experiment, following a 4 hour incubation, cycloheximide inhibited TNFa release in response to both the small IgGl anti-NIP complexes and LPS by 100% (Figure 26b). Treatment of adhered monocytes with actinomycin D inhibited the small IgGl anti-NIP complex-induced response by 90% and the LPS-induced response by 6 8 %. The presence of colchicine did not inhibit the small IgGl anti-NIP complex-induced response, but instead enhanced TNFa release by 28%. The LPS-induced response in the presence of colchicine was partially inhibited by 2 2 %. Presence of the MMP inhibitor significantly blocked the TNFa response to the small IgGl anti-NIP mAb complexes by 87% and the TNFa response to LPS by 77%. These results were confirmed when this experiment was repeated.

Summary The presence of cycloheximide abrogated TNFa release in response to LPS, the anti-FcyRIII mAb and the small IgGl anti-NIP complexes. However, actinomycin D and the MMP inhibitor had a stronger inhibitory effect upon both the anti-FcyRIII mAb and the small IgGl anti-NIP complex induced TNFa response compared with the LPS-induced response. Colchicine had a greater inhibitory effect upon LPS-induced TNFa release compared with the

177 anti-FcyRIII mAb response. Interestingly, colchicine enhanced the TNFa response following incubation with the small IgGl anti-NIP complexes. The inhibitors did not interfere with the TNFa ELISA.

Legend for Figure 26. The effect of inhibitors on TNFa release. a) Barchart showing the levels of TNFa released by cultures following a 4 hour incubation with the anti-FcyRIII mAb or LPS in the presence of medium, cycloheximide, actinomycin D, colchicine or the MMP inhibitor. Values are expressed as the mean ± S.D. from three separate experiments. b) Barchart showing the levels of TNFa released by cultures following a 4 hour incubation with the small IgGl anti-NIP complex or LPS in the presence of medium, cydohexamide, actinomycin D, colchicine or the MMP inhibitor. Data are from one representative experiment.

178 TNF (ng/m l) at 4 hours

0.002 V-0J004

179 h i £ * NJ o o 2 1 r □ □ □ L_ ON ON __ 4^ _ 1 _ - _ v -• _ 1 _ ■ mmsm m mm 180 L_ O __ \ *!,"/ _ V-- _ *!,"/ \ \ ; -v

O O 00 L_ TNF (ng/m l) at 4 hours -r _ ' 1 V 1 L_ ON ON O O r - __ L_ O O 4^ __ L_ o NJ NJ __ o o Medium Cycloheximide Actinomycin D Colchcine MMP inhibitor CHAPTER 4

DISCUSSION

181 The aim of this thesis was to test a hypothetical model proposing an IgG rheumatoid factor-triggered effector mechanism for inflammation in rheumatoid arthritis (Edwards & Cambridge, 1998). This model predicts that small immune complexes comprising self-associating IgG rheumatoid factors, owing to their small size, may evade complement- mediated clearance from the circulation and access tissue macrophages. The model further predicts that these immune complexes initiate inflammation within the synovium by preferentially binding to the surface im m unoglobulin receptor, FcyRIIIa. FcyRIHa crosslinking may trigger macrophage activation and the subsequent production of proinflammatory mediators such as TNFa and reactive oxygen species.

4.1. Fey receptor ligation by murine anti-FcyR mAb

4.1.1 Production of cytokines and free radicals following incubation with anti-FcyR mAb The initial objective of this study was to determine the ability of the macrophage receptor, FcyRIIIa to m ediate TN Fa release following receptor ligation. Previous work had shown the release of TNFa from human peripheral blood monocytes following non-specific Fey receptor ligation with IgG immune complexes (Debets et al., 1988; Polat et al., 1993). Additionally, the production of TNFa by monocytes was confirmed to be a result of Fey receptor crosslinking rather than the process of immune complex ingestion (Debets et al., 1988). Further work by Debets et al. (1990) investigated the ability of FcyRI and FcyRII to mediate TNFa release from monocytes. Their observations showed that TNFa release from monocytes was only achieved through FcyRI following receptor crosslinking with murine IgG2a in solid phase. Similarly, FcyRII only mediated TNFa release following receptor crosslinking with murine IgGl in solid phase after the monocytes were pretreated with either proteases or neurominidase. FcyRIHa crosslinking had also been shown to mediate TNFa production, however, this had only been demonstrated in natural killer cells (Anegdn et a l , 1988; Hendrich et ah, 1991).

182 Originally it was decided to repeat the protocol described by Debets et al. (1990) for FcyRIIIa. However, this raised complications. Monocytes required adherence to plastic for at least 24 hours for all three FcyR to be expressed. This is possibly why Debets et al. (1990) only studied TN Fa production mediated through FcyRI and FcyRII. As shown in this dissertation and by others (Kelley et al., 1987; Krause et a l, 1996-97), the adherence of monocytes to tissue culture plastic over 24 hours was itself a stimulus for TNFa production, as was the procedure of removing and reseeding monocytes following initial maturation. Therefore, TNFa production by pre-matured monocytes seeded into plates coated with either the anti-FcyRIII mAb or murine IgGl would be uninterpretable and it is unclear how Debets et al. (1990) avoided this. Upon further examination of the literature, work by Trezzini et al. (1990) suggested that for FcyRIIIa-mediated monocyte activation, multiple crosslinking might not be necessary. It was found that of the three anti- FcyR mAb, only the anti-FcyRIII mAb in fluid phase was able to trigger reactive oxygen species production from non-adhered monocytes (Trezzini et al ., 1990).

Indeed, when adhered human monocytes were incubated with anti-FcyR mAb in fluid phase, in the absence of secondary crosslinking, only the anti-FcyRIII mAb triggered TNFa release. Similarly, of the three anti- FcyR mAb, only that directed against FcyRIII triggered adhered monocytes to release IL-la and reactive oxygen species. These observations, in addition to those reported by Debets et al. (1990) and Trezzini et al. (1990), implied that FcyRI and FcyRII both required the crosslinking of multiple receptors, whilst only two or three FcyRIIIa receptors needed to be co­ ligated for the signalling of TNFa, IL-la and reactive oxygen species production.

It had also been decided to attem pt to m odulate FcyRIIIa expression and see how this may have affected levels of TNFa production following FcyRIIIa liagtion. TGFp has been reported to upregulate FcyRIIIa expression (Welch et al., 1990; W ong et al., 1991), however, preliminary results suggested that TGFp acts to inhibit FcyRIIIa expression in vitro. This conflict was further supported by the results presented in this thesis showing the presence of TN Fa to have no effect upon FcyRIIIa

183 expression, while others have shown this cytokine to downregulate receptor levels (Liao & Simone, 1994; Liao et al., 1994). In light of these varying reports, such studies were not pursued.

The anti-FcyRIII mAb would have the potential to ligate up to three FcyR. Two FcyRIIIa receptors could be ligated specifically through the mAb Fab regions, while a third FcyR could be non-specifically ligated through the antibody's Fc fragment (Figure 27). It therefore became clear that the anti- FcyRIII mAb might be a good surrogate for the specific type of complex that may be involved in rheum atoid arthritis i.e. a small IgG complex with two or three available FcyR binding sites.

Figure 27. A schematic diagram showing that an anti-FcyR mAb has the potential to ligate up to three Fey receptors.

Anti-Fey R mAb Anti-Fey R mAb Fab regions Fc fragment

Fey receptor —

4.1.2. FcyRIIIa crosslinking by anti-FcyRIII mAb andTNFa release As suggested by others (Debets et al., 1990), a true IgG Fc-FcyR interaction may be required for intracellular signalling to result in TNFa release. Alternatively, it has been suggested that signalling is dependent upon receptor aggregation and crosslinking rather than ligand binding (van de

184 Winkel & Capel, 1993). Using blocking antibodies to FcyRI or FcyRII it was found that anti-FcyRIII mAb-mediated TNFa release occurred independently of these two receptors, suggesting there was no additional recruitment of either FcyRI or FcyRII by the Fc region of the anti-FcyRIII mAb.

When the anti-FcyRI mAb used in these studies was originally generated it was found to completely inhibit the binding of opsonised erythrocytes to U937 cells and also murine IgG2a anti-T3 mAb induced T cell proliferation. However, it was also found that this anti-FcyRI mAb did not block the binding of monomeric IgG to U937 cells (Dougherty et al., 1987). This later observation casts some doubt over the ability of this monoclonal antibody to efficiently block FcyRI binding of soluble antibody, such as the Fc region of the anti-FcyRIII mAb. Although FcyRI binds murine IgG2a with high affinity, the anti-FcyRIII mAb was of subclass IgGl for which FcyRI has a relatively low affinity (Jones et a l, 1985). Additionally, in vivo FcyRI is normally saturated with monomeric IgG and to displace this requires a large immune complex (Tax & van de Winkel, 1990). It is possible that the FcyRI expressed by adhered human monocytes in these studies may have been saturated with human IgG. This rules out the ability of FcyRI to be recruited by the Fc region of the anti-FcyRIII mAb for the intracellular signalling of TNFa. However, this then raises the argument that the lack of an anti- FcyRI mAb-induced TNFa response may be due to the receptor being already saturated with human IgG. FcyRI expressed by adhered monocytes that have been cultured for at least 48 hours should be free of monomeric IgG owing to receptor recycling (Dr PK Wallace, personal communication). When adhered monocytes were cultured for 48 or 72 hours and subsequently incubated with the anti-FcyRI mAb, no TNFa response was observed (data not shown). Based on these arguments one may infer that FcyRI could neither signal following ligation by the anti-FcyRI mAb, nor be recruited by the anti-FcyRIII mAb for TNFa release in these studies.

Although FcyRI and FcyRII appeared to not be involved in the anti- FcyRIII m Ab induced TN Fa response, the Fc fragm ent of this m onoclonal antibody may have ligated another FcyRIIIa receptor. It was

185 found that neither F(ab ')2 nor Fab fragments of the anti-FcyRIII mAb were able to induce TNFa production. This could imply that for signalling and subsequent TNFa release, stimulation through FcyRIIIa requires recruitment of three receptors. Alternatively, ligation of only two FcyRIIIa receptors may be mandatory, but, if so, one interaction m ust involve binding through an Fc domain (Figure 28).

The crosslinking of either two or three FcyRIIIa receptors by the anti- FcyRIII mAb relies in both cases upon an interaction between the Fc region of the anti-FcyRIIIa mAb and a FcyRIIIa receptor. Kipps et al. (1985) showed that FcyRIIIa expressed by Killer cells was unable to m ediate ADCC in response to monomeric murine IgGl, suggesting a low affinity of FcyRIIIa for this m ouse subclass. However, recent data has show n that the Fc fragment of murine IgGl can bind FcyRIII. Kocher et al. (1998) showed that the Fc fragment of murine IgGl monoclonal antibodies directed against surface bound neutrophil granule proteins preferentially ligated the low affinity FcyRIEb following initial Fab binding. The interactions of these monoclonal antibodies are closely analogous with what appears to be happening with the anti-FcyRIII mAb, where one or two Fab interactions may be stabilising a further interaction through Fc, particularly since FcyRIIIa has a higher affinity for m onom eric IgG than FcyRIHb (Vance et al.f 1993). Moreover, since the anti-FcyRIII mAb would initially bind FcyRIIIa through its Fab region(s), the affinity of the anti- FcyRIII Fc region for subsequent FcyRIIIa ligation may be greater than that of the unbound antibody.

186 Figure 28. The signalling for TN Fa release through FcyRIIIa requires recruitm ent of either a) three FcyRIIIa receptors or b) two FcyRIIIa receptors where one interaction must involve binding through an Fc domain.

Fc fragment Fab region

4.1.3. FcyRIIIa crosslinking by anti-FcyRIII mAh and reactive oxygen species production Trezzini et al. (1990) found that anti-FcyRIII F(ab’)2 fragments were able to trigger monocytes to release reactive oxygen species of a magnitude similar to the response induced by the whole antibody. In this thesis it was found that the anti-FcyRIII F(ab ')2 fragments also triggered adhered human monocytes to generate reactive oxygen species, however, the response was markedly weaker than that induced by the whole antibody. Both the findings of Trezzini et al. (1990) and those reported in this thesis suggest that the FcyRIIIa mediated respiratory burst may be dependent upon the num ber of FcyRIIIa receptors crosslinked rather than an IgG Fc- FcyRIIIa interaction. These results suggest that the aggregation of two FcyRIIIa receptors may be sufficient for activation of the membrane-

187 bound NADPH-oxidase, whereas intracellular signalling and subsequent cytokine production is less readily attained.

FcyRI and FcyRII may still require multiple receptor crosslinking for a similar reactive oxygen species response. Studies by Anderson et al. (1986) have shown that reactive oxygen species production by U937 cells through FcyRI could be achieved using the anti-FcyRI mAb (32.2) which binds an epitope outside of the receptor ligand-binding site. However, a response was only achieved when this anti-FcyRI mAb was further crosslinked with a F(ab ')2 anti-murine immunoglobulin. This implies that the FcyRI-mediated reactive oxygen species production may have been dependent upon multiple FcyRI aggregation rather than an IgG Fc- FcyR interaction, further supporting the findings of this thesis.

4.1.4. Fey receptor crosslinking by anti-FcyR mAb and subsequent signalling It may be difficult to understand why FcyRIIIa should function differently from FcyRI and FcyRII in response to a binding site-specific ligand such as the murine anti-FcyR mAb, particularly since FcyRIIIa signal transduction is dependent upon the identical gamma chain homodimer that is utilised by FcyRI (Ra et a l, 1989; Ernst et al., 1993). All three Fey receptors can mediate similar functions. However, there are differences in both their binding capacities for human and murine IgG and their signalling pathways. FcyRIIa is the only Fey receptor to possess its own cytoplasmic signalling motif (van den Herik Oudijk et a l, 1995b) but it can also utilise the gamma chain for signalling (Masuda et a l, 1993). Owing to its low affinity for IgG, which can be increased following proteolytic activity, FcyRII has been suggested to act as a "bystander” receptor (Tax & van de Winkel, 1990). Additionally, recent crystal structure data suggests that the Fc region of a single IgG molecule can bind either one or two FcyRII receptors (Sondermann et a l, 1999a; Sonderm ann et al., 1999b). FcyRIIIa is only known to signal through the gamma chain homodimer (Wirthmuellar et ah, 1992), while FcyRI has been shown to signal through either the gamm a chain or by recruiting FcyRIIa, depending on the differentiation state of the cell (Melendez et al., 1998). These differentials between the three Fey receptors may have a significant

188 impact on their functional properties. Fey receptor clustering is known to result in the phosphorylation of the IT AM signalling motif within the cytoplasmic domains of the gamma chain and FcyRIIa. Therefore, the number of receptors crosslinked may determine the number of tryrosines that are phosphorylated and this in turn may influence the signalling potential of each Fey receptor. Fey receptor aggregation may therefore only result in the signalling of, for example TNFa release, through a particular FcyR if sufficient receptor numbers have been crosslinked.

Overall, these findings indicate that for the signalling of TNFa and IL- la release from adhered human monocytes, FcyRIIIa requires a maximum of three receptors to be crosslinked, while the crosslinking of only two FcyRIIIa receptors is required for reactive oxygen species production. FcyRI and FcyRII may require multiple receptor crosslinking for TNFa, IL-la and reactive oxygen species production. This would support the concept that a small IgG immune complex may have the potential to trigger the production of TNFa and other proinflammatory m ediators by macrophages through FcyRIIIa but not through FcyRI or FcyRII.

4.1.5. Effect of inhibitors on anti-FcyRIII mAb and LPS triggered TNFa release Both the LPS and anti-FcyRIII mAb induced TNFa responses were abrogated by cycloheximide and significantly inhibited by the presence of actinomycin D. This confirmed that the TNFa detected in supernatants required both mRNA transcription and protein translation, and was not due to mobilisation of intracellular stores of TNFa. The two major matrix metalloproteinases secreted by activated monocytes are gelatinase B (MMP-9) and interstitial collagenase (MMP-1) and gelatinase B is spontaneously released from macrophage-derived monocytes in vitro (Saren et a l, 1996). Although matrix metalloproteinases have been shown to process pro-TNFa into its mature form (Gearing et a l, 1994), there have been reports suggesting otherwise (Black et a l, 1996). The gelatinase-specific matrix metalloproteinase inhibitor significantly reduced both LPS and anti-FcyRIII mAb induced TNFa release from adhered monocytes. This indicated that the processing of membrane bound pro-TNFa was dependent upon this particular class of matrix

189 metalloproteinase. However, evidence from TACE knockout mice suggests that the processing of pro-TNFa is dependent upon this membrane-bound disintegrin-metalloproteinase in vivo (Black et al., 1997; Killar et al, 1999). Furthermore, TACE is known to be sensitive to MMP inhibitors (Black et ah, 19%). Therefore, the results presented in this thesis suggest that the gelatinase-specific inhibitor used in these studies also inhibited TACE thereby blocking TNFa release from adhered monocytes. Although not statistically significant, colchicine was found to reduce both LPS and the anti-FcyRIII mAb induced TNFa production by approximately 30%, suggesting some dependence upon microtubule polymerisation.

4.2. Fey receptor ligation by human IgG immune complexes IgG containing immune complexes have been shown to trigger human peripheral blood monocytes to produce TNFa (Polat et ah, 1993) and 11-1 (Chantry et ah, 1989). There have also been a number of studies comparing IgG immune complex characteristics with their binding to Fey receptors on human monocytes and neutrophils (Kurlander & Batker, 1982; Klaassen et ah, 1988; Huizinga et ah, 1989b; Voice & Lachmann, 1997). The effects of different IgG immune complexes on the activation of neutrophils, primarily the respiratory burst, through Fey receptors have been examined (Huizinga et ah, 1989a; Crockett-Torabi & Fantone, 1990; W alker et ah, 1991; Brunkhorst et ah, 1992; H undt & Schmidt, 1992). However, similar studies have not been performed on monocytes. It is difficult to extend the information regarding the Fey receptor binding of differing IgG immune complexes and subsequent neutrophil activation to monocytes since untreated neutrophils express only FcyRII and FcyRIIIb. Furthermore, although the neutrophil GPI-linked FcyRIDb may have the potential to signal, the pathways involved are distinct from the signalling m echanism of the transm em brane isoform, FcyRIIIa (Brunkhorst et ah, 1992; Horejsf et ah, 1999).

4.2.1. Human IgG anti-NIP immune complexes and TNFa release from adhered monocytes The formation of immune complexes using a multivalent antigen, such as the NIP coupled to BSA used in these studies, can be manipulated to alter lattice formation (Mannik, 1980). Small, soluble human IgG

190 immune complexes were prepared from human IgGl anti-NIP monoclonal antibodies in an excess of antigen (2.4NIP/BSA) and also in an excess of antibody. Insoluble complexes of human IgGl anti-NIP were prepared at molar equivalence. Only the soluble IgG anti-NIP mAb complexes prepared in an excess of antigen were able to trigger adhered human monocytes to release TNFa. Soluble and insoluble IgG immune complexes have been shown to activate the respiratory burst in neutrophils through different pathways (Crockett-Torabi & Fan tone, 1990), however, the lack of significant TNFa release in response to the insoluble IgG anti-NIP immune complexes and the soluble complexes prepared in an excess of antibody, was surprising. One possible explanation for this observation is that the insoluble complexes may have formed a lattice which had precipitated during preparation, and therefore may not have been able to activate sufficient number of cells needed to generate a detectable TNFa response. However, in relation to the proposed mechanism of pathogenesis of rheumatoid arthritis, it was the soluble IgG immune complexes that were of interest. The lack of response following incubation with the soluble complexes prepared in antibody excess may have been a result of the complex preparation. In light of previously described protocols (Crockett-Torabi & Fantone, 1990; Voice & Lachmann, 1997) it was decided to keep the IgG anti-NIP mAb concentration constant and alter the antigen's concentration to generate antibody and antigen excess. In retrospect, this may not have been the ideal methodology. In the case of the antibody excess preparation, there were possibly not enough complexes formed to have been able to trigger a TNFa response. Alternatively, the anti-NIP mAb prepared in antibody excess, unlike in antigen excess, may not have been suffciently crosslinked to have been able to elicit an FcyR-mediated response.

No obvious relationship was found between soluble IgGl anti-NIP complex size and the levels of TNFa released from adhered human monocytes. However, the subclass specificity of the small soluble immune complexes did influence the magnitude of the TNFa response. Fey receptors have differing affinities for each hum an IgG subclass. Small soluble IgGl and IgG3 anti-NIP immune complexes both evoked a strong TNFa response from adhered human monocytes. This is consistent with previous studies showing IgGl and IgG3 complexes to bind human

191 phagocytes (Kurlander & Batker, 1982; Klaassen et al., 1988; Huizinga et ah, 1989b). It has been shown by others that complexes of subclass IgG2 or IgG4 cannot efficiently bind human monocytes or neutrophils (Kurlander & Batker, 1982; Huizinga et ah, 1989b; Voice & Lachmann, 1997). Therefore, the observation that the small soluble IgG2 and IgG4 anti-NIP complexes were able to trigger TNFa release from adhered human monocytes was unexpected. The relative levels of TNFa released in response to each subclass did, however, correlate with the Fey receptor binding affinities of both FcyRII (Salmon et ah, 1992) and FcyRIII (van de W inkel & Capel, 1993) for hum an IgG.

4.2.2. Human IgG immune complex-induced TNFa release and the in vo lvem en t o f Fey receptors Blocking antibodies to each FcyR were used to determine which receptors were involved in small IgGl anti-NIP complex-induced TNFa release from adhered monocytes. Using blocking antibodies to FcyRI, FcyRII and FcyRIII, the TNFa response induced by these small, soluble IgGl immune complexes was found to be dependent upon both FcyRII and FcyRIIIa, where FcyRIIIa appeared to play a dom inant role. These im m une complex studies therefore support the hypothesis that small IgG immune complexes may trigger the production of TNFa from macrophages in rheumatoid arthritis, and implicate an important role for FcyRIIIa in this inflammatory process.

4.2.3. Effect of inhibitors on human IgG immune complex and LPS induced TNFa release As observed for both the LPS and anti-FcyRIII mAb-induced TNFa responses, the TNFa response induced by soluble small IgGl anti-NIP immune complexes was abrogated by the presence of cycloheximide and significantly inhibited by actinomycin D. This confirmed the soluble IgGl immune complex-induced TNFa response to be dependent upon both mRNA transcription and protein translation. Furthermore, the inhibitory effect of the MMP inhibitor, as for LPS and anti-FcyRIII mAb induced TNFa responses, again suggested the dependency upon gelatinase activity for TNFa processing. Interestingly, the presence of colchicine enhanced the small IgGl immune complex-induced TNFa release from adhered monocytes. This effect was unlikely to be due to the

192 prevention of Fey receptor internalisation, which may have possibly resulted in prolonged signal transduction. Work by others have shown that Fey receptor internalisation occurs independently of microtubule function and instead occurs by receptor diffusion into the plasma membrane (Michl et ah, 1983). Furthermore, colchicine was found to inhibit, rather then enhance, the anti-FcyRIII mAb induced TNFa response. However, colchicine has been shown by others to increase IL-ip production from LPS-stimulated monocytes (Allen et ah, 1991) and the production of platelet derived growth factor B mRNA by macrophages (Wangoo et al., 1992).

4.2.4. Human IgG anti-NIP immune complexes Although the above human IgG immune complex studies generated some encouraging data, there are a number of drawbacks in using these anti-NIP complexes. Owing to the nature of the antigen, there were uncertainties regarding complex size. The NIP/BSA antigen used to generate small complexes had an average NIP density of 2.4 molecules per BSA, forming a heteropopulation of monomers, dimers and trimers. Steric hindrance would possibly prevent the formation of tetramers. It was not possible to predict the precise proportions of monomers, dimers and trimers in each preparation. Routine HPLC analysis of the immune complexes prior to incubation with adhered monocytes was unfortunately not possible since the HPLC equipment used in these studies was based in a collaborative laboratory which could only supply restricted use. Furthermore, reagents were limited.

Another consideration to be made when interpreting the data is the stereochemistry of these anti-NIP complexes in relation to IgG rheumatoid factor complexes. The formation of an IgG anti-NIP dimer would be of a similar size to an IgG rheumatoid factor dimer. In addition, both types of dimers would have equal numbers of available FcyR binding sites. However, steric differences may affect the accessibility of these sites (Figure 29). Furthermore, it was impossible to control the position of the NIP molecules on the surface of the BSA. Therefore, the position of the two IgG anti-NIP antibodies in relation to each other could not been controlled (Figure 30).

193 Figure 29. Schematic diagram showing formation of a) human IgG anti- NIP dimer and b) human IgG rheumatoid factor dimer.

BSA with 2 NIP molecules

Avalible Fey R binding sites on Fc fragment

b)

Avalible Fey R binding site on Fc fragment

194 Figure 30. The stereochemistry of the IgG anti-NIP dimers could not be controlled.

4.3. H um an IgG rheum atoid factors

4.3.1. IgG rheumatoid factor mAb self-association Lu et al. (1993) generated the two human IgG rheumatoid factor mAbs used in these studies and showed that both the IgGl and the IgG2 rheumatoid factor mAb self-associated to form dimers following incubation at 37°C for 30 hours. The dissociation constant of the IgGl rheumatoid factor mAb was found to be 5.7 x 10'7M and that of the IgG2 rheumatoid factor mAb was 1.6 x 10_7M. When these experiments were repeated in this thesis, it was found that the self-association of both rheumatoid factor subclasses occurred spontaneously. The human IgGl rheumatoid factor mAb formed a heteropopulation of monomers, trimers and aggregates, while the IgG2 rheumatoid factor mAb self­ associated to form dimers and larger complexes. These immune complexes appeared stable both within the heteropopulations and following separation.

195 4.3.2. Human IgG rheumatoid factor-induced TNFa release TNFa production from adhered human monocytes was only detected in response to the IgGl rheumatoid factor mAb complexes, as expected. However, this result was only observed once, implying this response might be an artefact. Alternatively, the particular conditions during this experiment may have been optimal for immune complex formation and subsequent TNFa production. Although the HPLC data indicated that once formed these complexes appeared stable, as for the IgG anti-NIP complexes HPLC analysis prior to incubation with adhered monocytes was not feasible. Therefore, it was not possible to accurately determine the proportion of IgG rheumatoid factor dimers in each preparation. Moreover, the presence of fetal calf serum in cultures necessary for TNFa release from adhered human monocytes, may have influenced the stability of the IgG rheumatoid factor complexes.

These experiments were not pursued at this stage since there is a need for a better control over complex stability and size. Furthermore, there is the question of whether these rheumatoid factors are truly representative. Most rheumatoid factor mAb are of the class IgM since IgG rheumatoid factor mAb are difficult to generate and analyse in detail. This may be due to low levels of immunoglobulin secretion or the instability of EBV- transformed B cells (Lu et al, 1992). It is possible that the IgG rheumatoid factor mAb used in these studies, although generated from rheumatoid synovial cells, are not representative of "arthritogenic" rheumatoid factors.

4.4. TNFa and IL-la production by adhered human monocytes As already discussed, of the three anti-FcyR mAb only the anti-FcyRIIIa mAb induced TNFa and IL-la production from adhered human monocytes. It was interesting to note that IL-la time course differed from that for TNFa. The TNFa response peaked between 4 and 6 hours after stimulation, and subsequently declined with time. The IL-la response was only detectable from 10 hours following stimulation, as observed by others (Hazuda et ah, 1988), and then continued to rise with time. This lag observed for the IL-la response suggests a possible dependency upon the TNFa response.

