A Study of the Igm Interaction with Complement Using Mouse Igmligg2b Domain-Switched Hybrids

A Study of the IgM Interaction with Complement Using Mouse IgMlIgG2b Domain-switched Hybrids

Frieda Huey Chen

A thesis subrnitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Biochemistry University of Toronto

O Copyright by Frieda Huey Chen 1998 National Library Bibliothèque nationale l*I of Canada du Canada Acquisitions and Acquisitions et Bibliographie Services services bibliographiques 395 Wellington Street 395, rue Wellington Ottawa ON K1A ON4 Ottawa ON K1A ON4 Canada Canada

The author has granted a non- L'auteur a accorde une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan., distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la foxme de rnicrofiche/film, de reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. The further one pursues knowledge, the less one knows. - Lao Tzu

Dedicated to those who persevere in spite of it all. ABSTRACT

A study of the IgM interaction with complement using mouse IgM/IgGZb domain-switched hybrids

Degree of Doctor of Philosophy, 1998. Frieda Huey Chen Graduate Department of Biochemistry, University of Toronto

The IgM and IgG classes of immunoglobulins share an important bioiogical function that is distinct fiom that of other classes. When bound to antigen, both can initiate the classical complement pathway by interacting with Clq, a subunit of the fmt component of cornplement (Cl). To study the Clq binding properties of IgM in a form comparable to that of IgG, a 'monomeric' form of mouse IgM was prepared and its function was compared to that of a mouse IgG with the same antigen binding site. Since most of the wildtype IgM is secreted as pentarnen, monomenc IgM was isolated from a mutant (IgM P544G) which is secreted predominantly as monomers. The monomeric form of this mutant was not able to bind Clq or initiate cornpiement-mediated lysis (CML) on haptenated SRBC. Under the same conditions, the positive control (IgG2b) bound Clq and initiated CML in a dose-dependent manner. Because polymeric IgM isolated from diis mutant was able to bind Clq and initiate CML as well as wildtype IgM did, it is unlikely that the P544G mutation is directiy responsible for the inactivity of the mutant monomer. To understand why the monomeric IgM is inactive, mouse IgMlIgG2b hybrid mutants were created and analyzed for their Clq binding properties. Cfl, the intrinsically active C lq binding domain of IgG2b, did not bind C lq or initiate CML when placed in the monomeric IgM background (ppp). This suggests that the activity of an intrinsically active C lq binding domain is hidden by one or both of the neighbouring Cp2 and Cp4 domains. Cp3, the putative Clq binding domain of IgM, rernained inactive when it was placed in the non-inhibitory IgGZb background (yypy). This demonstrates that a preformed Clq binding site is not found in the Cp3 domain. Collectively, the results suggest that if the Cp3 contains the entire Clq binding site for IgM, the site is unformed and hidden. To detennine if one or both of the Cp2 and Cp4 domains are responsible for the inactivity of the ppyp hybrid rnonomer two hybrids (pmand lpyy) were created in which these domains were individudy replaced by their corresponding IgG2b dornains (y-hinge and Cy3). Both hybrid monomers were defective in initiating CML whereas only the pyyp hybrid, which has Cp4, was defective in Clq binding. The results show that Cp4 interferes with the interaction between C 1 and the Cy2 domain whereas Cp2 interferes with the subsequent activation of CL Cp2 and Cp4 may therefore have similar inhibitory effects on the Cpdomain in the IgM monomer. The transplantation of Cp3 together with Ck4 hto the IgG background permitted polymer formation. This polymer was able to bind Clq, although both the monomer and the polymer forms were not able to initiate CML;converseiy, all IgM polymers with a transplanted Cy2 domain were active in both Clq binding and CML and demonstrated apparent Kd values for Clq (1-2 x 10-~M) that were similar to that of wild-type IgM. This suggests that the affinity of the C lq binding sites in IgM are comparable to those in IgG2b. The findings reported in this thesis are discussed in relation to the activity of the IgM polymer and are consistent with the view that the 'star' to 'staple' conformational change resulting when IgM binds to antigen is necessary for the expression of the C lq binding site.

iii Acknowledgments

But there are two expensive forrns of education, either of which a parent rnay procure for his son by sending hirn as solitary pupil to a clergyman: one is, the enjoyment of the reverend gentleman's undivided neglect; the other is, the endurance of the reverend gentleman's undivided attention. -George Eliot frorn The Mill on the Floss

Having had the fortune to be a pupil of not one but three supervisors, I am happy to Say that I've enjoyed both their 'undivided attention' in the form of their guidance and support during the critical times, and their 'undivided neglect' in the form of the great freedom with which I was permitted to conduct my studies. Each member of this supervisory team or 'Triumvirate' - Dr. Robert Painter, Dr. David lsenman and Dr. Marc Shulman - has been a vital source of encouragement and direction for which 1 am greatly indebted. Over the years I've grown to appreciate the unique strengths of each. Dr. Painter has never ceased to impress me with his ability 'to see the forest from the trees', a talent that no doubt stems frorn his broad knowledge base - he is as versed in science as he is in Shakespeare. Dr. lsenman has always been tremendously generous with his time and technical expertise, and his tireless enthusiasm as a teacher is infectious. Dr. Shulman has contributed his thoroughness and ability to present alternative interpretations, traits that have made him valuable in maintaining the balance within the triumvirate. In addition to my supervisors. Dr. David Pulleyblank has also made helpful suggestions at my commîttee meetings and during the writing of this thesis. To Say that this work is a result of my efforts alone would be a gross distortion of the truth. Looking back on the whole process it is with humility that I recognize the contributions, both direct and indirect, of so many others and it is to thern that 1 wish to extend my thanks (hopefully without omissions). In the lab which became my home away from home, rny Iabmates becarne my surrogate family. Dr. Sudha Arya, who rnany of us refer to affectionately as 'mom', provided much in the way of emotional support. Her srnile and trademark laughter marked many mernorable days in the lab. Several fellow grad students who became my 'siblings' during this stage of my life and with whom 1 shared many experiences, successes, embarassments, arguments and at times miseries are Roger Ebanks, Nana Lee and John Chan. I thank Roger for sharing his philosophies, Nana for spinning her fairytales and John for being a free-spirit. Thanks go to Myrna Cohen-Doyle for her fabulous cottage parties and for sharing 'His Majesty' with me. I also owe thanks to the following for their friendship (in stream-of-consciousness order): Aiko Taniguchi-Sidle, Amir Khan, Fernando Rock, Maggie Everett, Yuan-Yuan Xu. Ranga Robinson, Navneet Aluwalia, Davinder Chawla, Shelley Hepworth and Owen Rowland. Shirley Furesz, Bhushan Nagar, Alp Oran, Erik Wiersma, Pat Bronskill, Augustin Nguyen, and members of the Dr. Isenman's, Dr. William's, Dr. PulleyblanKs, Dr. Klein's, Dr. Segall's and Dr. Bennick's labs past and present. I would also like to acknowledge the indispensable services provided by the staff in the departmental office - Anna, Hazel, Suzanne, Carol and Farhia and - and by the fifth floor caretaking staff. I must also thank several fnends outside of the gray building known as 'Med Scia for providing me with (too) many pleasant distractions. 1 will always remember jamming till the wee hours of the morning on some broken down piano with Wendy Lau, camping with Sue Choi, and Lucky Palace discussions with Deborah Villeneuve. Thanks to Nancy Ryu and Joseph Nachman for our musical partnerships. Thanks to Alma V. and Sing Hoo Yuen at Fudger House for their words of encouragement ('Why not? You should try!') and for being an 'appreciative audience'. Thanks to Nadia Saracoglu for her calming influence and for her 'spa' night hamrnock. I owe my parents many thanks for their care and concern throughout the years and for enduring my decision to pursue a doctoral degree. Finally 1 would like to express my deepest gratitude to two people who suffered the most at the hands of my true personality: Ann, who has been more friend than younger sister in a11 our years together and whose unwavering belief in me has helped me through the darker moments of self- doubt; and Steve, who has been supportive, patient. and above al1 understanding.

This work was supported by the Medical Research Council of Canada. The author also acknowledges the support provided by a Govemment of Ontario Graduate Scholarship.

The author thanks the copyright holders for the permission to reproduce certain figures and text in this thesis. Table of Contents

.. AB STR4CT ...... u ACKNOWLEDGMENTS ...... iv TABLE OF CONTENTS ...... vi LIST OF FIGURES ...... xi LIST OF TABLES ...... xiv AB B REVIATIONS ...... xv

CElAPTFCR 1 Introduction

1.1 General Introduction ...... 2

1.2 Immunoglobulin Structure ...... 3 IgG structure ...... 5 1.2.1.1 Quaternary structure ...... 5 1.2.1.2 Secondary and tertiary structure ...... 7 1.2.1.3 Primarystnicture ...... 10 1.2.1.4 Glycosylation ...... 12 IgM structure ...... 15 1.2.2.1 Quaternary structure ...... 16 1.2.2.2 J chah ...... 20 1.2.2.3 Secondary and tertiary structure ...... 22 1.2.2.4 Primary structure ...... 22 1.2.2.5 Glycosylation ...... 23 Phylogeny of immunoglobulins ...... 23 The 'Domain Hypothesis' ...... 26

1.3 The Classical Complement Cascade ...... 26 1.3.1 C 1 structure ...... 29 1.3.1.1 Clq subunit ...... 30 1.3.1.2 Clr and Cls ...... 33 1.3.1.3 TheClr2Cls2tetramer ...... 38 1.3.1.4 Arrangement of the C lr2C 1 s2 tetramer about C lq ...... 39 1.3.1.5 Nature of the interactions between the Clr2C ls;! tetramer and C lq ...... 42 Table of Contents (conttd)

1.3.2 Activation and regulation of C1 ...... 43 1.3.2.1 The relationship between C 1 binding and C 1 activation ..... 43 1.3.2.2 C 1 activation by immune complexes ...... 43 1.3.2.3 Autoactivation- of Cl ...... ,., ...... 45 1.3.2.4 Cl inhibitor ...... ,., ...... 45 1.3.2.5 The relationship between C 1 q binding and subsequent steps in the classical complement pathway ...... 47

1.4 The Interaction Between Irnmunoglobuiins and Cl ...... 48 The nature of the interaction ...... ,...... ,..,...... 48 1.4.1.1 IgM vs . IgG complement fixation and temperature dependence ...... 48 1.4.1.2 Electrostatic component of the IgG and C lq interaction ..... 48 1.4.1.3 Possible hydrophobie component of the interaction between IgG and C 1q ...... 50 1.4.1.4 Presence of polyethylene glycol affects the interaction ...... 50 1.4.1.5 Studies using radiolabeling of C lq and their implications ...... 50 linmunoglobulin binding site on Clq ...... 51 Cl binding site on IgG ...... 52 1.4.3.1 Evidence supporting the role of C-y2 domain ...... 52 1.4.3.2 TheroleoftheCy3 domain ...... 55 1.4.3.3 IgG subclasses differ in their interaction with complement ...... 55 1.4.3.4 Feanires which detemine these IgG subclass differences ...... 56 1.4.3.5 The role of the Asn 297 carbohydrate on C 1 binding and activation ...... 58 Cl binding site on IgM ...... 60 1.4.4.1 Evidence supporting the role of the Cp4 domain ...... 61 1.4.4.2 Evidence supporting the role of the Cp3 domain ...... 62 1.4.4.3 The role of the Asn 402 carbohydrate on C 1 binding and activation ...... 64 1.4.4.4 The role of J chain in regulating the interaction between IgM polyrner and complement ...... 64 Heterologous systems in the study of the interaction between IgG, IgM and C 1 ...... 65 Models of C1 binding and activation by IgG and IgM ...... 65

vii Table of Contents (cont'd)

1.5 Project Rationale ...... 71

CHAPTER2 Materials and Methods

DNA Prepration ...... 74 2.1.1 Transformation of competent HB 101 cells ...... 74 2.1.2 Preparation and transformation of electrocompetent ceils ...... 74 2.1.3 Purification of plasmid DNA ...... 75 2.1.4 DNA agarose gel electrophoresis ...... 75 2.1.5 hirification of DNA restriction fragments from agarose ...... 75 2.1.6 DNA sequencing ...... 76 2.1.7 Restriction enzyme digests and ligation reactions ...... 76

Construct Strategies ...... 77 2.2.1 Heavy chah vector constnicts ...... 77 2.2.2 Shuttle vectors ...... 79 2.2.3 Hybrid heavy chah plasmid vectors ...... 80 2.2.4 Heavy chah plasmid vectors for the P544G mutants ...... 89

Immunoglobulin Expression and Charactenzation ...... 93 2.3.1 CeII line ...... 93 2.3.2 Electroporation of MOPC 3 15.26 cells ...... 93 2.3.3 Preparation of the TNP-gelatin coating protein ...... 94 2.3.4 Metaboiic labeling...... 95 2.3.5 Immunoprecipitation ...... 95 2.3.6 SDS-PAGE...... 96

Purification of ImmunoglobuLins ...... 97 2.4.1 Optimization of purification conditions ...... 97 2.4.2 Affinity purification of immunoglobulins...... 98 2.4.3 Separation of the monomer species from the polymer ...... 99 2.4.4 Reduced and awlated ELISA ...... 99

Functional Assays ...... 101 2.5.1 Buffers ...... 101 2.5.2 Preparation of low density TNP-haptenated sheep red blood cells ...... 101

viii Table of Contents (cont'd)

Preparation of high-density TNP-haptenated sheep red blood celis ...... 102 Preparation of guinea pig complement for complement-mediated lysis (CML) assays ...... 102 CML assay for polymers ...... 102 CML assay for monomers ...... 103 C 1q purification ...... 103 Radiolabeling of C l q ...... 105 . . C l q bmding radioimmunoassay ...... 106

Effect of the P544G Mutation on Polymerization and Rationale for its Use in Studying the Activity of the IgM Monomer

Introduction ...... 109

Results ...... 112 3.2.1 Construction, expression and characterization of the IgM P544G derivative ...... 112 3.2.2 Activity of the IgM P544G polymer ...... 112 3.2.3 Analysis of the activity of the IgM P544G monomer ...... 116

Discussion ...... 121

Studies on the IgM/IgGlb Hybrid Imrnunoglobulins

4.1 Introduction ...... 125

4.2 Results ...... 128 4.2.1 Construction and expression of hybrid IgM/IgG2b immunoglobulins ...... 128 4.2.2 The Cy2 domain is not active in a monomeric IgM background ...... 136 4.2.3 The Cp3 domain, alone or together with Cp4, is not active in a monomeric IgG2b background ...... 143 4.2.4 CML activity of the polyrner fractions ...... 143 Table of Contents (cont'd)

4.3 Discussion ...... 153 4.3.1 Evidence that the Clq binding site in ychain is concealed ...... 154 4.3.2 Evidence that the C lq binding site is not forrned in Cp3 ...... 155 4.3 -3 Properties of the hybrid polymers ...... 156 4.3.4 Possible reasons for the enhanced Clq binding and CML activity of the pyyp polyrner ...... 157 4.3.5 Conclusion ...... 160

General Discussion and Future Perspectives

5.1 Generd Discussion ...... 162

5.2 Future Perspectives ...... 167

REFERENCES ...... 170 List of Figures

Figure 1.1 Proteins involved in the initiation step of the classical complement cascade ...... 4

Figure 1.2 The covalent structure of IgG ...... 6

Figure 1.3 A nbbon diagram of the C'y2 domain illustrates the immunoglobulin fold ...... 9

Figure 1.4 The carbohydrate structure of the human y-chah asparagine- iinked oligosaccharide...... 13

Figure 1.5 A schematic diamof the IgM and IgG monomeric subunits ...... 17

Figure 1.6 A mode1 of the IgM pentamer and types of disulfide bonding patterns ...... 19

Figure 1.7 The five carbohydrate structures of mouse p-chah asparagine- linked oligosaccharides ...... 24

Figure 1.8 The major pathways of complement activation ...... 28

Figure 1.9 The proposed structure for the Clq subunit of C 1 ...... 31

Figure 1.10 Electron micrograph and dimensions of the Clq molecule ...... 34

Figure 1.11 The proposed structure for the Clr and C 1s subunits of Cl ...... 36

Fie1.12 Models of the C 1 complex ...... 41

Figure 1.13 Putative location of residues in the Cp3 domain that are critical to the Clq binding site of IgM ...... 63

Figure 1.14 'Star' vs . 'staple' conformations of pentarneric IgM ...... 70

Figure 2.1 The heavy chain expression vectors for mouse IgG2b and IgM ...... 78

Figure 2.2 The constmct strategy for the yypp heavy chah expression vector ...... 81

Figure 2.3 The construct strategy for the myheavy chah expression vector ...... 82

Figure 2.4 The construct strategy for the plyp heavy chah expression vector ...... 84 List of Figures (cont'd)

Figure 2.5 The construct strategy for the pyyp heavy chah expression vector ...... 86

Figure 2.6 The consbnict strategy for the ppyy heavy chah expression vector ...... ,...... 87

Figure 2.7 The construct strategy for the pyyy heavy chain expression vector ...... ~...... ~...... 88

Figure 2.8 The constnict strategy for the ypyp P544G heavy chain expression vector ...... 90

Figure 2.9 The construct strategy for the pyyp P544G heavy chah expression vector ...... 9 1

Figure 2.10 The construct strategy for the IgM P544G heavy chain expression vector ...... 92

Figure 2.1 1 Device used to remove free DNP-glycine frorn the affinity- purified immunoglobuh ...... 100

Figure 3.1 Sucrose density gradient profde and a non-reducing SDS-PAGE gel of wildtype IgM and P544G mutant preparations ...... ,...... 113

Figure 3.2 CML and C lq binding activities of the IgM P544G mutant polymer vs. the wildtype IgM polymer ...... 114

Figure 3.3 Non-linear regression andysis of the Clq binding data for the wildtype IgM and P544G mutant polymers ...... 117

Figure 3.4 CML and C lq binding activities of the IgM P544G monomer vs. wildtype IgG2b monorner ...... 118

Figure 4.1 Schematic diagram of the constant regions of wildtype IgG2b, wildtype IgM and their domain-switched hybrid constructs ...... 129

Figure 4.2 Covalent species of IgG, IgM and the domain-switched hybrids ...... 130

Figure 4.3 Sucrose density gradient profile for IgG2b and C@-containing hybrids ...... 133 List of Figures (cont'd)

Figure 4.4 Sucrose density gradient profde for IgM and Cp3-containing hybrids ...... 135

Figure 4.5 Sucrose density gradient profde and a non-reducing SDS- PAGE gel of the ppyp hybrid and its P544G derivative ...... 137

Figure 4.6 Sucrose density gradient profde and a non-reducing SDS- PAGE gel of the pyyp hybrid and its P544G derivative ...... 138

Figure 4.7 CML and C lq binding activities of the CG-containing hybrid monomen vs. wildtype IgGZb monomers ...... 139

Figure 4.8 Non-hear regression analysis of the binding data for pyyy, pyyp, ppyp and IgG2b monomers ...... 142

Figure 4.9 CML and C lq binding activities of the Cp3-containhg hybrid monomers vs. wildtype IgG2b monomers ...... 145

Figure 4.10 CML and Clq binding activities of the Cp3-containing hybrid polymer yypp VS. that of wildtype IgM polymen ...... 148

Figure 4.1 1 CML and Clq binding activities of the Cy2-containing hybnd poiyrners, ppyp and pyp, vs. that of wildtype IgM polymers ...... 150

Figure 4.12 Non-Linear regression analysis of the binding data for pyyp, ppw, wpand IgM polymers ...... 152

Figure 4.1 3 Schematic diagram of C lq subunits binding to the Fcs disc of wildtype IgM or pp.~pentamer, pyyp pentamer and IgM hexamer ...... 159 List of Tables

Table 1.1 Hinge sequences for human and mouse IgG isotypes ...... Il

Table 4.1 Summary of apparent Kd values and number of C lq binding sites per we ...... 144

Table 5. I S urnmary of CML and C 1 q binding assay results for mouse wildtype IgM, wildtype IgG2b and their hybrids ...... 164 Ab breviations

Constant region Constant region of the heavy chah Constant region of the light chah Complement-mediated lysis DMSO Dimethylsuifoxide DNA Deoxyribonucleic acid DNP Dinitrophenyl Dithiothreitol Ethylenediaminetetraacetic acid Enzyme-Linked immunoadsorbent assay EM Electron microscopy FCS Fetd calf serum G418 Geneticin@ G-PBS Glucose in phosphate buffered saline GVB Gelatin veronal buffer HEPES N-2-Hydroxyethylpiperazine-NI-2-ethanesulfonic acid HI-FCS Heat-inactivated fetal cdf semm k Immunoglobulin MEM Minimal essentid medium NMR Nuclear magnetic resonance PAGE Pol yacry larnide gel electrop horesis PBS Phosphate buffered saline PBS-AZ 0.02% Sodium azide in phosphate buffered saline PMSF Phenylrnethylsulfonyl fluoride RT Room temperature SDS Sodium dodecyl sulfate SGVB Sucrose gelatin veronal buffer SOC Save our cells (medium) SRBC Sheep red blood ceus TCA Trichloroacetic acid TE Tris-EDTA buffer TNBS Trinitrobenzene suIfonic acid (Picryl sulfonic acid) TNP Trinitrophenyl CHAPTER 1

Introduction 1.1 GeneraI Introduction Healthy organisms stave off infections by rneans of their immune systems. An important part of this process involves a population of proteins that is able to differentiate among foreign substances and change in response to the invader and the number of challenges mounted by the invader. Immunoglobulins. or antibodies as they are comrnonly referred to, belong to this arm of the immune system known as 'adaptive' or 'specificf immunity. Using these proteins. the immune system is able to distinguish 'self from 'non-self. The term 'self, refers to tissues, cells, proteins and other structural materiais that are derived from a geneticdy identical source. By binding to what the immune system considers 'non-self, imrnunoglobulins target these foreign substances for destruction by elaborate biochernical pathways such as the complement system- Of the five known classes of irnmunoglobulins, only IgM and certain subclasses of IgG can initiate the complement pathway known as the classical complement cascade. Despite several decades of research in this field, scientists are far from having a complete picture of how this process occurs. Histoncally, a fundamental paradox in the field of irnrnunology was the coexistence of antibodies and effector systems. If antibodies are able to indiscriminately recruit and activate effector systems (e.g., complement), the potential for them to effect massive destruction in an organism is great. Fortunately it appears that in many cases antibodies are ineffectua1 unless they are bound to a target antigen. What then is the transformation brought about by this interaction that enables otherwise inert antibodies to activate such potent immune responses in a controiIed and localized way? In 1974, 1978 and again in 1983, Metzger summarized the experimentai evidence for and against the various models proposed. According to the 'associative' model, the antigen serves only to array the immunoglobulin molecules in such a way as to facilitate multivaient engagement of the first component of complement, an event which is believed to bring about its activation. To date, the bulk of the evidence supports this mode1 of antigen- mediated complement activation by the IgG class of immunoglobulins. In both the 'distortive' and 'aiiostenc' models, which have been invoked to explain the activation of complement by IgM, an interaction between the antigen and one end of the antibody results in the expression of a complement binding site elsewhere in the same molecule. Before these models are discussed in further detail, a clearer understanding of the structure of these immunoglobulins, the complement proteins involved and the mechanism by which complement is believed to be activated is necessary. What follows is a description of key proteins involved in this initiation step (see figure 1.1) with special emphasis on the structural similarities and differences between IgG and IgM and a subsequent discussion of these models and the project rationale.