196 Dinarello et al (1986) demonstrated the induction of IL-1 production by human mononuclear cells in response to human recombinant TNFa, as did Chantry et al (1989) from human monocytes. However, both studies tested for IL-1, rather than the individual isoforms. Nevertheless, this was further supported by the observation that the spontaneous production of IL-1 by isolated rheumatoid synovial mononuclear cells in vitro was TNFa dependent (Brennan et al, 1989). In agreement with these findings, the anti-FcyRIII mAb-induced IL-la production by adhered human monocytes was inhibited by the presence of a neutralising anti-TNFa mAb. However, recombinant human TNFa alone failed to induce IL-la production from the adhered human monocytes. This may imply that TNFa was involved in the release of intracellular IL-la that had already been synthesised following the ligation of FcyRIIIa, rather than the TN Fa itself being the stim ulus for IL- la production.

In contrast to the anti-FcyRIII mAb, LPS-induced IL-la production was not inhibited by the neutralising antibody to TNFa. LPS may itself trigger both IL-la synthesis and release therefore avoiding the dependency upon TNFa. Altenatively, this may be explained by the downregulation of both TNF receptors, TNF-RI and TNF-RII, together with the release of soluble TNF-RII, previously observed with LPS-mediated activation of monocytes (Leeuwenberg et al, 1994). TNF receptor expression by hum an monocytes and macrophages following FcyR-mediated activation has yet to be investigated. However, the expression of TNF receptors in rheumatoid synovium have been studied using immunostaining techniques. It was found that both TNF-RI and TNF-RII receptor expression levels were increased by rheumatoid synovial cells in comparison with controls and rheumatoid peripheral blood mononuclear cells (Deleuran et al, 1992; Brennan et al, 1992b). Furthermore, some of these cells were secreting TNFa (Deleuran et al, 1992).

The findings presented in this dissertation suggest that unlike LPS stim ulation, ligation of FcyRIIIa may favour the establishm ent of a pro- inflammatory cytokine loop initiated by TNFa, resulting in the release of other cytokines such as IL-la.

197 4.5. Rheumatoid factor, Fc receptors and rheumatoid arthritis The targeting of inflammation to synovium and other tissues in rheumatoid arthritis has previously been difficult to explain. Although circulating IgG rheumatoid factor-based immune complexes have long been suspected to be involved in the disease process, the effector mechanism has been unclear. The findings presented in this dissertation indicate that macrophage FcyRIIIa is the m ost likely Fey receptor to trigger TNFa, IL-la, and reactive oxygen species production following ligation by a small IgG-based im m une complex. Unlike FcyRI and FcyRII, FcyRIIIa expression by macrophages is tissue and site specific. A close correlation between macrophage FcyRIIIa expression and the location of both synovitis and extra-articular features has been demonstrated (Bhatia et al, 1998). The small (intermediate) IgG rheumatoid factor-containing immune complexes first described by Kunkel et al. (1961) may therefore be the pro-inflammatory stimulus for macrophage activation preceding T cell infiltration in the rheumatoid synovium. Once established, this inflammatory process may be subsequently amplified by the local generation of large complement-fixing complexes based on rheumatoid factors of all isotypes (Winchester et al., 1970; M annik & Nardella, 1995). Furthermore, the formation of large IgG rheumatoid factor complexes within the synovium may activate synovial macrophages through any of the three Fey receptors.

It may also be possible for IgA rheumatoid factor complexes to activate macrophages in rheumatoid arthritis through FcaR. This is supported by in vitro studies showing FcaR aggregation by IgA to induce TNFa, IL-lp and IL-6 release from human monocytes (Polat et al., 1993; Patry et al., 1995). Furthermore, FcaR expression by human monocytes can be upregulated by TNFa and IL-lp (Shen et al., 1994). However, although little is known about the tissue distribution of FcaR, an immunohistochemical study of tissues from both normal individuals and patients with rhuematoid arthritis has shown that FcaR is only expressed by gut macrophages (S Blades, personal communication). This suggests that IgA rheumatoid factor-based complexes may not activate macrophages through FcaR in rheumatoid arthritis.

198 In conclusion, the findings presented in this thesis, together with previous evidence (Edwards & Cambridge, 1998; Edwards et al, 1999) support a specific role for macrophage FcyRIIIa in the induction of inflammation induced by small circulating IgG rheumatoid factor immune complexes in rheumatoid arthritis.

4.6. B cell survival and IgG rheumatoid factor production in rheumatoid arthritis To complete the above model for the initiation and persistence of inflammation in rheumatoid arthritis, the continuous production of IgG rheumatoid factor requires explanation. The hypothesis has recently been extended to predict that IgG rheumatoid factors have the ability to perpetuate their own existence (Edwards et al., 1999). Moreover, the hypothesis suggests that pathogenic IgG rheumatoid factors do not arise from natural IgM rheumatoid factors as a result of a failure in the constraints that normally prevent these physiological autoantibodies from undergoing affinity maturation and class switching (Borretzen et al., 1994). Instead, the suggestion is that IgG rheumatoid factors with the potential for pathogenicity and natural IgM rheumatoid factors are derived from distinct origins.

B cell survival within the follicle centre requires positive survival signals in the form of T cell help for a source of cytokines and antigen complexed with the complement component, C3d (Lindhout et al., 1997; Fearon et al., 1995). A B lymphocyte recognising self should be deleted owing to a lack of T cell help (Bikoff, 1983). Autoreactive B cells might survive owing to the failure of T cell tolerance to self antigen. However, in the case of rheumatoid factor producing B cells, T cell help may be obtained without a failure in tolerance. As shown by Roosnek & Lanzavecchia (1991), antigen-antibody complexes containing foreign antigen can be efficiently taken up by rheumatoid factor B cells. The foreign antigen is then processed and presented, complexed with class II MHC, to T cells.

Normal individuals make physiological rheumatoid factors as part of a normal immune response. These autoantibodies are mostly low affinity IgM and are likely to be produced by B-la cells (Casali & Notkins, 1989b).

199 B-la cells produce "natural" autoantibodies which are often prevented from affinity maturation and class switching by restrictive mechanisms, possibly a result of the constitutive expression of the nuclear protein STAT-3 (Karras et ah, 1997). In contrast, rheumatoid factors from patients with rheumatoid arthritis display high levels of somatic hypermutation involving amino acid substitutions (Thompson et ah, 1995a). There is also evidence suggesting that rheumatoid factors from normal individuals and patients with rheumatoid arthritis are derived from distinct germ line gene families (Thompson et ah, 1994). Potentially pathogenic IgG rheumatoid factors may therefore be derived from conventional B-2 cells of bone marrow origin and may acquire rheumatoid factor activity by a chance mutation event during an immune response to an unrelated foreign antigen (Pulendran et ah, 1997). IgM antibodies against an irrelevant foreign antigen may therefore enter affinity maturation and class switching, during which a chance mutation may give rise to a high affinity IgG rheumatoid factor.

These IgG rheumatoid factors may have the ability to perpetuate their own production owing to their self-association properties (Pope et ah, 1974). For a B cell to survive, it must receive two positive signals. The first positive signal is from helper T cell derived cytokines (Lindhout et ah, 1997). The second is in the form of complexed antigen in association with C3d which provides a B cell with a survival signal by interacting with surface immunoglobulin and CR2 (Cooper et ah, 1990). Soluble antigen provides a negative survival signal to B cells within the germinal centre (Pulendran et ah, 1995) and therefore an autoreactive B cell should be deleted.

B cell survival is also regulated by the inhibitory receptor, FcyRIIbl. Co­ ligation of FcyRIIbl and B cell receptor (BCR) by complexed IgG provides a negative signal to the B cell, resulting in a downregulation of antibody production (Gergely & Sarmay, 1996). In contrast, the crosslinking of FcyRIIb2 results in receptor internalisation, delivering complexed IgG to lysosomes for antigen processing and presentation (Sandilands et ah, 1997). Since FcyRIIbl can only move within the plasma membrane and not mediate internalisation, FcyRIIbl crosslinking may also inhibit antigen processing and presentation to T cells (Minskoff et ah, 1998).

200 Normally a self reactive B cell may not recognise complexed antigen since the epitope that surface immunoglobulin (BCR) recognises is shared by free antibody. However, in the case of IgG rheumatoid factors, the antigen is the antibody. Therefore, while one epitope may be bound, the "antigen" can still present another to surface immunoglobulin. Furthermore, surface IgG rheumatoid factor may compete with FcyRIIb for IgG rheumatoid factor complexes. Thus positive signals may dominate and IgG rheumatoid factor producing B cells will survive and proliferate. Involvement of specific T cell reactivities is still possible, however, IgG rheumatoid factor production may be a self perpetuating process arising against the background of a normal T cell repertoire.

4.7. A new treatment for rheumatoid arthritis The hypothesis presented in this thesis provides the possibility of a new therapeutic strategy for rheumatoid arthritis. The pathogenic mechanism implicates IgG rheumatoid factors as the primary inflammatory trigger. Therefore, removal of these antibodies would prevent further disease progression. The obvious way to achieve this would be by destroying the rheumatoid factor-producing B cells. To specifically kill rheumatoid factor-producing B cells would be highly complicated. However, total B cell depletion would also result in the desired effect. Recently, a chimeric mouse human IgGl anti-CD20 monoclonal antibody has been generated (Reff et al., 1994). CD20 is a surface antigen expressed by all B cells from the stage of cytoplasmic \i heavy chain expression until plasma cell formation. It is not expressed by stem cells or early B cell precursors. CD20 is involved in the activation of cell cycle initiation and differentiation and ligation with the anti-CD20 monoclonal antibody results in B cell death. In recent trials, following a single infusion of the anti-CD20 monoclonal antibody, B cells were depleted from the peripheral blood, the bone marrow and the lymph nodes with no significant side effects. Over a period of time the B cell population then recovered to baseline levels, while immunoglobulin levels were maintained (Reff et al., 1994; Maloney et al, 1994).

Such a therapy would allow the repopulation of B lymphocytes, without the reappearance of the pathogenic rheumatoid factor producing B cells. Indeed such a trail is currently ongoing with positive results. Patients

201 with rheumatoid arthritis, following anti-CD20 treatment, have shown a dramatic and prolonged clinical improvement (Edwards, in preparation). This supports the hypothesis that rheumatoid arthritis is an antibody- driven disease.

4.8. Future work A number of questions regarding the structure and signalling properties of Fey receptors have been generated from the work presented in this thesis. Following the crystal structure studies of FcyRII, it would be of interest to generate similar data for FcyRIII. This may help to elucidate the relevance of IgG Fc valency for FcyRIIIa signalling. Alternatively, the generation of an anti-FcyRIII mAb consisting of an Fc fragment and a single Fab region may clarify whether the crosslinking of two or three FcyRIIIa receptors are required for the signalling of T N Fa production.

Recent data has shown that following crosslinking, FceRI (Stauffer & Meyer, 1997) and FcaR (Lang et al., in press) re-distribute to membrane glycosphingolipid-cholesterol rafts, which are rich in signalling molecules such as PTK. A collaboration is currently being organised with Dr W. Wade (Department of Immunology & Microbiology, Dartmouth College, Lebanon, NH) to determine whether Fey receptors, in particular FcyRIIIa, also enter rafts following crosslinking with murine anti-FcyR mAb.

LPS and FcyRIIIa appear to m ediate IL-la release from adhered monocytes through distinct pathways. FcyRIIIa-mediated IL-la release was found to be TNFa-dependent, while LPS-triggered IL-la release occurred independently of TNFa. It would therefore seem appropriate to stimulate adhered monocytes with both the anti-FcyRIII mAb and LPS in the presence of the anti-TNFa mAb to see if IL-la release could be restored. Furthermore, it may be possible to establish whether LPS activates intracellular signalling molecules distinct from those activated following FcyRIIIa ligation, such as specific mitogen-activated protein kinases or phospholipases. If such a molecule could be identified it may then be possible to inhibit its activation and observe the effects of this on LPS-mediated IL-la release in the presence of the anti-TNFa mAb.

202 Using an in vitro system, the work presented in this thesis supports a role for IgG rheumatoid factor complexes as a trigger of macrophage activation in rheumatoid arthritis. Human in vivo trials have demonstrated B cell depletion as a positive treatment for rheumatoid arthritis. The ultimate test would therefore be to demonstrate the presence and absence of circulating IgG rheumatoid factor complexes in patients with rheumatoid arthritis before and after anti-CD20 therapy. For this a highly sensitive and accurate method for the detection of IgG rheumatoid factor-based complexes is required. A novel approach to this problem is currently under investigation.

203 REFERENCES

204 Abo T, Tilden AB, Balch OM, Kumagai K, Traup GM, Cooper MD. Ethnic differences in the lymphocyte proliferative response induced by a murine IgGl antibody, Leu-4 to the T3 molecule. J Exp Med 1984;160:303-309.

Aho K, Heliovaara M, Reunanen A, Aromaa A, Leino A, Kurki P, Heikkila R, Palosuo T. Serum immunoglobulins and the risk of rheumatoid arthritis. Ann Rheum Dis 1997;56:351-356.

Albani S, Carson DA. A multistep molecular mimicry hypothesis for the pathogenesis of rheumatoid arthritis. Immunol Today 1996;17:466-470.

Allen JC, Kunkel HG. Hidden rheumatoid factors with specificity for native y globulins. Arthritis Rheum 1966;9:759-768.

Allen JM, Seed B. Isolation and expression of functional high-affinity Fc receptor complementary DNAs. Science 1989;243:378-381.

Allen JN, Herzyk DJ, Wewers MD. Colchicine has opposite effects on interleukin-1 beta and tumor necrosis factor-alpha production. Am J Physiol 1991; 261:L315-321.

Allen RC, Stjerholm RL, Steele RH. Evidence for the generation of an electronic excitation state(s) in human polymorphonuclear leucocytes and its participation in bactericidal activity. Biochem Biophys Res Com m um 1972;47:679-684.

Allen RC, Loose LD. Phagocytic activation of a luminol dependent chemiluminescence in rabbit alveolar and peritoneal macrophages. Biochem Biophys Res Commum 1976;69:245-252.

Alsalamen S, Firestein GS, Oez S, Kurrle R, Kalden JR, Burmester GR. Regulation of granulocyte macrophage colony stimulating factor production by human articular chondrocytes. Induction by both tumor necrosis factor-alpha and interleukin-1, downregulation by transforming growth factor beta and upregulation by fibroblast growth factor. J Rheum atol 1994;21:993-1002.

205 Alspaugh MA, Henle G, Lennette ET, Henle W. Elevated levels of antibodies to Epstein-Barr virus infection in sera and synovial fluids of patients with rheumatoid arthritis. J Clin Invest 1981;67:1134-1140.

Amigorena S, Bonnero TC, Drake JR, Choquet D, H unziker W, Guillet JG, Webster P, Sautes C, Mellman I, Fridman WH. Cytoplasmic domain heterogeneity and functions of IgG Fc receptors in B lymphocytes. Science 1992;256:1808-1812.

Anaya J, Diethelm L, Ortiz LA, Gutierrez M, Citera G, Welsh RA, Espinoza LR. Pulmonary involvement in rheumatoid arthritis. Semin Arthritis Rheum 1995;24:242-254.

Anderson CL, Abraham GN. Characterisation of the Fc receptor for IgG on a hum an macrophage cell line, U937. J Im m unol 1980;125:2735-2741.

Anderson CL. Isolation of the receptor for IgG from a human monocyte cell line (U937) and from human peripheral blood monocytes. J Exp Med 1982;156:1794-1805.

Anderson CL, Guyre PM, Whitin JC, Ryan DH, Looney RJ, Fanger MW. Monoclonal antibodies to Fc receptors for IgG on human mononuclear phagocytes. Antibody characterisation and induction of superoxide production in a monocyte cell line. J Biol Chem 1986;261:12856-12864.

Anderson CL, Looney RJ, Culp DJ, Ryan DH, Fleit HB, Utell MJ, Frampton MW, Manganiello PD, Guyre PM. Alveolar and peritoneal macrophages bear three distinct classes of Fc receptors for IgG. J Immunol 1990;145:196-201.

Anderson P, Caligiuri M, O'Brien C, Manley T, Ritz J, Schlossman SF. Fey receptor type III (CD16) is included in the £NK receptor complex expressed by human natural killer cells. Proc Natl Acad Sci USA 1990;87:2274-2287.

206 Andreesen R, Brugger W, Scheibenbogen C, Kreutz M, Leser HG, Rehm A, Lohr GW. Surface phenotype analysis of human monocyte to macrophage maturation. J Leukoc Biol 1990;47:490-497.

Anegbn I, Cuturi MC, Trinchieri G, Perussia B. Interaction of Fc receptor (CD16) ligands induces transcription of interleukin 2 receptor (CD25) and lymphokine genes and expression of their products in human natural killer cells. J Exp Med 1988;167:452-472.

Arend WP, Welgus HG, Thompson RC, Eisenberg SP. Biological properties of recombinant human monocyte-derived interleukin-1 receptor antagonist. J Clin Invest 1990;85:1694-1697.

Artandi SE, Calame KL, Morrison SL, Bonagura VR. Monoclonal IgM rheumatoid factors bind IgG at a discontinuous epitope comprised of amino acid loops from heavy-chain constant-region domains 2 and 3. Proc Natl Acad Sci USA 1992;89:94-98.

Assreuy J, Cunha FQ, Epperlein MM, Noronha-Dutra A, O'Donnell CA, Liew FY, Moncada S. Production of nitric oxide and superoxide by activated macrophages and killing of Leishmania major. Eur J Immunol 1994;24:672-676.

Auron PE, Webb AC, Rosenwasser LJ, Mucci SF, Rich A, Wolff SM, Dinarello CA. Nucleotide sequence of human monocyte interleukin 1 precursor cDNA. Proc Natl Acad Sci USA 1984;81:7907-7911.

Avrameas S. Natural autoantibodies: from ’horror autotoxicus' to 'gnothi seauton'. Immunol Today 1991;12:154-159.

Babior B. Oxidants from phagocytes: Agents of defence and destruction. Blood 1984a;64:959-966.

Babior B. The respiratory burst of phagocytes. J Clin Invest 1984b;73:599- 601.

Balkwill FR. Tumour necrosis factor. Brit Med Bull 1989;45:389-400.

207 Beekman B, van El B, Drijfhout JW, Ronday HK, Tekoppele JM. Highly increased levels of active stromelysin in rheumatoid synovial fluid determined by a selective fluorogenic assay. FEBS Lett 1997;418:305-309.

Benton HP, Tyler JA. Inhibition of cartilage proteoglycan synthesis by interleukin 1. Biochem Biophys Res Commun 1988;154:421-428.

Berken A, Benacerraf B. Properties of antibodies cytophilic for macrophages. J Exp Med 1967;123:119-144.

Bertolini DR, Ned win GE, Bringman TS, Sm ith DD, M undy GR. Stimulation of bone resorption and inhibition of bone formation in vitro by human tumour necrosis factors. Nature 1986;319:516-518.

Bevilacqua MP, Prober JS, Mendrick DL, Cotran RS, Grimbrone Jr MA. Interleukin 1 acts on cultured human vascular endothelium to increase the adhesion of polymorphonuclear leukocytes, monocytes and related leukocyte cell lines. J Clin Invest 1985;76:975-983.

Bhatia A, Blades S, Cambridge G, Edwards JCW. Differential distribution of FcyRIIIa in normal human tissues and co-localization with DAF and fibrillin-1: implications for immunological microenviroments. Im munology 1998;94:56-63.

Bianco NE, Dobkin LW, Schur PH. Immunological properties of isolated IgG and IgM anti-gamma-globulins (rheumatoid factors). Clin Exp Im m unol 1974;17:91-101.

Bignold LP, Martyn M, Basten A. Nodular pulmonary amyloidosis associated with benign hypergammaglobulinemic purpura. Chest 1980;78:334-336.

Bikoff EK. Tolerance for self IgG at the level of the Lyl+ T cell. J Exp Med 1983;158:1868-1880.

208 Birkedal Hansen H, Moore WG, Bodden M, Windsor LJ, Birkedal Hansen B, DeCarlo A, Engler JA. Matrix metalloproteinases: a review. Crit Rev Oral Biol Med 1993;4:197-250.

Black RA, Durie FH, Otten EC, Miller R, Slack JL, Lynon DH, Castner B, Mohler KM, Gerhart M, Johnson RS, Itoh Y, Okada Y, Nagase H. Relaxed specificity of matrix metalloproteinases (MMPs) and TIMP insensitivity of tumour necrosis factor-alpha (TNF-alpha) production suggest the major TNF-alpha converting enzyme is not an MMP. Biochem Biophys Res Com m un 1996;225:400-405.

Black RA, Rauch CT, Kozlosky CJ, Peschon JJ, Slack JL, Wolf son MF, Castner BJ, Stocking KL, Reddy P, Srinivasan S, Nelson N, Boiani N, Schooley KA, Gerhart M, Davis R, Fitzner JN, Johnson RS, Paxton RJ, March CJ, Cerretti DP. A metalloproteinse disintegrin that releases tumour-necrosis factor-a from cells. Nature 1997;385:729-733.

Blass S, Schuman F, Hain NAK, Engel J, Stuhlmiiller B, Burmester GR. p205 is a major target of autoreactive T cells in rheumatoid arthritis. Arthritis Rheum 1999;42:971-980.

Bona CA. V genes encoding autoantibodies. Molecular and phenotypic characteristics. Annu Rev Immunol 1988;6:327-358.

Bonagura VR, Artandi SE, Davidson A, Randen I, Agostino N, Thompson K, Natvig JB, Morrison SL. Mapping studies reveal unique epitopes on IgG recognised by rheumatoid arthritis-derived monoclonal rheum atoid factors. J Im m unol 1993;151:3840-3852.

Bonagura VR, Kwong T, Kenny T, Robbins DL, Morrison SL. The specificity of synovial IgM rheumatoid factors (RF) for genetically engineered IgG antibodies is not affected by the method used to immortalise RF-producing B cells. Scand J Immunol 1999; 49:106-111.

209 Borretzen M, Randen I, Zdarsky E, Forre 0, Natvig JB. Control of autoantibody affinity by selection against amino add replacements in the complementarity-determining regions. Proc Natl Acad Sci USA 1994;91:12917-12921.

Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utlizing the principle of protein-dye binding. Anal Biochem 1976;72:248-254.

Bredius RGM, Fijen CAP, de Haas M, Kuijper EJ, Weening RS, van de W inkel JGJ, Out TA. Role of neutrophil FcyRIIa (CD32) and FcyRIIIb (CD16) polymorphic forms in phagocytosis of human IgGl- and IgG3- opsonised bacteria and erythrocytes. Immunology 1994;83:624-630.

Brennan FM, Chantry D, Jackson A, Maini R, Feldmann M. Inhibitory effect of TNFa antibodies on synovial cell interleukin-1 production in rheumatoid arthritis. Lancet 1989;2:244-247.

Brennan FM, Zachariae COC, Chantry D, Larsen CG, Turner M, M aini RN, Matsushima K, Feldmann M. Detection of interleukin-8 bioactivity in synovial fluids from patients with rheumatoid arthritis and production of IL-8 mRNA by isolated synovial cells. Eur J Immunol 1990a;20:2141-2144.

Brennan FM, Chantry D, Turner M, Foxwell B, Maini RN, Feldmann M. Detection of transforming growth factor-^ in rheumatoid arthritis synovial tissue. Clin Exp Immunol 1990b;81:278-285.

Brennan FM, Maini RN, Feldmann M. TNFa - a pivotal role in rheum atoid arthritis. Br J Rheum atol 1992a;31:293-298.

Brennan FM, Gibbons DL, Mitchell T, Cope AP, Maini RN, Feldm ann M. Enhanced expression of TNF receptor mRNA and protein in mononuclear cells isolated from rheumatoid arthritis synovial joints. Eur J Im m unol 1992b;22:1907-1912.

210 Brennan FM, Browne KA, Green PA, Jaspar JM, Maini RN, Feldmann M. Reduction of serum matrix metalloproteinase-1 and matrix metalloproteinase-3 in rheumatoid arthritis patients following anti­ tumour necrosis factor-alpha (cA2) therapy. Br J Rheumatol 1997;36:643- 650 .

Bresnihan B, Alvaro-Garcia JM, Cobby M, Doherty M, Domljan Z, Emery P, Nuki G, Pavelka K, Rau R, Rozman B, Watt I, Williams B, Aitchison R, McCabe D, Musikic P. Treatment of rheumatoid arthritis with recombinant human interleukin-1 receptor antagonist. Arthritis Rheum 1998;41:2196-2204.

Brockhaus M, Schoenfield HJ, Schlaeger EJ, Hunziker W, Lesslauer W, Loetscher H. Identification of two types of tumour necrosis factor receptors on human cell lines by monoclonal antibodies. Proc Natl Acad Sci USA 1990;87:3127-3131.

Broker B, Edwards JCW, Fanger MW, Lydyard PM. The prevalence and distribution of macrophages bearing FcyRI, FcyRII and FcyRIII in synovium . Scand J Rheum atol 1990;19:123-135.

Brooks DG, Qui WQ, Luster AD, Ravetch JV. Structure and expression of hum an IgG FcRII (CD32). Functional heterogeneity is encoded by the alternatively spliced products of multiple genes. J Exp Med 1989;170:1369- 1385.

Brown CM, Longhurst C, Haynes G, Plater-Zyberk C, Malcolm A, Maini RN. Immunoglobulin heavy chain variable gene utilization by B cell hybridomas derived from rheumatoid synovial tissue. Clin Exp Im m unol 1992;89:230-238.

Brown PA, Nardella FA, Mannik M. Human complement activation by self-associated IgG rheumatoid factors. Arthritis Rheum 1982;25:1101- 1107.

211 Browning JL, Ngam-ek A, Lawton P, DeMarinis J, Tizard R, Chow EP, Hession C, O'Brine-Greco B, Foley SF, Ware CF. Lymphotoxin a novel member of the TNF family that forms a hetermeric complex with lymphotoxin on the cell surface. Cell 1993;72:847-856.

Briiggemann M, W illiams GT, Bindon Cl, Clark MR, W alker MR, Jefferis R, Waldmann FI, Neuberger MS. Comparison of the effector functions of human immunoglobulins using a matched set of chimeric antibodies. J Exp Med 1987;166:1351-1361.

Brunkhorst BA, Strohm eier G, Lazzari K, Weil G, Melnick D, Fleit HB, Simons ER. Differential roles of FcyRII and FcyRIII in immune complex stim ulation of hum an neutrophils. J Biol Chem 1992;267:20659-20666. Buchan G, Barrett K, Turner M, Chantry D, Maini RN, Feldmann M. Interleukin-1 and tumour necrosis factor mRNA expression in rheumatoid arthritis: prolonged production of IL-la. Clin Exp Immunol 1988;73:449-455.

Butler DM, Malfait AM, Mason LJ, Warden PJ, Kollias G, Maini RN, Feldmann M, Brennan RM. DBA/1 mice expressing the human TNF- alpha transgene develop a severe, erosive arthritis: Characteristics of the cytokine cascade and cellular composition. J Immunol 1997;159:2867-2876.

Calzada-Wack JC, Frankenberger M, Ziegler-Heitbrock HW. Interleukin- 10 drives human monocytes to CD16 positive macrophages. J Inflamm 1996;46:78-85.

Cambier JC, Pleiman CM, Clark MR. Signal transduction by the B cell antigen receptor and its coreceptors. Annu Rev Immunol 1994;12:457-486.

Cambier JC. Daeron M, Fridman WH, Gergely J, Kinet JP, Klausner R, Lynch R, Malisseu B, Pecht J, Reinherz E, Ravetch JV, Reth M, Samelson L, Sandor M, Schreiber A, Seed B, Terhorst C, van de Winkel JGJ, Weiss A. New nomenclature for the Reth motif (or ARH1/TAM/ARAM/YXXL). Immunol Today 1995;16:110.

212 Campbell IK, Piccoli DS, H am ilton JA. Stim ulation of hum an chondrocyte prostaglandin E2 production by recombinant human interleukin-1 and tumour necrosis factor. Biochim Biophys Acta 1990;1051:310-318.

Campbell IK, Ianches G, Hamilton JA. Production of macrophage colony- stimulating factor (M-CSF) by human articular cartilage chondrocytes. Modulation by interleukin-1 and tumor necrosis factor. Biochim Biophys Acta 1993;1182:57-63.

Capra JD, Winchester RJ, Kunkel HG. Hypergammaglobulinemic purpura. Studies on the unusual anti-y-globulins characteristic of the sera of these patients. Medicine 1971;50:125-138. Carson DA, Lawrance S, Catalano MA, Vaughan JH, Abraham G. Radioimmunoassay of IgG and IgM rheumatoid factors reacting with hum an IgG. J Im m unol 1977;119:295-300.

Carson DA, Chen PP, Fox RI, Kipps TJ, Jirik F, Goldfien RD, Silverman G, Radoux V, Fong S. Rheumatoid factor and immune networks. Annu Rev Im m unol 1987;5:109-126.

Carson DA. Rheumatoid factors. In: Textbook of rheumatology. Ed; Kelley WN, Harris ED, Ruddy S, Sledge CB. Saunders (Philadelphia) 1993:155-163.

Carswell EA, Old LJ, Kassel R, Green S, Fiore N, Williamson B. An endotoxin-induced serum factor that causes necrosis of tumours. Proc Natl Acad Sci USA. 1975;72:3660-3670.

Casali P, Notkins AL. CD5+ B lymphocytes, polyreactive antibodies and the human B-cell repertoire. Immunol Today 1989a;l0:364-368.

Casali P, Notkins AL. Probing the human B-cell repertoire with EBV: Polyreactive antibodies and CD5+ B lymphocytes. Annu Rev Immunol 1989b;7:513-535.

213 Ceuppens JL, Meurs L, Van Wauwe JP. Failure of OKT3 monoclonal antibody to induce lymphocyte mitogenesis: a familial defect in monocyte helper function. J Im m unol 1985a;134:1498-1502.

Ceuppens JL, Bioemmen FJ, Van Wauwe JP. T Cell unresponsiveness to the mitogenic activity of OTK3 antibody results from a deficiency of monocyte Fc gamma receptors for murine IgG2a and inability to cross­ link the T3-Ti complex. J Im m unol 1985b;135:3882-3886.

Ceuppens JL, Baroja ML, Van Vaeck F, Anderson CL. A defect in the membrane expression of the high affinity 72 kD Fc receptor on phagocytic cells in four healthy individuals. J Clin Invest 1988;82:571-578.

Chantry D, Winearls CG, Maini RN. Feldmann M. Mechanism of immune complex-mediated damage: induction of interleukin 1 by immune complexes and synergy with interferon-y and tumor necrosis factor-a. Eur J Im m unol 1989;19:189-192.

Charles P, Elliott MJ, Davis D, Potter A, Kalden JR, A ntoni C, Breedveld FC, Smolen JS, Ebert G, de Woody K, Feldmann M, Maini RN. Regulation of cytokines, cytokine inhibitors and acute-phase proteins following anti-TNF-alpha therapy in rheumatoid arthritis. J Immunol 1999;163:1521-1528.

Charpentier B, Hiesse C, Lantz O, Rerran C, Stephens S, O'Shaugnessy D, Bodmer M, Benoit G, Bach J, Chatenoud L. Evidence that anti-human tumor necrosis factor monoclonal antibody prevents OKT3-induced acute syndrome. Transplantation 1992;54:997-1002.

Chu CQ, Field M, Feldmann M, Maini RN. Localization of tumour necrosis factor a in synovial tissues and at the cartilage-pannus junction in patients with rheumatoid arthritis. Arthritis Rheum 1991a;34:1125- 1132.

Chu CQ, Field M, Abney E, Zheng RQH, Allard S, Feldmann M, Maini RN. Transforming growth factor pi in rheumatoid synovial membrane and cartilage/pannus junction. Clin Exp Immunol 1991b;86:380-386.

214 Clark IM, Powell LK, Hazleman BL, Cawston TE. The measurement of collagenase, TIMP and collagenase-TIMP comples in synovial fluid from patients with osteoarthritis and rheumatoid arthritis. Arthritis Rheum 1993;36:372-379.

Clark MR, Liu L, Clarkson SB, Qry PA, Goldstein IM. An abnormality of the gene that encodes neutrophil Fc receptor III in a patient with systemic lupus erythematosus. J Clin Invest 1990;86:341-346.

Clark MR, Stuart SG, Kimberly RP, Ory PA, Goldstein IM. A single am ino acid distinguishes the high-responder from the low-responder from of Fc receptor II on hum an monocytes. Eur J Im m unol 1991;21:1911-1916.

Clark RA, Abraham GN, Kyle R, Vaughan JH. Gamma globulin complexes in rheumatoid arthritis and hypergammaglobulinemic purpura. J Rheum atol 1974;1:54-65.

Clarkson SB, Ory PA. CD16. Developmentally regulated IgG Fc receptors on cultured hum an monocytes. J Exp Med 1987;167:408-420.

Clynes R, Dumitru C, Ravetch JV. Uncoupling of immune complex formulation and kidney damage in autoimmune glomerulonephritis. Science 1998;279:1052-1054.

Clynes R, Maizes JS, Guinam ard R, Ono M, Takai T, Ravetch JV. Modulation of immune complex-induced inflammation in vivo by the coordinate expression of activation and inhibitory Fc receptors. J Exp Med 1999;189:179-185.

Cohen IR, Cooke A. Natural autoantibodies might prevent autoimmune disease. Im m unol Today 1986;7:363-364.

Cohen L, Cheek RL, Hadler JA, Yount WJ, Eisenberg RA. The subclass distribution of human IgG rheumatoid factor. J Immunol 1987,*139:1466- 1471.

215 Cohen PL, Sharp S, Kulezycki Jr A. Human monocytes, B lymphocytes and non-B lymphocytes each have structurally unique Fey receptors. J Im m unol 1983;131:378-383.

Colotta F, Dauer SK, Sims JE, Mantovani A. The type II "decoy" receptor. A novel regulatory pathway for interleukin-1. Immunol Today 1994;15:262-266.

Conn DL, Hunder GG, O'Duffy D. Vasculitis and related disorders. In: Textbook of rheumatology. Eds; Kelly WN, Harris Jr ED, Ruddy S, Sledge CR. Saunders (Philadelphia) 1993:1077-1102.

Conway SC, Creed FH, Symmons DP. Life events and the onset of rheumatoid arthritis. J Psychosom Res 1994;38:837-847.

Cooper NR, Bradt BM, Rhim JS, Nemerow GR. CR2 complement receptor. J Invest Dermatol 1990;94:112S-117S.

Corper AL, Sohi MK, Bonagura VR, Steinitz M, Jefferis R, Feinstein A, Beale D, Taussig MJ, Sutton BJ. Structure of human IgM rheumatoid factor Fab bound to its autoantigen IgG Fc reveals a novel topology of antibody-antigen interaction. Nat Struc Biol 1997;4:374-381.

Crockett-Torabi E, Fantone JC. Soluble and insoluble immune complexes activate human neutrophil NADPH oxidase by distinct Fey receptor- specific m echanism s. J Im m unol 1990;145:3026-3032.

Cromartie WJ, Craddock JG, Schwab JH, Anderle SK, Yang CH. Arthritis in rats after systemic injection of streptococcal cells or cell walls. J Exp Med. 1977; 146:1585-1602.

C unningham TJ, Medcalf RL, Mathews JD, M uirden KD. Platelet activity in sera from patients with rheumatoid vasculitis. Ann Rheum Dis 1986;45:15-20.

216 da Silva JAP, Hall GM. The effects of gender and sex hormones on outcome in rheumatoid arthritis. Baillifere's Clinical Rheumatology 1992;6:193-219.

DaeronM. Fc receptor biology. Annu Rev Immunol 1997;15:203-234.

Dahlgren C. Polymorphonuclear leucocyte chemiluminescence induced by formyl-methionyl-leucy 1-phenylalanine and phorbol myristate acetate: Effects of catalase and superoxide dismutase. Agents Actions 1987;21:104- 112.

Danis VA, March LM, Nelson DS, Brooks PM. Interleukin-1 secretion by peripheral blood monocytes and synovial macrophages from patients with rheumatoid arthritis. J Rheumatol 1987;14:33-39.

Darville T, Tabor DR, Theus SA, Jacobs RF. Surface matrix binding alters murine peritoneal mononuclear phagocyte TNF-a and IL-6 induction. Im m unol Invest 1992;21:539-552.

Dauer SK, Fanslau W, Jacobs C, W augh S, Sims JE, W idm er MB. Interleukin-1 antagonists. Therapeutic Immunol 1994;1:113-122.

Davies B, Edwards SW. Chemiluminescence of human bloodstream monocytes and neutrophils: an unusual oxidant(s) generated by monocytes during the respiratory burst. J Biolumin Chemilumin 1992;7:229-238.

Dayer JM, Beutler B, Cerami A. Cachectin/ tumor necrosis factor stimulates collagenase and prostaglandin E2 production by human synovial cells and dermal fibroblasts. J Exp Med 1985;162:2163-2168.

Dayer JM, Horwitz DA, Davies KJ. Potential of hypochlorous acid and N- chloramines in cartilage breakdown by phagocytic cells. Free Rad Biol Med 1993;15:637-643. de Haas M, Koene HR, Kleijer M, de Vries E, Simsek S, van Tol MJD, Roos D, von dem Borne AEGKr. A triallelic Fey receptor type IIIA

217 polymorphism influences the binding of human IgG by NK cell FcyRIIIa. J Im m unol 1996;156:2948-2955. de Jong R, Loenan WA, Brouwer M, van Emmerik L, de Vries EF, Borst J, van Lier RA. Regulation of expression of CD27, a T cell specific m em ber of a novel family of membrane receptors. J Immunol 1991;146:2488-2494. de Vries N, Ronningen KS, Tilanus MG, Bouwens Rombauts A, Segal R, Egeland T, Thorsby E, van de Putte LB, Brautbar C. HLA-DRI and rheum atoid arthritis in Israeli Jews: sequencing reveals that DRB1*0102 is the predominant HLA-DRI subtype. Tissue Antigens 1993;41:26-30.

Debets JMH, van der Linden CJ, Dieteren IEM, Leeuwenberg JFM, Buurman WA. Fc-receptor cross-linking induces rapid secretion of tumor necrosis factor (cachectin) by human peripheral blood monocytes. J Im m unol 1988;141:1197-1201.

Debets JMH, van de Winkel JGJ, Ceuppens JL, Dieteren IEM, Buurman WA. Cross-linking of both FcyRI and FcyRII induces secretion of tumor necrosis factor by human monocytes, requiring high affinity Fc-FcyR interactions. Functional activation of FcyRII by treatment with proteases or neuram inidase. J Im m unol 1990;144:1304-1310.

Dechanet J, Merville P, Durand I, Banchereau J, Miossec P. The ability of_ synoviocytes to support terminal differentiation of activated B cells may explain plasma cell accumulation in rheumatoid synovium. J Clin Invest 1995;95:456-463.

Deleuran BW, Chu CQ, Feild M, Brennan RM, Mitchell T, Brockhaus M, Feldmann M, Maini RN. Localization of TNF receptors in the synovial tissue and cartilage/pannus junction in rheumatoid arthritis: Implication for local actions of TNFa. Arthritis Rheum 1992;35:1170-1178.

Deleuran BW, Lemche P, Kristensen M, Chu CO, Field M, Jensen J, Matsushima K, Stengaardpedersen K. Localization of interleukin-8 in the synovial membrane, cartilage-pannus junction and chondrocytes in rheumatoid arthritis. Scand J Rheumatol 1994;23:2-7.

218 Dembic Z, Loetscher H, Gubler U, Pan YC, Lahm HW, Gentz R, Brockhaus M, Lesslaver W. Two human TNF receptors have similar extracellular, but distinct intracellular domain sequence. Cytokine 1990;2:231-237.

Deo YM, Graziano RF, Repp R, van de Winkel JGJ. Clinical significance of IgG Fc receptors and FcyR-directed immunotherapies. Immunol Today 1997;18:127-135.

Dinarello CA, Cannon JG, Wolff SM, Bernheim HA, Beutler B, Cerami A, Figari IS, Palladino Jr. A, O'Connor JV. Tum or necrosis factor (cachectin) is an endogenous pyrogen and induces production of interleukin 1. J Exp Med 1986;163:1433-1450.

Dinarello CA. Interleukin-1 and interleukin-1 antagonism. Blood 1991;77:1627-1652.

Dinarello CA. Biologic basis for interleukin-1 in disease. Blood 1996;87:2095-2147.

Dougherty GJ, Selvendran Y, Murdoch S, Palmer DG, Hogg N. The human mononuclear phagocyte high-affinity Fc receptor, FcRI, defined by a m onoclonal antibody, 10.1. Eur J Im m unol 1987;17:1453-1459.

Duits AJ, Bootsma H, Derksen RHWM, Spronk PE, Kater L, Kallenberg CGM, Capel PJA, Westerdaal NAC, Spierenburg GTH, Gmelig-Meyling FHJ, van de W inkel JGJ. Skewed distribution of IgG Fc receptor Ila (CD32) polymorphism is associated with renal disease in systemic lupus erythematosus patients. Arthritis Rheum 1995;39:1832-1836.

Duke O, Panayi GS, Poulter L, Janossy G. An immunohistological analysis of lymphocyte populations and their microenviroment in the synovial membranes of patients with rheumatoid arthritis using monoclonal antibodies. Clin Exp Immunol 1982;49:22-29.

219 Dusi S, Della Bianca V, Donini M, Nadalini KA, Rossi F. Mechanisms of stimulation of the respiratory burst by TNF in nonadherent neutrophils: its independence of lipidic transmembrane signalling and dependence on protein tyrosine phosphorylation and cytoskeleton. J Immunol 1996;157:4615-4623.

Edwards JCW, Sedwick AD, Willoughby DA. Membrane properties and esterase activity of synovial lining cells. Ann Rheum Dis 1982;41:282-286.

Edwards JCW, Wilkinson LS. Immunohistochemistry of rheumatoid synovium. In: Mechanisms and models in rheumatoid arthritis. Eds; Henderson B, Edwards JCW, Pettipher ER. Academic Press (London) 1995;133-152.

Edwards JCW, Blades S, Cambridge G. Restricted expression of FcyRIII (CD16) in synovium and dermis: implications for tissue targeting in rheum atoid arthritis (RA). Clin Exp Im m unol 1997a;108:401-406.

Edwards JCW, Leigh RD, Cambridge G. Expression of molecules involved in B lymphocyte survival and differentiation by synovial fibroblasts. Clin Exp Im m unol 1997b;108:407-414.

Edwards JCW, Cambridge G. Rheumatoid arthritis: The predictable effect of small immune complexes in which antibody is also antigen. Br J Rheum atol 1998;37:126-130.

Edwards JCW, Sutton BJ. What does IgG look like? Br J Rheumatol 1998;37(abstract suppl 1):7.

Edwards JCW, Cambridge G, Abrahams VM. Do self-perpetuating lymphocytes drive human autoimmune disease? Immunology 1999;96:188-196.

Edwards SW. In: Biochemistry and physiology of the neutrophil. Cambridge University Press 1994.

220 Eisenberg SP, Evans RJ, Arend WP, Verderber E, Brewer MT, Hannum CH, Thompson RC. Primary structure and functional expression from complementary DNA of a human interleukin-1 receptor antagonist. Nature 1990;343:341-346.

Elliott MJ, Maini RN, Feldm ann M, Long Fox A, Charles P, Katsikis P, Brennan FM, Walker J, Bijl FI, Ghrayeb J, Woody JN. Treatment of rheumatoid arthritis with chimeric monoclonal antibodies to tumour necrosis factor a. Arthritis Rheum 1993;36:1681-1690.

Epperlein MM, Nourooz-Zadeh J, Jayasena SD, Hothersall JS, Noronha- Dutra A, Neild GH. Nature and biological significance of free radicals generated during bicarbonate hemodialysis. J Am Soc Nephrol 1998;9:457- 463.

Erikstein BK, Smeland EB, Blomhoff HK, Funderud S, Prydz K, Lesslauer W, Espevik T. Independent regulation of 55-kDa and 75-kDa tumor necrosis factor receptors during activation of peripheral blood B lymphocytes. Eur J Im m unol 1991;21:1033-1037.

Ermel RW, Kenny TP, Chen PP, Robbins DL. Molecular analysis of rheumatoid factors derived from rheumatoid synovium suggests an antigen-driven response in inflamed joints. Arthritis Rheum 1993;36:380-388.

Ernst LK, van de Winkel JGJ, Chiu IM, Anderson CL. Three genes for the human high affinity Fc receptor for IgG (FcyRI) encode four distinct transcription products. J Biol Chem 1992;267:15692-15700.

Ernst LK, Duchemin Am, Anderson CL. Association of the high affinity receptor for IgG (FcyRI) with the gamma subunit of the IgG receptor. Proc Natl Acad Sci USA 1993;90:6023-6027.

Fang Y, Xu C, Fu YX, Hoiers VM, Molina H. Expression of complement receptors 1 and 2 on follicular dendritic cells is necessary for the generation of a strong antigen-specific IgG response. J Immunol 1998;160:5273-5279.

221 Fanger MW, Shen L, Graziano RF, Guyre PM. Cytotoxicity mediated by human Fc receptors for IgG. Immunol Today 1989;3:92-99.

Fearon DT, Carter RH. The CD19/CR2/TAPA-1 complex of B lymphocytes linking natural to acquired immunity. Annu Rev Immunol 1995;13:127-149.

Field M, Chu C, Feldmann M, Maini RN. Interleukin-6 localisation in the synovial membrane in rheumatoid arthritis. Rheumatol Int 1991;11:45-50.

Fijen CAD, Bredius RGM, Kuijper EJ. Polymorphism of IgG Fc receptors in menigococcal disease. Risk marker in complement deficient patients. A nn Intern Med 1993;119:636-641.

Feinstein A, Munn EA. Conformation of the free and antigen-bound IgM antibody molecules. N ature 1969;244:1307-1309.

Fink PC, Piening U, Fricke PM, Deicher H. Platelet aggregation and aggregation inhibition by differing antiglobulins and antiglobulin complexes from patients with rheumatoid arthritis. Arthritis Rheum 1979;22:896-903.

Finkel T. Oxygen radicals and signaling. Curr Opin Cell Biol 1998;10:248- 253.

Fleit HB, Wright SD, Unkeless JC. Human neutrophil Fey receptor distribution and structure. Proc Natl Acad Sci USA 1982;79:3275-3279.

Fontana A, Hengartner H, Weber E, Fehr K, Grob PJ, Cohen G. Interleukin-1 activity in the synovial fluid of patients with rheumatoid arthritis. Rheumatol Int 1982;2:49-53.

Force WR, Walter BN, Hession C, Tizard R, Kozak CA, Browning JL, Ware CF. Mouse lymphotoxin-b receptor. Molecular genetics, ligand binding and expression. J Im m unol 1995;155:5280-5288.

222 Fox DA. The role of T cells in the immunopathogenesis of rheumatoid arthritis: New perspectives. Arthritis Rheum 1997;40:598-609.

Franklin EC, Holman HR, Miiller-Eberhard HJ, Kunkel HG. An unusual protein component of high molecular weight in the serum of certain patients with rheumatoid arthritis. J Exp Med 1957;105:425-438.

Gamble JR, Harlan JM, Klebanoff SJ, Vadas MA. Stimulation of the adherence of neutrophils to umbilical vein endothelium by human recombinant tumor necrosis factor. Proc Natl Acad Sci USA 1985;82:8667- 8671.

Gardner DL. Rheumatoid arthritis. In: Pathological basis of the connective tissue diseases. Edward Arnold (London) 1992:444-526.

Garrod AB. The nature and treatment of gout and rheumatic gout. London, Walton and Maberly, 1859.

Gearing AJH, Beckett P, Christodoulou M, Churchill M, Clements JM, Davidson AH, Drummond AH, Gallaway WA, Gilbert R, Gordon LJ, Leber TM, Mangan M, Miller K, Nayee P, Owen K, Patel S, Thomas W, Wells G, Wood LM, Woolley K. Processing of tumour necrosis factor- alpha precursor by metalloproteinases. Nature 1994;370:555-557.

Gergely J, Sarmay G. Fey receptors in malignancies: friends or enemies? Adv Cancer Res 1994;64:211-245.

Gergely J, Sarmay G. FcyRII-mediated regulation of hum an B cells. Scand J Im m unol 1996;44:1-10.

Gerszten RE, Garcia-Zepeda EA, Lim YC, Yoshidal M, Ding HA, Gimbrone MA, Luster AD, Luscinskas FW, Rosenzweig A. MCP-1 and IL- 8 trigger firm adhesion of monocytes to vascular endothelium under flow conditions. N ature 1999;398:718-723.

223 Gessl A, Willheim M, Spittler A, Agis H, Krugluger W, Boltz-Nitulescu G. Influence of tumour necrosis factor-a on the expression of Fc IgG and IgA receptors, and other markers by cultured human blood monocytes and U937 cells. Scand J Im m unol 1994;39:151-156.

Goldschmidt TJ, Andersson M, Malmstrom V, Holmdahl R. Activated type II collagen reactive T cells are not eliminated by in vivo anti-CD4 treatment. Implications for therapeutic approaches on autoimmune arthritis. Immunobiology 1992;184:359-371.

Gosselin EJ, Brown MF, Anderson CL, Zipf TF, Guyre PM. The monoclonal antibody 41FI16 detects the Leu 4 responder form of human FcyRII. J Im m unol 1990;144:1817-1822.

Gowen M, Wood DD, Ihrie EJ, McGuire MKB, Russell RG. An interleukin-1 like factor stimulates bone resorption in vitro. Nature 1983;306:378-380.

Gray PW, Barrett K, Chantry D, Turner M, Feldmann M. Cloning of a human tumour necrosis factor (TNF) receptor cDNA and expression of recombinant soluble TNF-binding protein. Proc Natl Acad Sci USA 1990;87:7380-7384.

Gregersen PK, Silver J, Winchester WJ. The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum 1987;30:1205-1213.

Guyre PM, Morganelli PM, Miller R. Recombinant immune interferon increases immunoglobulin G Fc receptors on cultured human mononuclear phagocytes. J Clin Invest 1983;72:393-397.

Hallgren HM, Buckley 3rd CE, Gibertsen VA, Yunis EJ. Lymphocyte phytohemagglutinin responsiveness, immunoglobulins and autoantibodies in ageing humans. J Im m unol 1973;111:1101-1107.

224 Halliwell B. Superoxide-dependent formation of hydroxyl radicals in the presence of iron salts. It's role in degradation of hyaluronic acid by a superoxide-generating system. FEBS Lett 1978;96:238-242.

Hanemaaijer R, Sorsa T, Konttinen YT, Ding Y, Sutinen M, Visser H, van Hinsbergh VW, Helaakoski T, Kainulalnen T, Tonka H, Tschesche H, Sato T. Matrix metalloproteinase-8 is expressed in rheumatoid synovial fibroblasts and endothelial cells. Regulation by tumor necrosis factor alpha and deoxycycline. J Biol Chem 1997;272:31504-31509.

Hannestad K. Presence of aggregated gamma-G-globulin in certain rheumatoid synovial effusions. Clin Exp Im m unol 1967;2:511-529.

Hardy RR, Hayakawa K, Shimizu M, Yamasaki K, Kishimoto T. Rheumatoid factor secretion from human Leu-1+ B cells. Science 1987;236:81-83.

Hartnell A, Kay AB, Wardlaw AJ. IFN-y induces expression of FcyRIII (CD16) on hum an eosinophils. J Im m unol 1992;148:1471-1478.

Harris Jr ED. Rheumatoid arthritis. Pathophysiology and implications for therapy. N Engl J Med 1990;322:1277-1289.

Harrison D, Phillips JH, Lanier LL. Involvement of a metalloproteinase in spontaneous and phorbol ester-induced release of natural killer cell- associated Fc gamma RIII (CD16-II). J Im m unol 1991;147:3459-3465.

Harworth C, Brennan FM, Chantry D, Turner M, Maini RN, Feldmann M, Kishimoto T. Expression of granulocyte macrophage colony- stimulating factor (GM-CSF) in RA: regulation by tumour necrosis factor alpha. Eur J Im m unol 1991;21:2575-2579.

Hastings MJ, Petricevic I, Williams AJ, Cole PJ, Easmon CSF. The effect of culture media on the production and measurement of luminol dependent chemilum inescence. Br J Exp Pathol 1982;63:147-153.

225 Hay FC, Nineham LJ, Perumal R, Roitt IM. Intra-articular and circulating immune complexes and antiglobulins (IgG and IgM) in rheumatoid arthritis: correlation with clinical features. Ann Rheum Dis 1979;38:1-7.

Hazuda DJ, Lee JC, Young PR. The kinetics of interleukin 1 secretion from activated monocytes. Differences between interleukin 1 alpha and beta. J Biol Chem 1988;263:8473-8479.

Hedberg H. Studies on synovial fluid in arthritis. I. The total complement activity. II. The occurrence of mononuclear cells with in vitro cytotoxic effect. Acta Med Scand Suppl 1967;479:1-137.

Hedfors E, Klareskog L, Lindblad S, Forsum U, Lindahl G. Phenotypic characterisation of cells within subcutaneous rheumatoid nodules. Arthritis Rheum 1983;26:1333-1339.

Henderson B, Revell P, Edwards JCW. Synovial lining cell hyperplasia in rheumatoid arthritis: dogma and fact. Ann Rheum Dis 1983;47:348-349. Hendrich C, Kuipers JG, Kolanus W, Hammer M, Schmidt RE. Activation of CD16+ effector cells by rheumatoid factor complex. Role for natural killer cells in rheumatoid arthritis. Arthritis Rheum 1991;34:423- 431.

Hibbs ML, Bonadonna L, Scott BM, McKenzie IF, Hogarth PM. Molecular cloning of a human immunoglobulin G Fc receptor. Proc Natl Acad Sci USA 1988;85:2240-2044.

Hibbs ML, Selvaraj P, Carpen O, Springer TA, Kuster H, Jouvin MHE, Kinet JP. Mechanisms regulating expression of membrane isoforms of FcyRIII (CD16). Science 1989;246:1608-1611.

Hibbs ML, Tolvanen M, Carpen O. Membrane proximal Ig-like domain of FcRIII (CD16) contains residues critical for ligand binding. J Immunol 1994;152:4466-4474.

226 Holoshitz J, Klajman A, Drucker I, Lapidot Z Yaretzky A, Frenkel A, van Eden W, Cohen IR. T lymphocytes of rheumatoid arthritis patients show augmented reactivity to a fraction of mycobacteria cross-reactive with cartilage. Lancet 1986; 2: 305-309.

Holt ME, Ryall MET, Campbell AK. Album in inhibits hum an polymorphonuclear leukocyte luminol-dependent chemiluminescence: evidence for oxygen radical scavenging. Br J Exp Pathol 1984;65:231-241.