1.2 Immunoglobulin Structure The overall subunit structure varies little from immunoglobulin class to class. In general, imrnunoglobulins are covalent complexes containing two pairs of distinct polypeptides known as the light (or L) and heavy (or H) chains. A single naturally occumng irnmunoglobulin will have only one of two types of Light chains, r or A, and one of several types of heavy chains. lmmunoglobulins are defined by their heavy chains and five types, y,p>,cr, and E correspondhg respectively to the IgG, IgM, IgD, IgA, and IgE classes have been identified. Certain classes such as IgG are further subdivided into irnrnunologically distinct but structurally related subclasses. For example, the human IgG class is subdivided into the IgG1, IgG2, IgG3 and IgG4 subclasses. In normal human sera IgG is the most abundant immunoglobulin at a serum concentration of 8-20 rng/ml, followed by IgA and IgM at 0.5-2 rng/ml each, IgD at 0.4 mghl and IgE at 0.002 mghl (Spiegelberg, 1974). A description of human IgG1, the most abundant IgG subclass, will be given followed by a description of IgM using the structure of IgGl as the reference point. Figure 1.1. Proteins involved in the initiation step of the classical complement cascade. Shown are the small sphere models for IgG, IgM, Clq, Clr, C 1 s and cI-Inh from the solution scattering analyses of Perkins et al. (1 990, 1991). C lq, Clr and Cls are complement proteins. ~hhis a regulator of the classical complement cascade (see text for details). 1.2.1 IgG structure 1.2.1.1 Quatemary structure Human IgGl is a covalently Linked tetramer which has a sedimentation coefficient of 6.7s. It consists of two identical heavy chains and two identical light chains of MW 50 000 and 25 000 respectively (see figure 1.2). Each light chain is paired with one of the heavy chains through a single disulfide bridge. This minimal unit is termed the 'halfmer'. Two halfmers joined by any number of inter-heavy chain disulfide bridges, two in the case of IgGl, form the basic monomeric subunit. Whereas each light chain is made up of two discrete globular domains, VL and CL, comprising the N-terminal and C-terminal halves respectively, each heavy chain of IgG (Le. the y-chain) consists of four such domains, VH, Cyl, Cy2 and Cy3. In IgG 1, the H-L interchain disulfide bond links cysteine 220 of the y-chah, situated between the Cyl and CyZ domains, to cysteine 214 located C-terminal to the CL domain of the light chain (Jeske and Capra, 1984; Kabat, 1987). In this respect, human and mouse IgGl are considered exceptions rather than the rule as most immunoglobulin heavy chains are linked to the light chah at position 13 1 between the VH and CHI domains. Another exception is the A2m(l) allotype of human IgA2 in which the light and heavy chains are held together only by non- covalent interactions and the light chains thernselves are linked to each other. A stretch of amino acid residues termed the 'hinge' lies between the Cyl and Cy2 domains of human IgG1. In X-ray crystallographic studies, these hinge residues are marked by a region of low electron density indicating that they form a highiy dynarnic structure (Marquart et al., 1980). Two pairs of cysteine residues in the hinge region participate in the inter-heavy chain disulfide bonds. The 'genetic hinge' and the various globular domains of the imrnunoglobulin, with the exception of IgA, are encoded by separate exons. The 'structural hinge' is an extension of the genetic hinge by the inclusion of seven Cy2 exon-encoded amino acid residues ending at Pro 238, the first discernible amino acid in the X-ray structure of IgGl (Deisenhofer, 198 1). Since the hinge heavy light chah chain \

intradornain ' S-S bond

\ Papain site inter heavy -Pepsin site and light chah 1

fragment 1

Figure 1.2. The covalent structure of IgG. Shown is the tetrameric H2L2 structure of human IgG1. The unfilled bars represent the variable domains of both the heavy and light chahs. The shaded bars represent the constant domains of the light chains. The location of the heavy chain constant domains Cyl, Cpand Cy3 are indicated on the heavy chain to the right. The inter heavy and light chah disulfide bonds and the inter heavy chah disulfide bonds are depicted as shaded lines. The intradomain disuIfide bonds are also shown, represented by the shaded arcs. The solid bars represent the constant domains of the heavy chains. Enzymatic digestion of the IgG with papain, whose major cleavage site occurs N-temiinal to the first inter heavy chah disulfide bond, results in the fragmentation of the molecule into two Fab fragments and one Fc fragment. The pepsin site is located C-terminal to the inter heavy chah disulfide bonds. As a result, pepsin cleavage results in the formation of a Fab'î fragment. region of the molecule is less compact than the other domains, it is particularly susceptible to proteolytic cleavage by many enzymes. Papain, one such protease, cleaves the y-chain between the Cyl domain and the first interheavy chah disulfide bond to produce two functionaliy distinct fragments, Fab and Fc, at a ratio of two to one. The fragment which consists of the VL, CL,VH, and Cyl domains retains the antigen binding activity of the original molecule, hence the name 'Fab' for 'fragment - antigen binding'. The structural hinge pennits the IgG molecule considerable segmental flexibility alIowing the Fab regions to move independenùy of each other and of the Fc region. This was clearly demonstrated in the electron micrographs of IgG which showed that the angle of the Fab arms with respect to each other could extend anywhere from O' to 180' (Valentine and Green, 1967). The effector functions, such as complement activation and binding to the Fc receptor, reside within the remaining Cy2 and C'y3 domains which make up the 'Fc' fragment or

'fragment - crystailizable' based on the observed ability of this fragment to form crystalline precipitates when dialyzed against water at 2°C (Porter, 1973).

1.2.1.2 Secondary and teliiary structure Our current understanding of the secondary structures in these globular domains is based upon the detailed crystal structures determined for human IgG Fab fragments, Fc fragments, Bence-Jones light chah dimers and whole IgG (Schiffer et al., 1973; Amzel et al., 1974; Huber et al., 1976; Marquart et al., 1980; Deisenhofer, 1981; Sutton and Phillips, 1983; Ely et al., 1985; Scbiffer et ai., 1989; Harris et al., 1992). The recent X- ray structure solution for the complete munne IgG2a molecule (Harris et al., 1992) shows that the structure of the imrnunoglobulin domains deterrnhed for IgG fragments and iight chain dimers are consistent with the domain structures determined for a complete molecule. Both variable and constant region domains fold into antiparallel P-pleated sheets that are secured at the centre by an intrachain disulfide bond and a hydrophobic tryptophan 'pin'. Studies of reduced immunoglobulin domains have indicated that the intrachain disulfide bond is important for maintainhg the stabiiity of the domain (Isenman et ai., 1975; Goto et al., 1979). The 'immunoglobulin fold', as it is calIed (Poljak, 1973; Schiffer et al., 1973), resembles a flattened cylinder. Constant region domains have a three-stranded outward- facing 'Q' face and a four-stranded inward-facing 'fi' face as depicted in figure 1.3. Variable domains have two additionai strands on the face that face inward instead of outward and have a looser three-dimensional structure. Significant non-covalent interactions between polypeptide chains of the IgGl rnolecule help the tetramer maintain its quatemary structure under conditions of mild reduction and akylation (Romans et al., 1977). Trans-interactions, for example, occur between homologous domains and involve hydrophobic interactions, electrostatic interactions, Van der Waals forces, hydrogen bonds, and sait bridges. Substantial contacts have been found between the VL and VH domains, between the CL and Cyl domains, and between the two Cy3 domains (reviewed by Padlan, 1994). In the human IgGl protein

Kol, for example, the interaction between the CL and Cyl domains is stabilized by the sait- bridges connecting Glu 123 (CL) to Lys 2 13 (Cyl) and Glu 124 (CL) to Lys 147 (Cyl). Pairs of Cy3 domains are held together by interactions involving as many as twenty residues from each domain. In IgG, Cy2 is the only domain which does not participate in inter-chain interactions. The reason for this is that Cy2 harbours a bulky carbohydrate branch at Asn 297. This moiety covers the hydrophobic 'fx' face and prevents contact between the C'y2 domains.

Some interactions have been found between neighbourïng dornains of the same chah These cis-interactions involve smalier contact areas than trans-interactions. The interface between Cy2 and Cy3, for exarnple, involves a total of seventeen residues and covers only 780 A2 whereas the interface between a pair of Cy3 domains covers a surface area of 2000 A2. Cis-interactions have also been reported between residues 13 1 to 139 of Cyl and the N-terminus of the hinge (Marquart et al., 1980). Face X 1x3 A (fx)

Figure 1.3. A ribbon diagram of the Cy2 domain illustrates the immunoglobulin fold. The shaded regions represent p-strands on the three-stranded, outward facing y-face whereas the unshaded regions represent p-strands on the four- stranded, inward-facing x-face. The P-strands of each face are numbered in the order of their appearance in the amino acid sequence starting fiom the N-terminus. Reproduced from Beale and Feinstein (1976). 1.2.1.3 Primry structure The entire amino acid sequence is known for al1 subclasses of IgG in human, mouse, rat, and rabbit. In humans, the subclasses are IgG1, IgG2, IgG3 and IgG4; in mouse, they are IgGl, IgG2, IgG2b and IgG3. The residue numben for IgG referred to in this thesis are based upon sequence alignments with the y-chain of the Eu human myelorna IgGl protein ('Eu assignment' of y-chain amino acid residues; Edelman and Gall. 1969).

As is the case for aü imrnunoglobulins, the fit 110 N-terminal amino acid residues of both light and heavy chains exhibit the highest degree of variabiiity, hence they are calIed the light and heavy chah variable domains, abbreviated 'VL'and 'VH'respectively. In general, both light and heavy chains contribute residues to the antigen binding site. Those residues are positioned by a scaffold of highly-conserved residues. In the primary sequence these conserved residues occur grouped together in four distinct blocks termed the framework regions. Altemating with the frarnework regions are four regions of hypervariability (Wu and Kabat, 1970). The first, second and fourth regions of variability (it is believed that the variability of the remaining hypervarïable region simply reflects the genetic polymorphism of the V-region) are called the 'Complementarity Detemiining Regions' or CDR's. They have been shown to be present in the antigen combining site and thus together define the antigen-binding specificity of the given Fab (Capra and Kehoe, 1975), although an antigen may not necessarily interact exclusively with residues of the CDR and may also interact with some part of the framework . Conversely, not al1 residues in the CDR are necessarily involved directly in antigen-antibody interaction. Among subclasses of human IgG, the primary sequence identity between homologous constant globular domains is greater than 95% (Kabat, 1987; Michaelsen et al., 1992). By cornparison, the hinge sequences have much less in common. Between the mouse IgGl and IgG2b subclasses, the percent identity between their respective Cyl, hinge, Cy2 and Cy3 domains are 854, 3195, 66% and 63% (Yamawaki-Kataoka et al.,

1981). As shown in Table 1.1 the hinge varies not only in sequence, but in length and Isotype HlNGE SEQUENCES <---- Genetic hinge --- > Upper hinge Core hinge Lower hinge Human IgGl EPKSCDKTHT CPP CP APELLGGP Human IgG2 ERK CCVECPP CP APPVAG P Human IgG3 ELKTPLGDTTHT CPRCP (EPKSCDTPPPCPRCP)3 APELLGGP Human IgG4 ESKYGGP CPS CP APEFLGGP Mouse IgG 1 VPRDCG CKPCI CT VPSEVS Mouse IgG2a EPRGPTIKP CPPCK CP APNLLGGP Mouse IgG2b EPSGPISTINP CPPCKECHK CP APNLEGGP Mouse IgG3 EPRIPKPSTPPGSS C P PGNILGGP

Table 1.1. Hinge sequences for human and mouse IgG isotypes. The hinge sequences are aligned relative to residues 2 16 to 238 of human IgG 1. The upper and core hinge sequences are encoded by at Ieast one exon and are collectively referred to as the 'genetic hinge'. The lower hinge residues are encoded by the 5' segment of the Cy2 exon. The genetic hinge and the lower hinge together form what is known as the 'structural hinge'. The upper, middle and lower hinges are defmed according to Bede and Feinstein (1976) and Endo and Arata (1985). number of interheavy chah disulfide bonds. Metzger used this observation in 1978 to argue against the dosteric mode1 of complement activation by IgG stating that "...if antigen binding to the Fab region could trigger a change in the Fc conformation directly, then the cntical link between the Fab and Fc regions might have been conserved." Another reason to take note of these differences in hinge structure is that it is important in a forthcoming discussion regarding the potentiai role of the hinge in regulating complement activation by the different IgG subclasses.

1.2.1.4 Glycosyiution As mentioned earlier, the carbohydrate groups on the second constant domains of IgG separate the opposing Cy2 domains within the same molecule from each other. The oligosaccharide, which accounts for 2-3% of the molecular weight, is linked to the conserved asparagine at residue 297 of the Asn-Ser-Thr triplet and varies not only from IgG to IgG, but even from one heavy chah to another within the same rnolecule. Initially, this heterogeneity and asymmetry made it mcult for researchers to determine the overall structure of this moiety, but advances in protein separation techniques coupled with glycosidase digestion analysis eventually allowed scientists to analyze the population of carbohydrate structures on rabbit and human IgG (Rademacher et al., 1983; Parekh et al., 1985). Parekh et al. showed that there are at least thirty distinct carbohydrate structures found on human IgG molecules that can be categorized into neutral, monosialylated and disialylated sugars. The existence of so many different variants of the CP-carbohydrate structure led them to regard the IgG 'rnolecule' as a population of glycoforms. An examination of these carbohydrate structures showed that they are ail tnincation variants of the structure shown in fi,we 1.4A, they are all biantennary cornplex-type oligosaccharides and they all contain the same core structure. The core stnicture, rnarked with "*" in figure 1.4A, consists of a trimannosyl unit P-linked to chitobiose (a dimer of N-acetyl-p-D- glucosamine). Some of the 'non-standard' features include a bisecting GlcNAc residue * Sia a(2,6) - Gal p(l14)GlcNAc P(1,2)-Man a(1,6) I * * a GlcNAc $(1,4)- Man P(1,4)-GlcNAc B(1,4)-GfcNAc P.

Fuc a(1,6) sis a(2.6)- Gal P(1,6)-GlcNAc f3(1,2)-Man a(1,3)

Figure 1.4. The carbohydrate structure of the human y-chah asparagine- linked oligosaccharide. Shown in A is the predominant carbohydrate structures of human IgG at Asn 297 (Parekh et ai., 1985). The "*" designates carbohydrate units which are common to aii structures found at this position. Fuc: fùcose. Gd: galactose, GlcNAc: N-acetylglucosamine~Man: mannose, and Sia: sialic acid. Also indicated are the LUikages. The three-dimensional orientation of the Asn 297 carbohydrate is shown in B with respect to the CyZ domains (Dweket al., 1984). The P-strands on the four-stranded inward facing x-faces are striped. linked P(1-4) to the central mannose unit and a fucose linked a(1-6) to the peptide-linked

GlcNAc. It has been suggested that the presence of the different carbohydrate structures in IgG reflect the intracellular availability of the various glycosyl transferases and glycosidases involved in the carbohydrate biosynthesis process (Parekh et al.. 1985). 1H NMR studies of the free oligosaccharide structures (Gettins et al., 1981) showed that each of the various linkages was limited to a single three-dimensional conformation with the exception of the a(l-6)linkage that forms one of the arms of the

biantennary structure. At this position, it was found that the arm could assume two conformations. X-ray crystallography of an intact Fc fragment (Sutton and Philiips, 1983) confmed the NMR structure determinations produced by Gettins et al. (198 1). Sutton and Phillips (1983) were able to further resolve the relationship between the carbohydrate structure and its associated immunoglobulin dornain (see figure 1.4B). According to these results the a(1-6) arm is immobiiized by its interaction with the hydrophobic residues Phe 243 and Pro 246 and the polar residue Thr 260 on the fint two P-strands of the four- stranded face of the attached domain. The a(1-3) arm of one oligosaccharide makes contact with the core of the opposite oligosaccharide strand. As a probable consequence of steric constraints, this particular arm always lacks a galactose unit. Glycosylation of the two Cy2 domains is therefore asyrnrnetric. The a(1-3) mof the other oligosaccharide projects into the space between the two domains formed by the interaction between the fit arm and the opposite domain and thus appears not to have the constrainü Iirniting the size of the first arm. It therefore follows that at least half of the Cy2 oligosaccharides should

lack the lengthening galactose unit. Analysis of carbohydrates in the rabbit IgG population has been consistent with this prediction (Rudd et ai., 199 1). Although there is great variation in the carbohydrate structures of IgG, the ratio of the various glycoforms in the polyclonal IgG population of normal human semm is surprisingly constant and only varies in certain disease states (Parekh et al., 1985). The levels of the highly galactosylated IgG species, for example, are consistently low in normal human sera but are elevated in the sera of patients suffenng from rheumatoid arthritis. Because of this finding, glycoform profiles are now being used in the diagnosis of this disease. Furthermore, circurnstantial evidence suggests that the highly-galactosylated IgG species is the antigen responsible for the condition. Indeed, when pathological immune complexes were isolated from arthritic patients they were found to contain only IgG and rheumatoid factor IgM (Winchester et al., 1971). The interaction between the IgM and the IgG has been locdized to the Cfl domain although there appear to be no changes in the peptide sequence of the IgG (Nardella et al., 198 1). Extensive study of this interaction has indicated that many of the amino acid residues on CFwhich bind to rheurnatoid factor IgM are also involved in the interaction between IgG and staphylococcal protein A (Nardella et al., 1985; Schroder et al., 1986). Together these results suggest that the carbohydrate is not directly involved, but instead dters the availability of cntical arnino acid residues in the imrnunoglobulin. Interestingly, the temporary remission of rheumatoid arthritis reported by arthritic women during pregnancy was found to coincide with the pregnancy-associated decrease in the relative level of the highly galactosylated IgG species (Rook et al., 199 1). Following delivery, the levels returned to pathological levels concomitant with post-parturn recurrence of disease symptoms. Although removd of the oligosaccharide moiety affects several important imrnunoglobulin functions such as binding to Fc receptor, binding to complement and antibody-dependent cellular cytotoxicity as well as its in vivo stability (Leatherbarrow et al., 1985; Tao and Morrison, 1989; Wawrzynczak et al., 1992; Wright and Morrison, 1994), determining its precise role in these effector function has been made difficult because of the important role it plays in the structure of the Fc. To date, no specific function has been assigned to the oligosaccharide moieties in irnmunoglobulins.

2.2.2 IgM structure By cornparison with IgG, IgM has severai distinctive features. Normal senim IgM exists primarily as a large 950 000 MW pentamer whereas IgG exists only as a monomer. Furthemore IgM pentamen are usualiy covalentiy associated with another polypeptide chain known as J chain. The IgM heavy chah (p-chain) has four globular constant domains (Cp1, Cp2, Cp3 and Cp4) as opposed to the y-chain which has only three (see

figure 1.5). The Cp4 domain of the p-chain has an additional C-terminal 19 amino acid sequence calied the 'p-tail'. Lastly, the p-chain is more heavily glycosylated than the y- chain.

1.2.3.1 Quaternary structure The predominant IgM species is a covalently Linked closed-ring polymer consisting of ten p-chahs, ten Iight chains and a single J chain. in human IgM, cysteine residues at

positions 337 in Cp2, 414 in Cp3 and 575 in the p-tail form the inter p-chain disulfide

bonds (Beale and Feinstein, 1969). C575 also forms the disulfide links between IgM and J chah (Koshland et al., 1985; Davis and Shulman, 1989b; and Frutiger et al., 1992). Inter heavy-chah disulfide bonds are believed to form between 'homo pairs' (e.g., between

C337 residues, but not between C337 and C414 residues) as deduced from studies involving the partial reduction of IgM, site-directed mutagenesis and X-ray crystal structure-based models of the IgM polymer (reviewed in Davis and Shulman, 1989b). Since studies suggest that strong non-covalent interactions exist between the Cp2 domains, the IgM monomenc subunit is therefore defmed by convention as the pair of HL halfmers that are joined by the 337 disdfide bond (reviewed in Davis and Shulman, 1989b).

It is not clear whether the 414 disulfide bond is normally formed in mouse IgM. Milstein et al. (1975) were unable to detect any disuEde bond formation at the 414 position of mouse IgM polymers even though they were able to demonstrate that both 337 and 575 disulfide bonds were formed. Furthemore it has been shown that the 414 cysteine residue is not necessary for polymerization since S414, a mutant IgM with a serine to cysteine substitution at that position is able to form covalent polymers @avis and Shulman, 1989b; Figure 1.5. A schematic diagram of the IgM and IgG monomeric subunits. Two of the distinguishing features of IgM are iliustrated - the additional constant domain and the ptail. The inter heavy chah disulfide bonds are shown as arrows and the carbohydrate moieties are represented by branched structures. Not shown are the disuKide bonds linking the light and heavy chahs. Wiersma et al., 1995). At the same time however both 1gM mutants S337 and S575 dso form covalent polymers which indicates that a disulfide bond can form between C414 residues. The extent of polyrnerization in these two mutants however is much less than that in wildtype IgM or the S414 mutant. Assuming that the serine to cysteine substitution at any one of the three positions has no effect on disulfide bond formation at the rernaining two positions, this observation suggests that the 414 disulfide bond is formed at a relatively low frequency in normal mouse IgM polymen. Davis and Shdman (1989b) have suggested a possible role for the 414 disulfide bond. Because the largest S414 polymer contains only five rnonomeric subunits, they postulated that the formation of the 414 disulfide bonds makes the growing polymer more

&id thereby decreasing the effkiency of ring closure and encouraging the formation of the larger hexarners. It is becoming evident that IgM exists as a mixture of polymers with several disulfide bonding pattems. Despite this heterogeneity, there exist preferred patterns. The terms 'in series' and 'parallel' are used to descnbe them (see figure 1.6B). Two disulfide bonds are 'parallel' if they connect the same two heavy chains but are 'in senes' if they connect one heavy chain to two different heavy chains. Both 414 and 575 disulfide bonds form pnmarily in series with the 337 disulfide as deduced from the partial reduction and site-directed mutagenesis studies (Milstein et al., 1975; Davis et al., 1989; Davis and Shulman 1989; Wieema and Shuiman, 1995). An example of a disuifide bonding pattern which satisfies these critena is shown in figure 1.6A. Interestingly, a recent study suggests that each p-chah is linked to three other p-chains in IgM hexamers (Wiersma et ai., submitted). The p-tail is necessary for IgM polymerization (Baker et al., 1987; Davis et al.,

1989a). Studies have indicated that it is also sufficient for polymerization since the transplantation of the p-tail onto the C-terminal end of the y-chah has resulted in the formation of IgG polymers (Smith et al., 1995; Sgrensen et al., 1996). In contrast, other B "parallel" "in series"

Figure 1.6. A modei of the IgM pentamer and types of disulfide bonding patterns. A, one proposed structure of pentameric IgM with the J-chah in which both 414 and 575 disulfide bridges are 'in series' with the 337 disulfide bridge (hidden by the overlap between the Cp2 domains). The diagrams in B illustrate two possible bonding patterns for the 414 cysteines in relation to the 575 disulfide bond. At the left, the Cys 414 disulfide bond is 'parallel' to the Cys 575 disulfide bond whereas at the right, the Cys 414 disulfide bonds are situated 'in series' with the Cys575 disulfide bond. Refer to the text for further detds. studies have indicated that the y-tail is not suficient for polymerization (Sitia et al., 1990; Wiersma et al., 1997). In particular Sitia et ai. (1990) observed that IgG with a p-tail was

secreted only as a monomer and furthemore most of the immunoglobulin was retained intracellularly. On the bais of this latter observation, they postuiated that the p-tail contains an endoplasrnic reticulum (ER) retention signal. Although S~rensenet al. (1996) were able to obtain IgG polymers by the C-terminal addition of the y-tail, they too

observed this intracellular retention of the mutant protein. Wiersma et al. (1997) have suggested that the differences in the results obtained by these groups is due to the source of the p-tail and to the length of the p-tail containing addition.

The extent of polymerization appears to be controiied by determinants other than the p-td. Sgrensen et al. (1996) showed that IgM with the a-tail of IgA, another

irnrnunoglobulin which also has a tailpiece but is secreted prirnarily as dimen and trimers. has a polymerization profile characteristic of wildtype IgM and not the IgA. These results suggest that although the y- and a-tails are necessary for polymerization, other stmctural

features of the remaining molecule determine the extent of polymerization.

1.2.2.2 J chuin The J chah is a 137 residue polypeptide having an approximate molecular weight of 15 000 and a single complex carbohydrate structure (Brandtzaeg, 1985; Koshiand, 1985). The J chah amino acid sequences have been determined for human, mouse and rabbit, and are highly conserved among these species (Mole et al., 1977; Max and Korsmeyer, 1985; Matsuuchi et al., 1986; Hughes et al., 1990). The primary structure alternates between regions of high hydrophobicity and regions of high hydrophilicity. There are eight cysteine residues dispened throughout the length of this particularly acidic polypeptide. As yet, no three-dimensional structure has been determined for J chain. Based on the circular dichroism studies and cornputer assisted analysis of the primary sequence, a single domain mode1 of J chain with an eight-stranded antiparallel p-barre1 structure has been proposed (Zikan et al., 1985). A two-domain mode1 has also been proposed (Cam et al., 1982). Although J chah is disulfide linked to the p-tail via the C575 residue, additional

structures must be present for J chah incorporation. Smith et al. (1995) found that IgG with a p-tail addition was able to polymerize, but the polymers were not associated with J chain. To identiS the domain requirements for J chah incorporation, Wiersrna et al. (1997) assembled a panel of IgMlIgG hybrid immunoglobulins in which Cp1. Cp2-Cp3 or Cp4 domains were replaced by their IgG counterparts Cyl, y-Hinge-Cy21 and Cy3 respectively. When they examined the polymeric species of these hybrids for J chah content, they found that Cp4 but not Cpl, Cp2 or Cp3, was necessary for J chah incorporation. Since the glycosylation sequon in the p- and a-tails of the J chain- associated immunoglobulins IgM and IgA is highly conserved among species, Wiersma et al. (1997) speculated that the degree of J chah incorporation into the IgM polymer might also be influenced by the presence of the p-tail glycan. When they destroyed the Asn-X- Ser glycosylation sequon in the p-tail by mutating residues N563 and S565, they found that the resulting population of polymeric IgM had a reduction in J chah content. It is generally accepted that the J chain regulates the extent of IgM polymerization.

In the presence of J chain, IgM is synthesized and secreted primarily as J chah-associated pentamers whereas in the absence of J chain, hexarners are predominantly produced (Cattaneo et al., 1987; Randail et al., 1992; Davis and Shulman, 1989; Brewer et al. 1994; Niles et al., 1995). Recent studies have demonstrated that the regulatory effect of the J chain requires disulfide bond formation at the 414 position (Fazel et al., 1997). Fazel et al. found that the S414 mutant fonned pentarners and tetramers in the same relative proportion regardless of J chah presence. Wiersma et al. (submitted) have reported the existence of J chain-deficient wildtype IgM pentamer. This is an interesting fmding since it suggests that other factors such as interactions with intracellular polymerizing enzymes (Roth and Koshland, 198 1) are involved in regulating IgM polymer size.

- [Thecorresponding hybrid is also descnbed in this thesis under the name of 'pyp'. 1.2.2.3 Secondary and rertiary structure Thus far, there are no published X-ray crystal structures of IgM polymer, its monomeric subunits or its proteolytic fragments. The sole existing structural data are in the fom of EM studies and they provide only a low resolution picture of this class of immunoglobulin (Feinstein and Munn, 1969; Feinstein et al., 1971; Feinstein and Richardson, 198 1). Although discrete globuiar domains correspondhg to the constant and variable domains could be seen in the electron micrographs, further details of the structure couid not be discemed. When the arnino acid sequence was determined for the human IgM myeloma protein 'Ou' (Putnarn et al., 1973), Beale and Feinstein (1976) found evidence that the constant domains of IgM assume the same structure as the constant domains of IgG. Most importantly, both the periodicity of the cysteine residues involved in the intradornain disulfide bridges and the Trp 'pin' residue were maintained. Furthemore, although the sequence similarity between any pair of y- and p-domains was less than 3095, hydrophobie residues were found at aiternating positions dong the heavy chah sequence suggesting that the structure involved mostiy P-pleated sheets, as was the case for y constant domains. The sequence identity was highest between the Cp3 and the Cy2 domains, whereas the Cp4 domain best resembled the Cy3 domain of IgG. This cornparison was further supported by the positioning of the carbohydrate chahs. The Cp3 domain was glycosylated at Asn 402 which, upon sequence alignment, was analogous to the Asn 297 glycosylation site in human IgGl.

1.2.2.4 Prirnary structure The amino acid sequence has been determined for human, mouse, and sheep IgM (Putnam et al., 1973; Kehs, et al., 1982; Patri and Nau, 1992). The 'Ou' assignment of p- chah constant domain arnino acid residues will be used in this thesis (Putnam et al., 1973). It is narned derthe human IgM Ou myeloma protein that was isolated from a patient with macroglobulinemia. 1.2.2.5 Glycosylation IgM is one of the most highly glycosylated immunoglobulins. It contains five sites of glycosylation that account for 16% of the total p-chah mas. Using high resolution IH-

NMR,Anderson et al. have identified the residues to which the sugar moieties are attached and have also determined their structures (Anderson and Grimes, 1982; Anderson et al., 1985). The carbohydrates are attached to residues Asn 17 1 of Cpl, Asn 332 of CpZ, Asn

364 and Asn 402 boîh of Cp3, and Asn 563 of Cp4. As in the case of IgG, there is some degree of heterogeneity at each of these sites. Shown in figure 1.7 are the predominant structures. The carbohydrate at Asn 171 is a complex, highly siaiylated biantennary structure whereas those at positions 322, 364 and 402 are complex highly sialylated triantennary structures. The polysaccharide at position 563 is a high mannose structure. Although it is commonly held that the sugar at Asn 402 is analogous to the Asn 297 carbohydrate of IgG, the orientation of the Asn 402 sugar as well as the others with respect to their associated domains have not yet been determined. Secreted IgM monomers have differentiy processed sugars than do IgM polymes (Davis et al., 1989; Sitia et al., 1990). When analyzed by SDS-PAGE, the mobility of heavy chains derived from IgM monomers was less than that from polymers. Cals et al. (1996) showed that this decrease in mobility was due to the carbohydrate at position 563 of the Cp4 domain and furthemore demonstrated that this carbohydrate had a different branch structure and was more galactosylated than that from the polymer-denved heavy chains.

On the basis of these results, Cals et al. suggested that the Asn 563 of the monomer was less accessible than that of the polymer to the cellular machinery that trims and processes these carbohydrate moieties.

1.2.3 Phylogeny of immunoglobulins Predictions as to the origin and evolution of the immune system remain a highly speculative field. It has been suggested that the immunoglobulins appeared after the Figure 1.7. The five carbohydrate structures of mouse p-chain asparagine- Iinked oligosaccharides. Shown are the predominant structures at each glycosylation site for mouse MOPC L04E myeloma IgM protein as determined by IH NMR spectroscopy. The approximate locations of the glycans dong the heavy chain are also shown. The oligosaccharide at position Asn 402 is believed to be analagous to the oligosaccharide at position Asn 297 of IgG. Fuc: fucose, Gai: galactose, GlcNAc: N- acetylglucosamine, Man: mannose, NG: N-glycolylneurarninic acid. Also indicated are the linkages. Adapted from Anderson et al. (1985). NG a(2,6)-GaI/3(1,4)-GlcNAc $(1P)-Man a(1.6) \ Man p(l.4)-GlcNAc B(1 A)-GlcNAc $- NG a(2.6)-Gal P(1 ,~)~~CNACB(1.2)-Man a(1.3) 1 Fuc a(1.6) /

NG a(2.6)GaI P(1.4)-GlcNAc p(l.2)-Man a(1.6) \ NG a(2.6)-Gal f3(1,4)-GlcNAc p(1.4) Man P(1A)-WNAc B(1 .4)-GlcNAc p- Mma(l,3) 0 / > Fuc a(1.6) NG a(2.6)-Gal$(1,4)-GlcNAc f3(1,2)

NG a(2.6)-Ga1 p(1,4)-GlcNAc f3(l,a)-Man a(1.6) \ \ NG a(2.6)-Ga1 p(1.4)-GlcNAc p(1.4) Man $(1A)-GlCNAc p(1 ,4)-GIcNAc B- \ 1 / g' Man a(1 g3) Fuc a(1.6) NG a(2.6)-Ga1 p(1A)-GlcNAc p(1.2) /

NG a(2,6)-Gal P(1.4)-GlcNAc $(1,2)-Man a(1.6) \ NG a(2.6)-GalP(1 ,4)-GlcNAc p(1.4) \ Man p(lA)-GIcNAc p(1 .4)-GIcNAc ,Man a(1.3)~ Fuc a(1.6) 0 NG a(2.6)-Ga1 P(1,4)-GlcNAc B(1.2)

Man a(1,6) - -\ Man $(1 ,4)-GlcNAc b(1.4)-GIcNAc / Man a(1,3) vertebrates and invertebrates diverged since immunoglobulins have not yet been found in the latter. In vertebrates, the number of immunoglobulin classes ranges from two to nine depending upon the species or family. Membes of the elasmobranch family (e.g., sharks, skates and rays) have only two immunoglobulin classes, arnphibians have two or three classes, birds have three and mamals have the highest number with as many as nine classes (counting the subclasses of IgG and IgA) in some species such as humans. IgM is comrnon to dl species having immunoglobulins. In species having more than one immunoglobulin class, IgM is the only class of immunoglobulin which increases appreciably in titre in response to a prirnary challenge with antigen. It is therefore believed that the other immunoglobulin classes evolved subsequently to the appearance of IgM.

1.2.4 The 'Domain Eypothesis ' The observation that multiple copies of a - 110 amino acid sequence were present and highiy conserved in both the IgG heavy and light chain sequences led Edelman and Gall (1969) to propose what would become known as the 'Domain Hypothesis'. According to this hypothesis, the light chahs should have two confomationaily similar domains and the y-chahs should have four such domains (Edelman and Gall, 1969). They also suggested that each domain evolved to acquire a separate and independent function. The structurai aspect of this hypothesis has been shown to be me. The functional aspect of this hypothesis has, however, found lirnited application even though it has been useful in conceptualizing the division of labour among the domains of this multifunctional protein. One function for which Edelman's hypothesis does hold tme, however, is the initiation of the complement system by IgG which has been shown to be confined to the second constant domain of the gamma heavy chah ((29).

1.3 The Classical Complement Cascade Cornplement is the name given to a system of semm proteins involved in a cascade which when activated enables the organism to mount a response against its target. The best known pathways involving complement proteins are the classical complement cascade, named so because it was the fmt pathway discovered, and the alternative pathway. More recently a third pathway involving rnannan binding proteins (MBP) has also been described. The classical and alternative pathways, which are best explained with the aid of the diagram in figure 1.8, begin with separate initiation steps, proceed through a series of amplification steps which enhance the response, and fmaliy converge at a common series of temiinal steps. In these last steps, a vansmembrane pore known as the 'membrane attack cornplex' (MAC) is formed through the cellular membrane of the target celi, thereby killing it. Since the classicai pathway and its initiation by imrnunoglobulins is the focus of this thesis, what foiIows is a bief description of its components. The proteins of the classicd cornpiement pathway include several inactive precursors of proteolytic enzymes which become active upon peptide bond cleavage (i.e. they are zymogens). Two of these zymogens, Clr and Cls, are subunits of the first component of complement. Duhg initiation of the classicai complernent pathway, they

become cleaved to form the corresponding active serine proteases, cT~and C~S.C~S binds to and cleaves C4 the next protein in the cascade (the proteins were narned in order of discovery which explains why this second component of the classical pathway is named 'C4' and not 'C2') into C4a and C4b. Nascently activated C4b acylates nearby protein or carbohydrate by means of its newly exposed and highly reactive thioester bond. C2 is then recruited by C4b and cleaved into C2a and C2b by C~S.C2b is released whereas C2a remains cornplexed with C4b to form what is known as the classical pathway C3 convertase. C3 binds to the C3 convertase and is cleaved to C3a and C3b. C3b then binds to C4bC2a to fomi the heterotrimeric CS convertase (C4bC2aC3b). It cm also transacylate to the target surface by means of its newly exposed thioester and can act as a new site for the initiation of the alternative pathway. When CS is cleaved by the C5 convertase, complement proteins C6, C7 and C8 are sequentially recruited to the site which in turn Classical Complement Cascade

C3 C5 convertase convertase

Figure 1.8. The major pathways of complement activation. The classicd complement pathway and the alternative pathway are show converging with the formation of the membrane attack complex (MAC). "H20"marks hydrolyzed C3. facilitates the polymerization of C9 into membrane attack complexes. This cascade-like mechanism occurs also in the fibrinolysis and blood coagulation pathways (Ga1 et al., 1994). Formation of the membrane attack complex is not the only important outcome of

this pathway, but it is certainly one of the easiest to measure experimentally. In fact the CML assay (complement-mediated lysis) used in the course of this thesis project utilizes the release of hemoglobin from pores formed in lysing red blood ce11 targets to determine the extent of complement activation. Other outcornes of the classical complement pathway include the release of numerous soluble factors that serve important functions. Anaphylactic actions of C3a, C4a and C5a bring about the events associated with inflammation such as vasodilation and the chernotaxis of macrophages and other speciaked cells capable of phagocytosing cellular debris to the site. C3b, which is an important component of the C5 convertase, also functions as an opsonin by coating the target and enhancing phagocytosis. Although this is an intriguing system, the primary subject of this thesis is the activation of the frst step in the cascade and the reader is therefore referred to the reviews by Müiier-Eberhard (1988) and Cooper (1985) and a monograph by Law and Reid (1995) for more detail on the complement proteins downstream from this step.

1.3.1 CI structure The first component of complernent is a complex comprised of three distinct subunits, C lq, C lr and Cls, at a ratio of 1:2:2 whose assembly is ~a2+dependent (Lepow et al., 1963). Cl plays an important role in the initiation of the classical complement pathway. It recognizes and binds to activators such as immune complexes and recruits and induces activation of other complernent proteins. It has been demonstrated that the Clq subunit binds to activators of complement such as antigen-antibody complexes and as a result of these interactions induces the associated Clr and Cls zymogens to undergo proteolytic cleavage transfomiing them into their active forms and GIS. Studies using pMed preparations of Clr and Cls and their corresponding active foms have led Sim et al. (1977) to the conclusion that Clr is activated fmt and in tum activates Cls. C~Son the other hand activates complement proteins C4 and C2 (Haines and Lepow, 1964a,b; Nagaki and Stroud, 1969; Sakai and Stroud, 1974). ~iris not capable of activating C2 or C4 by itself (Sirn et al., 1977). What follows is an overview of Cl structure and function. For further details, the reader is referred to reviews written by several authors (Cooper, 1985; Arlaud et al., 1987; Ga1 el: al., 1989; Arlaud et al., 1993).

1.3.1.1 Clq subunit When purified Clq was subjected to SDS-PAGE analysis under reducing conditions, it was found to consist of three distinct polypeptide chains in equirnolar ratios (Reid et al., 1972; Reid and Porter, 1976). These chains have been designated A, B and C according to their relative electrophoretic mobilities, with A being the slowest moving band. Each chah contains approximately 200 amino acids and has a molecular mass of 23- 24 kD. Since the whole Clq has a molecular mass of 439 kD, it was calculated that Clq contains six each of these three peptides. When Clq was subjected to SDS-PAGE analysis under non-reducing conditions, only two electrophoreticaiIy distinct species were detected (Yonemasu and Stroud, 1972). These were later estimated to have molecular masses of 46-49 kD (Reid and Porter, 1976). Reduction of the larger species yielded A and B chains whereas reduction of the smaiier species yielded ody C chains. Furthemore, the ratio of the larger species to the smaiIer species was 2: 1. Using these results, Reid and Porter deduced that each Clq molecule is cornposed of six pairs of A-B disuIfide linked heterodirners and three pairs of disuifide linked C-Chomodimers (see figure 1.9). A detailed structure of C lq is not available, however a mode1 exists deduced from a combination of amino acid sequence information, susceptibility to proteolytic digestion, electron micrographs and carbohydrate analysis. Clq contains hydroxylysine (2% amino &€?S Mol. Wt. NoJClq

N 1-rn i C Chains NrC

N- N- J C

A-S I 52750 6 Es Subunits

A-S I B-S Structural S-C I subunits S-C A-S I B-S

Intact Molecule

Figure 1.9. The proposed structure for the Clq subunit of Cl. The Clq molecuie is comprised of three types of polypeptide chains named A, B and C which each weigh about 25 kD. The N-terminal regions of these chains are shaded and represent the half of the polypeptide which participates in the formation of the fibril-like stem. This region is coliagen-like and thus susceptible to collagenase digestion. The unshaded regions are those which make up the globular domain and are resistent to coliagenase. The various disulfide linkages are shown. The overall structure of Clq resembles a bouquet of flowers in electron rnicrographs. Adapted fiom Porter and Reid ( 1979). acid content), 4-hydroxyprohe (5% amino acid content) and a large number of glycine residues (Müller-Eberhard, 1968; Yonemasu et al., 1971 ; Calcott and Muller-Eberhard, 1972; Reid et al., 1972). It was also found that the carbohydrate moiety consisted of many neutral hexoses such as glucose and that the glucose was part of a disaccharide with a boalactose unit (Yonemasu et ai., 1971; Calcott and Müller-Eberhard, 1972; Reid and Porter, 1975). These features were quite suggestive since they were characteristic of the basement membrane protein collagen (Calcott and Müller-Eberhard, 1972). Moreover, earlier experiments had shown that the Clq molecule was pariially susceptible to enzymatic digestion by coliagenase (Reid et al., 1972; Knobel et al., 1974). These observations led Reid and Porter (1976) to postdate that C lq had a triple helical structure. In agreement with this prediction both the Clq and its collagenous stalks are capable of inhibiting collagen-mediated platelet aggregation (Cazenave et al., 1976; Suba and Csako, 1976; Wautier et al., 1977). When the amino acid sequences of the three chains A, B, and C were finally determineci (Reid, 1983; Loos et al., 1989; Petry et al., 199 1; Sellar et al., 199 1; Petly et al., 1992) this prediction was strengthened by the discovery that the first 80 residues of all three chains contained the Gly-X-Y motif and that many of the hydroxlysine residues in the sequence were substituted with the disaccharide glucosyl-galactose. Since these were the hallmarks of the collagen structure it was deduced that the N-termini of the three chains formed the corresponding triple helical structure or fibril. It was also found that both the A-B and C-C disulfide bonds are located near the N-termini and in fact the latter appear to be well-situated to hold adjacent fibrils together and stabilize the six-membered fibril bundle (Reid and Porter, 1975; Reid and Porter, 1976). Midway through the senes of Gly-X-Y repeats, at about position 40, the regularity of the sequence is broken in ail three chains. In the A-chain, one of the triplets is augmented by the insertion of a threonine residue, in the C-chain an alanine replaces the glycine at position 36 (Reid and Porter, 1976; Reid, 1977; Reid and Thompson, 1978; Shinkai and Yonemasu, 1979), and if the B-chah sequence is aligned with the A- and C-chains to mawnize hornology, the additional triplet of the B- chah occurs in this very point. These deviations in the amino acid sequence suggest that the triple helix bends in this place. The sequence of the remaining 110 amino acids is strongly similar to the C-temiinal globular-type portions of the type VIII and type X collagens (Brass et al., 1992) and these residues are therefore believed to assume an overall globular conformation. Ln direct contrast to the stalles, the globular heads, as the C-terminai structures of Clq are known, arc resistant to digestion by collagenase (Reid et al., 1972; Knobel et al., 1974) but are instead susceptible to limited proteolytic digest by pepsin (Wiger and Natvig, 1972; Reid and Porter, 1976). The overall proposed mode1 for Clq - that of a bunch of a half dozen tulips (Reid and Porter, 1976) where the 'stems' represent the collagenous stalks and the 'flower heads' represent the globular heads, including the suggested 'bend' - have been corroborated by electron micrographs as shown in figure 1.10 (Shelton et al., 1972; KnobeI et al., 1975; Burton et al., 1980) and more recent neutron scattering studies (Perkins et al., 1984; Perkins, 1989). According to these studies, the length of the central fibril-like stalk is 112 A (see figure 1.10). The radiating strands connecting the central tmnk to the globular heads measure 115 A each. The diameter of the central sta.is 45 A whereas the six comecting strands each have a diameter of 15 A. This is consistent with a central stalk that is composed of six of the thinner stalks arranged in six-fold symmetry. The globular heads, which have dimensions of 50 A x 70 A, are held at an angle from the axis of the central fibril. The base-ann angle is estimated by Perkins (1984) to be 45' f 5' (using an arm length of 140 A), but the data are also consistent with angles ranging from 15' to 65'. This apparent flexibility may be of importance in C 1 activation.

1.3.1.2 Clr and Cls The Clr and Cls subunits share many features and are similar in amino acid Figure 1.10. Electron micrograph and dimensions of the Clq molecule. An electron micrograph (top; Knobel et al., 1975) and a schematic diagram of the Clq molecule (bottom) are shown. The dimensions have been determined by various methods (Perkins et al., 1984). sequence. Both are zymogens that exist as single polypeptide chains of approximate molecular mass 83 kD. Upon activation, Clr and Cls become cleaved at an Arg-Ile bond

(Arlaud and Gagnon, 1985; Spycher et al., 1986) into two chains, A (65 kD) and B (27 kD) as depicted in figure 1.1 1. The two chains remain associated through at least one interchain disulfide bond. The complete amino acid sequence has been determined for both

Clr (Sim et al., 1977; Journet and Tosi, 1986; Leytus et al., 1986) and Cls (Mackinnon et al., 1987; Tosi et al., 1987). A cornparison of their sequences showed that they had 408 amino acid identity and that al1 the cysteine residues were conserved (Tosi et al., 1987). Furthermore, the B-chah sequence of both Clr and Cls are simüar to the sequence of other senne proteases and have been shown to contain the classicai serine protease-active site with a trypsin-like specificity (Arlaud et al., 1982; Carter et al., 1983). This is consistent with that fact that both Clr and Cls cleave their respective substrates at an arginine residue. The amino acid sequence for both C lr and C 1s proenzymes can be divided into six distinct subregions (see figure 1-11). Subregions 1 and II have been shown to participate in the interaction between Clr and Cls (Gals et al., 1994; Cseh et ai., 1996). They also appear to play a role in modulating the activity of the distal protease domain. Subregion Ei contains a 100-residue repeat sirnilar to that in subregion 1. This repeat has recently been found in a hurnan bone morphogenic protein and in a developmentally regulated protein in Xenopus Zaevis but the significance of this similarity is not known (Bork et al., 1991). Subregion II resembles the epidemal growth factor (EGF) domain. This is consistent with the observation that EGF-domains are usudly exposed and are believed to participate in protein-protein interactions (Appella et al., 1988; Rees et al., 1988). EGF-domains are aIso found in proteinases of the coagulation system. Hydroxyasparagines are found in the EGF-motif of both Clr and Cls but whereas Clr is completely hydroxyiated, hydroxylation is only partiaiiy complete in Cls (Arlaud et al., 1987; Thielens et ai., 1990). Subregions N and V contain 60 amino acid repeats (known as short consensus repeats, A chain B chain

Figure 1.11. The proposed structure for the Clr and Cls subunits of Cl. These two subunits are durnbbell-shaped (shown at the bottom) when visualised by electron microscopy. Shown at the top is the general two-chah structure of the activated fomof Clr and Cls, their corresponding proteolytic fragments and their domain and subregion structures. The regions in black are believed to be the proteolytic domains whereas the shaded regions are believed to mediate the Clr-Cls interactions as well as to bind Ca2+. The locations of the carbohydrate structures in Clr and Cls are marked by fded and empty diamonds respectively. Adapted fiom Villiers et al. (1985). SCR or complement control protein repeats, CCP) that have previously been identified in other complement and semm proteins such as C2, Factor B, Factor H, C4b-binding protein, DAF and CRI. Since these proteins ail bind to C3b or C4b or both (Lintin and Reid, 1986; Dierich et al., 1988), it is likely that the IV and V regions of the C lr and Cls are involved in the recruitrnent of these complement components. These two subregions are closely associated with the B-chain and are involved in Clr dimerization. A double- stranded anti-parallel P-sheet structure for the SCR has been proposed based on 1H NMR studies of a Factor H SCR domain (Barlow et al., 1991). Subregion VI contains the catalytic site. ~irhas a slightly higher carbohydrate content (9.4%) than C~Swhich has 7.1% (Sim et al., 1977). Both have two Asn glycosylation sites in their respective A-chains, but ~irhas two additional sites in its B-chain (see figure 1.1 1). The carbohydrate contains mannose, galactose, N-acetylglucosarnine and siaiic acid (Sim et al., 1977; Arlaud and Gagnon, 1983). Although no detailed linkage structures are available, a partial determination of the N-linked oligosaccharide of C~Sby electron spray ionization mass spectrometry has been published (Petillot et al., 1995). The carbohydrate structure at position Asn 159 was shown to be a cornplex-type biantennary, bisialylated oligosaccharide, whereas the structure at Asn 391 occurred in two foms: a biantennary bisialylated or a triantemary trisialylated structure. It was also shown that the carbohydrate in the latter position is heterogeneoas and that there are at least three distinct major carbohydrate structures at this site. No specific role for the carbohydrates in the function of Cl has been proposed yet other than the possibility that they stabilize the otherwise hydrophobie stalks. In fact, a mutant human recombinant Cls protein expressed in insect cells with an Asn to Gln mutant at position 159 was able to form a fully functional Cl complex when reassembled with human semm Clr and Clq (Luo et al., 1992). Since such a protein lacks the carbohydrate at position 159 and is incompletely glycoslyated at position 39 1 - it is rnissing siaiic acid and possibly more at this latter position as a result of having been produced in insect ceii

culture - this suggests that in Cts, the carbohydrate moiety at position 159 is not crucial to the hnction of ihis protein and that the glycan at position 39 1 is not restricted to a specific structure.