Hopkins SJ, Meager A. Cytokines in synovial fluid. EL. The presence of tumour necrosis factor and interferon. Clin Exp Im m unol 1988;73:88-92.

Horejsf V, Drbal K, Cebecauer M, Cerny J, Brdicka T, Angelisova P, Stockinger H. GPI-microdomains: a role in signalling via immunoreceptors. Im m unol Today 1999;20:356-361.

Houssiou FA, Devogelaer JP, van Damme J, de Deuxchaisnes CN, van snick J. Interleukin-6 in synovial fluid and serum in patients with rheumatoid arthritis and other inflammatory arthritides. Arthritis Rheum 1988;31:784-788.

Hrncir Z, Tichy M, Salavec M, Vavrina J. Immunoglobulins A, G and M in synovial fluid in rheumatoid arthritis. Reactive synovitis of local origin and in post mortem synovial fluid. A nn Rheum Dis 1972;31:325- 329.

Huber H, Douglas SD, Fudenberg HH. The IgG receptor, an immunological marker for the characterisation of mononuclear cells. Immunology 1969;17:7-21.

Huber H, Fudenberg HH. Receptor sites of human monocytes for IgG. Int Arch Allergy 1968;34:18-31.

Huber H, Polley MJ, Linscott WD, Fudenberg HH, Muller-Eberhard HJ. Human monocytes: distinct receptor sites for the third component of complement and for im m unoglobulin G. Science 1968;162:1281-1283.

227 Huizinga TWJ, van der Schoot CE, Jost C, Klaassen R, Kleijer M, von dem Borne AEGKr, Roos D, Tetteroo PAT. The Pi-linked receptor FcRIII is released on stim ulation of neutrophils. Nature 1988;333:667-669.

Huizinga TWJ, van Kemenade F, Koenderman L, Dolman KM, von dem Borne AEGKr, Tetteroo PA, Roos D. The 40-kDa Fc gamma receptor (FcRII) on neutrophils is essential for the IgG-induced respiratory burst and IgG-induced phagocytosis. J Im m unol 1989a;142:2365-2369.

Huizinga TWJ, Kerst M, N uyens JH, Vlug A, von dem Borne AEGKr, Roos D, Tetteroo PAT. Binding characteristics of dimeric IgG subclass complexes to hum an neutrophils. J Im m unol 1989b;142:2359-2364.

Huizinga TWJ, de-Haas M, Kleijer M, Nuijens JH, Roos D, von dem Borne AEGKr. Soluble Fey receptor III in human plasma originates from release by neutrophils. J Clin Invest 1990a;86:416-423.

Huizinga TWJ, Kleijer M, Tetteroo PA, Roos D, von dem Borne AEGKr. Biallelic neutrophil NA-antigen system is associated with a polymorphism on the phosphoinositol-linked Fey receptor III (CD16) Blood 1990b;75:213-217.

Huizinga TWJ, Kuijpers RW, Kleijer M, Schulpen TW, Cuypers HT, Roos D, von dem Borne AEGKr. Maternal genomic neutrophil deficiency leading to neonatal isoim m une neutropenia. Blood 1990c;76:1927-1932.

Hulett MD, Witort E, Brinkworth RI, McKenzie IFC, Hogarth PM. Multiple regions of human FcyRII (CD32) contribute to the binding of IgG. J Biol Chem 1995;270:21188-21194.

Hundt M, Schmidt RE. The glycocylophosphatidylinositol-linked Fey receptor III represents the dominant receptor structure for immune complex activation of neutrophils. Eur J Im m unol 1992;22:811-816.

Jacoby RK, Jayson MIV, Cosh JA. Onset, early stages and prognosis of rheumatoid arthritis: a clinical study of 100 patients, with 11-year follow up. Br Med J 1973;2:96-100.

228 Janossy G, Duke O, Poulter LW, Panayi G, Bofill M, Goldstein G. Rheumatoid arthritis: A disease of T lymphocyte/macrophage imm unoregulatation. Lancet 1981;2:839-842.

Jarvinen P, Aho K. Twin studies in rheumatic diseases. Semin Arthritis Rheum 1994;24:19-28.

Jobbling SA, Auron PE, Gurka G, Webb AC, McDonald B, Rosenwasser LJ, Gehrke L. Biological activity and receptor binding of human prointerleukin-lp and subpeptides. J Biol Chem 1988;263:16372-16378.

Johhansson A, Dahlgren C. Characterisation of the luminol-amplified light generating reaction induced in human monocytes. J Leukoc Biol 1989;45:444-451.

Jones BM, Cheng IKP, Wong RWS, Kung AWC. CD5-positive and CD5- negative rheumatoid factor-secreting B cells in IgA nephropathy, rheumatoid arthritis and Grave's disease. Scand J Immunol 1993;38:575- 580.

Jones DH, Looney RJ, Anderson CL. Two distinct classes of Fc receptors on a human monocyte line (U937) defined by differences in binding of m urine IgG subclasses at low ionic strength. J Im m unol 1985;135:3348- 3353.

Jones MG, Shipley ME, Hearn JP, Hay FC. Specificity of rheumatoid factors in relation to the disease state in rheumatoid arthritis. Ann Rheum Dis 1990;49:757-762.

Jonsson T, Arinbjarnarson S, Thorsteinsson J, Steinsson K, Geirsson AJ, Jonsson H, Valdimarsson H. Raised IgA rheumatoid factor (RF) but not IgM RF or IgG RF is associated with extra-articular manifestations in rheumatoid arthritis. Scand J Rheumatol 1995;24:372-375.

Kantor A. A new nomenclature for B cells. Immunol Today 1991;12:388.

229 Karras JG, Wang Z, Huo L, Howard RG, Frank DA, Rothstein TL. Signal transduction and activator of transcription-3 (STAT3) is constitutively activated in normal, self-renewing B-l cells but only inducibly expressed in conventional B lymphocytes. J Exp Med 1997;185:1035-1042.

Kauanko IC, Bates EJ, Ferrante A. Mechanisms of human neutrophil- mediated cartilage damage in vitro: the role of lysosomal enzymes, hydrogen peroxide and hypochlorous acid. Immunol Cell Biol 1989;67:321-329.

Keffer J, Probert L, Cazlaris H, Georgopoulos S, Kaslaris E, Kioussis D, Kollias G. Transgenic mice expressing hum an tumour necrosis factor: a predicative genetic m odel of arthritis. EMBO J 1991;10:4025-4031.

Kelley JL, Rozek MM, Suenram CA, Schwartz CJ. Activation of human blood monocytes by adherence to tissue culture plastic surfaces. Exp Mol Pathol 1987;46:266-278.

Kerst JM, Van de Winkel JGJ, Evans AH, de Haas M, Slaper-Cortenbach IC, de Wit TP, von dem Borne AEGKr, van der Schoot CE, van Oers RH. G-CSF induces hFcyRI (CD64 antigen) positive neutrophils via an effect on myeloid precursor cells. Blood 1993;81:1457-1464.

Killar L, White J, Black R, Peschon J. Adamalysins. A family of metzincins including TNF-alpha converting enzyme (TACE). Ann NY Acad Sci 1999;878:442-452.

Kindt GC, van de Winkel JGJ, Moore SA, Anderson CL. Identification and structural characterisation of Fc gamma-receptors on pulmonary alveolar macrophages. Am J Physiol 1991;260:L403-411.

Kipps TJ, Parham P, Punt J, Herzenberg LA. Importance of immunoglobulin isotype in human antibody-dependent, cell-mediated cytotoxicity by murine m onoclonal antibodies. J Exp Med 1985;161:1-17.

Kipps TJ. The CD5B cell. Adv Im m unol 1989;47:117-185.

230 Klaassen RJL, Goldscmeding R, Tetteroo PAT, Von dem Borne AEGKr. The Fc valency of an immune complex is the decisive factor for binding to low-affinity Fey receptors. Eur J Im m unol 1988;18:1373-1377.

Klaassen RJL, Ouwehand WH, Huizinga TW, Engelfriet CP, von dem Borne AEGKr. The Fc-receptor III of cultured human monocytes. Structural similarity with FcRIII of natural killer cells and role in the extracellular lysis of sensitized erythrocytes. J Im m unol 1990,144:599-606.

Kleiner Jr DE, Stetler-Stevenson G. Structural biochemistry and activation of matrix metalloproteinases. Curr Biol 1993;5:891-897.

Kocher M, Edberg JC, Fleit HB, Kimberly RP. Antineutrophil cytoplasmic antibodies preferentially engage FcyRIIIb on human neutrophils. J Im m unol 1998;161:6909-6914.

Koene HR, Kleijer M, Algra J, Roos D, von dem Borne AEGKr, de Haas M. FcyRIIIa-158V/F polymorphism influences the binding of IgG by natural killer cell FcyRIIIa, independently of the FcyRIIIa-48L/R/H phenotype. Blood 1997;90:1109-1114.

Koene HR, Kleijer M, Swaak AJG, Sullivan K, Bijl M, Petri MA, Kallenberg CGM, Roos D, von dem Borne AEGKr, de Haas M. The FcyRIIIA-158F allele is a risk factor for systemic lupus erythematosus. Arthritis Rheum 1998;41:1813-1818.

Kohsaka H, Carson DA, Miyasaka N. Formation of peripheral imm unoreceptor repertoire for antigens. Arthritis Rheum 1998;41:1911- 1918.

Koopman G, Keehnen RM, Lindhout E, Newman W, Shimizu Y, van Seventer GA, de Groot C, Pals ST. Adhesion through LFA-1 (C D lla/ CD18)-ICAM-1 (CD54) and the VLA-4 (CD49d)-VCAM-l (CD106) pathways prevents apoptosis of germinal centre B cells. J Immunol 1994;152:3760-3767.

231 Kraan MC, Versendaal H, Jonker M, Bresnihan B, Post WJ, 't Hart BA, Breed veld FC, Tak PP. Asymtomatic synovitis preceeds clinically manifest arthritis. Arthritis Rheum 1998;41:1481-1488.

Krause A, Kamradt T, Burmester GR. Potential infectious agents in the induction of arthritides. Curr Opin Rheum atol 1996;8:203-209.

Krause SW, Kreutz A, Andreesen R. Differential effects of cell adherence on LPS-stimulated cytokine production by human monocytes and macrophages. Immunobiology 1996-97;196:522-534.

Kriegler M, Perez C, Defay K, Albert I, Lu SD. A novel form of TNF/cachectin is a cell surface cytotoxic transmembrane protein: Ramifications for the complex physiology of TNF. Cell 1988;53:43-53.

Kunkel HG, Franklin EC, Miiller-Eberhard HJ. Studies on the isolation and characterisation of rheumatoid factors. J Clin Invest 1959;38:424-434.

Kunkel HG, Miiller-Eberhard HJ, Fudenberg HH, Tomasi TB. Gamma globulin complexes in rheumatoid arthritis and certain other conditions. J Clin Invest 1961;40:117-129.

Kunkel HG, Tan EM. Autoantibodies and disease. Adv Immunol 1964;4:351.

Kurlander RJ, Batker J. The binding of human immunoglobulin G1 monomer and small, covalently cross-linked polymers of immunoglobulin G1 to human peripheral blood monocytes and polymorphonuclear leukocytes. J Clin Invest 1982;69:1-8.

Kurosaka M, Ziff M. Immunoelectron microscopic study of the distribution of T cell subsets in rheumatoid synovium. J Exp Med 1983;158:1191-1210.

Kurosaki T, Ravetch JV. A single amino acid in the glycosyl phosphatidylinositiol attachment domain determines the membrane topology of FcyRIII. Nature 1989;342:805-807.

232 Kurosaki T, Gander I, Ravetch JV. A subunit common to an IgG Fc receptor and T-cell receptor mediates assembly through different interactions. Proc Natl Acad Sci USA 1991;88:3837-3841.

Kyle RA, Gleich GJ, Bayrd ED, Vaughan JH. Benign hypergammaglobulinemic purpura of Waldenstrom. Medicine 1971;50:113-123.

Lambert IA, Schofield JB, Amlot P, Van Noorden S. Protection of germinal centres from complement attack: decay-accelerating factor (DAF) is a constitutive protein on follicular dendritic cells. A study in reactive and neoplastic follicles. J Pathol 1993;170:115-120.

Lang AK, Macht LM, Kirwan JR, Wraith DC, Elson CJ. Ability of T cells from patients with rheumatoid arthritis to respond to immunoglobulin G. Imm unology 1999;98:116-122.

Lang ML, Wade WF, Shen L. y-chain dependent recruitment of tyrosine kinases to membrane rafts by the human IgA receptor FcaR. J Immunol, in press.

Lanier LL, Le AM, Phillips, Warner NL, Babcock GF. Subpopulations of human natural killer cells defined by expression of the Leu-7 (HNK-1) and Leu-11 (NK-15) antigens. J Im m unol 1983;131:1789-1796.

Lanier LL, Cwirla S, Yu G, Testi R, Phillips JH. Membrane anchoring of a human IgG Fc receptor (CD16) determined by a single amino acid. Science 1989a;246:1611-1613.

Lanier LL, Yu G, Phillips JH. Co-association of CD3£ with a receptor (CD16) for IgG Fcon hum an natural killer cells. Nature 1989b;342:803-805.

Larsen CG, Zachariae C, Oppenheim JJ, Matsushima K. Production of monocyte chemotactic and activating factor (MCAF) by human dermal fibroblasts in response to interleukin 1 or tumour necrosis factor. Biochem Biophys Res Com mun 1989;160:1403-1408.

233 Le-Bouteiller PP, Mishal Z, Lemonnier FA, Kourilsky FM. Quantification by flow cytometry of HLA class I molecules at the surface of murine cells transformed by cloned FPL A genes. J Im m unol Methods. 1983:61:301-315.

Leeuwenberg JF, Dentener MA, Buurman WA. Lipopolysaccharide- mediated soluble TNF receptor release and TNF receptor expression by monocytes. Role of CD14, LPS binding protein, and bactericidal/permeability-increasing protein. J Im m unol 1994;152:5070- 5076

Levine PR, Axelrod DA. Rheumatoid factor isotypes following im m unization. Clin Exp Rheumatol 1985;3:147-149.

Levy PC, Utell MJ, Fleit HB, Roberts NJ Jr, Ryan DPI, Looney RJ. Characterisation of human alveolar macrophage Fc gamma receptor III: a transmembrane glycoprotein that is shed under in vitro culture conditions. Am J Respir Cell Med Biol 1991;5:307-314.

Liao G, Simon SR. Temporal down-regulation of FcyRIII expression and Fey receptor-mediated phagocytosis in human monocyte-derived macrophages induced by TNF-a and IL-lp. J Leukoc Biol 1994;55:702-710. Liao G, Simone J, Simon SR. Paracrine downregulation of FcyRIII in human monocyte-derived macrophages induced by phagocytosis of nonopsonised particles. Blood 1994;83:2294-2304.

Lindhout E, Koopman G, Pals ST, de Groot C. Triple check for antigen specificity of B cells during germinal centre reactions. Immunol Today 1997;18:573-577.

Locke IC. PhD thesis: Studies on the Streptodormase component of the Medicoment varidase. University of Portsmouth, 1994.

Loetscher H, Pan YC, Lahm HW, Gentz R, Brockhaus M, Tabuchi H, Lessiauer W. Molecular cloning and expression of the human 55 kD tumor necrosis factor receptor. Cell 1990;61:351-359.

234 Lomedico PT, Gubler U, Hellmann CP, Dukovich M, Giri JG, Pan YC, Collier K, Semionow R, Chua AO, Mizel SB. Cloning and expression of murine interleukin-1 cDNA in Escherichia coli. Nature 1984;312:458-462.

Lonnemann G, Endres S, van der Meer JW, Cannon GJ, Koch KM, Dinarello CA. Differences in the synthesis and kinetics of release of interleukin 1 alpha, interleukin 1 beta and tumor necrosis factor from hum an m ononuclear cells. Eur J Im m unol 1989;19:1531-1536.

Looney RJ, Abrahams GN, Anderson CL. Human monocytes and U937 cells bear two distinct Fc receptors for IgG. J Im m unol 1986a;136:1641-1647.

Looney RJ, Ryan DH, Takahashi K, Fleit HB, Cohen HJ, Abraham GN, Anderson CL. Identification of a second class of IgG Fc receptors on human neutrophils. A 40 kilodalton molecule also found on eosinophils. J Exp Med 1986b;163:826-836.

Lu EW, Deftos M, Tighe H, Carson DA, Chen PP. Generation and characterisation of two monoclonal self-associating IgG rheumatoid factors from a rheumatoid synovium . Arthritis Rheum 1992;35:101-105.

Lu EW, Deftos M, Olee T, Huang D, Soto-gil RW, Carson DA, Chen PP. Generation and molecular analyses of two rheumatoid synovial fluid- derived IgG rheumatoid factors. Arthritis Rheum 1993;36:927-937.

LugerTA, Charon JA, Colot M, Micksche M, Oppenheim JJ. Chemotactic properties of partially purified human epidermal cell-derived thymocyte- activating factor (ETAF) for polymorphonuclear and mononuclear cells. J Im m unol 1983;131:816-820.

Lui YJ, Xu J, de-Bouteiller O, Parham CL, Grauard G, Djossau O, de Saint Vis B, Lebecque S, Banchereau J, Moore KW. Follicular dendritic cells specifically express the long CR2/CD21 isoform. J Exp Med 1997;185:165- 170.

235 Maeda M, Van Schie RCAA, Yiiksel B, Greenough A, Fanger MW, Guyre PM, Lydyard PM. Differential expression of Fc receptors for IgG by monocytes and granulocytes from neonates and adults. Clin Exp Im m unol 1996;103:343-347.

Maini RN, Zvaifler N. Rheumatoid arthritis. In: Rheumatology. Eds; Kippel JH, Dieppe PA. Mosby ( St Louis) 1998;2:5.1.1-5.30.2.

Male D, Roitt IM. Introduction to the immune system. In: Immunology. Eds; Roitt IM, Brostoff J, Male D. Mosby (London) 1993:1.1-1.12.

Maloney DG, Liles TM, Czerwinski DK, Waldichuk C, Rosenberg J, Grillo-Lopez A, Levy R. Phase I clinical trial using escalating single-dose infusion of chimeric anti-CD20 monoclonal antibody (IDEC-C2B8) in patients with recurrent B-cell lymphoma. Blood 1994;84:2457-2466.

Mannik M. Physicochemical and functional relationships of immune complexes. J Invest Dermatol 1980;74:333-338.

Mannik M, Nardella FA, Sasso EH. Rheumatoid factors in immune complexes of patients with rheumatoid arthritis. Springer Semin Im m unopathol 1988;10:215-230.

Mannik M. Rheumatoid factors in the pathogenesis of rheumatoid arthritis. J Rheumatol 1992;19(suppl 32):46-49.

Mannik M, Nardella FA. IgG rheumatoid factors and self-association of these antibodies. Clin Rheum Dis 1995;11:551-572.

Mantovani L, Wilder RL, Casali P. Human rheumatoid B-la (CD5+ B) cells make somatically hypermutated high affinity IgM rheumatoid factors. J Im m unol 1993;151:473-488.

Mapp PI, Brootveld MC, Blake DR. Hypoxia, oxidative stress and rheumatoid arthritis. Brit Med Bull 1995;51:419-436.

236 Marlor CW, Webb DL, Bombara MP, Greve JM, Blue ML. Expression of vascular cell adhesion molecule-1 in fibroblast-like synoviocytes after stimulation with TNF. Am J Pathol 1992;140:1055-1060.

Marsh CB, Wewers MD, Tan LC, Rovin BH. Fey receptor cross-linking induces peripheral blood mononuclear cell monocyte chemoattractant protein-1 expression. Role for lymphocyte FcyRIII. J Immunol 1997;158:1078-1084.

Masuda M, Roos D. Association of all three types of Fc gamma receptor (CD64, CD32, and CD16) with a gamma-chain hom odim er in cultured human monocytes. J Im m unol 1993;151:7188-7195.

Matic M, Simon SR. Tumour necrosis factor release from lipopolysaccharide stimulated human monocytes. Lipopolysaccharide tolerance in vitro. Cytokine 1991;3:576-583.

Mclnnes IB, Liew FY. Interleukin 15: a proinflammatory role in rheumatoid arthritis synovitis. Im m unol Today 1998;19:75-79.

McNally JA, Bell AL. Myeloperoxidase-based chemiluminescence of polymorphonuclear leukocytes and monocytes. J Biolumin Chemilumin 1996;11:99-106.

Melendez A, Floto RA, Cameron AJ, Gillooly DJ, Harnett MM, Allen JM. A molecular switch changes the signalling pathway used by the FcyRI antibody receptor to mobilise calcium. Curr Biol 1998;8:210-221.

Mellman I, Koch T, Healey G, Hunziker W, Lewis V, Plutner H, Miettinen H, Vaux D, Moore K, Stuart S. Structure and function of Fc receptors on macrophages and lymphocytes. J Cell Sci 1988;Suppl 9:45-65.

Meyer JM, Evans TI, Small RE, Redford TW, Han J, Singh R, Moxley G. HLA-DRB1 genotype influences risk for and severity of rheumatoid arthritis. J Rheumatol 1999;26:1024-1034.

237 Michl J, Pieczonka MM, Unkeless JC, Bell GI, Silverstein SC. Fc receptor modulation in mononuclear phagocytes maintained on immobilised immune complexes occurs by diffusion of the receptor molecule. J Exp Med 1983;157:2121-2139.

Milgrom F. Development of rheumatoid factor research through 50 years. Scand J Rheum atol Suppl 1988;75:2-12.

Miller KL, Duchemin AM, Anderson CL. A novel role for the Fc receptor y subunit: enhancem ent of FcyR ligand affinity. J Exp Med 1996;183:2227- 2233.

Ming WJ, Bersani L, Mantovani A. Tumor necrosis factor is chemotactic for monocytes and polymorphonuclear leucocytes. J Immunol 1987;138:1469-1474.

Mino T, Sugitama E, Taki H, Kuroda A, Yamashit N, Maruyama M, Kobayashi M. Interleukin-la and tumour necrosis factor a synergistically stimulate prostaglandin E2-dependent production of interleukin-11 in rheumatoid synovial fibroblasts. Arthritis Rheum 1998;41:2004-2013.

Minskoff SA, Matter K, Mellman I. Fc gamma RH-bl regulates the presentation of B cell receptor-bound antigens. J Im m unol 1998;161:2079- 2083.

Miossec O, Naviliat M, Dupuy d'Angeac A, Sany J, Banchereau J. Low levels of interleukin-4 and high levels of transforming growth factor beta in rheumatoid synovitis. Arthritis Rheum 1990;33:1180-1187.

Mizel SB, Dayer JM, Krane SM, Mergenhagen SE. Stimulation of rheumatoid synovial cell collagenase and prostaglandin production by partially purified lymphocyte activating factor (interleukin 1). Proc Natl Acad Sci USA 1981;78:2474-2477.

Monteiro RC, Kubagawa H, Cooper MD. Cellular distribution, regulation and biochemical nature of an Fca receptor in humans. J Exp Med 1990;171:597-613.

238 Monteiro RC, Cooper MD, Kubagawa H. Molecular heterogeneity of Fca receptors detected by receptor-specific monoclonal antibodies. J Immunol 1992;148:1764-1770.

Moore S, Ruska K, Peters L, Olsen NJ. Associations of IgA and IgA- rheumatoid factor with disease features in patients with rheumatoid arthritis. Im m unol Invest 1994;23:355-365.

Moore TL, Dorner RW. Rheumatoid factors. Clin Biochem 1993;26:75-84.

Moreland LW, Pratt PW, Mayes MD, Postlethwaite A, Weisman MH, Schnitzer T, Lightfoot R, Calabrese L, Zelinger DJ, Woody JN, Koopman WJ. Double-blind, placebo-controlled multicentre trial using chimeric monoclonal anti-CD4 antibody, cM-T412, in rheumatoid arthritis patients receiving concomitant methotrexate. Arthritis Rheum 1995;38:1581-1588.

Morgan A, Jones ND, Nesbitt AM, Chaplin L, Bodmer MW, Entage JS. The N-terminal end of the Ch 2 domain of chimeric human IgGl anti- HLA-DR is necessary for Clq, FcyRI and FcyRIII binding. Immunology 1995;86:319-324.

Morgan BP, Daniels RH, Williams BD. Measurement of terminal complement complexes in rheumatoid arthritis. Clin Exp Immunol 1988;73:473-478.

Morton HC, van Egmond M, van de Winkel JGJ. Structure and function of human IgA receptors (Fc alpha R). Crit Rev Im m unol 1996;16:423-440.

Moss ML, Jin SLC, Milla ME, Burkhart W, Carter HL, Chen WJ, Clay WC, Didsbury JR, Hassler D, Hoffman CR, Kost TA, Lambert MH, Leesnitzer MA, McCauley P, McGeehan G, Mitchell J, Moyer M, Pahel G, Rocque W, Overton LK, Schoenen F, Seaton T, Su JL, Warner J, Willard D, Becherer JD. Cloning of a disintegrin metalloproteinase that processes precursor tumour-necrosis factor-a. Nature 1997;385:733-736.

239 Moulton PJ. Inflammatory joint disease: the role of cytokines, cyclooxygenases and reactive oxygen species. Br J Biomed Sci 1996;53:317- 324.

Munthe E, Natvig JB. Characterisation of IgG complexes in eluates from rheumatoid tissue. Clin Exp Imm unol 1971;8:249-262.

Munthe E, Natvig JB. Complement-fixing intracellular complexes of IgG rheumatoid factor in rheumatoid plasma cells. Scand J Immunol 1972;1:217-229.

My hr KM, Raknes G, Nyland H, Vedeler C. Immunoglobulin G Fc- receptor (FcgammaR) IIA and IIIB polymorphisms related to disability in MS. Neurology 1999;52:1771-1776.

Nagase K, Woessner Jr JF. Matrix metalloproteinases. J Biol Chem 1999;274:21491-21494.

Nardella FA, Teller DC, Mannik M. Studies on the antigenic determinants in the self-association of IgG rheumatoid factor. J Exp Med 1981;154:112-125.

Nardella FA, Dayer JM, Roelke M, Krane SM, Mannik M. Self-associating IgG rheumatoid factors stimulate monocytes to release prostaglandins and mononuclear cell factor that stimulates collagenase and prostaglandin production by synovial cells. Rheum atol Int 1983;3:183-185.

Nardella FA, Teller DC, Barber CV, Mannik M. IgG rheumatoid factors and staphylococcal protein A bind to a common molecular site on IgG. J Exp Med 1985;162:1811-1824.

Natvig JB, Gaarder PI, Turner MW. IgG antigens of the Cy2 and Cy3 homology regions interacting with rheumatoid factors. Clin Exp Im m unol 1972;12:177-184.

240 Naume B, Shalaby R, Lesslauer W, Espevik T. Involvement of the 55- and 75-kDa tumor necrosis factor receptors in the generation of lymphokine-activated killer cell activity and proliferation of natural killer cells. J Im m unol 1991;146:3045-3048.

Nedwin GE, Naylor SL, Sakaguchi AY, Smith D, Jarrett-Nedwin J, Pennica D, Goeddel DV, Gray PW. Human lymphotoxin and tumour necrosis factor genes: structure, homology and chromosomal localization. Nucleic Acids Res 1985;13:6361-6373.