1.3.1.3 The CIr2CIs2 tetrarner Cls is dependent on Ca2+ for dimerkation whereas Clr can dimerize both in the presence and in the absence of Ca2+ (Valet and Cooper. 1974a; 1974b). Although dimes of the Clr proenzyme and the activated forms can be dismpted by lowering the pH, cir2 withstands lower pH values (Arlaud et al., 1980a; 1980b). When purified Clr and Cls are incubated together in the absence of Ca2+ they associate to form C 1r2C 1s trimers whereas

in the presence of ~a*+,C lr2C 1s2 tetramers are formed (Arlaud et al., 1980a; 1980b). This suggests that the formation of the ClrzC ls~tetramer involves both @+-dependent and ca2+-inde pendent interactions . The ch2dimer appears in electron micrographs as a pair of uneven dumbbell- shaped units which are joined near their larger domains (Villiers et al., 1985; Weiss et al., 1986). C~Ssubunits are sirnilar in shape to cT~subunits. Using an avidin-ferritin complex to mark the position of C~S,Weiss et ai. demonstrated that they assume the two end positions in the Cir2cis2 tetrarner and that this interaction is mediated by the smaller domains of both subunits. lllustrated in figure 1.1 1 is a schematic diagrarn of ~irand C~S and their corresponding proteolytic fragments. Upon prolonged incubation at 37'C, the A- chain of CT~undergoes further autolytic cleavage to yield the a-,P- and y-chahs. The a- and P-c hains are released whereas the y-c hain stay s disulfide-linked to the B chain (these latter two chains consist of the catalytic domain and subregions TV and V). According to electron micrograph images, the y-B fragments of remain associated as dimers but are unable to bind Ca*+ or to associate with GIS subunits (Villiers et al., 1985). These results indicate that the proteolytic B-chain portion of the cT~molecule is located in the larger globular domains and that the Cls and Ca2+ binding sites are located in the smaller globular domain. C~Slike ~ircm be further fragmented. Upon prolonged exposure to proteolytic enzyme, diree fragments are released: al, a2,and p. The remaining y-B fragment, which corresponds to the larger dornain of C~S,is unable to dimerize or to bind ~ireven in the presence of Ca2+. These results suggest that the interaction between Cls and Clr and the

dimerkation of the C~Sin the presence of Ca2+ are both mediated by detenninants in the smaller domain. Together the Clr and C 1s subunits form a linear structure C 1s-C lr-Clr-Cls. Villier's studies indicated that this complex is a Linear-shaped tetramer, however Weiss et al. (1986) showed that the centrai Clr2 dimer actuaily takes the form of an assymetrical 'XI. EIectron rnicroscopic images of the C 11-2dirner showed that the interaction occurs between the non-globular portions. Furthemore the interaction occurs nearest to the B chain-containing globular domain. Hydrodynamic studies by Perkins (1989) which use experimentally determined sedimentation coefficients to calculate the length of the corresponding molecule assuming that it has a cylindrical shape, show that the length of the intact tetramer behaves as if it were shorter in length than the sum of its parts. This result is more consistent with the 'XI model of Clr-C lr interaction than with the fully extended linear model.

1.3.1.4 Arrangement of the CIrzCIs2 tetramer about Clq In the assembled C 1 complex, Clq acts as the frame about which the Clr2C ls2 tetramer is arranged. Since the central region of chemically crosslinked Cl complexes appear to be more dense in electron micrographs than Clq alone, it was concluded that the Clr2Cls2 tetramer is centrally located (Sumg et al., 1982). Reid et ai. (1979) found that the collagenous stalks of Clq were able to inhibit the reconstitution of Cl. This demonstrates that the tetramer interacts with the stems and not the globular heads.

Ultracentrifugation snidies and hydrodynamic modeling of the C 1r2C 1s2 tetramer in the presence and in the absence of C lq indicate that Cl~Cls2is more tightly packed in the C 1 complex but could not provide further detaiis as to the arrangement of the tetramer about Clq (Perkins et al., 1984).

The 'Oz', 'SI, '8' and W' models (figure 1.12) have been proposed for the spatial arrangement of the subunits within Cl (Poon et al., 1983; Cooper, 1985; Arlaud et al.,

1987; Perkins, 1989). In the 'Smand '8' models the C lr2C 1s2 tetramers are interwoven through the Clq arms so that the final complex has two-fold symmetry. The catalytic

domains of ali four subunits are located inside the core defined by the Clq anns. The smaller domains through which the Clr and Cls (subregions I and II) interact are on the outside of this cone. The only difference between these two models is that in the '8' model each Clr and Cls pair wraps around only one arm whereas in the 'S' model. each pair wraps around two arms. In the 'Oz'and 'W'models, the tetramer is located outside the

Clq cone. In both models there is only one axis of symmetry. In the '02' model, the

tetramer forms two rings that encircle the stalk near the protmding arms. In the W' model, each subunit is arrayed dong a separate arm so that the tetramers occupy four adjacent arms. Other models are also possibie, however these four best satisQ the criteria outiined by Perkins (1989): (1) The model must be consistent with the electron micrographs in that the C 1r2C ls2 tetramer must form a more compact structure in the presence of C lq, (2) the catalytic centres of Clr and Cls must be in contact with each other in order for

autoactivation to occur, and (3) the sites through which Cl interacts with the proteins such

as C2, C4 and the Cl inhibitor must be accessible. Perkins also suggests that the experimentally observed parameters such as sedimentation coefficients and hydrodynamic curve data should agree with those calculated using approximated dimensions of the Cl complex. Studies by Perkins et al. (1984, 1989) indicate that the W' model gives the best fit to the hydrodynarnic curve data however the problem with interpreting results £rom this Figure 1.12. Models of the Cl cornplex. The six arms of the Clq molecule are represented by black sticks whereas the Clr2Cls2 tetramer is represented by white circles and tubes. Both the '8' and 'Srmodels are shown from beneath the cone defined by the C lq arms whereas the '02' and W' models are shown from an angle above the cone. type of study is that the validity of these cornparisons depends upon how accurately the dimensions are chosen for each Cl model. Because the Cl complex assembles and dissociates rapidly (Bartholomew and Esser, 1977), Perkins also suggests that a cornplicated assembly involving the weaving of the tetramer in and about the stems of Clq such as in the 'S' and '8' models, is not so likely.

1.3.1.5 Naiure of the interactions behveen the ClrzCIsz tetramer and Clq The interaction between C lq and the C lrzCls2 tetramer is temperature dependent. This suggests that the assembly of Cl involves hydrophobic interactions. At O°C the dissociation constant for the interaction between the proenzyme fom of C lr2C ls2 and C lq is 72 pM whereas at 20°C the dissociation constant is 3 pM (Tseng et al., 1997). Previous studies have reported Kd values of about 15 nM at the lower temperature (Siegel and Schumaker, 1983; Ziccardi, 1985). Since radioiodinated Clq was used in these earlier expenments, Tseng et al. have suggested that the two groups found a higher dissociation constant for the Cl complex because the iodination interfered with the interaction between C lq and the tetramer. The stability of the complex has also been shown to decrease with increasing ionic strength indicating that the interaction has an ionic component (Ziccardi, 1985). A study by Illy et al. (1993) suggests that the assembly of the Cl complex involves interactions between acidic residues on C 1r and lysine (hydroxylysine) and arginine residues on C l q. Illy et al. found that Clr2C ls2 treated with a carboxylate group-modifier (glycine ethyl ester in the presence of carbodiimide) did not form complexes with Clq. On the other hand, Clq treated with the carboxylate group-modifier was able to interact with untreated C lr2Cls2 tetramers, but Clq treated with either a lysine- and hydroxylysine-modifier (pyridoxyl-S'-phosphate, acetic anhydride or citraconic anhydride) or a lysine- and arginine-modifier (cyclohexanedione) was not. Iliy et ai. proposed that arginine at position 38 of the A-chain (A38) and hydroxylysines at positions B32, B65 and C29 of Clq mediate assembly of the Cl cornplex. This proposal was based on sequence similarities with mannan binding protein, another protein that interacts with the C lr2Cls2 tetramer.

1.3.2 Activation and regdation of CI 1.3.2.1 The relationship behveen Cl binding and CI activation Severai observations suggest that CI binding and C 1 activation are separate events. Furthemore, the binding of Cl by immunoglobulins does not necessarily lead to the activation of Cl. For example, fluid-phase IgG binds weakiy to Cl but is unabie to activate C 1. Studies have also shown that not al1 hapten-bound IgG can activate C 1 since as much as 80% of the Cl bound to IgG-coated erthythrocytes remains inactive (Borsos et al., 1968; Colten et al., 1969). The studies of Coiten et al. (1969) also showed that the binding of Cl was rapid and occurred over a wide range of temperatures whereas the activation of C 1 was slow and temperature dependent. Folkerd et al. ( 1980) demonstrated that the rate of activation did not correlate with the affinity of Clq for the activator. When the binding of Clq to immune complexes and to glutaraidehyde-aggregated IgG was studied, Folkerd et al. found that although the affinities for Clq were similar in both cases, the immune complexed IgG was far better at activating Cl than the chemically aggregated IgG. Studies have also indicated that the activation of Cl by ce11 bound IgG depends on the density and distribution of epitopes on the target surface whereas the binding of Cl does not (Borsos and Circolo, 1983; Circolo et al., 1985). This provides further evidence that binding does not necessarily result in activation and that the binding and activation of C 1 are separate processes.

1.3.2.2 CI activation by immune complexes Hanson et al. (1985) have previously established that Clq is a flexible molecule and have further speculated that this flexibility plays a role in the activation of Cl. The data summarized in the previous section suggests that the binding of Cl by immunoglobulins does not necessariiy lead to the activation of Cl. Studies with IgG aggregates and non- covalent polymers formed by the interaction between IgG and bivalent antigens show that Cl must bind to at least two IgG antigen-bound molecules to be hemolytically active

(Hyslop et al., 1970; Metzger, 1974; Metzger, 1978; Wright et ai., 1980; Smith et al., 1995). Similady, a single molecule of IgM, which potentidiy contains multiple Cl binding sites, is able to both bind and activate C 1 (Borsos and Rapp, 1965a). On the basis of these observations, researchen have proposed a distortional model for C 1 activation by immune complexes (Schumaker et al., 1986). According to this model, the Clq arrns become distorted when Clq interacts with an array of binding sites through two or more of its globular heads. As a consequence of this distortion, the attached C lr2Cls2 tetramer is rearranged so that the proteolytic domains of Clr and Cls are brought closer together. This facilitates autoactivation of the Ch2 proenzyme and activation of the CLs subunits. Upon activation, the interaction between the C lq and the tetramer weakens by an order of magnitude (Siegel and Schumaker, 1983). Schumaker et d. (1986) predict that the catalytic domains of C~Sare consequently released from the interior of the C lq cone and become accessible to cT-Inh, C4 and C2. Direct evidence supporting the role of Clq distortion was supplied by studies involving antibodies directed against the Clq 'heads' and 'stalks'. It was found that a monoclonai antibody specific for the Clq 'head' was able to activate Cl in solution (Kilchherr et al., 1986). Furthemore the Fab12 fragments from these antibodies were as effective as the whole molecule at activating Cl indicating that activation was not proceeding by the conventional Fc-mediated way (Kilchherr et al., 1987). Fab fragments which are only monovalent, though equaüy able to bind to the Clq 'heads' were not able to activate C 1. Another group found that an antibody directed against the stalks close to the Clq 'head' was able to activate Cl, but when the flexibility of the molecule was increased by mild reduction and alkylation, this ability was lost (Hoekzema et al., 1988). 1.3.2.3 Autoacîivation of CI Spontaneous activation of purified Cl can occur, although at a 5 to 7-times lower rate than when induced by activators such as immune complexes (Ziccardi, 1982b). More recently, Thielens et al. demonstrated that the autoactivation of the Clr subunit occurs in two ways: through an intrarnolecular mechanism and an intermolecular mechanism (Thielens et al., 1994). The presence of Ca2+ was shown to slow down the autoactivation of purifïed proenzyme Clr through the intramolecular mechanism although it had no effect on the interrnolecular activation. The latter was determined by measuring the activation of proenzyme Clr by activated Clr in the presence and absence of ~a2+.In the presence of

Clq, the inhibitory effect was partially released. This release was dependent upon the interaction between C lq and Clr because in the presence of 1,3 diaminopropane, an diphatic diamine known to disrupt the interaction between Clq and the tetramer, C lq was ineffectual. In the presence of ~a2+,the addition of proenzyme C 1s or the Clsa fragment to proenzyme C lr resulted in the formation of tetramers in which the C lr remained stable in the proenzyme form. It appears that the formation of the tetramer abrogates the autoactivation of Clr and that the signal for autoactivation originates from Clq and not Cls.

1.3.2.4 CI inhibitor To prevent aberrant activation, the classical complement pathway is tightly regulated by the CI inhibitor (CI-hh), a member of the superfamily of proteins known as 'serpins' or serine protease inhibitors. ~ï-~nhis found in normal human serum and serves several functions - it prevents non-immune Cl activation, mediates the dissociation of and C~S from Clq, and limits the proteolytic Lifetime of C~S. CI-hh is a single polypeptide chah of 478 amino acids with a calculated molecula. weight of 71 100 (Perkins, 1991). The 13 ~Iigosaccharidesof CI-1nh account for 49% of the protein's total molecular weight and make it one of the most highly glycosylated serum proteins. This rnay help explain why CZ-M appears to have a molecular weight of 105 000 when analyzed by SDS-PAGE and is still commonly quoted as having this higher mas. A structural rnodel for ~ï-Inhhas ken proposed based upon electron micrographs (Odermatt et al., 1981; Schumaker and Phillips, 1993) and is consistent with the observations from neutron scattering studies (Perkins and Smith, 1990; Perkins, 1993). According to this model, CI-1nh resembles a match having a length of 180 A. The first 113 amino acid residues form the rod-shaped portion of the CI-~nh,and the rernaining 365 arnino acid residues form the globular head. This unique structure may allow it to specifically access and inactivate the protease domains of the CI complex. To date, no other serpin has been found which can do so. CI-1nh has been shown to inactivate a rnolecule of CI by interacting with the and C~Ssubunits to fom two ~ir~Ts(~I-lnh)~complexes (Ziccardi and Cooper, 1979).

These complexes have a lower affinity for Clq than the cir2ck2tetramer and this hastens its dissociation from Clq. SDS-PAGE analysis of the resulting complexes showed that both ~irand C~S are covalently linked to separate ci-1nh molecules. During the inactivation of CT by ci-~nh,the Arg 444-Thr 445 peptide bond within the C-terminal reactive centre of ~1-1nhis irreversibly cleaved (Saivesen et al., 1985). The ability of hydroxylamine to dissociate the complex suggests that a covalent ester bond is formed between one of these residues and the catalytic domain of Cir or C~S(Sirn et al., 1979). The mechanism by wtich ~i-~nhprevents the spontaneous activation of Cl has been elucidated by Bianchino et al. (1988). According to their studies, the spontaneous activation of Cl fits a model comprised of two reactions, one involving the unimolecular autoactivation of Cl, the other involving the birnolecular activation of Cl by CI. By ensuring that the Cl preparation was devoid of ci, they were able to show that ci-Inh prevented only the birnolecular reaction and that the slow unimolecular autoactivation of Cl continued to proceed in the presence of the inhibitor. This latter point suggests that cT-Inh does not interact with non-activated Cl. Ziccardi et ai. (1982b) were unable to detect any interaction between ci-1nh and fluid phase Cl when they were both present at p hy siological concentrations. At a semconcentration of about 137 pg/ml (Ziccardi and Cooper, 1980), ci-~nh exists in excess of Cl at a ratio of 7: 1. This is apparently more than enough inhibitor to serve its purpose (Ziccardi, 198 l), however when the levels of CI-Inh fail to 25% of the normal concentration a severe dinical disease known as angioneurotic ederna results. Hereditary cases are often treated with ~!-1nh infusions but unfortunately, rare cases of acquired ~T-lnhdeficiencies that arise from the production of autoantibodies against CI- Inh cannot be treated this way and are fatal. CI-1nh has also been shown in vitro to inactivate plasmin, plasminogen activator, clotting factors Xla and Wa(Hageman factor) and kallikrein @avis, 1988) linking the complement system to the blood coagulation and fibrinolysis pathways. Since cT-hh is a potent regulator of these important pathways this might explain the severe pathology of ci-hh deficiency. Inhibition of Cl activation is not lirnited solely to the action of the ~hh.More recently, a species of B-ce11 produced chondroitin sulfate proteoglycan (CSPG)was found that inhibits C 1 activation by binding to the C lq subunit and thus preventing the assembly of Clq with Clr and Cls (Kirschfink et al., 1997).

1.3.2.5 The relationship between Clq binding and subsequent steps in the classical complement pathway Using a matched set of rat irnmunoglobulins. Bindon et al. (1990) showed that different isotypes exhibited different degrees of interaction with complement proteins other than Clq. Although Bindon et ai. found that rat IgGl was better than IgG2a at binding

Clq, the situation was reversed when binding to whole Cl was examined. This shows that Clq binding does not necessarily correlate with Cl binding and suggests that the interaction between immunoglobulins and Cl may involve determinants on Clr and Cls.

Other isotype differences also occur with complement proteins further dong in the cornplernent pathway. IgG2c, though it binds approximately the sarne arnount of CI as does IgG2a, binds less than a fifth of the C4 and C3 bound by IgG2a. Moreover, although IgG2a binds C lq less efficiently than IgA it has approximately three times the Iytic activity of IgA. Conveaely IgGl and IgG2c which bind Clq better than IgG2a have undetectable Ievels of lytic activity.

1.4 The Interaction Between Immunoglobulins and Cl 1.4.1 The nature of the interaction 1.4.1.1 IgM vs. IgG complernentfiaiion and temperature dependence In complement fixation assays (Augener et al., 197 l), the complement binding activity of a sampIe is determined indirectly by measuring the arnount of hemolytic activity remaining after the interaction. IgG proteins are better able to 'fix' complement at 4OC than at 37°C (Sandberg and Stollar, 1966) whereas IgM molecules are better able to fix complement at the higher temperature (Cunnif and Stollar, 1968). This was later confiied by Poon et al. (1995) who showed that hybrid IgG and IgM molecules containing the yFc exhibit the temperature preference of IgG whereas those with the pFc exhibit the temperature preference of IgM.

1.4.1.2 Electrostatic component of the IgG and Clq interaction Both Burton et al. (1980) and Emanuel et al. (1982) have measured and reported the release of 9-12 salt ions upon binding of a single molecule of C lq to IgG aggregates. On the basis of these measurements, Burton et al. postulated that 12 salt bridges form as a result of interactions between IgG and Clq. These results are consistent with previous demonstrations by Hughes-Jones and Gardner (1978) and Lin and Fletcher (1978) that the IgG-C 1 interaction is dependent on ionic strength. Hughes-Jones and Gardner found that as the ionic strength was decreased from p=0.20 to p=û. IO, the interaction between heat- aggregated IgG and Clq was noticeably enhanced. A plot of the functional &inity against the square root of the ionic strength suggested that three pairs of oppositely charged residues were involved in the binding site. Lin and Fletcher studied the same interaction over a larger ionic strength range (zero to 0.6 M NaCl or p=û to 0.6) and carne to the same conclusion. Both groups also initiated studies on the inhibitory effects of various charged compounds. A review of the ionic compounds studied is provided by Van Schravendijk and Easterbrook-Smith (1985). Hughes-Jones and Gardner found that the poly-ions

dextran sulphate, polyglutaminic acid and polylysine were all capable of inhibithg the Clq- IgG interaction again implicating an ionic component. Lin and Fletcher showed that heparin, 1,7 diamino heptane and suramin were also inhibitory. More importantly, the latter anion was shown to bind directly to Clq. These results are consistent with studies showing that DNA, another negatively charged molecule, is able to interfere with Clq-IgG

binding in the same way (Burton et al., 1980; Van Schravendijk and Dwek, 1982). Von

Schravendijk further characterïzed the DNA-Clq interaction to reveai that it was both heat and pH sensitive. Also, the inability of DNA to interact with pepsin-treated C lq indicated that the majorïty of the interaction involved the globular heads. The ability of Clq to bind to negatively charged DNA suggested that the interaction was mediated by positively charged residues in Clq. Chemical modification of the arginine residues with I,2 cyclohexanedione was found to affect the interaction whereas modification of carboxyl residues with methyl acetamine did not. The interaction between Clq and IgG is also heat and pH sensitive and globular subunits are known to bind to IgG (Hughes-Jones and Gardner, 1979). Furthemore, chernical modification of another positively charged residue, lysine, by acetic anhydride or methylacetimidate in C1q also decreases the interaction (Emanuel et al., 1982). Since these features resemble those of the Clq-DNA interaction it is likely that the binding sites for IgG and DNA on Clq overlap. 1.4.1.3 Possible hydrophobic component of the interaction between IgG and Clq 1-anilino-8-naphthdenesulfonate(ANS) is well known as a potent inhibitor of the Clq-rabbit IgG immune complex interaction. According to Alcolea et al. (1986), the interaction between ANS and Clq is stronger than that between ANS and the Fc fragment of IgG. This suggests that ANS prevents the IgG from binding Clq by binding to Clq itself. Since the ANS-C lq interaction is a hydrophobic one, Alcolea et al. suggest that the C lq-IgG interaction also has a hydrophobic component.

1-4.1.4 Presence of polyethylene glycol affects the interaction The binding of radiolabeled Clq to insoluble rabbit IgG immune complexes increases as the concentration of polyethylene glycol (PEG)in the solution increases (Wines and Eas terbrook-Smith, 1988). After C lq binding sites were chernically destroyed using the histidine modifier diethyl pyrocarbonate (DEPC),the IgG immune complexes were no longer able to bind to Clq in the presence of PEG. This finding suggests that the presence of PEG does not creaie novel Clq binding sites but enhances the binding between Clq and existing sites. Furthemore, the increase in the affinity constants was found to Vary logarïthmically with the increase in PEG concentration which was in agreement with the excluded volume effect.

1.4.1.5 Studies using radiolabeling of Clq and their implications A cornparison of labeling methods and their subsequent effect(s) on Clq binding activity can reveal information about the type of functional groups involved at various stages of hemolysis. Radiolabeling of Clq by lactoperoxidase without loss of hemolytic activiq can be done. The radioactivity in this radiolabeled Clq is contained solely in the smaller subunit, later found to correspond to the disulfide-linked Cchain dimer (Heusser et al., 1973). Upon heavy radiolabeling using the lactoperoxidase method of iodination, the hemolytic activity of the Clq was reduced. Since this method of iodination selectively labels the C-chain (Tenner et al., 1981), this result suggests that the C-chah is important for activation and agrees with the subsequent results by Heinz et al. (1984)who found that an antibody directed against the C-chah of Clq inhibits activation. It was not determined whether this interfered with the interaction between Clq and IgG or the tetramer.

1.4.2 Imrnunoglobulin binding site on CIq Since the IgG-Clq interaction has a strong ionic component, charged residues are likely to be involved. The existing data gathered from chemical modification studies are in agreement with this prediction. Together, they suggest that positively charged arginine and histidine residues in the globular region of Clq are involved (Burton et al., 1980; Easterbrook-Smith, 1983; Comis and Easterbrook-Smith, 1985). Using cross-linking experiments, Wines and Easterbrook-Smith (1990)found that the interaction between Clq and rabbit IgG involved residues in al1 three chains of Clq. Arnino acid sequence detenninations for the three chains of Clq (Reid, 1983; Loos et al., 1989; Petry et al., 199 1; Sellar et al., 199 1; Petry et ai., 1992) soon facilitated the use of peptide inhibition to localize the binding site. These studies independently led Baurnan and Anderson (1990) and Jiang et al. (1992) to different conclusions. Whereas Bauman and Anderson found that a synthetic peptide spanning a region in the globular region of the Bshain was able to inhibit complex formation between C lq and IgG-containhg immune complexes, Jiang et ai. found that the interaction involves the C-chah preferentially. It still remains unciear whether the short peptides used in these studies mirnic the conformational site proposed by

Wines and Easterbrook-Smith ( 1990). Marques et al. (1993) repeated and expanded on the chemicd modification experiments. By isolating and analyzing enzymatically generated fragments following chemical modification, they showed that arginine residues in two clusters of residues in the globular domain of C lq, B 1 14-B 129 and A 162-(B 163)-C 156, were affected. Based on the predicted solvent accessibility of these residues, they proposed that the latter site coincides with the IgG binding site. 1.4.3 CI binding site on IgG There is a larger body of work airned at defining the Clq binding site on immunoglobulins than at defhng the immunoglobulin binding site on Clq. The earliest efforts to locate the Clq binding site on immunoglobulins were conducted on the IgG molecule. Prompted by the Domain Hypothesis, which suggested that the different functions of the immunoglobulin molecule might be cornpartmentalized into the discrete regions of the molecule known as domains, researchers began by deconstmcting the IgG molecule. Complement fixing activity was fxst associated with the Fc fragment and not the Fab fragment of rabbit IgG (Ishizaka et al., 1962). Indeed, this was Iater confmed in the human IgG 1 subclass whereby on a molar basis, aggregated Fc fragment of IgG 1 was as efficient as the intact protein in fixing ci whereas the Fab fragment was shown to have no such activity (Ishizaka et al., 1967).

1.4.3.1 Evidence supporting the role of C@ domain Snidies of carefully controlled proteolytic digests of rabbit IgG suggested that the portion of the molecule that interacts with Cl is located near the N-terminus of the Fc fragment (Utsumi, 1969). Utsumi found that the complement fwng activity of Fc could be destroyed by removing about fifty amino acids from the N-terminus. Conversely, efforts to detect complement fixing activity in the Fc' fragment, which corresponds to the Cy3 domains, met with faiiure suggesting too that the active portion of the molecule is likely to be found in the Cy2 domains (Irimajiri et al., 1968).