Nepom GT, Byers P, Seyfried C, Healey LA, Wilske KR, Stage D, Nepom BS. HLA genes associated with rheumatoid arthritis. Identification of susceptibility alleles using specific oligonucleotide probes. Arthritis Rheum 1989;32:15-21.

Newkirk MM, Lemmo A, Rauch J. Importance of the IgG isotype, not the state of the glycosylation in determining human rheumatoid factor binding. Arthritis Rheum 1990;33:800-809.

Newkirk MM, Rauch J. Binding of human monoclonal IgG rheumatoid factors to Fc is influenced by carbohydrate. J Rheum atol 1993;20:776-780.

Niedel JE, Kuhn LJ, Vandenbark GR. Phorbol diester receptor copurifies with protein kinase C. Proc Natl Acad Sci USA 1983;80:36-40.

Nophar Y, Kemper O, Brakebusch C, Englemann H, Zwang R, Aderka D, Holtmann H, Wallach D. Soluble forms of tumor necrosis factor receptors (TNF-Rs). The cDNA for the type I TNF-R cloned using amino acid sequence data of its soluble form encodes both the cell surface and a soluble form of the receptor. EMBO J 1990;9:3269-3278.

Oliver SJ, Brahn E. Combination therapy in rheumatoid arthritis: The animal model perspective. J Rheumatol 1996;23:56-60. Opplinger IR, Nardella FA, Stone GC, Mannik M. Human rheumatoid factors bear the internal image of the Fc binding region of staphylococcal protein A. J Exp Med 1987;166:702-710.

241 Ory PA, Clark MR, Kwoh WW, Clarkson SB, Goldstein IM. Sequences of complementary DNAs that encode the NA1 and NA2 forms of Fc receptor III on hum an neutrophils. J Clin Invest 1989a;84;1688-1691.

Ory PA, Goldstien IM, Kwoh EE, Clarkson SB. Characterisation of polymorphic forms of FcRIII on human neutrophils. J Clin Invest 1989b;83:1676-1681.

O'Sullivan FX, Gay RE, Gay G. Spontaneous arthritis models. In: Mechanisms and models in rheumatoid arthritis. Eds; Flenderson B, Edwards JCW, Pettipher ER. Academic Press (London) 1995;471-483.

Parekh RB, Dwek RA, Sutton BJ, Fernandes DL, Leung A, Stanworth D, Rademacher TW. Association of rheumatoid arthritis and primary osteoarthritis with changes in the glycosylation pattern of total serum IgG. Nature 1985;316:452-457.

Parekh RB, Roitt I, Isenberg D, Dwek R, Rademacher T. Age-related galactosylation of the N-linked oligosaccharides of human serum IgG. J Exp Med 1988;167:1731-1736.

Pascual V, Capra JD. VH4-21, a human VH gene segment over­ represented in the autoimmune repertoire. Arthritis Rheum 1992;35:11- 18.

Passlick B, Flieger D, Ziegler-Heitbrock HH. Identification and characterisation of a novel monocyte subpopulation in human peripheral blood. Blood 1989;74:2527-2534.

Patry C, Herbelin A, Lehuen A, Bach JF, Monteiro RC. Fca receptors mediate release of tumour necrosis factor-a and interleukin-6 by human monocytes following receptor aggregation. Immunology 1995;86:1-5.

Pearson CM. Development of arthritis, periarthritis and periostitis in rats given adjuvants. ProcSoc Exp Biol 1956;91:95-101.

242 Pekin TJ, Zvaifler NJ. Hemolytic complement in synovial fluid. J Clin Invest 1964;43:1372.

Pennica D, Ned win GE, Hatflick JS, Seeburg PH, Derynck R, Palladino MA, Kohr WJ, Aggarwall BB, Goeddel DV. Human tumour necrosis factor: precursor structure, expression and homology to lymphotoxin. Nature 1984;312:724-729.

Perkins SJ, Nealis AS, Sutton BJ, Feinstein A. Solution structure of human and mouse immunoglobulin M by synchrotron X-ray scattering and molecular graphics modelling. A possible mechanism for complement activation. JM ol Biol 1991;221:1345-1366.

Perussia B, Dayton ET, Lazarus R, Fanning V, Trinchieri G. Immune interferon induces the receptor for monomeric IgGl on human monocytic and myeloid cells. J Exp Med 1983;158:1092-1113.

Perussia B, Trinchier G. Antibody 3G8, specific for the human neutrophil Fc receptor, reacts with natural killer cells. J Im m unol 1984;132:1410-1415.

Peschon JJ, Slack JL, Reddy P, Stocking KL, Sunnarborg SW, Lee DC, Russell WE, Castner BJ, Johnson RS, Fitzner JN, Boyce RW, Nelson N, Kozlosky CJ, Wolf son MF, Rauch CT, Cerretti DP, Paxton RJ, March Q, Black RA. An essential role for ectodomain shedding in mammalian development. Science 1998;282:1281-1284.

Pettipher EJ, Higgs GA, Henderson B. Interleukin-1 induced leucocyte infiltration and cartilage proteoglycan degradation in the synovial joint. Proc Natl Acad Sci USA 1986;83:8749-8753.

Pfefferkorn LC, Yeaman GR. Association of IgA-Fc receptors (FcaR) with

FcyRIy2 subunits in U937 cells. J Im m unol 1994;153:3228-3236.

Pike RM, Sulkin SE, Coggeshall HC. Serological reactions in rheumatoid arthritis. II. Concerning the nature of the factor in rheumatoid arthritis serum responsible for increased agglutination of sensitized sheep erythrocytes. J Im m unol 1949;63:447-463.

243 Plater-Zyberk C, Maini RN, Lam K, Kennedy TD, Janossy G. A rheumatoid arthritis B cell subset expresses a phenotype similar to that in chronic lymphocyte leukaemia. Arthritis Rheum 1985;28:971-976.

Platonov AE, Shipulin GA, Vershinina IV, Dankert J, van de Winkel JGJ, Kuijper EJ. Association of human Fc gamma Rlla (CD32) polymorphism with susceptibility to and severity of meningococcal disease. Clin Infect Dis 1998;27:746-750.

Polat GL, Laufer J, Fabian I, Passwell JH. Cross-linking of monocyte plasma membrane Fca, Fey or mannose receptors induces TNF production. Imm unology 1993;80:287-292.

Pope RM, Teller DC, Mannik M. The molecular basis of self-association of antibodies to IgG (rheumatoid factors) in rheumatoid arthritis. Proc Nat Acad Sci USA 1974;71:517-521.

Pope RM, Teller DC, Mannik M. The molecular basis of self-association of IgG-rheumatoid factors. J Im m unol 1975;115:365-373.

Porges AJ, Redecha PB, Doebele R, Pan LC, Salmon JE, Kinberly PR. Novel Fc gamma receptor I family gene products in human mononuclear cells. J Clin Invest 1992;90:2102-2109.

Powell RJ, Leyland AM, Pound JD, Bossingham DH. An improved assay for IgG rheumatoid factor. Its value in the diagnosis of rheumatoid arthritis. J Rheum 1985;12:427-431.

Pulendran B, Kannourakis G, Nourl S, Smith KGC, Nossal GJV. Soluble antigen can cause enhanced apoptosis of germinal-centre B cells. Nature 1995;375:331-334.

Pulendran B, Van Driel R, Nossal GJ. Immunological tolerance in germinal centres. Im m unol Today 1997;18:27-32.

244 Qui WQ, de Bruin D, Brownstein BH, Pearse R, Ravetch JV. Organisation of the human and mouse low-affinity FcyR genes: Duplication and recombination. Science 1990;248:732-735.

Ra C, Jouvin ME, Blank U, Kinet JP. A macrophage Fey receptor and the mast cell receptor for IgE share an identical subunit. Nature 1989;341:752- 754.

Rademacher TW, Parekh RB, Dwek RA, Isenberg D, Rook G, Axford JS, Roitt I. The role of IgG glycoforms in the pathogenesis of rheumatoid arthritis. Springer Sem in Im m unopathol 1988;10:23-49.

Radoiu N, Cleveland RP, Leon MA. Peripheral T lymphocytes from rheumatoid arthritis patients recognise determinants on light chains and/or Fab fragments of im m unoglobulin G. J Clin Invest 1982;70:329- 334.

Ravetch JV, Perussia B. Alternative membrane forms of FcyRIII (CD16) on human natural killer cells and neutrophils. Cell type-specific expression of two genes that differ in single nucleotide substitutions. J Exp Med 1989;170:481-497.

Ravetch JV, Kurosaki TS. A single amino acid in the glycophosphatidylinositol attachment domain determines the_ membrane topology of FcRIII. Nature 1989;342:805-807.

Ravetch JV, Kinet JP. Fc receptors. Annu Rev Im m unol 1991;9:457-492.

Reff ME, Carner K, Chambers KS, Chinn PC, Leonard JE, Raab R, Newman RA, Hanna N, Anderson DR. Depletion of B cells in vivo by a chimeric mouse human monoclonal antibody to CD20. Blood 1994;83:435-445.

Reth M. Antigen receptor tail clue. Nature 1989;338:383-384.

Reveille JD. The genetic contribution to the pathogenesis of rheumatoid arthritis. Curr Opin Rheumatol 1998;10:187-200.

245 Richter J, Andersson T, Olsson I. Effect of tumor necrosis factor and granulocyte/macrophage colony-stimulating factor on neutrophil degranulation. J Im m unol 1989;142:3199-3205.

Richter J, Gullberg U, Lantz M. TNF-induced superoxide anion production in adherent human neutrophils involves both the p55 and p75 TNF receptor. J Im m unol 1995;154:4142-4149.

Robbins DL, Wistar R Jr. Comparative specificities of serum and synovial 19S IgM rheumatoid factors in rheumatoid arthritis. J Rheumatol 1985;12:437-443.

Robbins DL, Feigal DW Jr, Leek JC. Relationship of serum IgG rheumatoid factor and IgM rheumatoid factor and disease activity in rheumatoid arthritis. J Rheumatol 1986;13:259-562.

Robbins DL, Benisek WF, Benjamini E, Wistar R Jr. Differential reactivity of rheumatoid synovial cells and serum rheumatoid factors to human immunoglobulin G subclasses 1 and 3 and their CH3 domains in rheumatoid arthritis. Arthritis Rheum 1987;30:489-497.

Robbins DL, Kenny TP, Snyder LL, Ermel RW, Larrick JW. Allotypic dependency of the sepcificity and avidity of human monoclonal IgM rheumatoid factors derived from rheumatoid synovial cells. Arthritis- Rheum 1993;36:389-393.

Roberts-Thompson PJ, McEvoy R, Langhans T, Bradley J. Routine quantification of rheumatoid factor by rate nephelometry. Ann Rheum Dis 1985;44:379-383.

Roosnek E, Lanzavecchia A. Efficient and selective presentation of antigen-antibody complexes by rheumatoid factor B cells. J Exp Med 1991;173:487-489.

Rose HM, Regen C, Peausco E, Lipman MO. Differentiating agglutination of normal and sensitized sheep erythrocytes by sera of patients with rheumatoid arthritis. Proc Soc Exp Biol Med 1948;68:1-6.

246 Rosenfeld SI, Looney RJ, Leddy JP, Phipps DC, Abraham GN, Anderson CL. Human platelet Fc receptor for IgG. Identification as a 40-kD membrane protein shared by monocytes. J Clin Invest 1985;76:2317-2322.

Ruddy S, Britton MC, Schur PH, Austin KF. Complement components in synovial fluid: activation and fixation in seropositive rheumatoid arthritis. Ann N Y Acad Sci 1969;168:161-172.

Rudge SR, Pound JD, Bossingham DH, Powell RJ. Class specific rheumatoid factors in rheumatoid arthritis: response to chrysotherapy and relationship to disease activity. J Rheum atol 1985;12:432-436.

Sabharwal UK, Vaughan JH, Fong S, Bennett PH, Carson DA, Curd JG. Activation of the classical pathway of complement by rheumatoid factors: assessment by radioimmunoassay for C4. Arthritis Rheum 1982;25:161- 167.

Sack U, Kinne RW, Marx T, Heppt P, Bender S, Emmrich F. Interleukin-6 in synovial fluid is closely associated with chronic synovitis in rheumatoid arthritis. Rheum atol Int 1994;14:83-84.

Saklatvala J, Pilsworth LM, Sausfiled SJ, Gavrilouic J Heath JK. Pig catabolin is a form of interleukin 1. Cartilage and bone resorption, fibroblasts make prostaglandin and collagenase, and thymocyte- proliferation in augmented in response to one protein. Biochem J 1984;224:461-466.

Saklatvala J, Sarsfield Y, Townsend Y. Purification of two immunologically different leucocyte proteins that cause cartilage resorption, lymphocyte activation and fever. J Exp Med 1985;162:1208- 1222.

Saklatvala J. Tumour necrosis factor a stimulates resorption and inhibits synthesis of proteoglycan in cartilage. Nature 1986;322:547-549.

247 Salmon JE, Edberg JC, Kimberly RP, Mensa E, Ryan R. Fey receptor HI on human neutrophils. Allelic variants have functionally distinct capabilities. J Clin Invest 1990;85:1287-1295.

Salmon JE, Edberg JC, Brogle NL, Kimberly RP. Allelic polymorphisms of human Fey receptor IIA and Fey receptor IIIB. Independent mechanisms for difference in hum an phagocytic function. J Clin Invest 1992;89:1274- 1281.

Samuelsson J, Forslid J, Hed J, Palmblad J. Studies of neutrophil and monocyte oxidative responses in polycythaemia vera and related myeloproliferative disorders. Br J Haematol 1994;87:464-470.

Sanders LAM, van de Winkel JGJ, Rijkers GT, Voorhorst-Ogink MM, de Haas M, Capel PJ, Regers BJ. Fey receptor Ila (CD32) heterogeneity in patients with recurrent bacterial tract infection. J Infect Dis 1994;170:854- 860.

Sandilands GP, Macpherson SA, Burnett ER, Russell AJ, Daunie I, Macsween RNM. Differential expression of CD32 isoforms following alloactivation of hum an T cells. Im m unology 1997;91:204-211.

Saren P, Welgus HG, Kovanen PT. TNF-alpha and IL-1 beta selectively induce expression of 92-kDa gelatinase by human macrophages. J* Imm unol 1996;157:4159-4165.

Sasso EH, Barber CV, Nardella FA, Yount WJ, Mannik M. Antigenic specificities of human monoclonal and polyclonal IgM rheumatoid factors. The Cy2-Cy3 interface region contains the major determinants. J Imm unol 1988;140:3098-3107.

Saxne T, Palladino MA Jr, Heinegoud D, Talal N, Wollheim FA. Detection of tumour necrosis factor alpha but not tumour necrosis factor beta in rheumatoid arthritis synovial fluid and serum. Arthritis Rheum 1988;31:1041-1045.

248 Schall TJ, Lewis M, Koller KJ, Lee A, Rice GC, Wong GHW, Gatanaga T, Granger GA, Lentz R, Raab H, Kohr WJ, Goeddel DV. Molecular cloning and expression of a receptor for human tumor necrosis factor. Cell 1990;61:361-370.

Schmitt DA, Hanau D, Bieber T, Dezutter-Dambuyant C, Schmitt D, Fabre M, Pauly G, Cazenave JP. Human epidermal Langerhans cells express only the 40-kilodalton Fey receptor (FcRII). J Im m unol 1990;144:4284-4290.

Scholl PR, Geha RS. Physical association between the high affinity IgG receptor (FcyRI) and the y subunit of the high affinity IgE receptor (FceRI). Proc Natl Acad Sci USA 1993;90:8847-8850.

Schrohenloher RE. Characterisation of the gamma-globulin complexes present in certain sera having high titres of anti-gamma globulin activity. J Clin Invest 1966;45:501-512.

Schur PH. Clinical features of SLE. In: Textbook of rheumatology. Eds; Kelly WN, Harris Jr ED, Ruddy S, Sledge CR. Saunders (Philadelphia) 1993:1017-1042.

Sedmak DH, Davis HH, Singh U, van de Winkel JGJ, Anderson CL. Distribution of IgG Fc receptor antigens in placenta and on endothelial cells in man: An im m unohistochem ical study. Am J Pathol 1991;138:175-~ 181.

Selvaraj P, Rosse WF, Silber R, Springer TH. The major Fc receptor in blood has a phophatidylinositol anchor and is deficient in paroxysmal nocturnal haemoglobuinuria. Nature 1988;333:565-567.

Senior BW, McBride PD, Morley KD, Kerr MA. The detection of raised levels of IgM to Proteus mirabilis in sera from patients with rheumatoid arthritis. J Med Microbiol 1995;43:176-184.

249 Serino G, Pitingolo F, Granatieri C, Maresca V, Invernizzi F, Zara C, Zanussi C. Hypergammaglobulinemic purpura of Waldenstrom: characterisation of circulating immune complexes. Acta Haemat 1983;69:152-157.

Shen L, Collins JE, Schoenborn MA, Maliszewski CR. Lipopolysaccharide and cytokine augmentation of human monocyte IgA receptor expression and function. J Im m unol 1994;152:4080-4086.

Shmerling RH, Delbanco TL. The rheumatoid factor: An analysis of clinical utility. Am J Med 1991;91:528-534.

Short CL. The antiquity of rheumatoid arthritis. Arthritis Rheum 1974;17:193-205.

Simmons D, Seed B. Fey receptor of natural killer cells is a phospholipid- linked membrane protein. Nature 1988;333:568-570.

Sims JE, Painter SL, Gow IR. Genomic organisation of the type I and type IIIL-1 receptors. Cytokine 1995;7:483-490.

Singer JM, Plotz GM. The latex fixation test. I. Applications for the serological diagnosis of rheumatoid arthritis. Am J Med 1956;21:888-892.

Snorrason E. Landre-Beauvais and his Goutte Asthenique Primitive. Act Med Scand 1952;142(suppl 266):115-118.

Sondermann P, Fluber R, Jacob U. Crystal structure of the soluble form of the human Fc gamma-receptor lib: a new member of the im m unoglobulin superfamily at 1.7 A resolution. EMBO J. 1999a;18: 1095- 1103.

Sondermann P, Jacob U, Kutscher C, Frey J. Characterisation and crystallization of soluble human Fc gamma RII (CD32) isoforms produced in insect cells. Biochemistry 1999b;38:8469-8477.

250 Spies T, Morton CC, Nedospasor SA, Fiers W, Pious D, Strominger JL. Genes for the tumor necrosis factors alpha and beta are linked to the human major histocompatibility complex. Proc Natl Acad Sci USA. 1986;83:8699-8702.

Stastny P. Association of the B-cell alloantigen DRw4 with rheumatoid arthritis. N Engl J Med 1978;298:869-871.

Stauffer TP, Meyer T. Compartmentalized IgE receptor-mediated signal transduction in living cells. J Biol Chem 1997;139:1447-1454.

Stengelin S, Stamenkovic I, Seed B. The isolation of cDNAs from two distinct human Fc receptors by ligand affinity cloning. EMBO J 1988;7:1053-1059.

Stone GC, Sjobring U, Bjorck L, Sjoquist J, Barber CV, Nardella FA. The Fc binding site for streptococcal protein G is in the Cy2-Cy3 interface region of IgG and is related to the sites that bind staphylococcal protein A and hum an rheum atoid factors. J Im m unol 1989;143:565-570.

Strieter RM, Kunkel SL, Showell HJ, Remick DG, Phan SH, Ward PA; Marks RM. Endothelial cell gene expression of a neutrophil chemotactic factor by TNF-alpha, LPS and IL-1 beta. Science 1989;243:1467-1469.

Stuart SG, Trounstine M, Vaux D, Koch T, Martens C, Mellman I, Moore KM. Isolation and expression of cDNA clones encoding a human receptor for IgG (FcyRII). J Exp Med 1987;166:1668-1684.

Sturrock RD, Sharma JN, Buchanan WW. Evidence of rheumatoid arthritis in ancient India. Arthritis Rheum 1977;20:42-44.

Swedler W, Wallman J, Froelich CJ, Teodorescu M. Routine measurement of IgM, IgG and IgA rheumatoid factors: high sensitivity, sepcificity, and predictive value for rheumatoid arthritis. J Rheumatol 1997;24:1037-1044.

251 Sylvestre DL, Ravetch JV. Fc receptors initiate the Arthus reaction: redefining the inflammatory cascade. Science 1994;265:1095-1098.

Sylvestre DL, Clynes R, Ma M, W arren H, Carroll MC, Ravetch JV. Immunoglobulin G-mediated inflammatory responses develop normally in complement-deficient mice. J Exp Med 1996;184:2385-2392.

Tak PP, van der Lubbe PA, Cauli A, Daha MR, Smeets TJM, Kluin PM Meinders AE, Yanni G, Panayi GS, Breed veld FC. Reduction of synovial inflammation after anti-CD4 monoclonal antibody treatment in early rheumatoid arthritis. Arthritis Rheum 1995;38:1457-1465.

Taketazu F, Kato M, Gobi A, Ichijo H, ten Dijke P, Itoh J, Kyogoku M, Ronnelid J, Miyazono K, Heldin CPI, Funa K. Enhanced expression of transforming growth factor beta's and transforming growth factor beta type II receptor in the synovial tissues of patients with rheumatoid arthritis. Lab Invest 1994;70:620-630.

Tamm A, Kister A, Nolte KU, Engelbert J, Schmidt RE. The IgG binding site of human FcyRIIIB receptor involves CC' and FG loops of the m em brane-proxim al domain. J Biol Chem 1996;271:3659-3666.

Tamm A, Schmidt RE. The binding epitopes of human CD16 (FcyRIII) monoclonal antibodies. Implications for ligand binding. J Immunol _ 1996;157:1576-1581.

Tanimoto K, Moritoh T, Azuma T, IToriuchi Y. Detection of IgG rheumatoid factor by concanavalin A treatment and complement fixation with IgG rheumatoid factor. Ann Rheum Dis 1976;35:240-245.

Tartaglia LA, Goeddel DV. Two TNF receptors. Immunol Today 1992;13:151-153.

Tausk F, Gigli I. The human C3b receptor: function and role in human diseases. J Invest Dermatol 1990;94:141S-145S.

252 Tax WJM, Willems HW, Reekers PPM, Capel PJA, Koene RAP. Polymorphism in mitogenic effect of IgGl monoclonal antibodies against T3 antigen on hum an T cells. N ature 1983;304:445-447.

Tax WJM, van de Winkel JGJ. Human Fey receptor II: a standby receptor activated by proteolysis. Immunol Today 1990;11:308-310. te Velde A A, Huijbens RJ, de Vries JE, Figder CG. IL-4 decreases Fc gamma-R membrane expression and Fc gamma R-mediated cytotoxic activity of hum an monocytes. J Im m unol 1990;144:3046-3051.

Teitsson I, Withington RH, Seifert MH, Valdimaisson H. Prospective study of early rheumatoid arthritis. I. Prognostic value of IgA rheumatoid factor. Ann Rheum Dis 1984;43:673-678.

Thomas R, Lipsky PE. Presentation of self peptides by dendritic cells. Possible implications for the pathogenesis of rheumatoid arthritis. Arthritis Rheum 1996;39:183-190.

Thompson BM, Mundy GR, Chambers TJ. Tumour necrosis factors alpha and beta induce osteoblastic cells to stimulate osteoclast bone resorption. J Im m unol 1987;138: 775-779.

Thompson KM, Randen I, Borretzen M, Forre 0, Natvig JB. Variable- region gene usage of human monoclonal rheumatoid factors derived from healthy donors following immunisation. Eur J Immunol 1994;24:1771-1778.

Thompson KM, Borretzen M, Randen I, Forre 0, Natvig JB. V-gene repertoire and hypermutation of rheumatoid factors produced in rheumatoid synovial inflammation and immunized healthy donors. A nn NY Acad Sci 1995a;764:440-449.

Thompson KM, Borretzen M, Natvig JB. New data on rheumatoid factors in health and disease. Rheumatology in Europe 1995b;24(suppl 2):165-167.

253 Thould AK, Thould BT. Arthritis in Roman Britain. Br Med J 1983;287:1909-1911.

Tomana M, Schrohenloher RE, Koopman WJ, Alarcon GS, Paul WA. Abnormal glycosylation of serum IgG from patients with chronic inflammatory diseases. Arthritis Rheum 1988;31:333-338.

Trentham DE, Townes AS, Kang AH. A utoim m unity to type II collagen: an experimental model of arthritis. J Exp Med 1977;146:857-868.

Trezzini C, Jungi TW, Spycher MO, Maly FE, Rao P. Human monocytes CD36 and GDI 6 are signalling molecules. Evidence from studies using antibody-induced chemiluminescence as a tool to probe signal transduction. Immunology 1990;71:29-37.

Tuomi T. Which antigen to use in the detection of rheumatoid factors? Comparison of patients with rheumatoid arthritis and subjects with 'false-positive' rheumatoid factor reactions. Clin Exp Immunol 1989;77:349-355.

Tyler JA, Articular cartilage cultured with catabolin (pig interleukin 1) synthesizes a decreased number of normal proteoglycan molecules. Biochem J 1985;227:869-878.

Unkeless JC, Scigliano E, Freedman VH. Structure and function of human and murine receptors for IgG. Annu Rev Immunol 1988;6:251- 281.

van Jaarsveld CHM, Otten HG, Jacobs JWG, Kruize A A, Brus HLM, Bijlsma JWJ. Is there an indication for HLA-DR typing for individual patients with rheumatoid arthritis? Curr Opin Rheumatol 1998a;16:483- 488.

van Jaarsveld CHM, Otten HG, Jacobs JWG, Kruize A A, Brus HLM, Bijlsma JWJ. Association of HLA-DR with susceptibility to and clinical expression of rheumatoid arthritis: re-evaluation by means of genomic tissue typing. Br J Rheum atol 1998b;37:411-416.

254 van Schaardenburg D, Lagaay AM, Otten HG, Breeveld FC. The relation between class-specific serum rheumatoid factors and age in the general population. Br J Rheum atol 1993;32:546-549. van Vugt MJ, Heijen AF, Capel PJ, Park SY, Ra C, Saito T, Verbeek JS, van de Winkel JGJ. FcRy-chain is essential for both surface expression and function of hum an FcyRI (CD64) in vivo. Blood 1996;87:3593-3599. van de Winkel JGJ, Anderson CL. Biology of human immunoglobulin G Fc receptors. J Leukoc Biol 1991;49:511-524. van de W inkel JGJ, Ernst LK, Anderson CL, Chiu IM. Gene organisation of the human high affinity receptor for IgG, FcgRI (CD64). Characterisation and evidence for a second gene. J Biol Chem 1991;266:13449-13455. van de Winkel JGJ, Capel PJA. Human IgG Fc receptor heterogeneity: molecular aspects and clinical implications. Immunol Today 1993;14:215- 221. van de W inkel JGJ, De W it TPM, Ernst LK, Capel PJA, Ceuppens JL. Molecular basis for a familial defect in phagocyte expression of IgG receptor I (CD64). J Im m unol 1995;154:2896-2903. van den Berg WB, Joosten LAB, Helsen M, van de Loo FAJ. Amelioration of established murine collagen-induced arthritis with anti- IL-1 treatment. Clin Exp Im m unol 1994;95:237-243. van den Broek MF, van de Langerijt LG, van Bruggen MC, Billingham ME, van den Berg WB. Treatment of rats with monoclonal anti-CD4 induces long-term resistance to streptococcal cell wall-induced arthritis. Eur J Im m unol. 1992; 22: 57-61. van den Herik-Oudijk, Ter Bekke MWH, Tem pelm an MJ, Capel PJA, van de Winkel JGJ. Functional differences between two Fc receptor ITAM signalling motifs. Blood 1995a;86:3302-3307.

255 van den Herik Oudijk IE, Capel PJ, van der Bruggen T, Van de Winkel JGJ. Identification of signaling motifs within human Fc gamma Rlla and Fc gamma Rllb isoforms. Blood 1995b; 85:2202-2211. van der Lubbe PA, Dijkmans BAC, Markusse HM, Nassander U, Breeedveld FC. A randomised, double-blind, placebo-controlled study of CD4 monoclonal antibody therapy in early rheumatoid arthritis. Arthritis Rheum 1995;38:1097-1106.

Vance BA, Huizinga TWJ, Ward well K, Guy re PM. Binding of monomeric human IgG defines an expression polymorphism of Fo/RIII on large granular lymphocyte/natural killer cells. J Immunol 1993;151:6429-6439.

Venables P. Epstein-Barr virus infection and autoimmunity in rheumatoid arthritis. Ann Rheum Dis 1988;47:265-269.

Villiger PM, Terkeltaub R, Lotz M. Monocyte chemoattractant protein-1 (MCP-1) expression in human articular cartilage. Induction by peptide regulatory factors and differential effects of dexamethasone and retinoic acid. J Clin Invest 1992;90:488-496.

Vincenti MP, Coon Cl, Brinckerhoff CE. Nuclear factor kappa B/p50 activates an element in the distal matrix metalloproteinase -1 promoton. in interleukin-1 beta stimulated synovial fibroblasts. Arthritis Rheum 1998;41:1987-1994.

Voice JK, Lachmann PJ. Neutrophil Fey and complement receptors involved in binding soluble IgG immune complexes and in specific granule release induced by soluble IgG immune complexes. Eur J Im m unol 1997;27:2514-2523.

Waage A, Kaufmann C, Espevik T, Husby G. Interleukin -6 in synovial fluid from patients with arthritis. Clin Immunol Immunopathol 1989;50:394-398.

256 Waaler E. A factor in human serum activating the specific agglutination of sheep blood corpuscles. Abstracts 3rd Internat Cong Microbiol 1939;NY:351.

Waaler E. On the occurrence of a factor in human serum activating the specific agglutination of sheep blood corpuscles. Acta Path Microbiol Scand 1940;17:172-188.

Walker BAM, Hagenlocker BE, Stubbs Jr EB, Sandborg RR, Agranoff BW, Ward PA. Signal transduction events and FcyR engagement in human neutrophils stimulated with immune complexes. J Immunol 1991;146:735-741.

Wainstein E, Edberg J, Csernok E, Sneller M, Hoffman G, Keystone E, Gross W, Salmon J, Kimberly R. FcyRIIIB alleles predict renal dysfunction in Wegener's Granulomatosis (WG). Arthritis Rheum 1996;39:S210.

Waldenstrom J. Kliniska metoder for p&visande av hyperproteinami och delas praktiska v'Arde fdr diagnostiken. Nord Med 1943;20:2288-2295.

Walport MJ. Complement. In: Immunology. Eds; Roitt IM, Brostoff J, Male D. Mosby (London) 1993:12.1-12.16.

Walport MJ, Davis KA, Morley BJ, Bott M. Complement deficiency and_ autoim munity. A nn NY Acad Sci 1997;815:267-281.

Wangoo A, Haynes AR, Sutcliffe SP, Sorooshian M, Shaw RJ. Modulation of platelet derived growth factor B mRNA abundance in macrophages by colchicine and dibutyryl-cAMP. Mol Pharmacol 1992;42:584-589.

Warderdam PAM, van de Winkel JGJ, Gosselin EJ, Capel PJA. Molecular basis for a polymorphism of human Fey receptor II (CD32). J Exp Med 1990;172:19-25.

257 Warderdam PAM, van de Winkel JGJ, Vlug A, Westerdaal NAC, Capel PJA. A single amino acid in the second Ig-like domain of the human Fey receptor II is critical for hum an IgG2 binding. J Im m unol 1991;147:1338- 1343.

Ware CF, Craue PD, Vanarsdale TL, Andreus JL, Grayson MH, Jerzy R, Smith CA, Goodwin RG. Tumor necrosis factor (TNF) receptor expression in T lymphocytes. Differential regulation of the type I receptor during activation and effector T cells. J Im m unol 1991;147:4229-4238.

Warnock MG, Goodacre JA. Cryptic T-cell epitopes and their role in the pathogenesis of autoimmune diseases. Br J Rheumatol 1997;36:1144-1150.

Webb AC, Collins KL, A vron PE, Eddy RL, Nakai H, Byers MG, Hay ley LL, Henry WM, Shaws TB. Interleukin-1 gene (IL-1) assigned to the long arm of human chromosome 2. Lymphokine Res 1986;5:77-85.

Welch, GR, W ong HL, W ahl SM. Selective induction of FcyRIII on human monocytes by transforming growth factor-p. J immnuol 1990;144:3444-3448.

Wilkinson LS, Edwards JCW. Binding of antibodies raised against tumour necrosis factor alpha (TNF alpha) to blood vessels and macrophages in inflamed synovial tissue. Rheumatol Int 1991;11:19-25.

Wilkinson LS, Edwards JCW, Poston RN, Haskard DO. Expression of vascular cell adhesion molecule -1 in normal and inflamed synovium. Lab Invest 1993;68:82-88.

Williams AF, Barclay AN. The immunoglobulin superfamily - domains for cell surface recognition. Ann Rev Immunol 1988;6:381-405.

W illiams DG, Moyes SP, Mageed RA. Rheum atoid factor isotype switch and somatic mutation varients within rheumatoid arthritis synovium. Immunology 1999;98:123-136.

258 Williams Jr RC. Rheumatoid factors and other serum components associated with rheumatoid arthritis. In: Rheumatoid arthritis as a systemic disease. Saunders (Philadelphia) 1974:162.

Williams Jr RC. Rheumatoid factors: Historical perspective, origins and possible role in disease. J Rheumatol 1992a;19(suppl 32):42-45.

Williams Jr RC, Malone CC, Tsuchiya N. Rheumatoid factors from patients with rheumatoid arthritis react with p 2 -microglobulin. J Im m unol 1992b;149:1104-1113.

Williams Jr RC, Malone CC. Rheumatoid-factor-reactive sites on CH 2 established by analysis of overlapping peptides of primary sequence. Scand J Im m unol 1994;40:443-456.

Williams RO, Feldmann M, Maini RN. Anti-tumor necrosis factor ameliorates joint disease in murine collagen-induced arthritis. Proc Natl Acad Sci USA 1992;89:9784-9788.

Winchester RJ, Agnello V, Kunkel HG. Gamma globulin complexes in synovial fluids of patients with rheumatoid arthritis. Partial characterisation and relationship to lowered complement levels. Clin Exp Im m unol 1970;6:689-706.

Winchester RJ. The molecular basis of susceptibility to rheumatoid arthritis. Adv Im m unol 1994;56:389-466.

Wingfield P, Pain RH, Craig S. Tumor necrosis factor is a compact trimer. FEBS Lett 1987;211;179-186.

W irthm ueller U, Kurosaki T, M urakam i MS, Ravetch JV. Signal transduction by FcRIII (CD16) is mediated through the y chain. J Exp Med 1992;175:1381-1390.

Woessner Jr JF. Matrix metalloproteinases and their inhibitors in connective tissue remodelling. FASEB 1991;5:2145-2154.

259 Wong HL, Welch GR, Brandes ME, Wahl SM. IL-4 antagonises induction of Fc gamma RIII (CD16) expression by transforming growth factor-beta on hum an monocytes. J Im m unol 1991;147:1843-1848.

Wong WW. Structural and functional correlation of the human complement receptor type 1. J Invest Dermatol 1990;94:645-675.

W right SD, Ramos RA, Tobias PS, Ulevitch RJ, M athison JC. CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS biding protein. Science 1990;249:1431-1433.

Wu J, Edberg JC, Redecha PB, Bansal V, Guyre PM, Coleman K, Salmon JE, Kimberly RP. A novel polym orphism of FcyRIIIa (GDI 6 ) alters receptor function and predisposes to autoimmune disease. J Clin Invest 1997;100:1059-1070.

Xu WD, Firestein GS, Taetle R, Kaushansky K, Zvaifler NJ. Cytokines in chronic inflammatory arthritis. II. Granulocyte macrophage colony- stimulating factor in rheumatoid synovial effusions. J Clin Invest 1989;83:876-882.

Yelamos J, Garcia-Lorano JR, Moreno I, Aguilera I, Gonzalez MF, Garcia A, Nunez-Roldan A, Sanchez B. Association of HLA-DR4-Dwl5 (DRB1*0405) and DR10 w ith rheum atoid arthritis in a Spanish population. Arthritis Rheum 1993;36:811-814.

Youinou PY, Moroow JW, Lettin AWF, Lydyard PM, Roitt IM. Specificity of plasma cells in the rheumatoid synovium I. Scand J Immunol 1984;20:307-315.

Young DA, Lowe LD, Clasr SC. Comparison of the effects of IL-3, granulocyte-macrophage colony-stimulating factor, and macrophage colony-stimulating factor in supporting monocyte differentiation in culture. Analysis of macrophage antibody-dependent cellular cytotoxicity. J Im m unol 1990;145:607-615.

260 Yuan J. Transducing signals of life and death. Curr Opin Cell Biol 1997;9:247-251.

Ziegler-Heitbrock HWL, Passlick B, Fleiger D. The monoclonal antimonocyte antibody My4 stains B lymphocytes and two distinct monocyte subsets in human peripheral blood. Hybridoma 1988;7:521-527.

Ziegler-Heitbrock HWL, Ulevitch RJ. CD14: Cell surface receptor and differentiation marker. Immunol Today 1993;14:121-125.

Ziff M. The rheumatoid nodule. Arthritis Rheum 1990;33:761-767.

Zimmerman RJ, Chan A, Leaden SC. Oxidative damage in murine tumor cells treated in vitro by recombinant human tumor necrosis factor. Cancer Res 1989;49:1644-1648.

261 APPENDICES

262 Appendix 1.

Relevant publications

Edwards JCW, Cambridge G, Abrahams VM. Do self-perpetuating B lymphocytes drive human autoimmune disease? Immunology 1999;97:188-196.

Abrahams VM, Cambridge G, Lydyard PM, Edwards JCW. Induction of TNFa by adhered hum an monocytes: A key role for FcyRIIIa in rheumatoid arthritis. Accepted for publication in Arthritis Rheum.

263 Immurwi gy 1999 97 188 196 REVIEW ARTICLE

Do self-perpetuating B lymphocytes drive human autoimmune disease?

J. C. W. EDWARDS, G. CAMBRIDGE & V. M. ABRAHAMS Centre for Rheumatology, Department of Medicine, University College London, UK

SUMMARY

Normal immunological memory is thought to be underpinned by T lymphocytes. However, in rheumatoid arthritis there are indications that T-lymphocyte control has been subverted by self- perpetuating B lymphocytes. Potential mechanisms in other autoimmune states are less clear, but a number of observations suggest that misappropriation of immunological memory by B lymphocytes may be a common feature of human autoantibody-associated disease. Put simply, autoantibodies drive their own production. If so, the availability of safe B-lymphocyte-depleting agents provides a potential means for reversal of autoimmunity.

HOW DO AUTOREACTIVE B LYMPHOCYTES cartilage structures, including growth plate. Rheumatoid SURVIVE? arthritis targets synovium, and also structures such as pericar­ dium and alveoli.6 Cartilage is only affected if adjacent to Whatever its significance, the presence of autoantibodies synovium. remains the one piece of hard evidence about human autoim­ The second problem is that most T-lymphocyte trans­ munity which has to be built into a pathogenic mechanism. ferrable animal models require immunization with a massive * Current dogma states that a B lymphocyte can only avoid antigen and/or adjuvant load. In human autoantibody- early death if it obtains two positive survival signals; cytokines associated diseases, unlike rheumatic fever or post-dysenteric from helper T lymphocytes and, in the follicle centre, antigen, arthritis,7 there is rarely any evidence for recent exposure to a complexed to the complement fragment C3d.1,2 In theory, particular foreign antigen. Rheumatoid arthritis occurs at autoreactive B lymphocytes cannot survive because of T- random during adult life, with little or no geographic or lymphocyte anergy to self and because any positive survival temporal clustering. In the absence of evidence of recent signal obtained from complexed self antigen is outweighed by antigen exposure, it is difficult to see why a failure of a negative signal from uncomplexed antigen. Yet B lympho­ T-lymphocyte tolerance should occur in mid life. What seems cytes recognizing one or more of about 50 self proteins do more likely is that the random onset of disease reflects a survive in some individuals, and may be associated with random process in the immune system; we suggest that disease.3 immunoglobulin gene mutation is the obvious candidate. The prevailing view is that autoreactive B-cell survival is secondary to a failure in the maintenance of T-lymphocyte self-tolerance, induced by an external ‘trigger’ immunogen. AUTOANTIBODIES MAY DRIVE THEIR OWN This view derives from early ideas about cross-reactivity PRODUCTION between bacterial and tissue-specific antigens in rheumatic As noted by Pulendran et al.,8 autoantibodies may arise by fever.4 However, this may be a very misleading model for random mutation during immune responses to any antigen. spontaneous autoimmunity. The autoantigens subsequently recognized need not resemble Enthusiasm for loss of T-lymphocyte tolerance in autoim­ the original antigen. A single amino acid substitution may munity was bolstered by animal models transferrable by T cause dramatic changes in the conformation of the antigen- lymphocytes, such as collagen II arthritis.5 However, there are binding site. In most cases dramatic changes in antibody two problems with these models. The pattern of disease reflects specificity will lead to loss of affinity for foreign antigen and the antigen used and, in virtually every case, fits poorly with death of the B lymphocyte. However, if the new antigen- the putative human equivalent. Collagen II arthritis affects binding site interacts with a self antigen in such a way as to generate positive survival signals, the B lymphocyte may Received 13 O ctober 1998; revised 13 January 1999; accepted survive and proliferate (Fig. 1). 13 January 1999. Many of the autoantigens associated with rheumatic disease Professor J. C. W. Edwards, University College London Centre are proteins implicated in survival, differentiation and function for Rheumatology, Arthur Stanley House, 40-50 Tottenham Street, of immune cells; notably immunoglobulin G (IgG) Fc, Clq, London W1P 9PG, UK. and nucleoproteins such as topoisomerase-1 and the leucocyte

188 © 1999 Blackwell Science Ltd

264 Autoimmunity and self-perpetuating B lymphocytes 189

B-ccIl clone producing immunoglobulin lo foreign antigen

Random immunoglobulin gene mutation

B-cell clone producing autoantibody -> z>

Autoantibody

Facilitation of B-cell survival in the context Facilitation of B-cell of T-cell help survival in follicle centre

Tissue pathology

Figure 1. A generalized mechanism for autoantibody-associated disease. differentation-related oncogene product dek.9-12 Antibodies to individuals tend to come from a common germ line. They these antigens may have a particular opportunity to upset show evidence of immunoglobulin gene mutation, but regulatory mechanisms. If an antibody, by modifying the mutations tend to be silent.16 In contrast, RF from RA subjects normal function of one of these proteins, stimulated or mim­ derive from a wide range of immunoglobulin germ line genes icked T-lymphocyte cytokine activity, whether directly or and show a high frequency of substitution mutations, indicat­ indirectly, and shifted the balance of unbound and complexed ing affinity maturation. antigen, then its parent clone could become self-perpetuating It is possible that physiological RF come from the B-l (Fig. 1). The best characterized example of such a mechanism subset of B lymphocytes,17 associated with ‘natural’ autoanti­ is in rheumatoid arthritis.13 Subsets of rheumatoid factors may bodies and that pathogenic RF arise from the B-2 subset. B-l keep their parent B-lymphocyte clones alive by disturbing the cells show constitutive expression of the nuclear protein STAT3 control of both T-lymphocyte help and survival signals given and distinct regulatory mechanisms in B-l cells may relate to by immune complexes in the follicle centre. If rheumatoid the blocks to affinity maturation and class switching of physio­ factors can, perhaps other autoantibodies can. logical RF. However, B-l-derived antibodies can show both these features and the means by which subtypes of RF may persist, discussed later, do not necessarily require either a B-l RHEUMATOID FACTORS AND THEIR RELATIONSHIP or B-2 origin. TO DISEASE Class switching, high affinity and fine specificity may all Approximately 80% of subjects with rheumatoid arthritis (R A) be important in RF pathogenicity. A critical test of this develop circulating antibodies to IgG Fc, or rheumatoid concept comes from hypergammaglobulinaemic purpura of factors (R F).9 Seronegative cases occur, but in many, RF are Waldenstrom ( HPW )18 in which high levels of polyclonal RF’ present in joints. Seronegative cases probably also comprise occur in the absence of a recognizable stimulus, as in RA, but conditions which can be clinically indistinguishable from RA, without arthritis. This raises the issue of RF as inflammatory such as spondarthropathies. RF are present in the circulation m ediators. of a few normal individuals and can also be induced by immunization to foreign antigen or by chronic infection.14 RF AND INFLAMMATION This dictates that if RF are pathogenic in RA, pathogenic potential must be restricted to a subset of RF. The pathogenic significance of RF has been doubted because There are recognized differences between RF from normal of the difficulty in finding a RF-based effector mechanism and RA subjects.9 ‘Physiological’ RF from normal subjects which explains the pathology of RA. However, new infor­ are chiefly of IgM class and low affinity. RF in RA are of all mation on the immunological microenvironment in synovium classes, and are structurally and genetically distinct, as shown has led to the identification of a plausible effector pathway. by analysis of immunoglobulin VH regions.1’’ RF from normal IgG RF exist in the circulation of rheumatoid subjects as

© 1999 Blackwell Science Ltd. Immunology, 97, 188 196

265 190 J. C. W. Edwards et al. oligomers and predominantly dimers.19 These dimers fix com­ restricted to IgG3.29 This would be consistent with an inability plement poorly and, consequently, can escape clearance by of IgG3 RF dimers to induce signalling via FcyRIIIa. red cell complement receptors (CR1). Unlike IgM-based com­ The above discussion only relates to the initiation of plexes, IgG dimers are small enough to pass out of the inflammation in synovium by circulating complexes. The domi­ circulation and access tissue macrophages. Histological studies nant involvement of synovium in RA is likely to involve a indicate that macrophage activation is the initial event in RA series of secondary events. Synovial fibroblasts are unusually synovium, preceding T-lymphocyte accumulation.20 responsive to cytokines such as TNF-a in terms of the induc­ Recent studies indicate that high-level expression of the tion of expression of molecules involved in B-lymphocyte IgG Fc receptor FcyRIIIa is restricted to macrophages in survival: vascular cell adhesion molecule-1 (VCAM-1), decay tissues affected by RA.21 Of the three classes of FcyR, FcyRIII accelerating factor and complement receptor 2.30 Induction of is believed to be particularly important for the binding of the expression of these moleculess on synovial subintimal small, and specifically dimeric, complexes and the consequent fibroblasts by cytokines released from FcyRIIIa + macrophages generation of mediators such as tumour necrosis factor-a would explain the survival of B lymphocytes in RA synovium, (TNF-a).22 FcyRI expression is up-regulated by interferon-y23 with follicle formation in some cases. A proportion of these B and may be chiefly involved in a mature inflammatory lymphocytes are known to generate RF.9 High concentrations response. Its high affinity allows it to bind free, locally of IgG RFs within the confines of the joint are associated with synthesized antibody to a specific pathogen and interact with the formation of larger, complement-fixing complexes, which, the pathogen subsequently, via multiple receptors. FcyRII is together with IgM RF-based complexes, are likely to amplify of low affinity and is implicated in binding large complexes, the inflammatory process.19 again via multiple receptors.23 A specific role for FcyRIIIa in response to small complexes RF AND T-LYMPHOCYTE HELP is supported by the fact that a monoclonal antibody to FcyRIIIa in free soluble form is capable of inducing release In order to obtain T-lymphocyte help, B lymphocytes normally of both TNF-a and reactive oxygen species, whereas mono­ take up antigen bound to surface immunoglobulin and present it to a T lymphocyte. It is recognized that B lymphocytes that clonal antibodies to FcyRI and FcyRII do not induce TNF-a carry surface antibody to IgG Fc, i.e. RF, can obtain help release under the same conditions.24 TNF-a release has only without requiring T lymphocytes responsive to IgG.15 By been achieved via FcyRI and FcyRII by secondary cross- taking up IgG attached to a non-self antigen they can present linkage following ligation of the receptors by IgG Fc.25 This that antigen to a responsive T lymphocyte (Fig. 4).31 suggests that FcyRI and FcyRII may require multiple cross- RF-producing B cells survive for short periods in everyone, if linking in order to induce signalling but that FcyRIIIa may foreign antigen is available, as after immunization (Fig. 5a). induce signalling in response to a soluble ligand capable of However, the RF produced is mostly IgM. Moreover, there is engaging two, or at most three (two Fab- and one Fc-based little evidence of immunoglobulin gene mutations leading to interactions), receptors. This would be closely analogous to amino acid substitution.16 There seems to be a block to class the situation for an IgG RF dimer with which additional switching and affinity maturation for RF in normal individuals, non-RF IgG molecules may be in dynamic association. suggesting that the acquisition of T-cell help is not enough to On this basis it can be argued that RA is precisely the allow RF-secreting B lymphocytes to survive long-term. This inflammatory state that IgG RF dimers should be expected to suggests that there is a protective mechanism operating in the generate. There are, however, further issues about IgG sub­ follicle centre. class. Voice and Lachmann26 recently highlighted the impor­ tance of subclass and complex size in interactions between complexed IgG and Fc and complement receptors. IgGl and SURVIVAL IN THE FOLLICLE CENTRE IgG3 RF dimers may interact differently with FcyRIIIa. Small The survival of B lymphocytes in follicle centres depends on IgG3-based complexes induce FcyRIIIa-dependent TNF-a the balance between positive and negative signals provided by production only slightly less well than equivalent IgGl-based antigen in different forms (Fig. 6). B-lymphocyte clones recog­ complexes in our hands (Abrahams, in preparation). However, nizing soluble autoantigens should not survive because they self-association may raise specific steric considerations. One should receive a negative survival signal from uncomplexed possibility relates to the long hinge of IgG3 (Fig. 2). Modelling antigen. However, antigen available in the form of an immune studies predict that the Fc receptor binding sites on an IgGl complex can give a positive survival signal. This signal is dimer would be approximately 75 A apart, whereas for IgG3 modulated by the attachment of C3d molecules to the antigen dimers the distance would be approximately 200 A with the following complement fixation.2 The combined interactions hinge extended.27'28 There is debate about whether the cysteine- between C3d and its receptor, CR2, and antigen with surface rich segment of the IgG3 hinge is normally extended or immunoglobulin lead to a positive signal which is amplified compressed. However, steric interference in a self-associated tenfold for each C3d molecule attached to the complex. The dimer may reduce the likelihood of a compressed configur­ situation is further modulated by FcyRIIb which binds large ation. A distance of 75 A would allow direct apposition of the immune complexes and provides a negative signal.32 FcyRIIIa y-chains responsible for signalling, but 200A would Although IgG RF exist in the circulation in rheumatoid not (Fig. 3). This provides one possible explanation for the subjects largely as non-complement-fixing dimers, this prob­ absence of synovitis in subjects with HPW. ‘Intermediate’ ably reflects both their relatively low concentration in serum complexes, consisting of self-associated RF, exist in HPW and the preferential survival of non-complement-fixing com­ sera, but in patients with no arthritis they were found to be plexes. At sites of IgG RF synthesis, as in rheumatoid syn-

© 1999 Blackwell Science Ltd, Immunology, 97, 188-196

266 Autoimmunity and self-perpetuating B lymphocytes 191

FcR binding site

IgGl RF dimer

lgG3 RF dimer

• 6

Figure 2. Comparison of IgGl and IgG3 RF dimers showing the dilference in hinge length.

c75 A IgG l RF

IM -fl

la cv R llla lacyR Ilia

Figure 3. A suggested basis for how the long hinge of lgG3 may inhibit IgG3 RF dimers from bringing the signalling y-chains of FcyRIIIa into apposition. ovium, higher concentrations of IgG RFlead to the formation tigen could keep itself alive by generating sufficient antibody of complement-fixing multimers,18 with the potential to to provide a supply of complexed antigen. However, it appears provide RF-specific B lymphocytes with a positive survival that the regulation of positive and negative signals is such that signal (Fig. 5b). the negative signal from soluble antigen normally dominates. In theory, any B-cell clone recognizing a soluble autoan- Two aspects of this regulation may fail in the case of

© 1999 Blackwell Science Ltd. Immunology, 97. 188 196

267 192 ./. C. W. E dw ards et a!.

RF-specific B cell

Figure 4. The basis o f‘bystander’ 1-cell help to RF-spccific B lymphocytes.

RI'-secreting clones. An isolated autoreactive B-lymphocyte IgM RF are also produced, to the extent that they have tended clone will normally have to compete for the same epitope on to obscure the significance of IgG RF (Fig. 5). an autoantigen with its own secreted immunoglobulin. Thus it may be unable to bind complexed antigen through surface B LYMPHOCYTES AM) MEMORY IN OTHER immunoglobulin except if other clones reactive with other AUTOIMMUNE DISEASES epitopes on the same autoantigen are present. This does apply to RF-secreting clones (Fig. 5b) because an IgG RF molecule The ways in which B-lymphocyte clones might become self- that masks an antigenic epitope on IgG provides the same perpetuating in other autoimmune states cannot be traced as epitope on itself. Secondly, surface RF may compete with fully as in rheumatoid arthritis. However, several observations FcyRI lb for large complexes, reducing the negative signal and proposed mechanisms which appear to fit such a frame­ from this receptor.32 It would appear that, when it comes to work will be briefly reviewed. preventing survival of IgG RF-secreting B lymphocytes, there Davies and colleagues have suggested that the formation are weaknesses in regulatory mechanisms in the follicle centre of antibodies to Clq is a crucial early step in the common as well as in the context of T-lymphocyte help. sporadic form of systemic lupus.10 Anti-Clq antibodies can Multimeric IgGl- and IgG3-based complexes are both trigger complement consumption. In rare genetically deter­ capable of fixing complement and RF of both subclasses are mined cases complement components are primarily deficient. likely to enhance RF-specific B-lymphocyte survival. This may In either case, limited availability of complement components contrast with the relative capacity of their dimers to cross-link may both impair clearance of immune complexes and nuclear FcyRIIIa and initiate synovitis and would explain why in material normally cleared via complement and, by limiting HPW RF production is perpetual (multimer-driven), but generation of C3d, reduce the efficiency of B-lymphocyte arthritis (dimer-initiated) does not occur (Fig. 5c). clonal selection, allowing the genesis of autoantibodies. IgG3 also differs from other IgG subclasses in lacking the Anti-Clq antibodies have the potential to interfere with Ga epitope on Fc, a common recognition site for RF.33 Even the rules of antibody-antigen interactions in a way similar to within subclasses, pathogenic potential may depend on the RF. An anti-Clq-specific B lymphocyte can endocytose any detailed stereochemistry of the CDR antigen interaction. antigen complexed with C lq and present it to T cells. Clq-anti- These considerations may help explain the variable anatomical Clq complexes could disturb regulation in the follicle centre and temporal patterns seen in different cases of the disease. in a similar way to RF polymers. The kinetics of self­ The prediction is that IgG RF capable of supporting perpetuation of anti-Clq-specific B lymphocytes would differ survival of their parent B-lymphocyte clone would also assist from RF-specific B lymphocytes because the clone would not survival of RF-specific clones secreting other isotypes. This generate both antigen and antibody locally, e.g. in a joint. would explain why in both RA and HPW large amounts of Clq anti-Clq complexes would probably also be loo big to

(

268 Autoimmunity and self-perpetuating B lymphocytes 193

(a)

lg genejn rearrangement

Bystander T cell help m je st t IgM RF of homogeneous via surface RF binding germ line gene origin: complexed foreign MSB ■ often transient a n tig e n s <3 1 w

block to affinity maturation and class switching: limited persistence

( b ) ASIRk nH s 1 ■ W Further random J y lg gene m utation

Bystander T cell help via surface RF binding complexed foreign secreted antigens ■ IgGl RF

large IgGl RF small immune complex affinity maturation complexes at mediated inflammation and persistence of * sites of local any RF-specific B synthesis cell supported by positive signal i §5Wy from complexes RF of all subclasses of if extreme: hyperglobulinaemic heterogeneous origins ■=> purpura/ vasculopathy

KigureS. Physiological RF in normal individuals (a); RF in rheumatoid arthritis (b); and RF' in HPW (c). cross vessel walls, suggesting that the synovitis of lupus may persist. In what way does their surface or secreted immuno­ be due to immune complexes present as a result of mechanisms globulin make them able to survive within the thymic environ­ downstream of complement depletion. ment? AChR are present on myoid cells which are found in In myasthenia gravis B lymphocytes generating antibodies close contact with professional antigen-presenting cells in to nicotinic acetyl choline receptors (AChR) survive in the myasthenic, but not normal, thymus.34 Could anti-AChR thymus.34 The interesting question may be not why B lympho­ modulate myoid cell function in such a way that the thymic cytes of this specificity arise, but why, once generated, they environment becomes favourable to B-lymphocyte survival?