Chernical cleavage of a murine IgG2a using CNBr narrowed the complement fwng activity of the immunoglobulin to a 62-residue fragment containhg the amino terminal of the Fc fragment (Kehoe and Fougereau, 1969). This fragment was later sequenced and found to correspond to residues 253-3 14 within C@ of the y-chah (Kehoe et al., 1974). Smailer fragments of the Fc fragment were also obtained by proteolytic digestion. Separate Cy2 and Cy3 domain fragments were generated through the use of pH controlled trypsin digestion of Fc (Euerson et al., 1972). Of the two fragments tested, only the Cy2- containing fragment was able to fix complement with an activity comparable to whole Fc (Ellerson et al., 1972). Subsequent detailed analysis of this fragment revealed that it contained two disulfide-linked polypeptides of differing lengths. The longer of the two polypeptides spans residues Th 223 through to Lys 338 (Eu-numbering) whereas the shorter peptide, which also begins at Thr 223, ends at Lys 248. The putative Cy2 location for the Clq binding site of IgG was arrived at independently by those studying the interaction between IgG and other proteins which interact with the Fc region. Mutual inhibition of the binding of Clq and protein A to rabbit IgG immune complexes implied that their respective binding sites were near each other. This provided further support for the Cy2 location of the putative Clq binding site on IgG since the protein A binding site had already been locaiized by crystallization studies (Deisenhofer, 1981) to a region spanning Cy2 and the N-terminal portion of the Cy3 domain.

Burton et al. (1980) postulated that the Clq binding site was located within Q2 and i33, the two C-terminal P-strands of the Cy2 domain. They based their analysis on the existing data from X-ray crystal structure determinations of the Fc fragment, chernicd modification studies and inhibitor studies. Furthemore the solvent accessibility, sequence conservation, and charge of each residue in this region were considered. A chernical modification study of IgG by the sarne group later confirmed this hypothesis. Emanuel et al. (1982) showed that a modified lysine residue on the fy2 strand at position 322 of the Cy2 domain was responsible for the decrease in Clq binding by lysine-modified IgG

(Emanuel et al., 1982). On the basis of this finding, they proposed that Lys 322 and other nearby residues serve as the Clq binding site. Peptide inhibition studies were also performed to identiQ the amino acid residues involved in the Clq binding site using synthetic peptides that corresponded to various sequences spanning the Cy2 domain (Lukas et ai., 1981; Prystowsky et ai., 1981; Takada et al., 1985). The results of these studies were somewhat misleading since many of the tested peptides were able to inhibit the interaction between Clq and IgG. An explanation for the many positive results obtained is that the amino acid residues do not necessarily hold the same three-dimensional conformation in these peptides as they in do within the intact Cy2 domain. A hinctionai group that is sequestered in the intact protein would be fully exposed in a short peptide sequence. Furthemore short peptides not found in the sequence of the y-chah, such as polyglutamic acid and polylysine have been shown to inhibit the Clq-IgG interaction merely because they are charged and can interfere with the ionic component of the interaction (Hughes-Jones and Gardner, 1978). Using site-directed mutagenesis, Duncan and Winter (L988) created a panel of twenty-two mouse IgGZb molecules in which single residues in the Cy2 domain were replaced with alanine, a srnail neutral arnino acid. Since the interaction between IgG and Clq is dependent on ionic strength, residues that had either charged or polar side chahs were selected for mutation. Duncan and Winter aiso took into consideration other factors including solvent accessibility, conservation arnong hemolyticaily active IgG isotypes and presence on peptides reported to inhibit the IgG-Clq interaction. By examining the ability of the mutants to activate cornplement, they determined that an E3 18, K320 and K322 motif was necessary for this hinction. They confirmed this finding by showing that a peptide spanning these three residues was able to inhibit the binding of Cl to IgG2b. These residues lie on the S2 B-strand in accordance with the prediction by Burton et al.

(1980). The argument for these residues is further strengthened by the observation that in the three-dimensional structure of the Cy2 domain their functional groups project into the same region forming a localized cluster. Further studies comparing paired C$ domains with monomerized domains indicate that the Clq binding site is an associative one in which both domains contribute cooperatively to the interaction (Udaka et al., 1986), and that the pair is required for the activation of Cl (Utsumi et al., 1985).

1.4.3.2 The rule of the Cy3 domain Studies using fragments of rabbit IgG generated by the enzyme plasmin found that Facb, an H2L2 tetramer which lacks the Cy3 domains, was able ro both bind and activate complement in much the same way as the intact immunoglobuh, configthe belief that the C@ domain is not necessary for complement activating activity (Colomb and Porter,

1975; Arlaud et al., 1976)- Although relatively little is known of the role of the Cy3 domain in Cl binding and activation, there has been some evidence linking the Cy3 domain to the stabilization of the

C 1 and IgG interaction by interactions with the C lr2C ls2 subunits. According to a study by Okada and Utsumi (1989), immune complexes comprised of rabbit F(acb)2 fragments2 but retaining the full C Lq binding afhity of intact QG,did not bind the entire C 1 molecuie as strongly as did immune complexes containing whole IgG. SVnilarly Ovary et al. (1976) had found that although rabbit Facb anti-DNP were able to fix complement when bound to DNP-BSA, Fab fragments of guinea pig IgG directed against the C@ domain of rabbit IgG were able to inhibit complement fixation by rabbit IgG.

1.4.3.3 IgG subclasses differ in their interaction with cornplement The human IgG subclasses IgGI, IgG2, IgG3 and IgG4 have been compared for their abilities to bind human and guinea pig CI in CI fixation assays (Augener et al., 197 1) and to complex with the human CIq subunit in ultracentrifugation studies (Schumaker et al., 1976). In both studies IgG3 was the most active subclass followed by the IgG1, IgG2 and the IgG4 subclasses in order of decreasing activity. Immunoglobulins from other species have also been studied for subclass

2~arneas die Facb fragment rekrred to by Colomb and Porter (1975), ArIaud et al. (1976) and Ovary et al. (1976). The 'F(acb)$ notation refers to the tetrameric (H2L2)fonn as opposed to the dimeric (HL) form which Okada and Utsumi also studied. differences. Studies on mouse IgG indicate that the Clq binding and complernent fixing activities were highest for IgG2a and IgG2b and very low for IgGl (Leatherbarrow and Dwek, 1984; Oi, et aU, 1984). Direct measurements of the interaction between IgG and Clq have not been reported for rat IgG, however approximations have been obtained through the measurement of complement consumption by aggregated irnmunoglobulin (Fust et al., 1980). IgG2b was best and IgGl was poorest at consuming homologous complement. When a heterologous complement source was used, slightly different profdes resulted. Rat IgG2c failed to consume human cornplement whereas rat IgGl failed to consume guinea pig complement. Finally, the most active species from the different species were comparable in complement fixation and activation. Rat IgG2b was found to be as effective as the human IgGl and IgG3 isotypes in a study of complement-mediated lysis using both hurnan and guinea pig complement sources (Bindon et al. ,1988). Mouse IgG2a and IgG2b isotypes were comparable to human IgGl and IgG3 in the fixation of

guinea pig complement (Dangl et al., 1988).

1.4.3.4 Features which determine these IgG srdxlass diferences When Duncan and Winter (1988) published their study pinpointing the three cntical residues for cornplement binding (E318, K320 and K322) an important question was raised. Why were IgG isotypes such as human IgG4 unable to initiate cornplement- mediated lysis even though they contained al1 three residues? Isenman et al. (1975) had previously demonstrated that the Fc fragment of IgG4 was as effective as the Fc fragment of IgGl in fixing Cl thereby providing evidence that the site did exist but that its expression was modulated by the presence of structural features outside of the Fc region. Circumstantial evidence suggested that the hinge region provided this modulating

feature. First of aii, the hinge was the most variable structure arnong the IgG subclasses, in ternis of amino acid sequence and length. Secondiy, reports existed that correlated hinge flexibility with the capacity to fix complement (Oi et al., 1984; Schneider et al., 1988). If, as these experiments suggested, the hinge acts as a spacer between the Fab arms and the effector sites on the Fc portion of the imrnunoglobulin, it foilows that a hemolyticaiIy active IgG subclass would be less effective if its hinge region were either shortened or rendered less flexible. It therefore came as a surprise when a mutant of the highly active human IgG3 isotype, which contained only fifteen of the sixty-two exon-encoded hinge residues, was found to be as effective at binding Clq as was the wiid-type IgG3 molecule (Michaelsen et al., 1990). Although Michaelsen et al. also found that an IgG3 mutant missing the entire genetic hinge was unable to bind Clq, they could not determine whether this defect was due to the loss of the hinge as a spacer or to the loss of the interheavy chah disuifide bond. In a subsequent study, Brekke et ai. (1993a,b) created a hinge-deletion mutant in which the entire genetic hinge was replaced by a short four-amino acid hinge (Ala-Ala-Cys-Ala) containing a cysteine residue for the interheavy chain disulfide bond. This mutant was found to be more active than wildtype IgG3. Alternatively, Shopes (1993) found that a mutant IgG with its Fab anns tethered together was as effective at initiating complement activity as the more flexible wildtype control. Together these results suggest that the length and flexibility of the hinge does not play a determining role in the complement binding activity of the IgG subclasses. Evidence was also accumulating to suggest that the features which defined the activity of an IgG isotype resided within the Cy2 domain as opposed to the hinge.

Clackson and Winter (1989) found that most of the complement binding activity of the mouse IgG2b isotype could be transferred to the inactive IgGl isotype with the Cy2 domain alone. Further studies involving hinge-swapped mutants suggested that neither the hinge nor the Cyl domain of an active isotype could confer complement binding activity to an inactive isotype. Both mutant human IgG4 with an IgG3 hinge and mutant IgG4 with an IgG3-Cyl domain were inactive (Tan et al., 1990; Norderhaug et al., 1991). Conversely IgG3 retained its activity when either its hinge or its Cyl domain was substituted with that of IgG4. The role of the inter heavy and light chain disulfide bonding pattern was dso explored and the resuits of the corresponding experiments suggest that this too was not responsible for the isotype differences. When Brekke et al. (1993a) changed the inter heavy and light chah disuifide bonding pattem of human IgG3 to that of IgGl, they found that the mutant IgG3 exhibited complement-mediated Lysis characteristic of IgG3 and not IgG 1. To locate features in C@ which were responsible for the isotype differences, Tao et al. (199 1) constmcted partial domain-switch mutants of human IgG 1 and IgG4. The major determinant was found to occur near the three critical residues in the carboxy terminal region of Cy2 spanning residues 292-340. This result was confumed by Greenwood et al. (1993). Within this region IgGl and IgG4 differ only at residues 296, 327, 330 and 331. Based on site-directed mutagenesis studies of these residues, a proline to serine substitution at position 33 1 was identified as the primary reason for the inability of the IgG4 isotype to activate the complement pathway (Tao et al., 1993; Xu et al., 1994). Incidentally, the proline at the corresponding position had been mutated in the earlier study by Duncan and Winter (1988), but it appean that the proline to glycine substitution that they chose to make was not inactivating. Burton (1992) had postulated that the IgG4 Fc fra,omnt was able to fix complement (Isenman et al., 1975) because a structural "lesion" which prevented this interaction in whole IgG4 was sensitive to the presence of the Fab arms. The studies by Tao et al. and Xu et al. indicated that the Ser 33 1 residue rnay be part of this lesion.

1.4.3.5 The role of the Asn 297 carbohydrate on CI binding and activation Two observations suggest that the y-chah glycan at position 297 plays a role in the IgG-C lq interaction. The carbohydrate moiety occurs on the putative C lq binding domain and furthemore prevents direct contact between adjacent C$ domains thereby influencing the accessibility of residues in this domain (Deisenhofer et al., 1976; Sutton and Phillips, 1983). The role of the Asn 297 glycan in regulating Cl binding by IgG has therefore been extensively studied. Using site-directed mutagenesis, Tao and Morrison (1989) were able to alter the tripeptide glycosylation sequence (Asn-X-ThrLys) at position 297 and produce specifically aglycosylated human IgG. Both a conservative Asn to Gin mutation and two nonconservative Asn to His and Asn to Lys mutations were made at the point of carbohydrate attachrnent. In three mutants, the C lq binding was either decreased or Iost entirely. These results confirmed earlier experiments which had also shown that the carbohydrate was important for the IgG-Clq interaction. Mouse IgG2b and IgG2a produced by cells treated with tunicamycin and therefore devoid of carbohydrate exhibited a three to four-fold reduction in complement-mediated lysis and Clq binding activity as compared to the saine irnmunoglobulins produced in the absence of tunicarnycin (Nose and Wigzell, 1983; Leatherbarrow et al., 1985). Similarly, glycosidase treated rabbit and human IgG were found to be less effective than the untreated controls (Winkelhake, 1980; Koide et al., 1977). Antigen binding was unimpaired by the glycosidase treatment indicating that the loss of activity was not due to the inability of the altered molecule to bind to the target. Furthermore they demonstrated that the gross structure of the junction between domains Cy2 and Cy3 was not significantly altered since the treated molecule was as effective as the untreated molecule in binding to protein A. Circumstantial evidence suggested that C lq does not interact directly with the y- chah glycan. In one study, the lectin concanavalin A bound to the carbohydrate moiety of IgG but did not affect the binding of Clq to IgG (Van Schravendijk and Easterbrook- Smith, 1985). In another study, all of the polysaccharides tested were unable to inhibit the interaction between Clq and IgG even though several of the charged polypeptides studied were found io be inhibitory (Emanuel et al., 1982). Furthermore it seemed unlikely that a glycan with so many possible structures codd be involved directly in this interaction. A recent study provides evidence that the carbohydrate structure affects the positioning of the Cy2 domains with respect to each other and that this in tum affects the availability of the amino acid residues involved in the Clq-IgG interaction. White et ai. (1997) showed that the orientation of the oligosaccharide chain and not its structure determines the complement binding activity of IgG. They found that two murine monoclonal IgG2a antibodies which shared identical constant region domains and had the same carbohydrate composition differed in their interactions with Clq, C3b and C4 complement proteins. Analysis of the two immunoglobulins showed they had different lectin-binding profües. men they tested the sensitivity of the carbohydrate moieties to peptide N-glycosidase F digestion and reactivity with P- 1,4-galactosyltransferase they found that the difference in the lectin binding properties of the two monoclonal antibodies was a function of N-glycan accessibility. It was also found that the IgG1 with the less accessible carbohydrate structure was more effective at classical complement pathway activation. Because a sequestered carbohydrate structure must necessarily coincide with a more exposed C-y2 domain, these findings provide further evidence that an important role of the carbohydrate group is to hold the pair of domains in a specific fixed orientation by acting as a 'spacer' (Winkelhake et al., 1980).

1.4.4 CI binding site on IgM C 1 binding activity is in the Fc region of IgM. Most of the early attempts to localize the Cl binding sites in the IgM molecule involved fragment studies. Plaut and Tomasi found that like IgG, IgM is susceptible to proteolytic cleavage (Plaut and Tomasi, 1970). Furthemore, under carehilly controlled temperature conditions, they were able to obtain a 342 kD tryptic fragment of human IgM corresponding to Fcps, a pentamer of Cy3- and Cp4-containing Fcp units. Complement fixation assays on the non-aggregated Fcp5 suggested that a cornplement binding site was localized within this fragment (Plaut et al., 1972). They also found that this fragment was consistently better at fixing complement than the parent pentamer in several IgM species, with efficiencies ranging from 10 times to as much as 3 1 times that of the intact molecule. This may well have been one of the first indications that the Cl fixing sites are norrnally inaccessible in IgM and interaction with antigen is required for their exposure. Bubb and Conradie (1976) added to the emerging picture of Cl fixation by IgM with studies which coafmed the activity of Fcps, but contrary to the results of the previous group. they found that. on a molar basis, the activity was less than that of the intact molecule. By measuring the ionic strength dependency of the interaction between Clq and uncomplexed IgM, Poon et al. (1985) were able to determine the number of salt ions released upon the binding of Clq and calculated that six salt bridges are formed as a result of this interaction. Wright et ai. (1988) performed a similar measurement but on complexed IgM. They found that nine sait bridges are formed and suggested that the interaction involves the formation of three salt bridges between each of three globular heads and three sites on the complexed IgM polymer. These studies indicate that charged amino acid residues are involved in the C lq and IgM interaction as was the case for the C lq and IgG interaction.

1.4.4.1 Evidence supporting the role of the Cp4 domain

As methods improved, it becarne possible for researchers to isolate intact Cp3 and Cp4 domains. Such fragments were tested for CI fwng activity. The main conclusion drawn frorn these studies is that the complement fixing activity of the Fcp5 fragment could be accounted for entirely by the CF~domain and that no activity was present in the Cp3 domain (Bubb and Conradie, 1976; Hurst et al., 1976; Bubb and Conradie, 2977). In particular a fragment of the Cp4 domain containing residues 468-491 disulfide linked to residues 519-546 was found to fix cornplement (Hurst et al., 1976). Since the first 24 amino acids were contained within the overlap between this fragment and another complement-€hg CNBr fragment, they suggested that the Cl fwng site lay within this stretch of amino acids. This was later confmed by Johnson and Thames (1976) who found that a synthetic peptide conesponding to residues 487-49 1 at the C-terminal end of this 24-amino acid stretch was able to fx complement. 1.4.4.2 Euidence supporting the role of the Cp3 domain Drawing conclusions from the peptide inhibition studies was problematic since it could not be proven that the small peptides retained their native three-dimensional structure. As a result, researchers continued the search for the CI binding domain using other experimentd avenues, and despite the conclusions derived from the earlier studies, these experiments suggested that the Cp3 domain contained the CI binding site. Leptin and Melchers (1983) established a panel of monoclonal antibodies with specificities for each of the four p constant domains. It was discovered that the ody two antibodies in this set capable of inhibiting the binding of Cl to IgM were directed against the Cy3 domain. At about the same time, Shulrnan et al. (1982 and 1986) were able to isolate two ce11 lines which produced mutant IgM with normal hapten binding capacity but which were defective in initiating the complement cascade. Upon further characterization, it was found that both mutant phenotypes were attributable to single point mutations in the Cp3 domain, S406N and P436S. It was further shown that the P436S defect resulted from the lack of binding to Clq (Wright et al., 1988). Arya et al. have produced the most comprehensive set of site-directed mutants to date on residues of the Cp3 domain (Arya et ai., 1994). According to this report, residues important for CML activity are localized within two distinct clusters at opposite ends of the domain. These residues are depicted in figure 1.13 in a proposed model of Cp3. One cluster occurs on the loop joining @2 to Q3 and includes residues P434, P436, and D432, whereas the second cluster includes residues D417 and C414 on the loop connecting fx4 to fy2, and D356 on the bend comecting fxl to fx2. The P436 and D432 residues in this first cluster lie on the penphery of the Fcg disc according to a model of the IgM pentamer generated from synchrotron X-ray scattering data (Perkins et al.,l99 1). Since mutations in this second cluster affect both polyrnerization and CML activity of IgM, Arya et al. (1994) suggest that these residues do not directly contact Clq, but are instead important for rnaintaining the correct local conformation of the Cp3 domain. 'fx' face

Figure 1.13. Putative location of residues in the Cp3 domain that are critical to the Clq binding site of IgM. This mode1 of the Cp3 domain is based on the X-ray structure of the Cy2 domain (Deisenhofer et al., 198 1). The B-pleated sheet strands making up the face of the domain are striped and numbered from right to left. The residues have been located on the mode1 by reference to sequence identities between p (Ou)and y (Eu)heavy chains. Adapted fiom Arya et al. (1994). 1.4.4.3 The role of the Am 402 carbohydrate on CI binding and activation

As was the case for IgG, the role of the carbohydrate moiety in complement activation has also been studied. However, unlike IgG, IgM has more than one glycosylation site (figure 1.7), with five carbohydrate chahs dispersed throughout the constant domains of the p.-chain. Of these sites, Asn 402 is believed to be equivalent to the Asn 297 of IgG. Shulman et al. (1986) isolated a mutant IgM which was defective in complement- mediated lysis. Chmcterization of this mutant showed that it contained a Ser to Asn mutation at 406 and that this point mutation caused hyperglycosylation of the heavy chah Wright et al. (1988) further studied this mutant and determined that the abnormal glycosylation occurred at position 402 and that the defect could be reversed by processing the IgM in the presence of deoxymannojirimycin. Protein produced in the presence of deoxymannojirimycin contains a high mannose oligosaccharide structure similar to that at position 563 in place of the complex oligosaccharide chah that is normally present on Asn 402. Deletion of the oligosaccharide at position 402, achieved by mutating the glycosylation sequence Asn-X-ThrLys at position 402, showed that aglycosylation at this site did not alter complement binding so much as the ability of the IgM to polymenze (Muraoka and Shulman, 1989). The role of the carbohydrate in IgM is therefore similar to its role in IgG. It probably serves its purpose by maintaining the structural integrity of the Fc region necessary for Clq binding and does not to directly contact Clq during binding and activation.

1.4.4.4 The role of J chain in regulating the interaction between IgM polyrner and complement It has been shown that IgM hexamers are from six to twenty times more efficient at initiating the complement pathway than IgM pentamers (Davis and Shulrnan, 1989b; Randall et al., 1990; Fazel et al., 1997). Since hexamers do not contain J chain, this indicates that the J chain is not directly involved in the irnmunoglobulin-complement interaction. The studies by Randall et al. suggest that the role of the J chah is to regulate the size of the polymer, which in turn affects the potency of IgM in complement-mediated lysis. One explanation for this observed effect is that the increased valency may increase the avidity of antigen or complement binding. Another possible reason is discussed later in

the discussion section of Chapter 4.

1.4.5 Heterologous systems in the study of the interaction between IgG, IgM and CI Although mouse immunoglobulins are abundant and comrnonly studied, mouse cornplement proteins are not as readily available. On the basis of the evolutionary conservation of complement proteins and their sequences, it has been a cornrnon practice to use complement proteins from a heterologous source if proteins from a homologous source cannot be obtained. Guinea pig complement, for example, is often used as a source for studying interactions between complement and mouse or human immunoglobulins. When interpreting the results of such studies, it is often a concem that the heterologous system does not accurately represent the homologous system. Studies have indicated that immunoglobulins have binding preferences for cornplement from specific sources (Fust et al., 1977; Fust et al., 1980; Zollinger and Mandrell, 1983; Sasaki and Yonemasu, 1984) but other factors can also give rise to misleading reports of such preferences. When soluble antigen is used, human IgM does not activate guinea pig complement, however when particulate antigen is used, human IgM does activate guinea pig complement (Van der Zee et al., 1986).