© 1999 Blackwell Science Ltd, Immunology, 97. 188 196

269 194 J. C. W. E dw ards et al.

(c) B cell committed to production of IgGl to foreign antigen Further random in I L lg gene mutation T cell help via surface RF binding complexed Hi secreted foreign antigens lgG3 RF 1 0 0 formation of no small immune affinity maturation large lgG3 RF complex mediated and persistence of complexes at inflammation any RF-specific B sites of local cell supported synthesis B by positive signal § from complexes (?class switching 0 to IgGl still V ineffective)

lgG3 RF plus IgM RF of if extreme: hyperglobulinaemic heterogeneous origins purpura/ vasculopathy

Figure 5c. R F in HPW

C'R2 C3d Autoantigen with epitope

| FcyRIIh AAb, autoantibody; RF, rheumatoid factor

Aiitoantigern specific kB cell i

Figure 6. Relative potency of survival signals lor autoantigen-specific and RF-specific B cells. Cross-hatching of elements indicates reduced availability because of competition by other ligands.

1999 Blackwell Science Ltd, Immunology. 97, 188 196

270 Autoimmunity and self-perpetuating B lymphocytes 195

There have to be answers to these questions; answers likely to lymphocytes with an anti-CD20 antibody. If enough self- make additional reasons for loss of T-lymphocyte tolerance to perpetuating autoreactive cells could be destroyed the impli­ AChR redundant. cation is that autoimmunity would collapse and lymphoid In multiple sclerosis T-cell responses to myelin antigens tissue be repopulated by B lymphocytes innocent of any are often invoked. However, in contrast to experimental autoimmune tendency. Anti-CD20 therapy is licenced for allergic encephalomyelitis, the pathological domain is not that lymphoma and, apart from transient symptoms due to cytoly- of myelin, but of the blood-brain barrier.35 The only estab­ sis, is well tolerated.41 Up to 95% of B lymphocytes can be lished immunological abnormality is the persistent survival of cleared with repopulation in 100 days. This is in stark contrast a few B-lymphocyte clones in the central nervous system to anti-T-lymphocyte therapy, for which the justification is (CNS). The key question is how these clones live within the unclear and which, even if successful, carries a risk of long­ CNS, generating the same oligoclonal immunoglobulin bands term immunodeficiency. Anti-T-lymphocyte strategies have over many years. CNS damage may simply reflect intolerance been disappointing. Interestingly, pan-lymphocyte depletion of high levels of immunoglobulin, which induce macrophage in RA with anti-CD52 produced benefit limited approximately phagocytosis of myelinin vitro. Recent immunohistochemical to the period of B lymphopenia42 suggesting that autoreactive studies indicate that initial activation of microglia occurs in B-lymphocyte depletion did not reach the putative threshold. the absence of T lymphocytes.36 B-lymphocyte survival in the Anecdotal reports suggest that RA is about as amenable to CNS should be precluded by the absence of stromal cell long-term remission as lymphoma. Significant long-term ligands such as VCAM-1. However, if a B lymphocyte were remission rates in lymphoma probably require maximal doses to generate surface or secreted immunoglobulin which over­ of depleting antibody (2-3 g is recommended for anti-CD20) came the need for such ligands it might survive in the CNS, combined with conventional cytotoxic agents.43 Specific anti- without being exposed to the competitive pressures of a lymphoid organ. B-cell therapy of this type has not been explored. It might just Two other aspects of autoimmunity of recent interest may produce long-term cure. be relevant to the potential for autoreactive B lymphocytes to self-perpetuate. It is now accepted that a proportion of autoan­ tibodies can penetrate cells, alter cell function and even cause REFERENCES cell death.37 During cell death, availability of autoantigen is 1. L i n d h o u t E., K o o p m a n G ., P a l s S.T. & d e G r o o t C. (1997) altered by changes in cell compartmentalization.38 This raises Triple check for antigen specificity of B cells during germinal the possibility that in conditions such as systemic sclerosis B centre reactions. Immunol Today 18, 573. lymphocytes generating antibodies to antigens such as 2. F e a r o n D.T. & C a r t e r R.H. (1995) The CD19/CR2/TAPA-1 topoisomerase-1 perpetuate their existence by causing cell complex of B lymphocytes: linking natural to acquired immunity. death during division, with release of autoantigen in a form Annu Rev Immunol 13, 127. suitable both for induction of a T-cell response and for 3. R o i t t I.M. (1991) Principles of autoimmunity. In: Clinical providing a positive signal to follicular B cells. Mechanisms Immunology, (eds. J. Brostoff, G. K. Scadding, M. Male & I. M. of this sort may be relevant to the wide variety of syndromes R oitt) C hapter 4, 4 1 10. Gow er M edical Publishing, London. associated with antibodies to nucleic acid-binding proteins.39 4. G l y n n L.E. (1975) Rheumatic Fever. In: Clinical Aspects of In other autoantibody-associated disorders mechanisms Immunology, (cds G.H. Gell, R. R. A. Coombs & P. J. L a c h m a n n ) involving autoreactive B-lymphocyte self-perpetuation become Chapter 38, p. 1079. Blackwell Scientific Publications, Oxford. more difficult to identify. They are likely to be diverse, since 5. C h i o c c h i a G ., Boissier M .C., M anoury B. & Fournier C. (1993) T cell regulation of collagen-induced arthritis in mice. the most plausible mechanism is different in each condition so Immunomodulation of arthritis by cytotoxic T cell hybridomas far analysed. In sarcoidosis, for instance, autoantibodies may specific for type II collagen. Eur J Immunol 23, 327. mimic lectins.40 However, there are almost certainly some 6. M atteson E.L., C o h e n M .D. & C o n n D .L. ( 1 9 9 8 ) Rheumatoid conditions, probably including type I diabetes, in which arthritis; clinical features and systemic involvement. In: autoantibodies are secondary to a truly T-lymphocyte-driven Rheumatology; (ed. J.H. K l i p p e l & P.A. D i e p p e ) , pp. 5.4.1.4.8. mechanism. Mosby, London. 7. T o i v a n e n A. (1998) Reactive arthritis and Reiter’s syndrome; A DIGITAL ANALOGY history and clinical features. In: Rheumatology; (Eds J.H. K l i p p e l & P.A. D i e p p e ) , pp. 6.11.1.11.8. Mosby, London. The case has been made for autoimmune disease being driven, 8. Pulf.ndran B., Van Drif.l R. & N o s s a l G.J. (1997) not by a primary failure of T-lymphocyte tolerance but by Immunological tolerance in germinal centres. Immunol Today antibody. As for the genome or computer software, the true 18, 27. danger to the immune system may be a chance mutation in 9. C a r s o n D.A. (1993) Rheumatoid factor In: Textbook of an information string which converts ‘data’ to ‘command’. As Rheumatology, (eds. W. N. Kelley, E. D. H arris, S. R uddy & a transcribed sequence of DNA may become a lethal stop C . B. Sledge), pp. 155 W. B. Saunders, Philadelphia. 1 0 . D a v i e s K.A. ( 1 9 9 6 ) Complement, immune complexes and systemic codon and a misread floppy disk may lead to a system crash, lupus erythematosus. Br J Rheumatol 35, 5. if an antibody becomes a B-lymphocyte growth promoter it 11. G u l d n e r H .H ., Szostecki C., Vosberg H.P., L a k o m e k H .J., may be very difficult for the immune system to re-establish P e n n e r E. & B a u t z F.A. (1986) Scl 70 autoantibodies from control. scleroderma patients recognize a 95 kDa protein identified as DNA topoisom erase I. Chromosome 94, 132. THERAPEUTIC IMPLICATIONS 12. M u r r a y K .J., S z e r W ., G r o m A.A. el al. (1997) Antibodies to the 45 kDa DEK nuclear antigen in pauciarticular onset juvenile If self-perpetuating B lymphocytes exist, their therapeutic rheumatoid arthritis and iridocyclitis: selective association with implications are enormous. It is now possible to ablate B M HC gene. J Rheumatol 24, 560.

© 1999 Blackwell Science Ltd, Immunology, 97, 188 196

271 196 J. C. W. Edwards et al.

1 3. E d w a r d s J.C.W . ( 1 9 9 9 ) Is rheumatoid factor relevant. In: domain of chimeric human IgGl anti-HLA-DR is necessary for Rheumatoid Arthritis; Questions & Uncertainties (eds. H. Bird & Clq, FcgRI and FcgRIII binding. Immunology 86, 319. M. L. S n a i t h ) Chapter I. Blackwell Scientific, Oxford. 28. E d w a r d s J.C.W . & S u t t o n B.J. (1998) What does IgG look like? 14. W e l c h M .J., F o n g S., V a u g h a n J. & C a r s o n D. (1983) Increased Br J Rheumatol 37 (Abstracts Suppl) 1, 7. frequency of rheumatoid factor precursor B lymphocytes after 29. Serino G., Pitingolo F., G ranatieri C. et al. (1983) immunization of normal adults with tetanus toxoid. Clin Exp Hypergammaglobulinemic purpura of Waldenstrom: character­ Immunol 51 , 299. isation of circulating immune complexes. Acta Haemal 69, 15. T h o m p s o n K.M ., R a n d e n I., B r r e t z k n M., Forre O. & N atvig 152. J.B. (1994) Variable region gene useage of human monoclonal 30. E d w a r d s J.C.W ., L e ig h R .D . & C a m b r i d g e G. (1997) Expression rheumatoid factors derived from healthy donors following immun­ of molecules involved in B lymphocyte survival and differentiation ization. Eur J Immunol 24, 165. by synovial fibroblasts. Clin Exp Immunol 108, 407. 16. B o r r e t z e n M ., Randen I., Zdarsky E., F orre O., N atvig J.B. 31. R o o s n e k E. & Lanzavecchia A. (1991) Efficient and selective & T h o m p s o n K.M. (1994) Control of autoanlibody affinity by presentation of antigen-antibody complexes by rheumatoid factor selection against amino acid replacements in the complementarity- B cells. J Exp M ed 173, 487. determining regions. Proc Natl Acad Sci USA 91, 129171. 32. S a r m a y G ., K o n e z G., P e c i n t I. & G e r g f . r l y J. (1997) Fc gamma 17. K a r r a s J.G ., W a n g Z., H u o L., H o w a r d R.G ., F r a n k D.A. & receptor type I lb induced recruitment of inositol and protein R o t h s t e i n T.L. (1997) Signal transducer and activator of phosphatases to the signal transductory complex of human B transcription-3 (STAT3) is constitutively activated in normal, self- cells. Immunol Lett 57, 159. renewing B-l cells but only inducibly expressed in conventional B 33. J e e f e r is R. (1995) Rheumatoid factors, B cells and immunoglob­ lymphocytes. J Exp M ed 185, 1035. ulin genes. Br Med Bull 51, 312. 18. Capra J.D., W inchester R.J. & Kunkel H.G. (1971) 34. K i r c h n e r T., H o p p e F., S c h a l k e B. & M uller-H erm elink H.K. Hypergamma-globulinemic purpura. Studies on the unusual (1988) Microenvironment of thymic myoid cells in myasthenia anti-y-globulins characteristic of the sera of these patients. gravis. Virchows Arch B Cell Pathol Incl Mol Pathol 54, 295. Medicine 50 , 125. 35. Tourtellotte W .W ., W a l s h M .J., Baumhefner R.W ., 19. M a n n i k M. & N a r d e l l a F.A. (1985) IgG rheumatoid factors Staugaitis S.M. & S h a p s h a k P. (1984) The current status of and self association of these antibodies. Clin Rheumatic Dis multiple sclerosis intra-blood-brain-barrier IgG synthesis. Ann N 11, 551. Y Acad Sci 436, 52. 20. V ersendaal H., J o n k e r M ., Bresnihan B. et al. (1999) 36. G a y F.W ., D r y e T.J., D i c k G.W . & E s i r i M.M. (1997) The Asymptomatic synovitis precedes clinically manifest arthritis. application of multifactorial cluster analysis in the staging of Arthritis Rheum in press. plaques in early multiple sclerosis. Identification and characteriz­ 21. B h a t i a A., B l a d e s S., C a m b r i d g e G. & E d w a r d s J.C.W. (1998) ation of the primary demyelinating lesion. Brain 120, 1461. Differential distribution of FcyRIIIa in normal human tissues and 37. Alarcon-Segovia D ., Ruiz-A rguelles A. & L l o r e n t e L. (1996) co-localization with DAF and fibrillin-1: implications for immuno­ Broken dogma: penetration of autoantibodies into living cells. logical microenviroments. Immunology 94, 65. Immunol Today 17, 163. 22. K l a s s e n R .J.L., Goldschem eding R., T e t t k r o o P.A.T. & V o n 38. V a is h n a w A .K ., M c N a l l y J.D . & E l k o n K.B. (1997) Apoptosis D e m B o r n e A.E.G.K. (1988) The Fc Valency of an immune in the rheumatic diseases. Arthritis Rheum 40, 1917. complex is the decisive factor for binding to low-affinity Fey 39. C r a f t J. & H a r d i n J.A. (1993) Anlinuclear antibodies. In: receptors. Eur J Immunol 18, 1373. Textbook of Rheumatology (eds. W. N. Kelley, E. D. Harris, 23. van De W inkel J.G .J. & C a p e l P.J.A. (1993) Human IgG Fc S. R uddy & C. B. Sledge), pp. 164. W . B. S a u n d e r s , receptor hetcrogenity: molecular aspects and clinical implications. Philadelphia. Immunol Today 14, 215. 40. P i l a t t e Y., T i s s e r a n d E.M ., G r e f f a r d A., B ig n o n J. & L a m b r e 24. A b r a h a m s V.M., C a m b r i d g e G. & E d w a r d s J.C.W. (1998) C.R. (1990) Anticarbohydrate autoantibodies to sialidase-trealed FcyRIIIa mediates TNFa secretion by human monocytes/macro­ erythrocytes and thymocytes in serum from patients with pulmon­ phages. Br J Rheumatol 37 (Abstracts Suppl) 1, 90. ary sarcoidosis. A m J M ed 88, 486. 25. D e b e t s J.M .H ., Van De W inkel J.G .J., C e u p p e n s J.L., D i e t e r e n 41. M a l o n e y D .G ., G rillo-Lopez A .J., B o d k in D .J. et al. (1997) I.E.M . & B u r m a n W.A. (1990) Cross-linking of both FcyRI and IDEC-C2B8: results of a phase I multiple-dose trial in patients Fcyll induces secretion of tumor necrosis factor by human mono­ with relapsed non-Hodgkin’s lymphoma. J Clin Oncol 15, 3266. cytes, requiring high affinity Fc FcyR interactions. J Immunol 42. I s a a c s J.D ., W a t t s R.A ., H a z l e m a n B.L. et al. (1992) Humanised 144, 1304. monoclonal antibody therapy for rheumatoid arthritis. Lancet 26. V o i c e J. & L a c h m a n n PJ. (1997) Neutrophil Fc gamma and 340, 748. complement receptors involved in binding soluble IgG immune 43. M cLaughlin P., Grillo-Lopez A .J., L i n k B.K. et al. (1998) complexes and in specific granule release induced by soluble IgG Rituximab chimeric anti-CD20 monoclonal antibody therapy for immune complexes. Eur J Immunol 27, 2514. relapsed indolent lymphoma: half of patients respond to a four- 27. M o r g a n A., J o n e s N .D ., N e s b i t t A .M ., C h a p l i n L., B o d m e r dose treatment program. J Clin Oncol 16, 2825. Additional data M.W. & E m t a g e J.S. (1995) The N terminal end of the CH2 on file, R oche Products, Welwyn, U K.

© 1999 Blackwell Science Ltd, Immunology, 97, 188-196 INDUCTION OF TNF-a PRODUCTION BY

ADHERED HUMAN MONOCYTES :

A KEY ROLE FOR FcyRIIIa IN RHEUMATOID

ARTHRITIS

VIKKIM. ABRAHAMS, GERALDINE CAMBRIDGE, PETER M. LYDYARD AND JONATHAN C.W. EDWARDS

Supported by Celltech Therapeutics Ltd, Berkshire, UK.

Vikki M. Abrahams, BSc, Geraldine Cambridge, BSc PhD, Jonathan C.W. Edwards, MD, Centre for Rheumatology, University College London, London, UK; Peter M. Lydyard, BSc PhD Department of Immunology, University College London, London, UK

Address for correspondence: Professor JCW Edwards, UCL Centre for Rheumatology, 4th Floor, Arthur Stanley House 40-50 Tottenham Street, London W1P 9PG, England

Telephone: 44 (0)171 380 9215 Fax: 44 (0)171 380 9278 email: [email protected]

273 Relevance to rheumatologists

Monoclonal antibodies to immunoglobulin receptors (FcyR) on human macrophages were used to probe the mechanisms by which IgG rheumatoid factor-based immune complexes might induce cytokine production. Evidence was obtained for specific involvement of the receptor FcyRIIIa in the production

of the cytokines tumour necrosis factor alpha and interleukin 1 alpha, supporting previous evidence for a key role for this receptor in the pathogenesis of rheumatoid arthritis.

274 control. After washing twice with buffer the cells were incubated with fluorescein isothiocyanate-(FITC) conjugated goat F(ab ) 2 anti-mouse Ig (DAKO, Cambridge, UK) at 4°C for 45 minutes. Following two washes with buffer the cells were fixed with 2% paraformaldehyde in PBS and analysed by flow cytometry using a FACScan (Becton Dickenson) with Win MDI software (Microsoft). Culture and stimulation of monocytes. Freshly isolated monocytes (5 x

105 /ml) were seeded into 96 well plastic tissue culture plates (Becton Dickenson) and incubated at 37°C for 24 hours. The culture supernatants were removed and replaced with fresh growth medium to ensure that cytokine levels were at baseline. The adherent human monocytes were then incubated with the anti-

FcyR mAb. Medium alone and isotype-matched controls were used as negative controls. Lipopolysaccharide (LPS) from Escherichia coli 026:B6 (Sigma; in the absence of polymixin B), and phorbol myristate acetate (PMA; Sigma) were both used as positive controls. Possible LPS contamination of reagents was excluded by performing all incubations in the presence of polymixin B (10 ng/ ml; Gibco BRL), which consistently abrogated the LPS-induced TNFa response. Culture supernatants were then collected at various time points and the cell-free samples stored at -70°C. TNFa ELISA. Cell-free culture supernatants were loaded onto 96 w eir maxisorb plates (Gibco BRL) previously coated with murine IgGi anti-human

T N F a mAb and blocked with 0.02M Bes (pH 7; Sigm a)/1% BSA. Hum an recombinant T N F a (National Institute for Biological Standards Control,

Hertfordshire, UK) standards were used to calibrate the assay. The plate was incubated for 1 hour at room temperature with continuous agitation. Following aspiration of the wells, the captured T N F a was detected by incubating with a sheep anti-human T N F a polyclonal antibody. Following aspiration the captured TNFa was then detected with horseradish peroxidase (HRP) labelled anti-sheep IgG (H & L). The plate was incubated for 30 minutes at room temperature with

275 F(ab )2 and Fab fragment binding to FcyRIIIa was tested by flow cytometry and saturating binding concentrations were 2 0 ng/ ml and 10 ng/ml respectively.

Isolation of monocytes from whole blood. Citrated venous blood was collected and buffy coat prepared following sedimentation with 6 % Dextran. This was layered onto Histopaque (density 1.077 g/ml; Sigma) and centrifuged at 400g for 30 minutes. The mononuclear cell-containing layer was removed, washed six times with RPMI-1640 medium (Gibco BRL, Paisley, UK) and the mononuclear cells resuspended in RPMI-1640 medium supplemented with 10% heat inactivated fetal calf serum (Sigma). The cell suspension was transferred into plastic petri dishes (Bibby Sterilin Ltd, Staffordshire, UK) and incubated at

37°C/5% CO 2 for 2 hours. Nonadherent cells were removed by washing six times with Hank's balanced salt solution (Gibco BRL). Adhered monocytes were gently harvested using a rubber policeman and resuspended in growth medium (RPMI- 1640 supplemented with 25 mM Hepes, 20 ng/m l gentamycin, 2 mM L- glutamine, 100 U/ml penicillin, 100 ng/ml streptomycin, 2% non-essential amino acids, 1 mM sodium pyruvate (all from Gibco BRL) and 10% fetal calf serum) and the cell concentration adjusted to 5 x 10 5 /m l.

Monocyte purity was determined by flow cytometric analysis of CD14 and

CD3 expression and shown to be greater than 95%. Trypan blue (Sigma) exclusion- showed monocyte viability to be greater than 98%.

Expression of FcyR by cultured adhered human monocytes. Freshly isolated monocytes (1 x 10 6 /ml) were seeded into 24 well plastic tissue culture plates (Becton Dickenson, Oxford, UK) and incubated at 37°C for 0, 24 or 48 hours. The cultured monocytes were then harvested by cooling on ice and gentle pipetting. FcyR expression was measured by indirect immunofluorescence flow cytometry. Briefly, the harvested cell suspension was washed twice with ice cold buffer (PBS, 1% bovine serum albumin (BSA), 0.1% sodium azide) and then incubated at 4°C for 45 minutes with an anti-FcyR mAb or an isotype-matched

276 MATERIALS AND METHODS

Antibodies. The anti-FcyRIII murine mAb-producing cell line, 3G8 (IgGi) and purified murine anti-FcyRII mAb, IV.3 (IgG 2b) were generous gifts from Dr M. Fanger (Dartmouth, Hanover, NH). The 3G8-producing cell line was used to generate purified anti-FcyRIII mAb from which F(ab )2 and Fab fragments were prepared. For all experiments, purified anti-FcyRIII mAb (3G8), purchased from Immunotech (Marseille, France) was used. The anti-FcyRI mAb, 10.1 (IgGi) was kindly donated by Dr N. Hogg (ICRF, London, UK). Saturating binding concentrations of the anti-FcyRI, anti-FcyRII mAb (both 10 jig/ml) and anti- FcyRIII mAb (20 jig/ml) were pre-determined by flow cytometry. Murine IgGi and IgG 2b were purchased from Southern Biotechnology Associates, Inc (Birmingham, Al). These antibodies were dialysed before use to remove azide. Horseradish peroxidase conjugated donkey anti-sheep Ig (H & L) was purchased from Jackson Immunoresearch Laboratories, Inc (West Grove, PA). A murine anti-human TNFa mAb (CB006, IgGi) and a sheep anti-human TNFa polyclonal antibody were kindly provided by Celltech Therapeutics (Berkshire, UK).

Preparation of 3G8 F(ab ) 2 and Fab fragments. Anti-Fey RIII mAb (3G8) was purified from culture supernatant by affinity chromatography. Briefly, the supernatant was adjusted to pH 8 and loaded onto a 2 ml Protein G-Sepharose 4B_ column (Sigma, Poole, UK) previously equilibrated with phosphate buffered saline (PBS, pH 8) at 4°C. Following washing with PBS, the mAb was eluted with 3M magnesium chloride. The purified mAb was then dialysed into PBS and concentrated.

F(ab )2 fragments of the purified anti-FcyRIII mAb were prepared by pepsin cleavage and Fab fragments were prepared by papain digestion (Pierce, Rockford, IL). The digests were purified using protein G-Sepharose 4B. The F(ab)2/Fab fragments were collected as flow through, dialysed into PBS and concentrated.

Purity of the F(ab ) 2 and Fab digests was confirmed by SDS-page. Anti-FeyRIII

277 There are three classes of receptor for IgG (FcyR) (13), all of which can mediate a number of effector functions including phagocytosis, antibody dependent cytotoxicity and the release of inflammatory mediators (14,15). Monocytes and macrophages constitutively express Fcy RII (CD32), which is functionally associated with large immune complexes. FcyRI (CD64), also functionally associated with large complexes, is only expressed at low levels, being inducible by interferon-y (15). Macrophage FcYRIIIa (CD16) has a restricted tissue distribution (16), being expressed at high level only in synovial intimal and other tissues, such as pericardium, involved in rheumatoid arthritis. Furthermore,

Fcy RIII has been shown to preferentially bind small immune complexes (17). These considerations led to the hypothesis that FcYRIIIa has a distinct functional role in the generation of inflammation by small immune complexes, in the context of rheum atoid arthritis (12,18). It has been dem onstrated that ligation of F c y R I and F c y R I I can result in T N F a production by human monocytes

(19). However, T N F a production through F c y R I was only observed in response to murine IgG 2 a in solid phase, while F c y R I I only evoked a response to murine IgGi in solid phase following treatment with proteolytic enzymes. Human

FcYRIIIa has also been shown to mediate T N F a production but data have only been available for natural killer cells (20,21). In this study we have investigated the ability of monocyte-derived macrophages to produce T N F a in vitro following ligation of each of the three human F c y R , using murine monoclonal antibodies (mAb) to each receptor as surrogates for small immune complexes.

278 Macrophage-derived tumour necrosis factor alpha (T N F a ) is thought to be a dominant mediator of synovitis in rheumatoid arthritis (1). The stimulus for the production of T N F a by macrophages is not known. Recent debate has focussed on signals generated as part of putative T cell responses to articular or microbial antigens (2). However, synovial intimal macrophage activation precedes, and is subsequently spatially dissociated from, T cell infiltration (3,4). In addition, it is difficult to integrate mechanisms based on T cell responses with the high levels of circulating rheumatoid factors and frequent presence of extra- articular features in rheumatoid arthritis (5). For many years, rheumatoid factor-based immune complexes were considered the likely inflammatory stimulus in rheumatoid synovium (6,7). However, two factors made this view unpopular. The large complement-fixing circulating complexes thought to be most phlogistic, do not deposit in vessel walls in rheumatoid arthritis in the way seen in lupus nephritis. Nor are they likely to be able to cross endothelium from the circulation to access and activate synovial macrophages. Consequently, immune complexes have been viewed as having only a secondary role in rheumatoid arthritis, when generated locally within synovium already infiltrated with lymphoid cells. Recently, it has become clear that complement fixation within the_ circulation is anti-phlogistic, since it results in immune complex clearance rather than inflammation ( 8). The pathogenic potential of circulating complexes may actually be enhanced by a failure to fix complement. IgG rheumatoid factor have the ability to self-associate and rheumatoid arthritic sera contain, amongst others,

"intermediate" IgG rheumatoid factor based complexes (9,10) which, owing to their small size, fail to fix complement and can, therefore, escape clearance ( 1 1 ).