1.4.6 Modeis of CI binding and activation by IgG and IgM

The nature of the interaction between CI and IgG has been well-documented. Association constants on the order of - 104 M-1 have been reported for the interaction between C1 and monomeric IgG (Sledge and Bing, 1973; Müller-Eberhard, 1975; Hughes-

Jones, 1977) and of - 108 M-1 for the interaction between Cl and chemicaily cross-linked IgG (Burton, 1985). Throughout these studies, there is every indication that the Cl binding site of uncomplexed monomeric IgG is available, although the binding is measurably weaker. What is evident however, is that although uncornplexed IgG can bind Cl, it is unable to activate it. Borsos and Rapp (1965a and 1965b) have shown that the minimal IgG-antigen unit required for binding and activation of Cl includes a pair of IgG molecules. This is consistent with the distortional mechanism of Cl activation explained in a previous section. According to this model, unless two or more of the Clq globular heads are simultaneously engaged, activation will not occur since the distortional signal necessary for Cl activation cannot be created. Unanchored monomeric IgG then could not conceivably effect this outcorne because of its inability to bring about this distortion in Clq. IgG molecules aggregated by heat treatment, chemical cross-linking, and genetically engineered disulfide bridged polymers with as few as two monomenc subunits are able to initiate complement lysis in the absence of antigen (Hyslop et al., 1970; Metzger, 1974; Metzger, 1978; Wright et al., 1980; Smith et ai., 1995). These findings suggest that the role played by antigen is to bnng two Clq binding sites into close association. This 'associative' model has corne to be accepted as the mechanism by which antigen-bound IgG binds and activates C 1. In contrat to IgG, IgM is already multivalent with respect to Clq binding sites. Depending upon whether one or two sites exist on each H2L2 subunit and depending upon the stenc availability of these sites, an IgM pentamer may have as many as ten potential Clq binding sites. A single molecule of IgM should therefore be able to both bind and activate Cl. Akhough this has been confmned for antigen-bound IgM (Borsos and Rapp, 1965a) a study suggests that this is not possible for uncomplexed IgM. Poon and Schumaker (199 1) showed that, under conditions of physiological ionic strength, soluble IgM did not interact with Clq. Similarly circulating IgM does not bind or activate Cl in vivo. This is important from a physiological standpoint. If circulating IgM could interact with Cl, the complement systern would be indiscnminately activated and consumed causing systemic tissue damage in the process. Two models, the 'aliosteric' model and the 'distortional model' (not to be confused with the distortional model of Cl activation) have ken proposed to exphin the lack of CL binding by uncomplexed IgM and its subsequent activity in the presence of antigen. According to the allosteric model, antigen-binding induces a signal to be transmitted through the Fab to the Clq binding site putatively located in Cp3, causing it to be expressed. Evidence produced by Brown and Koshland (1975. 1977) showing that hapten binding at the Fab end of the rnolecule was able to induce conformational changes in the distally located Fc region supported this model. In one study they showed that IgM complexed with monovalent or multivdent antigen was not recognized as well by a J chain specific antibody as was uncomplexed IgM. This suggests that the J chah epitopes near the Fc portion of IgM are somehow affected by antigen binding. Strangely enough, the binding of monovalent hapten had no effect on 3 chah epitope expression. In a related study, IgM complexed to monovalent antigen was found to fix complement as well as multivdent antigen indicating that the ability of the IgM to fx complement directly resulted from filling of the antigen combining site. In a review on this subject, Metzger (1978) pointed out some inconsistencies between these two studies. Two moles of determinant per mole of irnmunoglobulin, for example. gave 50% of the maximal effect in the J chah study, but 10 moles of deteminant per mole of UMiunoglobulin produced no effect in the fmation assays. Furthemore, Metzger calculated that only 1% of the antigen combining sites were occupied in the second experiment when the hapten concentration was adjusted to give maximal cornplement fixation. Similarly only 0.02% of the antigen binding sites could be occupied at 50% maximal complement fixation. Brown and Koshland's data are also inconsistent with the reported ability of Fcps fragments to fix complement (Bubb and

Conradie, 1976). In the same article, Metzger provides more arguments against a generai allostenc model of immunoglobulin and complement interaction. If the ailosteric mechanism held true for both IgG and IgM, one would expect that the peptide sequence responsible for conveying the signai from the Fab to Fc, the hinge in the case of IgG and the Cp2 domain in the case of IgM, would be better conserved than it is. Subsequent to

Brown and Koshland's studies, Karush et al. (1979) published a report containing contradicting results. Using lactose linked to one or two of the cysteine residues of p2- microglobulin as monovalent and multivalent antigens respectively, they showed that only IgM bound to the latter could consume complement. Attempts to measure long range conformational changes in IgG and IgM using fluorescence depolarization (Nezlin et al., 1973; Holowka and Cathou, 1976a; Holowka and Cathou, 1976b) and circular dichroism (Cathou and Dorrington, 1974) have not been fruidul. According to the distortive model, the multivalent binding of hapten forces an otherwise rigid structure to take on an alternate confornation thereby revealing a previously concealed site or creating a new one. What distinguishes this model from the allosteric model is that several antigen combining sites must be occupied for distortion to occur.

Experimental evidence has shown that severai Fab arms must be antigen-bound before IgM can fix complement and suggests that the distortive model is applicable to IgM (Beale and Fazakerley, 198 1). By controlling the peptic digestion of anti-Salmonella porcine IgM, Beale and Fazakerley were able to generate severai different populations of IgM with an average number of Fab arms ranging from zero to ten. They found that the population containing mostly three- and four-armed antibodies could bind and agglutinate Salmonella, but could not fu< complement even though antibodies with more arms could. These results suggest that four or more arms must be engaged to cause the necessary conformational change in IgM. The distortional model is fbrther supported by electron micrograph studies which show that bound and fixe IgM assume different conformations (Feinstein and Munn, 1969). According to these studies, soluble IgM pentamer exists in a fully extended 'star' conformation whereas IgM bound to bacterial flagella has a 'staple' conformation (see figure 1.14). In the 'star' conformation the Fab arms are fully extended and Lie within the plane defmed by the Fcps disc, but in the 'staple' conformation, dl five Fab amis are bent away from the Fcp disc toward the antigenic surface. Borsos et al. (1981) found that the ability of ceU-bound IgM to induce complement-mediated lysis depends upon the density of surface-bound antigen on the target. By a titration expenment, they found that at low enough densities, no hernolysis of the target ceil resulted. The authors suggest that at low densities the average distance between hapten sites exceeds the span attainable by a single IgM molecule. Under such conditions, IgM would only bind monovalently and would therefore not undergo the 'star to staple' conformational change visualized by Feinstein and MUM (1969). More recently, a paper by Ohishi et al. (1995) showed that membrane IgM (bound via a C-terminal transmembrane anchor not present in secreted IgM) which normally does not bind C 1, cm be induced to activate complement if cross-iinked by an anti-lambda antibody (Ohishi et al., 1995). The results from the proteolytic, EM, titration and cross-linking studies provide convincing evidence for the distortional mechanism of IgM binding to Cl. It is interesting to note that in order to prevent circulating IgG and IgM from activating complement, nature has evolved different safety measures which can only be by-passed by antigen binding (or non-biologically relevant artificiai means). In the one case, IgG must be polyrnerized in order to function, in the other case IgM, whose polymer state is necessary to overcome the inherently low binding affinities of the antigen-combining sites in this primary response irnmunoglobuiin, requires further distortional changes. A 'Sbr Conformation' 'Staple Conformation'

Fab ams

Figure 1.14. 'Star' vs. %tapie9 conformations of pentameric IgM. Shown are A, the schematic diagrams; B, molecular models; and C, electron micrographs of these two foms. The staple form of IgM is shown bound specifically to bacterial flagella. The electron micrographs and models are fkom Feinstein and Munn (1969) and Feinstein et al., (1971). 1.5 Project Rationale The focus of this thesis stems from the cumulative research in this laboratory which for many years has been aimed at understanding the structurai basis for the various functions of immunoglobulins. Our more immediate interests have shifted from the earlier studies of the IgG molecule and its interactions with various effector fûnctions, to the less- studied but even more complex IgM molecule. In paaicular, the goal of ihis project was to further our understanding of the nature of the interaction between IgM and the Fust component of complernent.

As will be detailed in the introduction to Chapter 3, our iab bas previously made use of a proline to glycine mutation at position 544 of mouse IgM to create an IgM monomer in usable quantities. Despite being arrayed on a haptenated red blood ce11 surface, the rnonomeric mutant IgM was unable to initiate complement-mediated lysis, although a rnouse IgG2a control was active under the çame conditions. Further studies showed that the monomer IgM array was defective because it could not bind Clq (Taylor et al., 1994). To explain these observations concerning the inactivity of rnonomeric IgM, we proposed the following three possibilities:

(1) The Clq binding site of monomer IgM is inherently active, but is not exposed,

(2) the Clq binding site of monomeric IgM is exposed, but not active, or (3) the Clq binding site of rnonomeric IgM is neither active nor exposed.

These possibilities imply that the cis interactions of a p domain rnay affect its Clq-binding potential. Our approach to differentiating among these possibilities was to engineer a set of IgMRgGZb mouse hybrids in which various combinations of the comparable domains are interchanged using recombinant DNA technology. The reader may recall that the isolated Cy2 domain has been shown to be as active as the entire IgG molecule (Ellerson et al., 1972). It thus follows that the Cy2 is fully active in an isolated domain and equally importantly, that the structural environment within which it is found, namely the remabhg

IgG dornains, permits the expression of the binding site within this domain. By transplanthg the putative Cl binding domain of IgM into the permissive IgG background and studying the activity of the resulting monomer, we can determine whether or not the domain is 'inherentiy' active. By the term 'inherent' we mean that is has a preformed binding site. Conversely, by placing the active Cl binding domain of IgG ioto the corresponding IgM background, we can determine whether or not this environment affects the expression of the transplanted domain. By individually replacing the flanking domains with their IgG counterparts and studying the effects of these hybrid environrnents on the Cy2 domain, we cm hirther refine our picture in terms of the contributions of the individual domains. To ensure that differences in the activities among the generated hybrid imrnunoglobulins were a direct result of differences in their respective constant regions, they were engineered to contain identical V-regions with the moderately high affinity MOPC 3 15 V-region being chosen (Kafor TNP - 5 x 106 M-1). Because the intrinsic affïnity for the hapten TNP is a hundred fold greater for the MOPC 3 15 V-region pair than for the Sp6 V-regions present in the P544G mutant IgM whose Cl binding activity as a monomer had previously been found to be lacking (Taylor et al., 1994), as a preliminary to the IgGlIgM hybrid experiments presented in Chapter 4, it was important to fvst establish whether the lack of Clq binding by P544G IgM monomer was aiso seen when the higher affinity MOPC 3 15 V-region was employed. This question is addressed in Chapter 3. Finally, since some of the hybrid molecules will exist as polymers, a further objective is to elucidate the role of polymerization in the expression of the Cl binding site of IgM. This will be discussed in conjunction with the monomer results in Chapter 4. CHAPTER 2

Materials and Methods 2.1 DNA Preparation 2.1.1 Transformation of competent RB101 cells Transformation of HB 101 ceils made cornpetent by the calcium chlonde method (Life Technologies), was performed as directed by the supplier. Briefly, 100 pl of competent ceils were thawed on ice and mixed with the source of DNA (purified or in a two-piece ligation mix, 10 pl volume or less). After a 30 minute incubation on ice. the cells were heat-shocked at 42°C for 40 seconds, retumed to ice for another 2 minutes and then left to recover at 37°C in one ml of LB before they were selected on LB-ampicillin (150 mgll) agar plates.

2.1.2 Preparation and transformation of electrocompeten t cells

Briefly, 600 ml of LB (2% w/v tryptone, 1% w/v yeast, 1% w/v NaCl, 0.2% w/v glucose, pH 7.2) was inoculated with HB 101 cells and grown until the absorbance at 600 nm was between 0.5 and 1.0. The cells were then immediately chilled on ice for 30 minutes and harvested by centrifugation at 5000 rpm for 10 minutes at 4°C (RC-5 Superspeed Sorvall Refngerated Centrifuge; Du Pont Instruments). AU subsequent steps were performed on ice or at 4°C. Cells were resuspended in a liter of ice cold Hz0 and spun down. This wash step was repeated once with 500 ml of H20 and once with 20 ml of 10% glycerin. Finally, the cells were resuspended in a volume of 2 ml 10% glycerin, subaliquots were made and the cells were stored at -70'C until use. Transformation of three-piece ligation mixtures was performed by electroporation since the efficiency of transformation by the calcium chloride method was too low. Before the ligation mixtures could be used for transformation they were dialyzed to remove any salts using type VS 0.025 pM membrane (Millipore) dialysis discs. Specifically, four @ of each mixture was spotted ont0 the top of the dialysis disc which was then fioated atop a reservoir of sterile water (one liter) that was being gently stirred. After four hours at RT, the liquid was then transferred fi-om the top of the disc to Eppendorf tubes, vacuum dned and redissolved in 20 pI of Hz0 . Meanwhile, electrocompetent cells were thawed at RT and transferred to ice. 10 pl of DNA was mixed thoroughly with 40 pl of the cells in a sterile 0.2 cm cuvette and pulsed in a Bio-Rad E. coli Pulser Transformation Apparatus (Bio-Rad Laboratories) at a setting of 25 pF and 2.5 kV. Time constants for al1 the saruples, ranging from 5.1 to 5.4 mec, were within the acceptable limits. Pulsed cells were immediately transferred to one ml of SOC medium and placed in a 37'C incubator to recover for 1 hour, with gentle rocking, before they were plated on selective media.

2.1.3 Purification of pkasmid DNA Both small scale and large scale plasmid DNA was prepared by the aikaline lysis method of R. Treisman (Sarnbrook et al., 1989). Transformed bacteria grown in TB were harvested, lysed and the DNA prepared as previously described. Bacterial RNA was removed by digestion with ribonuclease 1 "A" from bovine pancreas (Pharmacia Biotech). Large scale preparations of DNA were subjected to an extra purification step by isopycnic centrifugation through a cesium chloride gradient containing 0.4 mg/d ethidium bromide.

2.1.4 DNA agarose gel electrophoresis Gel electrophoresis through agarose slab gels, prepared using the TAE buffer system described by Sambrook et al. (1989), was used for determining the size and quantity of a DNA sample as well as for separating different-sized DNA fra,gnents for purification. Gels prepared with 0.8-1.5% agarose in 1 x TAE buffer containing 0.25 pghl ethidium brornide were run at a constant voltage (100V)for 20 minutes or more until separation was achieved. Ethidium bromide stained nucleic acid bands were visualized by ultraviolet Iight.

2.1.5 Purifcation of DNA restriction fragments front agarose DNA fragments separated by agarose gel electrophoresis were purified by a modification of the gel explosion technique previously described (Chan, 1996). Slices of agarose containing the DNA fragment were minced using a razor blade and transferred to an Eppendorf tube containing an equal volume of TE-saturated phenol (pH 8.0). After vortexing, the tube and its contents were snap-chilled in a dry icelethanol bath (less than 5 minutes) and spun in a rnicrofuge for 5 minutes at 13000 rpm. The aqueous phase was extracted once with an equal volume of a 1: 1 phenol chloroform, twice with a haIf volume of chloroform and the DNA was precipitated from the aqueous solution by adding a half volume of 7.5 M ammonium acetate and a 2.5 times the total volume of ice cold ethanol.

The precipitate formed after the solution was cooled for 20 minutes at -70°C was collected by centrifugation and washed twice with 70% ethanol.

2.1.6 DNA sequencing Double-stranded plasmid DNA was sequenced by the dideoxy sequencing method using the T%equencingTM Kit (Pharmacia Biotech) according to the manufacturer's instructions. To screen clones for the point mutation that corresponds to the amino acid at position 544 of the wildtype IgM Cp4 domain, we sequenced clones using the forward primer S'CCTGCAGACATCAGT3' and confirmed the results by sequencing the same clones with the reverse primer STCTCACTCTGACATGG3'.

2.1.7 Restriction enzyme digests and ligation reactions Restriction enzymes and modifying enzymes used in these strategies were obtained from New England Biolabs and Boehringer Mannheim and the reactions were generally carried out under conditions detennined by the manufacturer using the supplied buffers. DNA fragments were ligated with T4 DNA ligase. The vector and the insert were ligated at a 1: 1 molar ratio starting with 200 ng of the vector and the correspondhg molar amount of the insert in a 10 pl volume using 200 units of ligase (New England Biolabs unit definition). Ligation mixtures of fragments with compatible ends were incubated overnight at lS°C whereas blunt-end Iigation mixtures were incubated at RT. The efficiency of blunt- end ligation was fbrther improved by using ten times the amount of enzyme as in the sticb-end reactions (high concentration T4 DNA ligase, New England Biolabs). We were able to fil1 in the 5' overhangs resulting from the restriction digests by incubating 20 pi of the DNA sample with a solution containing 3 pl of IO x polymerase buffer, 3 pl of 10 rnM Dm,3 pi of a mix containing 5 mM of each dNTP, and 1 pl of the Klenow hgment (1Ulpl)for 15 min at 37'C. After the reaction was cornplete, the enzyme was inactivated by phenol/chloroform extraction.

2.2 Construct Strategies 2.2.1 Heavy chain vector constructs The starting IgM and IgG2b heavy chain vector constructs, pSV2neo-VH-Cp and pSV2neo-V~-Cflb respectively, were generously provided by Dr. Michel Klein (University of Toronto) and have been previously descnbed (Rinfret et al., 1990). Briefly, these pSV2neo-based vectors contain the cloned v~315 segment, the IgH enhancer and the corresponding Balblc germiine sequence for the IgM and IgG2b heavy chah constant domains (figure 2.1). The creation of heavy chah vectors for the IgMAgG2b hybrids required multi-step strategies that involved partial digests, extraneous site bock-outs, subcloning and two- or three-piece ligations. In several cases, an important restriction site occurred elsewhere in the sarne plasmid. Two different methods were used to specificdly utilize one of the two available restriction sites. In one method, these plasmids were partially digested so that the majority of the digested products were the result of a single cleavage event, approximately half of which were cleaved at the desired position. The linearized plasrnid was purified from the mixture which included uncut plasmid and srnaller fragments resulting from double cleavage. After ligating in the insert and transformation, differentiation between the two possible products was achieved by "diagnostic" restriction digest analysis. In cases neo

IgH enhancer , 7

Figure 2.1. The heavy chain expression vectors for rnouse IgG2b and IgM. S hown are the pSv2-V~-CyZb(top) and pS V2-VH-Cp @ottom) plasmid vectors encoding the complete heaw chahs for IgG2b and IgM respectively. In both plasmids, the region flanked by the 5' Xba I and the EcoR 1restriction sites contains the IgH enhancer as well as the DNA sequence for the MOPC 3 15-denved VH-region (VH3 15). The pSV2-V~Cflb plasrnid was created by inserthg the 5.5 kb EcoR 1bhnt-endedlSal 1 fragment containing the germline Cy2b constant region sequence (Balbk allele) into the Xho 1bluntendedlSa1 1 site of the modified psV2-V~315 vector. The pSv2V~Cpplasrnid was created by ligating the 4.5 kb Sa1 1 fragment containing the Balbk Cp gene into the Sa1 1 site of the pSV2- VH~15 vector. Additional restriction sites shown were used in the creation of the pBSM13+(C@-Cp4), pBSM1 3+(Cp2-Cp4)inc-ad pUC 19(Cy1-Cy3) shuttle vectors. where other complications prevented us from using partial digestion effectively, we ernployed the technique of extraneous site knock-out. This is similar to partial digestion except that aftcr the linearized DNA is punfied, it is treated with the Klenow fragment and nucleotides to fili in the 5' overhangs. When these blunt-ended fragments were religated with T4 ligase, the enzyme recognition sequence at the site was not regenerated (this is tnie of the restriction sites for which this was done), thereby knocking out that site. After the cloning step, restriction digest analysis was then used to differentiate between the two possible products and the correct product could then be subject to enzyme digest and subsequent steps. For enzymes with more than one additional site, we subcloned smaller fragments of the heavy chah sequences into cloning vectors to render the critical restriction site unique. The shuttle vector constructs which resulted from these manipulations wilI be described first.

2.2.2 Shuttle vectors

The pBSM13+(Cp2-Cp4) vector was previously cons~cted(Arya et al., 1994) by subcloning the 2.5 kb BamH VKpn 1restriction fragment of pSV2neo-VH-Cp, containing part of Cp2 and ail of the Cp3 and Cp4 domains (see figure 2.1), into the BamH YKpn 1 site of the pBSM13' (Stratagene) plasmid vector. The pBSM13+(Cp2-Cp4 vector, which contains the cornplete Cp2 domain in addition to the Cp3 and Cp4 dornains, was created by subcloning the 2.8 kb Ec1136 WKpn I restriction fragment of pSV2neo-VH-Cp. (see figure 2.1) into the blunt- ended BamH I site and Kpn I site of the pBSM13+ vector.

The pUC19(Cyl-Cy3) plasmid was constructed by inserting the 4.5 kb Mun 1 bhnt-endeasal I fragment of the starting vector pSV2neo-V~-Cy"b, which contains domains Cyl, Cy2 and Cy3 (see figure 2.1), into the Ec1136 WSal 1 site of the pUC19 plasmid vector. 2.2.3 Eybrid heavy chain plasmid vectors The strategy used to create the vector corresponding to the heavy chain of yypp, pSV2neo-V~-Cyl-Hy-C@-Cp4,is shown in figure 2.2. The CyZ and Cy3 dornains were cut from the plasmid pUC19(Cyl-C'@) by digestion with Ec1136 II and replaced with the 1.2 kb blunt-ended Hind III fragment fiom pBSMl3+(Cp2-Cp4) which contained the Cp3 and Cp4 domains. This step generated an intermediate plasmid pUC 19(Cyl-Hy-Cp3 - Cp4) from which the 4.5 kb fragment containing domains Cyl, Hy, Cp3 and Cp4 was

excised using EcoR 1 and Sa1 1. Substitution of the fragment for the corresponding Mun

YSal 1 fragment of pSV2neo Cy;?b~,, WO (Mun 1 and EcoR I ends are compatible), a variant of the starting wildtype construct in which the second Mun 1 site that was 3' of the

C@ domain and 5' of the Sa1 I site was knocked out. yielded the desired vector . The strategy used to create the vector corresponding to the heavy chah of wy, pSV2ne0-V~-Cyl-Hy-C~3-Cy3,is depicted in figure 2.3. First, the Cp4 domain was

removed from the shuttle vector pBSM13+(Cy2-Cp4)by digestion with Nhe 1and Mun 1 enzymes and the remaining ends were blunt-ended. Next, the 458 bp blunt-ended

Eco0 109 fragment containing Cy3 was inserted into the resulting site to produce the intermediate plasmid pBSM 13+(Cp2-Cp3-C@). From this intemediate, we isolated a

blunt-ended Hind III fragment containing both Cy3 and Cy3 domains and inserted these into the Ec1136 II site of pUC19 (Cyl-CP) in place of the CyZ and Cfl domains to create pUC19(Cyl-Hy-Cp3-Cy3). Finally, the EcoR YSal 1 fragment of pUC19(CyI-Hy-Cp3- Cy3) was substituted for the Mun I/Sal 1 fragment of the starting wildtype construct

pSV2neo Cy'b~~~K/O. The two-step strategy for the construction of heavy chah vector pSV2neo-V~-cpl- Cp2-Cs-Cp4, corresponding to the ppyp hybnd is shown in figure 2.4. In the fust step, the 880 bp Ec2136 Wblunt-ended Eco0109 fragment of pSV2neo-V~-C@b,that contains Cy2, was ligated into the blunt-ended AfZ WMun 1 digested shuttle vector pBSM13+(Cp2- Cp4)inc-in place of the Cp3 domain sequence. The resulting intermediate pBSM13+(C@ I 1. digestwim enzymes

1. aiest mth t a est mlh enzyme 2 mat with &Tz. i. &tww Blunt-end

YT4 DNA Ligase

Figure 2.2. The construct strategy for the yypp heavy chah expression vector. Shown is a schematic representation of the two-step constmct strategy for pSV2neoV~CylHyCp3Cp4. The ysonstant domain exons (solid) and the y-constant domain exons (shaded) are showo. Details of the strategy are given in the text. E: EcoR 1, Ec: Ecl136 II, H: Nind m,M: Mun 1, S: Sa1 1. Figure 2.3. The construct strategy for the yypy heavy chah expression vector. Shown is a schematic representation of the three-step construct strategy for pSV2neoV~CylHyCp3Cy3. The p-constant domain exons (solid) and the y-constant domain exons (shaded) are shown. The "*" denotes a site that was made unique by a site knock-out procedure. Details of the strategy are given in the text. E: EcoR 1, Ec: Ecl 136 II, H: Hind HI, M: Mun 1, N: Nhe 1, O:Eco0109 1, S: SalI. I 1. O' est mth enzymes 2 &now Bluntgnd 3. Treat vmt? AUc Phos I I 1. O* est GUIenzyme 1. Digest with an 2. &QW ~lunt-end 2 Treat wrth alk %: V v

14 DNA bgase -T- -T4 DNA -9 pUC19 Cyl -Hy-Q3-Q3

I I 1- Oigest Wh enzymes 1. Digest mth enzymes 2 Treat with aIk phas. I 1.0 est wi!h enzymes 2 knowWunt-end t. oigestmth 0 3. Treat nnth 81k phos. 2 Klenow bluntgnd 1 ûiiest with €c 1

T4 DNA LigaSe

T;i DNA Ligase b-- Y

Figure 2,4. The construct strategy for the ppyp heavy chain expression vector. Shown is a schematic representation of the three-step constnict strategy for pSV2neoV~Cp1Cp2Cy2Cp4. The p-constant domain exons (solid) and the y-constant domain exons (shaded) are shown. Details of the strategy are given in the text. A: AIfII, Bt: BstE Ii, Ec: Ed 136 n, M: Mun 1, O:Eco0109 1. C@-Cp4)inc. was then cut with BstE II and the 1.2 kb fragment containing part of Cp2, d

of Cy2 and part of Cp4 was ligated into the corresponding BstE II site of the original wfld- type pSV2neo-VH-Cp vector resulting in the formation of pSV2neo-VH-Cp1-Cp2-Cfl- Cp4. Positive clones were tested for the correction orientation of the insert by restriction

digest analysis with BamH 1. Correct orientation clones were expected to have a BamH 1 restriction pattern similar to that of pSV2neo-VH-Cp Illustrated in figure 2.5 is the strategy used to create the pSV2neo-VH-Cyl-Hy-Cy2- Cp4 vector, which encodes for the p'yyc~heavy chain. Briefiy, the 3 13 bp BstE II blunt- ended/ Bst XI restriction fragment containing the y-hinge was isolated from the pUC 19(Cyl -Cy3) shuttie vector and substituted for the Cp2 domain of the Ed136 WstXI-digested pSV2neo-VH-Cp1 -Cp2-C@-Cp4. The strategy used to create the vector corresponding to the heavy chain of ppyy, pSV2neo-V~-Cpl-Cp2-Cr+!-CP,is shown in figure 2.6. The two pieces (both were approximately lkb in size) of the BstX I/Snl I digest of pUC19(Cyl-Cy3), which together contain the Cs, Cy3 and the remaining 5' end of the originally cloned IgG2b fragment,

were ligated to the 8 kb Iarger fragment of the BstX VSal 1 digested pSVZneo-VH-Cpl-

Cp2-Cy2-Cp4~,1wo vector (the extraneous Sa1 I site 5' of the Cp1 domain was knocked out pior to this step) to create pSV2neo-V~-Cpl-Cp2-C@-C@. Figure 2.7 represents the strategy used to create the vector corresponding to the heavy chain of pyyy, pSV3neo-VH-Cp 1-Hy-Cy2-Cy3. First pUC19(Cyl-Cy3) was digested to completion with BstE II and Sal 1. To determine which of the resulting two fragments contained the Hy, Cy2 and Cy3 domains, a sample of the mixture was treated with Sac 1and the fragments were examined for susceptibility to Sac 1cleavage (the correct

fragment contains three Sac 1 sites whereas the other fragment contains none). Thus determined, the 4 kb band was purified and iigated to the BstE WSal I site of pSV2neo-VH- Cpl-Hy-CyL-Cp4sd (a modified form of ~SV~~~OVH-C~~-H~-CII;!-C~~in which the second Sai I site 5' of the Cpl domain was knocked out) in place of the last three domains. I I 1. Digest with enzymes 1.oiWithl3t 2 Treat with ak phos. 2 iüenow blunlgnd 3DigestmViBx

+T4 DNA Ligase

Figure 2.5. The construct strategy for the pyyp heavy chah expression vector. Shown is a schematic representation of the construct strategy for pSV2neoV~- CplHyCy2Cp4. The p-constant domain exons (solid) and the y-constant domain exons (shaded) are shown. Details of the strategy are given in the text. B t: BstE II, Bx: BstX 1, Ec: Ecl 136 Il[. I 1. Dgest mtt, enzymes 2 Tmat wmv akphos.