They are also small enough to cross endothelium, with the resulting potential to bind to immunoglobulin receptors on tissue macrophages ( 1 2 ).

279 Objective . Small IgG rheumatoid factor immune complexes may provide the trigger for macrophage-derived TNFa production in rheumatoid arthritis. Immune complexes may bind to any of three IgG Fc receptors (FcyR). Therefore, the ability of monocyte-derived macrophages to produce TNFa was examined following ligation of each of the three human FcyR, using murine monoclonal antibodies (mAb) to each receptor as a model for small immune complexes. Methods. Adhered human monocytes expressing all three FcyR were incubated with murine anti-FcyR mAb directed against FcyRI, FcyRII or FcyRIII. Supernatants were collected at various time points and tested for the presence of TNFa and IL-la by ELISA. Results. The anti-FcyRIII mAb induced adhered human monocytes to release TNFa. However, F(ab )2 and Fab fragments of the anti-FcyRIII mAb failed to induce TNFa production. TNFa was undetectable following incubation with the anti-FcyRI or anti-FcyRII mAb. Furthermore, blocking FcyRI or FcyRII had no effect on the levels of TNFa released in response to the anti-FcyRIII mAb. Of the three anti-FcyR mAb, only the anti-FcyRIII induced IL-la production from adhered human monocytes and this was inhibited by the presence of a neutralising anti-TNFa mAb. Conclusions. This study suggests a dominant role for FcyRIIIa in the induction of both TNFa and IL-la production by human macrophages in rheumatoid arthritis following receptor ligation by small immune complexes. The signalling of TNFa production, may require either the ligation of three FcyRIIIa receptors or only two FcyRIIIa receptors, where one interaction must involve binding via an Fc domain. Additionally, IL-la production following FcyRIIIa ligation appears to be dependent upon the presence of TNFa.

280 continuous agitation. The plate was then washed four times with PBS and tetramethylbenzidine (TMB) peroxidase substrate (Sigma) added. The reaction was stopped with 2M sulphuric acid and optical densities read at 450 nm.

Interleukin-1 alpha (IL-la) ELISA. Cell free supernatants were loaded onto

96 well maxisorb plates precoated with a mouse anti-human IL-la mAb.

Recombinant human IL-la was used as a standard. The assay was then performed as instructed by the manufacturer (R & D Systems, Oxfordshire, UK). Statistical analysis. Significance was determined using unpaired T-test with Fisher's correction for small sample size.

RESULTS

Expression of FcyR by adhered human monocytes. When isolated human monocytes were allowed to adhere to plastic for 48 hours, an increase in FcyRIIIa expression was seen (from 10% to 50% cells positive for FcyRIIIa). FcyRI and

FcyRII expression levels did not change over the period of incubation (30% and

90% cells positive for FcyRI and FcyRII respectively) (Figure 1). In other experiments, CD 14 levels declined after 24 hours (data not shown) therefore adherence for 24 hours was chosen for subsequent experiments.

TNFa release from adhered monocytes following incubation with anti-

FcyR mAb. When adhered human monocytes were incubated with the anti-

FcyRI, anti-FcyRII or anti-FcyRIII mAb (Figure 2) TNFa was detected only in those cultures incubated with the anti-FcyRIII mAb. The response was concentration dependent (Table 1) and polymixin B resistant. Levels of TNFa produced in response to the anti-FcyRIII mAb peaked at 4 hours and then declined with time.

T N F a was undetectable in the supernatants of cultures incubated with medium alone, mouse IgGj or mouse IgG 2b- T N F a release was induced by both LPS (in the

281 absence of polymixin B) and PM A. The LPS induced TNFa response peaked at 6 hours and then declined with time, while the PMA response rose continuously. All three anti-FcyR mAb block ligand binding and recognise epitopes in or close to their receptor's binding site (22-24). Therefore, in order to determine whether any additional FcyR recruitment was occurring during the anti-FcyRIII response, anti-FcyRI and anti-FcyRII mAb were used to block receptor binding.

Adhered human monocytes were preincubated at 37°C for 30 minutes with either medium alone or the anti-FcyRI mAb or the anti-FcyRII mAb at saturating concentrations (1 0 jig/ml), as determined by indirect immunofluorescence and flow cytometry. The anti-FcyRIII mAb was then added to all cultures. Blocking either FcyRI or FcyRII receptors had no significant effect on the levels of TNFa released in response to anti-FcyRIII mAb stimulation (Figure 3).

Effect of anti-Fc 7 RIIIa F(ab )2 and Fab fragments on TNFa secretion by monocytes. To determine whether TNFa release in response to the anti-FcyRIII mAb was dependent upon an Fc-FcyRIIIa interaction, adhered monocytes were incubated with F(ab ) 2 or Fab fragments of the anti-FcyRIII mAb. Neither the anti-

FcyRIII F(ab )2 or Fab fragments induced adhered monocytes to secrete TNFa

(Figure 4). LPS (in the absence of polymixin B) also induced adhered monocytes

to release TNFa as observed previously (1.076 ± 0.57 ng/m l).

In a separate series of experiments, when adhered monocytes were preincubated with F(ab )2 fragments of the anti-FcyRIII mAb at binding saturating levels (20 ng/ml), the whole anti-FcyRIII induced TNFa response was inhibited by 69.0 ± 6.2%.

Effect of various inhibitors on TNFa release. To determine whether the presence of TNF a in supernatants was due to the release of preformed or newly synthesised cytokine, cultures were pre-treated with cycloheximide, actinomycin

282 D or colchicine. As seen in Figure 5, cyclohexamide, an inhibitor of protein translation, significantly inhibited TNFa secretion in response to the anti-Fcy RIII mAb (99.9 ± 0.1%; p < 0.01) and LPS (99.9 ± 0.3%; p < 0.01). Treatment of adhered monocytes with actinomycin D, which blocks transcription, also inhibited the LPS response (70.2 ± 7.8%) and significantly inhibited the anti-Fcy RIII mAb response (93.3 ± 6.2%; p < 0.01). The presence of colchicine, an inhibitor of microtubule integrity and secretory vesicle function, did not significantly inhibit the anti-FcyRIII mAb response (24.1 ± 23.0%) or the LPS response (37.3 ± 5.6%).

IL-la production by monocytes following incubation with anti-FcyR mAb. When adhered human monocytes were incubated with the anti-FcyRI, anti-

Fcy RII or anti-Fcy RIII mAb, only the anti-Fcy RIII mAb induced secretion of IL-1 a

(Figure 6 ). IL -la production in response to the anti-Fcy RIII mAb was not detectable for at least 11 hours incubation and rose gradually over the next 61 hours. I L -la was not detected in supernatants from cultures incubated with medium alone or mouse isotype. LPS (in the absence of polymixin B) also induced adhered monocytes to release IL-la (Figure 6 ), detectable from 10 hours after stimulation and again, levels rose with time. In addition, it was found that although the TNF a levels in supernatants were similar in the presence of LPS or the anti-FcyRIII mAb (data not shown), higher IL-la levels were found in response to LPS compared with anti-FcyRIII mAb. Inhibition of IL-la production by anti-TNFa mAb. It has previously been shown that IL-la production by rheumatoid synovial mononuclear cells may be dependent upon the presence of TNFa (25). Therefore, adhered human monocytes were incubated for 48 hours with either LPS or the anti-Fcy RIII mAb in the presence of increasing concentrations of a neutralising murine monoclonal anti-TNFa antibody (CB006). In the representative experiment shown in Figure 7a and b, IL-la secretion was inhibited by the anti-TNFa antibody when cultures were stimulated with the anti-Fcy RIII mAb (Figure 7a).

283 No such inhibition was seen in the presence of LPS (Figure 7b).

DISCUSSION

This study examined the ability of ligand binding-site specific murine anti-

FcyR monoclonal antibodies to induce the production of TNFa from human monocyte-derived macrophages, in the absence of secondary crosslinking. Of the three anti-FcyR mAb, only that directed against FcyRIIIa triggered adhered human monocytes to release TNFa. This suggests that FcyRI and FcyRII both require the crosslinking of multiple receptors, whilst only two or three FcyRIIIa receptors need to be co-ligated for signalling of TNFa production. This is consistent with previous observations that FcyRI and FcyRII ligation only results in TNF a release by human peripheral monocytes when multiple receptors were crosslinked by immobilised murine IgG (19). Both LPS and anti-Fcy RIII m Ab-induced TNFa responses were inhibited by cyclohexamide and actinomycin D. This confirmed that TNFa detected in supernatants required both transcription and protein translation^, and was not due to mobilisation of intracellular stores of TNFa. Colchicine reduced both LPS and anti-FcyRIII mAb-induced TNFa production by approximately 30%, suggesting some dependence upon microtubule polymerisation.

As suggested by others (19), a true IgG Fc-FcyR interaction may be required for intracellular signalling to result in TNFa production. Using blocking antibodies to FcyRI or FcyRII we found that anti-FcyRHIa mAb-mediated TNFa production was independent of these two receptors. Furthermore, neither F(ab ) 2 nor Fab fragments of the anti-Fcy RIII mAb were able to induce TNFa production. This could imply that for signalling and subsequent TNFa release, stimulation through FcyRIIIa requires recruitment of three receptors. Alternatively, ligation of only two receptors may be mandatory, but, if so, one interaction must involve binding via an Fc domain. This may be due to the additional requirement for

284 certain sequences or associated carbohydrate structures present within the Fc region to stabilise the interaction. IL-la is present at high levels in rheumatoid synovial fluids (26), and was the first pro-inflammatory cytokine to be implicated in rheumatoid synovitis, and particularly joint destruction (27). Therapeutic blockade of IL-1 is effective, but perhaps less strikingly so than for TNFa (28). As for TNFa release, only the anti-FcyRHIa mAb induced IL-la production from adhered human monocytes.

Furthermore, its inhibition by a neutralising anti-TNFa mAb, suggests that FcyRIIIa mediated IL-la production is dependent upon the presence of TNFa. In contrast, LPS-induced IL-la production was not inhibited by the neutralising antibody to TNFa. This may be explained by the downregulation of both TNF receptors TNF-R55 and TNF-R75, together with the release of soluble TNF-R75, previously observed with LPS-mediated activation of monocytes (29). Our novel findings suggest that ligation of Fey RHIa may favour the establishment of a pro- inflammmatory cytokine loop initiated by TNFa. This is very much in keeping with previous observations on rheumatoid synovial cell cultures suggesting a dominant role for TNFa (25). The targeting of inflammation to synovium and other tissues in rheumatoid arthritis has previously been difficult to explain. Although^ circulating IgG rheumatoid factor-based immune complexes have long been suspected to be involved in the disease process, the effector mechanism has been unclear. Our findings indicate that macrophage FcyRIIIa is the most likely FcyR to be triggered to produce TNFa following ligation by a small IgG-based immune complex. Ligation of FcyRIIIa, but not FcyRI or FcyRII, on monocytes, has also been shown to induce the generation of reactive oxygen species (30), which may contribute further to tissue injury in areas of Fey RHIa expression.

This conclusion is consistent with previous observations highlighting a possible role for FcyRIIIa in the pathogenesis of rheum atoid arthritis (12,18).

285 Unlike FcyRI and FcyRII, FcyRIIIa expression by macrophages is tissue and site specific. A close correlation between Fey RHIa expression and the location of both synovitis and extra-articular features has been demonstrated (16). The small IgG rheumatoid factor-containing ("intermediate") immune complexes first described by Kunkel et al (31) may therefore be the pro-inflammatory stimulus to macrophage activation preceding T cell infiltration in rheumatoid synovium.

This inflammatory stimulus may subsequently be amplified by T-B lymphocyte interaction and the local generation of large complement-fixing complexes based on rheum atoid factor of all isotypes (6,7). In conclusion, the pattern of induction of TNFa production by human monocyte/macrophages supports previous evidence for a specific role for FcyRIIIa in inflammation induced by small circulating immune complexes (12,18). Taken as a whole these findings suggest that the clinical syndrome of rheumatoid arthritis may reflect a pro-inflammatory mechanism based on such small immune complexes quite distinct from that seen in lupus nephritis and other forms of basement membrane-associated complex deposition.

ACKNOWLEDGEMENTS The authors would like to thank Celltech Therapeutics for financial support.

REFERENCES

1. Elliott MJ, Maini RN, Feldmann M, Long-Fox A, Charles P, Katsikis P, et al. Treatment of rheumatoid arthritis with chimeric monoclonal antibodies to tum or necrosis factor a. Arthritis Rheum 1993; 36:1681-90.

2. Mclnnes IB, Liew FY. Interleukin 15. A proinflammatory role in rheumatoid arthritis synovitis. Immunol Today 1998; 19: 75-9.

286 3. Kraan MC, Versendaal H, Jonker M, Bresnihan B, Post WJ, 't Hart BA, et al. Asymptomatic synovitis precedes clinically manifest arthritis. Arthritis Rheum.

1998;41:1481-8.

4. Kurosaka M & Ziff M. Immunoelectron microscopic study of the distribution of T cell subsets in rheumatoid synovium. J Exp Med 1983; 158:1191-1210.

5. Maini RN, Zvaifler N. Rheumatoid arthritis. In: Klippel and Dieppe PA, editors. Rheumatology. 2nd edition. London: Mosby; 1998;vol 2:p. 5.1.1-5.15.10.

6 . Winchester RJ, Agnello V, Kunkel HG. Gamma globulin complexes in synovial fluids of patients with rheumatoid arthritis. Partial characterisation and relationship to lowered complement levels. Clin Exp Immunol 1970;6:689-706.

7. Mannik M, Nardella FA. IgG rheumatoid factprs and self-association of these antibodies. Clin Rheum Dis 1995;11:551-72.

8 . Walport MJ, Davies KA, Morley BJ, Botto M. Complement deficiency and autoim m unity. Ann N Y Acad Sci 1997;815:267-81.

9. Pope RM, Teller DC, Mannik M. The molecualer basis of self-association of antibodies to IgG (rheumatoid factors) in rheumatoid arthritis. Proc Natl Acad Sci

USA 1974;71:517-21.

10. Nardella FA, Teller DC, Mannik M. Studies on the antigenic determinants in the self-association of IgG rheumatoid factor. J Exp Med 1981;154:112-25.

11. Brown PB, Nardella FA, Mannik M. Human complement activation by self-

287 associated IgG rheumatoid factors. Arthritis Rheum 198;225:1101-7.

12. Edwards JCW, Cambridge G, Abrahams VM. Do self-perpetuating B lymphocytes drive human autoimmune disease? Immunology 1999;97:188-96.

13. Unkeless JC. Function and heterogeneity of human Fc receptors for im m unoglobulin G. J Clin Invest 1989;83:355-61.

14. Fanger MW, Shen L, Graziano RF, Guyre PM. Cytotoxicity mediated by human Fc receptors for IgG. Immunol Today 1989;10:92-9.

15. van de Winkel JGJ, Capel PJA. Human IgG Fc receptor heterogenity: molecular aspects and clinical implications. Immunol Today 1993;14:215-21.

16. Bhatia A, Blades S, Cambridge G, Edwards JCW. Differential distribution of FcyRIIIa in normal human tissues and colocalisation with DAF and fibrillin-1. Immunology 1998;94:56-63.

17. Klaassen RJL, Goldschmeding R, Tetteroo PAT, von dem Borne AEGK. The Fc valency of an immune complex is the decisive factor for binding to low-affinity Fey receptors. Eur J Immunol 1988;18:1373-7.

18. Edwards JCW, Cambridge G. Rheumatoid arthritis: the predictable effect of small immune complexes in which antibody is also antigen. Brit J Rheumatol 1998;37:126-30.

19. Debets JMH, van de Winkel JGJ, Ceuppens JL, Dieteren IEM, Buurman WA. Cross-linking of both FcyRI and FcyRII induces secretion of tum our necrosis factor by human monocytes, requiring high affinity Fc-FcyR interaction.

288 Functional activation of FcyRII by treatment with proteases or neuraminidase. J Immunol 1990;144:1304-10.

20. Agegon I, Cuturi MC, Trinchieri G, Perussia B. Interaction of Fc receptor (CD16) ligands induces transcription of interleukin 2 receptor (CD25) and lymphokine genes and expression of their prducts in human natural killer cells. J Exp Med 1988; 167:452-72.

21. Hendrich C, Kuipers JG, Kolanus W, Ham m er M, Schmidt RE. Activation of CD16+ effector cells by rheumatoid factor complex. Role of natural killer cells in rheumatoid arthritis. Arthritis Rheum 1991;34:423-31.

22. Fleit HB, Wright SD, Unkeless JC. H um an neutrophil Fey receptor distribution and structure. Proc Natl Acad Sci USA 1982;79:3275-9.

23. Looney RJ, Ryan DH, Takahashi K, Fleit HB, Cohen HJ, Abaraham GN, et al. Identification of a second class of IgG Fc receptors on human neutrophils. A 40 kilodalton molecule also found on eosinophils. J Exp Med 1986;163:826-36.

24. Dougherty GJ, Selendran Y, Murdoch S, Palmer DG, Hogg N. The human mononuclear phagocyte high-affinity Fc receptor, FcRI, defined by a monoclonal antibody, 10.1. Eur J Immunol 1987;17:1453-9.

25. Brennan FM, Chantry D, Jackson A, Maini R N , Feldmann M. Inhibitory effect of T N F a antibodies on synovial cell interleukin 1 production in rheumatoid arthritis. Lancet 1989;2(8657):244-7.

26. Fontana A, Hengartner H, Weber E, Fehr K, Grob PJ, Cohen G. Interleukin-1 activity in the synovial fluid of patients with rheumatoid arthritis. Rheumatol

289 Int 1982;2:49-53.

27. Saklatavala J. T N F a stimulates resorption and inhibits synthesis of proteoglycan in cartilage. Nature 1986;322:547-9.

28. Bresnihan B, Alvaro-Garcia JM, Cobby M, Doherty M, Domljan Z, Emery Y, et al. Treatment of rheumatoid arthritis with recombinant human interleukin-1 receptor antagonist. Arthritis Rheum 1998;41:2196-2204.

29. Leeuwenberg JF, Dentener MA, Buurm an WA. Lipopolysaccharide LPS- mediated soluble TNF receptor release and TNF receptor expression by monocytes. Role of CD14, LPS binding protein, and bactericidal/permeability- increasing protein. J Immunol 1994;152:5070-6.

30. Trezzini C, Jungi TW, Spycher MO, Maly FE, Rao P. Human monocytes CD36 and CD16 are signalling molecules. Evidence from studies using antibody- induced chemiluminescence as a tool, to probe signal transduction. Immunology 1990;71:29-37.

31. Kunkel HG, Muleer-Eberhand HJ, Fendenberg H, Tomasi TB. Gamma globulin complexes in rheumatoid arthritis and certain other conditions. J Clin Invest 1960;40:117-29.

290 LEGENDS FOR FIGURES Figure 1. Effects of culturing adhered human monocytes on FcyR expression. Following adherence to plastic for i) 0, ii) 24 or iii) 48 hours at 37°C, hum an monocytes were then incubated with predetermined optimal binding concentrations of a) anti-Fcy RI mAb (10.1) or murine IgG! b) anti-FcyRII mAb

(IV.3) or murine IgG2b or c) anti-FcyRIII mAb (3G8) or murine IgGj. Surface bound antibody was detected using FITC conjugated goat F(ab )2 anti-mouse Ig and fluorescence intensity measured by flow cytometry. Data shown are from a representative experiment with fluorescent intensity along the X axis and number of events along the Y axis. Similar results were observed in two repeat experiments.

Figure 2. The time course of TNFa production by adhered human monocytes. Cultures were incubated at 37°C with anti-Fcy RI (10.1), anti-Fcy RII (IV.3), anti- FcyRIII (3G8), mouse IgG!, mouse IgG2b, all at 50 ng/ml, LPS (0.5 ng/m l in the absence of polymixin B), PMA (0.5 ng/ml) or medium alone. Cell free culture supernantants were collected at various time points over 24 hours and assayed for TNFa Data shown are from one representative experiment and similar time courses for TNFa were observed in five repeat experiments.

Figure 3. Effects of blocking Fey RI or Fey RII on anti-Fcy RIII mAb induced TNFa production. Adhered human monocytes were preincubated with anti-Fcy RI (10.1) or anti-Fcy RII (IV.3) at saturating binding levels or medium alone for 30 minutes at 37°C. After 4 hours subsequent incubation with anti-Fcy RIII (3G8) at 50 \ig/m l, supernatants were collected and assayed for the presence of TNFa. Values are expressed as the mean ± S.D. from three experiments.

Figure 4. Effects of anti-Fcy RIII mAb F(ab ) 2 or Fab fragments on TNFa

291 production. Adhered human monocytes were incubated at 37°C with anti-

FcyRni mAb (3G8) (50 pig/ml), F(ab )2 fragments of anti-FcyRIII (33 ng/ml) or Fab fragments of anti-FcyRIII (17 pig/ml). After 4 hours, cell-free supernatants were collected for measurement of TNFa concentration. Values are expressed as the m ean ± S.D. from three experiments.

Figure 5. Effects of various inhibitors on T N F a production. Adhered human monocytes were incubated at 37°C with anti-Fcy RIII (3G8) (50 pig/ml) or LPS (0.5 pig/ml) in the presence of medium alone, cyclohexamide (150 pig/ml), actinomycin D (2 pig/ml) or colchicine (1 pig/ml) (all from Sigma). After 4 hours, cell-free supernatants were collected and assayed for T N F a . Values are expressed as the m ean ± S.D. from three experiments.

Figure 6 . Time course of I l- la production by adhered human monocytes. Cultures were incubated with anti-Fcy RI (10.1), anti-Fcy RII (IV.3), anti-Fcy RIII (3G8), mouse IgGlr mouse IgG2b (all at 50 pig/ml), LPS (0.5 pig/ml, in the absence of polymixin B) or medium alone. Culture supernatants were collected at various time points over 72 hours and the cell free samples were then assayed for I l- la . Data shown are from one representative experiment and similar time courses for IL-la were_ observed in two repeat experiments.

Figure 7 a) & b). Effects of an anti-TNFa mAb on IL-la production. Adhered hum an monocytes were incubated at 37°C with a) anti-Fcy RID mAb (3G8; 50 pig/ml) or b) LPS (0.5 pig/ml in the absence of polymixin B) in the presence of a mouse anti-human TNFa mAb (CB006) at 0, 1, 10 or 100 pig/ml. Supernatants were collected following 48 hours incubation and tested for IL-la. Data shown are from one representative experiment and similar results were observed when this experiment was repeated.

292 Events Events _ Events 00 10 10 Figure 1 Figure aii) aiii) ai) ° ° ,yJw ., "Varv "Varv ., ,yJw FL1-H FL1-H FL1-H 102 LLI > LU UJ if) > d) c > 0) c 00 O O 00 o 00 bi) bii) —rVr-fV V r i— biii) FL1-H FL1-H 102 293 LU LU LU > 0) c t/> (D c= > O O 10 10 ciii) cii) ci) ° ° Tmr FL1-H FL1-H 102 TTTTTTTT TTTTTT1 ►n c ft) ho crc hJ o oj *3 *3 53 J3 0 < 00 •n 'Tfl ^Ti (_a (_a Ul o o n 1 ct N J P C a ^> C/5 & 5 o o CR I ' CfQ 5- 2- £ 294 TNF (ng/ml) CJ1 o o o o CJT o Time (Hours) 295 00 00 o TNF (ng/ml) at 4 hours P ON P t-1 4^ j O K o o o Medium Anti-FcyRI (10.1) Anti-FcyRII (IV.3) i-t n> C ►n Crq O n o CJ p Ln If* o 296 o TNF (ng/ml) at 4 hours N> o o o Anti-Fc yRIII F(ab)2 of Fab of anti-FcyRIII anti-FcyRIII • hrl I—- r> u> O 00 cn cn s. □ □ ■ 297 TNF (ng/ml) 0.002 /- + 0.002 /- + 0.004 0.001

Medium Cyclohexamide Actinomycin D Colchicine hi c h fD ON crq ► 3 h i n HH o M > k > 3 h-ri. h i n HH < l-H HH I > 3 H-* h i n ? 3 i— i HH 1—1 O 0 0 Q 00 .

m S. <» 09 <» S. a cr 3 ^ &pn ON c/i 5 2 3 3 298 IL-1 (pg/ml) N> 1 XI o o w o o CJ On o Time (H ours) 1 >—I • »T >—I c CTQ o 00 299 IL-1 (pg/ml)IL-1 J N> O O O O O O ^ fti o o o o

Anti-TNF mAb ( |1 g/m l) ►n i—*• C •-t N l rD C rq i oo _ I ON ON _ L_ 4^ __ 300 IL-1 (pg/ml) L_ hj ___ o o

Anti-TNF mAb ( |i g/ml) Appendix 2.

Calibration of the HPLC column

HPLC elution times of calibration markers of known molecular weights.

Calibration marker Molecular weight Elution time (kDa) (minutes) A lbum in 6 6 9.63 Alcohol dehydrogenase 150 9.20

Bamalyase 2 0 0 8.76 Apo ferritin 443 8.16 Thyroglobulin 669 7.26

301 Appendix 3.

Calibration of the FACS can

MFIs for each Fluorosphere bead population. Values are expressed as the m ean ± S.D. from three separate readings.

Molecules of FITC per bead Mean fluorescent intensity (MFI) 2500 14.647 ± 10.824 6500 36.767 ± 30.951 19000 104.96 ± 86.684 55000 283.87 ±233.15 150000 619.75 ±350.75

The equation of above standard curve was y = -4.4999 + 0.92352x, where: y = MFI m = -4.4999 c = 0.92352.

302 Appendix 4.

TN Fa ELISA

Optical densities for the IgGl rheumatoid factor mAb incubated at 37°C in the presence (+) or absence (-) of adhered monocytes.

Optical densities at 450nm

IgGl-RF mAb 2 Hrs 6 Hrs 24 Hrs + monocytes 1.085 0.964 1.034 - monocytes 1.063 1.196 1.841

303 Appendix 5.

Buffers and media

Bicarbonate buffer (BIC) 8g N a 2Co 3, 15.5g NaHCo 3 and make up to 10 litres with distilled water. Adjust to pH9.6.

Bradford assay dye 600mg of Coomassie blue G-250 in 1L of 2% perchloric add.

Citrate buffer (0.05M) 25.7ml 0.2M N a 2HPC>4,24.3ml 0.1M citric add and make up to 100ml with distilled water. Adjust to pH5.

Hank's Balanced salt solution (HBSS) Dilute 10X HBSS without phenol red (Gibco) 1:10 with distilled water.

Phosphate buffered solution (PBS) PBS 10X : 20g KC1, 20g KH2P0 4 ,114.8g N a 2HPo 4, 800g NaCl. Make up to 10 litres with distilled water. Dilute PBS 10X 1:10 and adjust to pH7.4.

PBS/Tween Dilute PBS 10X 1:10 and adjust to pH7.4. A dd 0.1% Tween 20 (Sigma)

304