Figure 2.6. The construct strategy for the ppyy heavy chain expression vector. Shown is a schematic representation of the construct strategy for pSV2neoV~- CplCp2CyZCy3. The yconstant domain exons (solid) and the y-constant domain exons (shaded) are shown. The "*" denotes a site that was made unique by a site knock-out procedure. Details of the strategy are given in the text. Bx: BstX 1, S: Sul 1. I 1. Oigest vmth enzymes 2 Treat mth ak. phos. \Y

-T4 DNA L$ase

Figure 2.7. The construct strategy for the yyyy heavy chain expression vector. Shown is a schernatic representation of the construct strategy for pSV2neoVH- CplHyCy2Cy3. The y-constant domain exons (solid) and the y-constant domain exons (shaded) are shown. The "*" denotes a site that was made unique by a site knock-out procedure. Details of the strategy are given in the text. Bt: BstE If, S: Sa1 1. 2.2.4 Heavy chain plasmid vectors for the P544G mutants In order to create pSV2neo-V~-Cyl-Cp2-Cfl-Cp4~~4~~(figure 2.8), the 1.5 kb Apa YKpn I fragment of plasmid 1119633, a mutant of pBSM13+(Cp2-Cp4) in which the codon at position 1862 has been changed from CCA to GGA resulting in a proline to glycine mutation at amino acid 544 of the corresponding IgM protein, was inserted into pSV2neo-VH-Cp1 -Cp2-Cy2-Cp4 that had been digested with Apa 1and partially digested with KpB 1(Arya et al., 1994). Clones for which the desired Kpn 1site had been cut and in which the insert was correctly oriented were determined by examining their BamH 1 restriction digest patterns. Those with patterns identical to that obtained by the digestion of the pSV2neo-VH-C~~-C@-C@-Cp4 were then sequenced to confm the presence of the P544G mutation using primers flanking the region. The P544G mutant form of pSV2neo-V~-Cpl-yH-Cy2-Cp4(figure 2.9) was created by repiacing the C@ and Cp4 domains with the Cy2 and Cp4p544~domains derived from pSV2neo-V~-Cpl-CpZ-Cy2-Cp4~544~using the enzymes BstX 1and Pvu 1. Two sets of ligations were performed since the identity of the two resulting fragments from each digest was unknown (we lcnew that the Pvu 1site was in the Amp gene, but because not ali of the constant domain containing insert was characterized we did not know how fa. it was from the BstX site). In one reaction, the large fragment of pSV2neo-V~-Cpl-yH- Cy2-Cp4 was Ligated to the srnall fragment of pSV2neo-V~-Cp1-Cpî-CyZ-Cp4p544~and in the other, the small fragment of pSV2neo-V~-Cpl-yH-Cy2-Cp4was Ligated to the large fragment of pSV2neo-VH-Cpl-Cfl-C@-Cp4p544~. Successfid clones were identified by their BamH 1 digest patterns. Those which resembled pSV2neo-VH-Cp 1-yH-Cy2-Cp4 were Mersequenced for P544G-containhg mutants. The P544G mutant of the starting vector pSV2neo-VH-Cy was created by replacing the Cy2 domain of pSV2neo-V~-Cpl-Cp2-C?f;!-Cp4p5#~with the Cp3 domain (figure 2.10). Cy3 and part of Cp4 were isolated from pBSM13+(Cp2-Cp4)i,.. using Sul 1, which has a unique site in the pBSM13+ cloning vector sequence, and Apo 1, which has a 1 I 1. Digest mîh enzymes 1. Digest with enzymes 2 Treat mth alk. Ohos.

Figure 2.8. The construct strategy for the ppyp P544G heavy chain expression vector. Shown is a schematic representation of the construct strategy for pSV2neoV~CplCp2CflCj~4p544~. The p-constant domain exons (solid) and the y- constant domain exons (shaded) are shown. Details of the strategy are given in the text. Ap: Apn 1, K: Kpn 1. Figure 2.9. The construct strategy for the pyyp P544G heavy chain expression vector. Shown is a schernatic representation of the constnict strategy for pSV2aeoV~Cp1HyC~2Cp4~~~~~. The p-constant domain exons (soiid) and the y- constant domain exons (shaded) are shown. Details of the strategy are given in the text. Bx: BstX 1, P: Pvu 1. Figure 2.10. The construct strategy for the IgM P544G heavy chah expression vector. Shown is a schematic representation of the construct strategy for pSV2neoV~Cpp544~-The p-constant domain exons (solid) and the y-constant exons (shaded) are show. Details of the strategy are aven in the text. Ap: Apa 1, Ec: Ec1136 II, S: Sa1 1. unique site in the Cp4 just 5' of the codon encoding residue 544. The 1.3 kb Sa1 1blunt- ended/Apa 1 hagrnent was Ligated to the 11 kb Ecl136WApa 1 fragment of pSV2neoV~- Cpl-CpZ-Cy2-Cp4 in place of the Cy2 and the 5' portion of Cp4.

2.3 Immunoglobulin Expression and Characterization 2.3.1 Ce12 line MOPC 3 15.26, a lambda-chah producing, heavy-chah loss murine myeloma ceil line (Mosmann et ai., 1979) was kindly provided by Dr. Michel Klein (University of Toronto). These cells were maintained in alpha-MEM medium (University of Toronto Media Preparation) supplemented with 2 mM of either L-glutamine or GlutaMAX-II (Life Technologies), 100 U/ml peniciilin, LOO pglml streptomycin, 9 mM HEPES buffer, and 9% heat-inactivated fetal calf serum (HI-FCS)(Life Technologies). Seiection medium included 600 pg/d of G418 (Life Technologies). Cells were grown in a 37OC incubator, 100% relative humidity, 5% COz. Cells lines were frozen in 10% DMSO and 90% HI- FCS and stored in liquid nitrogen below -70°C.

2.3.2 Electroporation of MOPC 315.26 cells MOPC 315.26 cells were grown to a density of 7-8 x 105 cellslml in complete media (alpha-MEM with 9% heat-inactivated low IgG FCS (Life Technologies), 2 mM GlutaMAX-II (Life Technologies), 100U/ml penicillin, 1ûûpg/ml streptomycin and 4.5 ml of an amino acidlbiotin-Bi2 solution). On the day of transfection, 50 ml of culture per transfection was centrifuged at 2000 rpm for 15 minutes at 4OC. The harvested cells were then washed two hesin ice-cold ~a2+-and ~gz+-freePBS and resuspended in 0.8 ml of ice-cold PBS. Fifty pg of iinearized DNA dissolved in 20 pl of lxTE was then added to the suspension on ice and carefuily mixed to avoid introducing air bubbles. A separate reaction using 20 pl of TE was used as a negative control. The mixtures were transferred to a 0.4 cm cuvette and using an electroporation device Gene Pulser II (Bio-Rad Laboratories), the cells were electroporated at a setting of 700V,25m (time constant for al1 successfid nuis were between 0.6 and 0.7 mec) and immediately retumed to ice for 10 minutes. These cells were then resuspended in 40 ml of complete non-selective media and aliowed to recover at 37°C in the CO2 incubator overnight. On the following day, the ovemight ce11 cultures were harvested by centrifugation and resuspended in 10 ml of selective media containing G418 (Life Technologies) at 600 pg/d. The cells were counted and their viability was checked (a viability between 20-70% was ideal for maximal DNA uptake). They were then resuspended in selection media at a density of 5 x 103 cells/ml and plated at 200jUwell in three 96-well tissue culture plates. Only one plate was used for the control. Cells were also resuspended at 1 x 104 cells/rnl and plated at 1 müwell in one 24-well plate. After about 10 days. the supernatant in those wells containing visible colonies was assayed for antibody production by ELISA using a TNP-gelatin capture and goat anti-mouse lambda alkaline phosphatase detection (Southem Biotechnology).

2.3.3 Preparation of the TNP-gelatin coating protein Five ml of a 20 mghl gelatin solution was added to a solution containing 50 mg of TNBS (1 ml of a 5%w/v stock solution, Sigma Chernical Co.) dissolved in 10 ml phosphate buffer pH 7.4 with 0.02% sodium azide (PBS-Az), and incubated for 30 minutes at room temperature with constant mixing. After the incubation, the protein solution was transferred to a length of dialysis nibing (12000 MW cutoff) and dialyzed at 37'C against PBS-Az with several changes until there was no yellow colour in the dialyzing buffer. The TNP-coupled gelatin was then aliquoted and stored at 4'C in the dark until used. TNP-gelatin prepared in this way was used for both ELISA and Clq binding assays. Microtitre 96-well polystyrene plates were coated with TNP hapten by incubahg the wells ovemight at RT with TNP-gelatin diluted five hundred fold in 0.01 M sodium borate pH 9.2. Approximately 5 x 106 cells, having a viability of greater than 85%. were harvested by cenmgation from the growth media (alpha-MEM) at 700 rprn, 7 minutes, 4OC in a RC- 3B Sorvail centrifuge (Dupont Instruments) and washed two times with 4 mi of Hank's Balanced salt solution. Cells were then resuspended in 500 pi of methionine- and cysteine- free lx Dulbecco's Modification of Eagle's Medium (Flow Laboratories) supplemented with 10% heat-inactivated Fetal Calf Serum (Life Technologies), 2 rnM GlutaMAX-II (Life

Technologies), 9 mM HEPES, 10pg/d peniciliin and 100 uniislml streptomycin. After starving the cells for 1 hour at 37OC, 100 wi of Trans3sS-label (ICN Phannaceuticals, Inc.) was added as a source of radiolabeled L-methionine and L-cysteine. The cells were then incubated for 4 hours at 37°C to permit the incorporation of radiolabeled amino acids into the irnmunoglobulin products and then chased with an equal volume (500 pi) of complete alpha-MEM media for another 1 hour.

2.3.5 Immunoprecipitation Following metabolic labeling, the supernatant was collected from cells that were pelleted by centrifugation at 13,000 rpm in a table top centrifuge. To 950 pl of the supernatant, 238 pl of 5x NTSE detergent buffer (50 mM Tris-HCI, 0.75 M NaCl, 2.5%, NP40, 0.1% NaN3, 5 mM, EDTA in PBS pH 7.2) was added so that the protein was in a solution of 1 x detergent buffer. PMSF and iodoacetamide were aiso added to a final concentration of 1 mM each to prevent proteolysis and aberrant disulfide bond formation. The whole mixture was pre-cleared at 4OC with 30 pl of a 10% w/v S. artreus suspension (IgSorb, Enzyme Center, Inc.) made in lx NTSE. After a one hour incubation during which the mixture was constantly mixed by rocking, S. aureus was pelleted by centrifugation. Rabbit anti-mouse p-chah and rabbit anti-mouse y-chain antibodies (Organon Teknika Corp.) were added to the supematants of IgM, ppw, pyyp, yypp, yypy and the P544G derivatives and the solution was left to incubate at 4OC overnight. Eluates of the pre-clearance S. aureus pellets were analyzed by SDS-PAGE gel and in al1 eight cases, no major protein bands were apparent indicating that there was no non-specific

binding to the S. aureus. Since mouse IgG2b, and the p.^ and ppyy hybrids each contain an intact yFc, they were able to bind directly to the S. aureus and were therefore not treated with a pre-clearance step or precipitating antibodies but were instead directly precipitated with S. aureus. After the incubation, complexes containing the mutant irnmunoglobulins were immunoprecipitated with 30 pl S. nureus at 4OC for 1 hour. The precipitates were pelleted by centrifugation, resuspended and washed five times in lx NTSE to remove ail traces of non-specifically bound protein. A 30 pl volume of 2x Sample Loading Buffer (0.125 M Tris-HC1 pH 6.8, 4% SDS, 20% glycerol and 0.2% bromophenol blue) was then added and the bound protein was eluted from the S. aureus by boiling for 5 minutes. The samples were then loaded ont0 a non-reducing 2.8/3.5 % SDS-PAGE gel described below.

2.3.6 SDS-PAGE SDS-PAGE was performed by running slab gels in the buffer system developed by Laemmli (1970) using a mini-gel apparatus (Bio-Rad Laboratories). To resolve large proteins ranging from 150 kD to 950 kD (approximate molecular weights of IgG2b monomers and IgM pentarners respectively) we prepared gels reinforced with 0.5% agarose in which the bottom half was composed of 3.5% polyacrylamide and the top half was composed of 2.8% acrylamide. No stacking gel was used. To visualize light and heavy chah of the immunoglobulins obtained by reduction, we prepared 10% separating gels overlaid with 5% siacking gels. Both the reduction of disulfide bonds and the incorporation SDS were achieved by boiling the protein with an equai volume of 2 x sample loading buffer containing 50 mM dithiothreitol for 5 minutes and immediately loading it onto the gel. After electrophoresis, the gel was fixed and stained in a 0.25% solution of Coomassie Blue R dissolved in 45% methanol, 45% water and 10% acetic acid with mixing (RT, 15 minutes) and destained in several washes of 10% methanol and 7.5% acetic acid in water. The gels were then dried onto 3bf~chromatography paper (Whanan Ltd.) using a slab gel dner (80°C, 1 hr). Prier to drying, gels containing P-emitting isotopes (as in the metabolic labeling studies), were rinsed in water and soaked in 1 M sodium salicylate (RT, 1 hr). For autoradiography (or autofluorography), preflashed Kodak XAR-5 film was exposed to the dried gel (18 hr) and developed in a Kodak X-OMAT developer.

2.4 Purification of Immunoglobulins

2.4.1 Optimization of purification conditions Wild-type and chimeric recombinant immunoglobulins were purified from culture supematants using immunoaffinity chromatography. To improve the yield of monomers we fist optimized the hapten density of the afTinity matrix. DNP-lysine was coupled to CNBr Sepharose CL-4B (Pharmacia Biotech) at hapten concentrations ranging from zero to 10 rnM according to the manufacturer's instructions. Two pl portions of each of the various beads (packed) were incubated ovemight at 4'C with 1 ml of culture supernatant containing 4 pg of IgGZb. IgG2b remaining in the supernatant after incubation was absorbed ont0 S. aureus to separate it from the other proteins in the supernatant, eluted with SDS-PAGEsarnple buffer, and run on a 7% non-reducing SDS-PAGE gel. The intensities of the resulting bands were visuaily compared to those of standards obtained by absorbing known amounts of IgG2b ranging from 500 ng to 4 pg to S. aurew. Using this method, we found that al1 the beads tested bound approximately the same amount of immunoglobulin. Next we eluted the bound irnmunoglobulin with 20 pi of 50 rnM DNP- glycine (near saturation) for 30 minutes at 37'C. The eluates were treated with 50 mM DTT in SDS-PAGEsample buffer, boiled, and run on a 7% SDS-PAGE gel dong with standardized amounts of mouse MOPC 141 IgG2br (Sigma). Again. by comparing the intensities of the corresponding protein bands, the amount of IgG2b that was eluted from the different beads was deterrnined. As a consequence, DNP-lysine coupled to Sepharose at a concentration of 2 mM was chosen for fuaher optimization steps. To minimize the exposure of the protein to the 37'C elution temperatures, we tested the yields using different elution tintes. We settled on an elution time of 15 minutes because we found that most of the protein was eluted during this tirne.

2.4.2 Affinity purification of immunoglobulins The hybrid proteins were purified using the pre-determined optimal conditions. Culture supernatants containing about 1 mg of total immunoglobulin were batch absorbed overnight onto 100-200 pl of 2 mM DNP-Sepharose CL43 beads in a large Bask at 4OC.

The beads were then collected by filtration through a sintered glass hmel and washed with 200 ml of cold PBS-Az containing 0.2 rnM PMSF. The beads were then transferred to a 2 ml plastic chromatography colurnn and drained by positive pressure from the top using a syringe. The antibodies were then eluted at 37°C for 15 minutes using 1 ml of 50 mM DM-glycine (ICN Pharmaceuticals, Inc.) prepared in PBS-Az pH 7.4 (pre-warmed to 37'C) containing 0.2 mM PMSF. The eluate was collected using the syringe to expel the liquid through the matrix. Elution was repeated once. Residual DNP-glycine was then removed by three consecutive passages (separated by 1 hour to allow antibody-bound hapten to equilibrate) through Dowex ion exchange resin (1-X8,200-400 mesh, Sigma) that had been preabsorbed with heat-inactivated Fetal Calf Serum and washed with PBS-Az until the absorbance at 280 nrn of the washes was less than 0.001 (250 p1 packed volume). The sampIe was simultaneously depleted of free hapten, filtered and collected by centrifugation at 800 rpm, room temperature for 10 minutes (IEC centra MP4R; International Equipment Co.) through an improvised device as illustrated in figure 2.1 1. The device consists of a 3 ml syringe barre1 tipped with a 0.45 3 pm filter (Millipore) inserted into a 15 ml polypropylene conical tube.

2.4.3 Sepuration of the monomer species from the polymer The monomeric species were separated from the polyrneric species using a sucrose density gradient. Between 100 pi and 1 ml of affinity purified immunoglobulin was layered gently on top of a 520% sucrose density gradient prepared in 0.1% PBA (bovine serum albumin in PBS-Az) of total volume 33 ml and was centrifuged for 18 hours at 22 000 rpm in a swinging bucket rotor at 4°C (SW28; Beckman Instruments). Approxlmately 400 pl (twenty drop) fractions were collected by pushing 30% sucrose through the bottom of the tube with a 60 ml syringe and syringe pump (mode1 341A; Sage) and collected from the top using a sucrose density gradient fractionator (mode1 184; Isco).

2.4.4 Reduced and alkylated ELISA The amount of imrnunoglobulin in each fraction was quantified by a TNP-gelatin coat capture ELISA. In order to eliminate the hapten binding advantage of the polymer fractions, al1 sarnples were reduced under non-denaturing conditions in 10 mM DTï (a 10x stock solution was prepared in IM Tris-HCI, pH 8) for 1 hour at RT and aikylated with 26 rnM iodoacetamide (30 min RT) before they were added to the hapten-coated wells according to a procedure by Wiersma and Shulman (1995). Reduced and alkylated gradient samples were fust diiuted to a tenth of their original concentrations then subjected to five-fold serial dilutions with PBS-Az containing 0.18 gelatin. In total, the sarnples were tested at four different concentrations. Affinity-purified sarnples that had been spectrophotornetrically quantified were similarly treated with reducing and alkylating reagents and used as standards beginning at a concentration of 1 mghl (two-fold seriai dilutions were performed in PBS-Az containing 0.1% gelatin). Following detection with the appropriate alkaline phosphatase-conjugated detection antibody (Jackson 15 ml polypropylene conical tube

3 ml syringe

eluate

7Dowex resin

0.45 prn filter tip

Figure 2.11. Device used to remove free DNP-glycine from the affinity- purified immunoglobulin. The immunoglobulin-containing DNP-lysine Sepharose column eluate was layered over the Dowex resin and the whoie device was subjected to centrifugation. ImmunoResearch Laboratories Inc.), anti-yFc for wildtype IgG2b, anti-p for wild-type IgM and equal quantities of both conjugates for the mutants, the weils were developed with a solution of 0.25 mghl p-nitrophenyl phosphate (Sigma 104 substrate in 5 mg tablet form, Sigma) in ELISA substrate buffer at RT until the yellow colour developed. The extent of the colorimetric reaction was measured by reading the absorbance at 405 nm (Mode1 450 Microplate Reader; Bio-Rad Laboratories).

2.5 Functional Assays 2.5. L Buffers In the CML and C lq binding assays, the following veronal (barbital) saline buffers

were used (Borsos and Rapp, 1965): GVB, gelatin veronal buffer containing 0.1 % geiatin (physiological ionic strength); SGVB, sucrose gelatin veronal buffer made isoionic with sucrose and containing 0.1% gelatin (low ionic strength, p = 0.06). PBS, phosphate buffered saline (0.01 M sodium phosphate and 0.15 M NaCI, pH 7.4) was used in the purification of the immunoglobulins. Haptenated-red blood cells were prepared and stored in glucose phosphate buffer (G-PBS, phosphate buffer containing 1% 13-glucose). ELISA substrate buffer contains 48.5 ml diethanolamine titrated with 65 ml of 1 N HCl, 100 mg sodium azide and 50 mg MgCl2 made up to a final volume of 500 ml with H20.

2.5.2 Preparation of low density TNP-haptenated sheep red blood cells Two ml of 50% Whole Sheep Red Blood Cells (SRBC) in Alsever's Solution (Cedarlane) were washed four times in 20 ml of PBS-Az pH 7.4 until the supematant was clear. The packed cells were then resuspended in a solution containing 20 mg of TNBS (400 pl of 5% TNBS,Sigma) in 10 ml of PBS-Az final pH 7.4, and gently mixed at room temperature. After twenty minutes of incubation, the cells were then spun down and washed with PBS-Az until the yellowish orange colour disappeared from the supematant. The fmai peliet was then resuspended in 10 ml of PBS-azide, used immediately or wrapped in foi1 and stored at 4°C for no more than three days until required.

2.5.3 Preparation of high-density TNP-haptenated sheep red blood cells Two ml of 50% Whole Sheep Red Blood Celis in Aisever's solution (Cedarlane Laboratories Ltd.) were washed four times in 20 ml of G-PBS (see Buffers section) and the final supernatant aspirated. 788 pl of a 5% TNBS (39.4 mg) solution (Sigma) was adjusted to neutral pH by the addition of a molar equivalent of 2M NaOH and made up to a 2.1 ml volume with 0.28 M sodium cacodylate pH 6.9 (final solution 18.8 mghl TNBS, pH 6.9, ionic strength 9 mS). Next, 300 pl of packed SRBC were added dropwise to the TNBS soiution. After gentle mixing at room temperature for 30 minutes, the cells were washed in cold G-PBS containing 1% heat-inactivated FCS until the washes became colourless. The hi&-density TNP-haptenated sheep red biood cells were resuspended in the wash buffer and used immediately or stored in the dark at 4OC for no more than two days.

2.5.4 Preparation of guinea pig complement for complement-mediated Zysis

(CML) assays Lyophilized guinea pig complement (Cedarlane Laboratories Ltd.) was reconstituted with 1 ml cold Hz0 and absorbed with high density TNP-SRBCfor 45 minutes on ice. It was then aiiquoted and stored at -70°Cuntil use.

2.5.5 CML assay for polymers

2 x 107 low-density TNP-haptenated SRBC were added to each of a series of test tubes containing two-fold dilutions of polymeric immunoglobulin in GVB (total volume 200 pi) and incubated for 30 minutes at room temperature with gentle shaking. One ml of cold GVB was then added to each tube and after a quick shake, the cells were pelleted and the supematants were discarded. The wash step was repeated once. The fmal ceii pellet was then resuspended in 250 pl GVB and put into an ice bath. One ml of guinea pig complement, diluted 1:200 in GVB, was added to the tubes. The samples were then immediately transferred to a 37°C waterbath to incubate, with occasional shaking, und detectable lysis occurred in the positive control (about 15-30 minutes). The samples were then immediately cooled on ice and the unlysed ceils were separated from the supernatant by a brief centrifugation step. The extent of ceIl lysis was quantitated by measunng the absorbante of the supernatant at 412 nm, and cornparhg this reading to that of the control, cells lysed in an equal volume of H20.

2.56 CML assay for monomers

2 x 107 high-density TNP-haptenated SRBC were added to irnmunoglobulin monomer samples senally diluted (two-fold) in 250 pl of GVB and incubated at room temperature for 30 minutes. The cells were then washed twice in cold GVB as above and resuspended in 250 pl GGVB. One hundred pl of guinea pig complement ( 1:50 dilution in GVB) was added to each tube. Incubation at 37 OC proceeded until lysis was visible in the positive IgG2b controls. Lysis was stopped with the addition of 1 mi of cold GVB buffer and the degree of lysis was measured as in the previous assay.

2.5.7 CI q purification Human Clq was prepared using ion exchange chromatography followed by gel filtration according to the method of Tenner et al. (1981). Al1 steps were performed between O and 4OC. Two units of frozen human plasma (-600 ml) were thawed and made up to a 5 mM concentration in EDTA by the addition of a 0.5 M EDTA, pH 8.0 stock solution. After pre-filtering the plasma through glass wool to remove particulate matter, it was then loaded onto a 5.6 x 18 cm Biorex 70 column preequiiibrated with starting buffer (82 mM NaCl, 2 mM EDTA, 50 rnM sodium phosphate, pH 7.3, ionic strength 11 mS) which was run at a rate of 30 rnühour. Once the plasma entered the column, the resin was then washed at a rate of 60 mUhour with starting buffer until the absorbance at 280 nrn returned to background levels. Clq was then eluted from the column using a linear 82 to 300 mM NaCl gradient. In total, one hundred 12 ml fractions were collected. Every third fraction was assayed for the presence of Clq by measuring its hemolytic activity against Ig-sensitized sheep red blood cells (SRBC)when supplemented with the rernaining components of the complement cascade in the form of human Clq depleted serum (Sigma). Briefly, IgM-SRBC were made by incubating 10 ml of 1 x 109 SRBC/ml with 30 pg of an IgM fraction of rabbit antibodies against SRBC (Organon Teknika) for 30 min at 37°C and washed once with GVB buffer. The cells were then resuspended in GVB buffer at a concentration of 5.24 x 108 ceUs/ml. Four pl of the column fraction was added to the following cocktail: LOO pl IgM-SRBC, 10 pl Clq- depleted semm (Sigma) and 300 pl GVB and the mixture was incubated at 37OC for about

30 minutes until lysis occurred. The reaction was stopped with the addition of 1 ml of cold GVB buffer and the absorbance of the supernatant was measured at 412 nm to determine the degree of lysis. The hemolytic activity of the C lq-depleted serum alone was subtracted as the background. Fractions found to contain significant arnounts of Clq were then pooled, adjusted to 33% saturation in ammonium sulfate with solid ammonium sulfate and left oveniight at 4°C. The protein precipitate was collected by centrifugation at 10,000 rpm for 30 minutes at 4OC (RC-5Superspeed Sorvall Refrigerated Centrifuge; Du Pont Instruments) and the pellet was redissolved in 5 ml of a high salt buffer, 0.5M NaCl, 0.05 M Tris, pH 7.2. The sarnple was further clarified by centrifugation for 45 minutes at 18,000 x g, 4 OC and then carefully loaded ont0 a 2 x 140 cm Biogel Column A5m which had been preequilibrated in the same buffer. The column was run at a rate of 15-20 mi per hour and 2.5 ml fractions were collected (200 in total). Every third fraction was assayed for the presence of C 1 q. As a result, fractions 10 1- 137, which contained the peak of Clq activity, were combined and solid ammonium sulfate was added to 33% saturation and mixed at 4'C overnight to precipitate the protein. The pellet was redissolved in 5 ml of buffer containing 0.5 NaCl and 0.05 M Tris pH 7.2 and dialyzed extensively against 6 Liten of the same buffer at 4OC. The final absorbance at 280 nm was read and the fmal concentration and total yield were determined using an extinction coefficient at 280 nm of 6.8 for a 1% solution of Clq in lcm path length cuvette. The fmal product was then portioned into 200 pg aliquots and stored at -70°C unid use.

2.5.8 Radiolabeling of CIq According to IIeusser et al. (1973), Clq rnay be enzymatically radiolabeled with lactoperoxidase without affecting its hemolytic ability. The following is the modified protocol used in Our laboratory. To 360 pl of purified Clq protein (about 200 pg in 0.5

NaCl and 0.05 M Tris, pH ), 10 pl of 0.2 M sodium phosphate buffer pH 7.4, 2 pl lactoperoxidase (1 mg/ml), 2 pl NaiuI (-100 pCi/pl) and 2 pl of a 1: 10 000 dilution of a H202(comrnercially available as a 30% stock solution) in 0.05 M phosphate pH 7.4 buffer were added in the given order and the reaction was allowed to proceed for 10 minutes on ice. The reaction was stopped with the addition of 1 ml of PBS-Az and the free iodide was removed by centrifuging the solution through a 1 ml column of DOWEX 1 beads (preabsorbed with gelatin) until the percent of free iodide, as deteri-nined by trichloroacetic acid precipitation of the protein (see below) was Iess than 5% of the total (usually only one pass was required). The radiolabeled protein was used stored at 4OC and used within 24 hours. To determine the percent of free counts, 5 pl of the Clq was added to 50 pl of a 5% skim rnilk powder solution and precipitated with 500 p1 of 8% trichloroacetic acid (TCA). After one minute of thorough eng,the protein that had precipitated out of the solution was pelleted by a brief high speed centrifugation. The percent free iodide was calculated using the following fonnula: cpm supernatant % free 1251 = cpm supematant + cprn pellet x 100%

2.5.9 CI q binding radiuimmunoassay To evaluate the binding of Clq to the various IgMngG2b hybrids, we used a modification of a previously described assay that measured the binding of 1x1-radiolabeled Clq to IgM/IgGZb hybrids bound to 96-well microplates coated with the capture antigen TNP-labeled gelatin (Taylor et al., 1994). Briefly, TNP-coupled gelatin was used to coat the welIs of OptiPlateTM96-well polystyrene microplates (Packard Instrument Company). The wells were next blocked with 1% gelatin in PBS-Tween for two hours at 37'C and used immediately after six washes with PBS-Az. Zero to 200 ng of the purified monomers (100 ng for polymers) diluted to 100 pl in GVB was added to the wells and incubated at roorn temperature for 4 hours. ELISA quantitation of the antibody remaining in the supematant indicated that greater than 90% of the antibody added was bound to the wells. The plates were then washed six times with PBS-Az before the addition of 3.0 x 10-8 M 1251-labeled Clq in

SGVB-0.05% Tween (100 pl). After a 5 hour incubation at 37OC, the Liquid was aspkated and the wells were washed five times with SGVB to remove any unbound radioactivity. 25 1.11 of Microscint-OTM(Packard Instrument Company) was added to each well and the cpm values were measured by a Packard Top Countm microplate scintillation and luminescence counter (Packard Instrument Company). Al1 values were corrected by subtracting the background detexmined from wells incubated without antibody. For those irnrnunoglobulins found to bind significantly to C lq in this assay, a second C lq binding assay was performed using the same incubation conditions except that the amount of antibody was fixed at 100 ng for monomers, 50 ng for polymers, and the concentration of radiolabeled Clq was varied from zero to 3 x 10-8 M. In order to determine the actual amount of Clq bound flexible 96-weli polyvinyl chloride plates (Becton Dickinson and Company) were used for this assay and the welis were cut out, directly counted and compared with a radiolabeled Clq standard of spectrophotometrically deteeedarnount. The apparent Kd values, the total number of binding sites/well (n) and estimates of the erron were determined by non-linear regression analysis of the hyperbolic curve, described by the equation

molecules C 1q bound = nK1qkree Ki + IClql ~ree obtained from a plot of the amount of C lq bound against the concentration of radiolabeled

C 1 q added. Fitting of the data was done using the program Regression Version M 1.23 (Blackwell Scientific). CHAPTER 3

Effect of the P544G Mutation on Polymerization and Rationale for its Use in Studying the Activity of the IgM Monomer

The contents of Chapters 3 and 4 have been pubiished collectively as a single paper in the Journal of Irnmunology (F. H. Chen, S. K. Arya, A. Rinfret, D. E. Isenman, M. J. Shulman and R. H. Painter (1997). "Domain-switched mouse IgMlIgGZb hybnds indicate individual roles for Cp2, Cp3 and Cp4 domains in the regulation of the interaction of IgM with complement C 1 q" .159,3354-3364). For presentation purposes and also to facilitate inclusion of some additional data not shown in the publication, the material has been split into two chapters. 3.1 Introduction One way to snidy the Cl binding site of IgM is to systematically introduce mutations at candidate binding residues. The choice of individual residues cm be based on their predicted locations in the folded protein or on the nature of their side chains. A residue that is expected to be accessible to solvent is more Likely to be involved in a protein- protein interaction than a residue that is expected to be buried within a hydrophobie core.

Similarly, a residue with a charged side-chah is likely to be surface-exposed and has the potential to be involved in an ionic interaction. Using this line of reasoning, Arya et al. (1994) created a panel of single-site mutagenized IgM. When the mutants were examuied by SDS-PAGEgel electrophoresis, it was found that several of them were impaired in polymerization. Since polymeric and monomeric IgM have different valencies, they would be expected to differ in complement activation. This expectation has previously been investigated. Augener et ai. (197 1) demonstrated that naturally occumng IgM monomers (7s) were fifteen times less efficient than polymers at binding whole Cl. By studying monometic subunits obtained through the rnild reduction and alkylation of polymenc IgM, Swanson et al. (1988) also found that the monomeric species could bind detectably to Clq although the data indicate that this binding was poorer than the binding of the polymenc species to C 1q. With this to consider it became evident that in the panel of IgM mutants studied by

Arya et al. (1994), a distinction needed to be made between the two classes of mutants which were found to be defective in complement activation - those for which the substituted amino acid directIy affected the binding site and those for which the substituted amino acid indirectly affected the activity of the molecule by altering its state of polymerization. Mutants falling under the latter category which produced only monomers necessitated the establishment of a monomeric IgM 'positive control'. Since the monomenc species accounted for oniy 3% of the total output of wildtype IgM, it was not an easy matter to produce a polymer-free monomer preparation. By serendipity one mutant, designated IgM P544G. produced a fuUy active polymer but was so defective in poIymerization that most of the secreted imrnunoglobulin was in the monomeric fom. It is probable that the spatial arrangement of the mutated proline residue at position 544 within the Cp4 domain interferes with the polymerization event mediated by the cysteine 575 residue. By taking advantage of the properties of this mutant, Taylor et ai. (1994) were able to isolate a sufficient quantity of this 'monomeric IgM' to test it for complement binding activity. Not only was the monomer hemolyticaily inactive but, contrary to the earlier findings by Augener et al. (197 1) and Swanson et al. (1988), Taylor did not detect any binding between the monomeric IgM and C 1 or Clq under conditions that pemiitted dose-dependent binding of CIq to the IgG control. Before we could embark on a study to determine the cause of this phenomenon, we had to determine whether or not the observed effect of the P544G mutation was reproducible in an IgM with a higher affinity antigen-binding site. The reason for this was two-fold. Firstly, Swanson et al. (1988) had argued that their ability to observe Clq binding by monomeric IgM was likely due to the high affinity (Kd - 10-10 M-1) their immunoglobulin exhibited for the fluorescein hapten that was used. The second, but more important, reason for using a different IgM was that Taylor's expenments cornpared an Sp6-derived mouse IgM monomer with an HY L .2-derived mouse IgG2a. Although both were specific for the TNP hapten, they nevertheless had different variable regions. We therefore chose to use a mouse IgM molecule with a V-region for which a matching complement activating IgG subclass was available. Rinfret et al. (1990) had previously subcloned the heavy-chain constant region genes for mouse IgM and IgG2b into a vector containhg the MOPC 3 15-derived VH-region. When these vectors were transfected into the recipient MOPC 315.26 ce11 line, a heavy-chah loss variant of the original IgA- producing plasmacytoma MOPC 3 15, they produced immunoglobulins with the same variable regions. Furthermore the hapten binding affnity of this variable region pair is a hundred-fold greater than those present in the immunoglobulin used in Taylor's study. The staaing point for this work was therefore to determine the hemolytic and C1q binding activities of P544G IgM rnonomer and polymer that possessed the relatively high affinity MOPC 3 15 combining site for the hapten TNP. 3.2 Results 3.2.I Construction, expression and characterizution of the IgM P544G derivative The effect of the proline to glycine substitution on the polymerization of an IgM

with a MOPC 3 15 derived V-region was analyzed. Mutant immunoglobulin from clones secreting the most protein was immunoprecipitated with anti-y-chah antibody and

fomalized Staphylococccrs aureus ceIls and subjected to electrophoresis on a non-reducing SDS-PAGE gel. An example of the electrophoretic pattern is shown in the inset to figure 3.1. In direct contrast to the wildtype IgM which consists primarily of covalently-linked polymer (pentarner and some hexamer), most of the secreted IgM P544G immunoglobulin nins with a mobility equivalent to that of the wildtype IgM monomeric species. To determine whether any polymers are formed by non-covalent interactions, affinity-purified IgM PS44G was subjected to ultracentrifugation through a sucrose density gradient. The profile of the resulting gradient is shown in figure 3.1. Again, the species corresponding to the peak centered about fraction number 19, most Iikely rnonomers, predominate. Using the area under the monomer and polymer peaks. (the polymer peak is centered at fraction number 41), we estimated that the monorner accounts for at least 90% of the total immunoglobulin output. A visual inspection of the SDS-PAGE gel inset shows that the two results are consistent. This means that there are no substantial non-covalent interactions. Our results are also in agreement with the previously reported values of 5% polymer and 95% monomer for the Sp6 version of the mutant IgM (Arya et al., 1994; Taylor et al., 1994).

3.2.2 Activity of the IgM P544G polymer The sucrose density gradient fraction (#41) containhg the IgM P544G polymer was assayed for its abiiity to mediate lysis of TNP-coated sheep red blood ceIls by guinea pig complement. Figure 3.2A shows the results of this experiment. I 30 u P544G mutant 1

0 10 20 30 40 50 60 Fraction Number

Figure 3.1. Sucrose density gradient profile and a non-reducing SDS- PAGE gel of wiidtype IgM and P544G mutant preparations. Affinity-purified immunoglobuIin was separated on a 5.20% sucrose density gradient and fractions were collected and analyzed for immunoglobulin content by a reduced and aikylated ELISA as descnied in Materials and Methods. Monomers came out in the eariy fractions centered about fraction 19 whereas the polymer emerged at fraction 41. Metabolically labeled immunoglobulin that was immunoprecipitated fiom ce11 culture supernatant was nin on a 2.8% - 3.5% SDS polyacrylamide gel (inset) under non-reducing conditions. As the lanes are fiom separate gels, the respective positions of conaols, IgM polymer (P) and IgG2b monomer (M), are indicated for each gel. Figure 3.2. CML and Clq binding activities of the IgM P544G mutant polymer vs. the wildtype IgM polymer. A, We compared the ability of the polymer fraction of the IgM P544G mutant to initiate CML on TNP-haptenated sheep red blood celis with that of wildtype IgM as descnied in Materials and Methods.

B, Cornparison of the C 1q binding activity of IgM P544G with IgM wildtype polymers. A fixed concentration of radiolabeled C lq was incubated with increasing amounts of TNP- bound polymer under low ionic strength conditions and the amount of specifically bound C 1q was determined after several washes as descnied in Marerials aud Methods. Results are presented as the rnean of duplicate values. Arnount of lmmunoglobulin Added (ng)

i0 4'0 60 80 160 Amount of lmmunoglobulin Added (ng) The polymer hction of IgM P544G was also tested for its Clq binding ability at a lower ionic strength to enhance the binding. Results of the Clq binding assay are shown in figure 3.2B. Swanson et al. (1988) had shown that the activity of IgM polymers was highly dependent on the hapten density. This fact was manifest in a plot of Clq binding versus density of hapten where a sharp peak appeared. In the fitassay, the amount of immunoglobulin added was varied and the amount of Clq added was fixed. This way we could ensure that if a negative result was obtained. it was not because the immunoglobulin was displayed at a suboptimal density. We did not fmd any noticeable shqdependence of Clq binding activity on the hapten density as did the other group. Therefore in the second assay, the amount of Ig was fixed and the concentration of Clq was varied. It is obvious from both the CML assay and the Clq binding assay that the IgM P544G polymers are indistinguishable from the wild-type IgM polyrner control. When the data was andyzed by non-linear regression to a hyperbolic curve, shown in figure 3.3, according to the equation

molecules Clq bound = nlClqjfree Kd + [Clqlfree we found that at ionic strength (p) = 0.06, the apparent Kd for the interaction between the polymer and Clq is 1.3 t 0.3 x 109 M and the nurnber of Ciq binding sites per well in both cases was 3.7 + 0.3 x 10~~.Both values were identical to those obtained with the IgM wildtype polymer.

3.2.3 Annlysis of the activity of the IgM PS44G monomer Since the IgM P544G polymer was fuUy active in both Clq binding and CML activity, we were able to use the corresponding monomer as a mode1 for the wildtype IgM monomer. To determine whether or not this IgM monorner with a MOPC 3 15-derived V- region is able to initiate complement-mediated lysis, we performed a hemolytic assay using

TNP-coated SRBC. Figure 3.4A shows the plot of the % Lysis venus the arnount of IgM k -e IgM -t IgM P544G 1

Free Cl q x IO* M

Figure 3.3. Non-linear regression analysis of the Clq binding data for the wildtype IgM and P544G mutant polymers. The hyperbolic curves are derived from a non-linear regression analysis of the binding data as descnbed in Materials and Methods. Figure 3.4. CML and Clq binding activities of the IgM P544G monomer vs. wildtype IgG2b monomer. A, CML activity of IgM P544G mutant monomer. We tested the ability of the monomer fraction of IgM P544G to initiate CML under conditions where IgG2b is able to do so. For the monomer CML assay, SRBC highly modified with TNP-hapten were used as targets. This assay is described in more detail in M~terlalsand Methods.

B, Cornparison of the CIq binding activiv of IgM P544G with wildtype IgG2b. A fixed concentration of radiolabeled CI q was incubated with increasing amounts of TNP-bound monomer under Low ionic saength conditions and the amount of specifically bound C lq was detemiined after several washes. Details of this assay are given in Marerials and Methods. Results are presented as the mean of duplicate values. Amount of lmmunoglobulin Added (ng)

Amount of lmmunoglobulin Added (ng) added for the P544G monomer and for the mouse IgG2b with a matching V-region (positive control). According to this plot, the IgM P544G monomer has no detectable hemolytic ability although the positive control was able to mediate lysis in a dosedependent manner- To detennine if the defect in hemolysis was at the stage of Clq binding, we performed C lq binding assays on the sucrose density gradient-separated IgM P544G monomer fractions. We thought that if the monomenc IgM has a poor affinity for Clq, then a cluster of densely packed IgM rnonomers may be required to produce a measurabie signal. By perfodng the Clq binding assay over a range of immunoglobulin densities, between zero to 200 ng of IgM monomer per well, we were able to ensure that any negative results were caused by the inactivity of the monomer as opposed to being a consequence of suboptimal immunoglobulin densities. Figure 3.4B shows the resulis of this experiment. The positive IgG2b control was able to bind Clq when as little as 10 ng of the immunoglobulin were added to the welis. We found however that the IgM P544G monomer was not able to bind detectably to Clq even at the maximum density tested (200 ng immunoglobulin addedlwell). Based on the collective results, we concluded that the IgM P544G monomer was not hernolytically active and that this inactivity stems from a defect in its ability to bind C lq. 3.3 Discussion In these experirnents we sought to determine whether the observation regarding the P544G derived monomeric IgM and its lack of binding to Clq could be repeated with an IgM monomer with a higher hapten afinity. This was of some concem as Swanson et ai.

(1988) suggested that the affnity of the antigen binding site is relevant to activity of the molecule. The allosteric mode1 of Cl activation by polymeric IgM implied by their results is to some extent addressed in the present chapter. In the present work, the TNP-specific

IgM has an intrinsic Ka of 2.5 x 106 M-l (Johnston et ai., 1974). This value is approximately a hundred-fold higher than that for the Sp6-derived IgM used by Taylor et al. (Muraoka and Shulman, 1989). It must be noted, however, that the hapten affïnity of our mutants is SUlower than that of the anti-fluorescein monomers used by Swanson et al. Our anaiysis of the IgM P544G mutants confirrns the major points set out in the studies by Arya et al. (1994) and Taylor et al. (1994). Fint, the proline to glycine mutation at position 544 of the pchain interferes with the polymerization of IgM. Since proline residues are ofien found where the carbon backbone of the protein changes direction, it is possible that proline 544 is involved in an important bend within the dornain structure.

Even ihough glycine residues are also found in bends, they are more flexible than prolines.

A glycine at the same position may therefore not be able to sustain the correct local structure. Consequentiy, the folding of the heavy chain is altered near the C-terminus and the C575 residue may no longer be in the correct position for forming the intersubunit disulfide bonds. Altematively, new side chains may be exposed which interfere with the domain interactions between adjacent subunits. It is also possible that there is an aiteration in the ER/Golgi apparatus retention signal which Sitia et ai. (1990) propose to reside within the p-tail and which retains the irnmunoglobulin in the endoplasrnic reticulum/Golgi apparatus until polymerization is complete. The second conclusion is that the P544G point mutation does not impair the activity of the polymer species. Arya et al. (1994) reported that the P544G polymer had 100% of the activity of the wildtype. Our analysis shows that this is also true of the MOPC3 15 V- region IgM P544G. Over the range of IgM added as well as for the different concentrations of Clq used in the binding assays, the CML activity and Clq binding results of the P544G polymer were similar to that of IgM wildtype. This shows that the proline 544 residue is not directly involved in the Clq binding site of IgM. It aiso strengthens the assumption that the IgM P544G rnonomer behaves like the wildtype IgM rnonomer. Finally, we demonstrated that the IgM P544G monomer is unable to bind Clq and is thus unable to activate complement under conditions where IgG2b monomers are able to activate complement. As already indicated, Swanson et al. (1988) reported that IgM monomer subunits prepared by reduction and alkylation of IgM, which had a high afinity for the fluorescein hapten, were able to bind Clq. The specific activity of the subunit preparation was not reported. It is possible that IgM prepared from reduced and aikylated polymers may have a different intramolecular disulfide bonding pattern than IgM which has been synthesized and secreted as monomers. We expect that the heavy chains of the IgM P544G monomer are held together by a disulfide bond at C337 whereas the monomeric subunits produced by Swanson et al. could aiso be held together by disulfide bonds at positions 414 and 575 depending on the susceptibility of the various interheavy chain disulfide bonds to reduction. An additional point addressed by Taylor's study which is beyond the scope of this thesis deais with the possibility that the IgM monomer is not a faimrepresentation of the rnonomer subunit within polymenc IgM. It is well hown that regions of a protein that are normally sequestered as result of polymerization can become glycosylated when polymerization does not occur. There was therefore reason to believe that the monomeric IgM might be hyperglycosylated by cornparison with its counterpart within the IgM polymer. In agreement with this prediction it has been shown that the electrophoretic mobility of heavy chains from monomeric IgM is Iess than that of heavy chains from polymeric IgM (Davis et al., 1989a; Cals et al., 1996). To determine whether or not the inactivity of the monomeric IgM was due to abnormal glycosylation, Taylor et ai. (1994) tested both monorneric and polymenc forms of IgM produced in the presence of deoxymannojirimycin, an oligosaccharide-processing inhibitor, for C lq binding activity. They found that the polymer retained its full activity whereas the monomer remained inactive. Therefore, the inactivity of the monomer was not due to the presence of additional carbohydrate structures. CHAPTER 4

Studies on the IgM/IgGlb Hybrid Irnmunoglobulins 4.1 Introduction The work described in the previous chapter established that the IgM P544G monorner is unable to bind Clq when aggregated by a haptenated surface and by analogy suggests that the wildtype IgM monomer is also unable to bind Clq. This is a surprishg finding since monomenc IgM contains dl the elements of polymeric IgM except for the intersubunit disulfide bonds. However, it must be remembered that uncomplexed polymeric IgM binds ody weakly to Cl and to Clq as compared to its binding when complexed with antigen (Feinstein et al., 1983; Poon et al., 1985; Ziccardi, 1985). This suggests that there is a connection between the inactivity of the aggregated IgM monomer and the uncomplexed IgM polymer. Since the observed inactivity of the monomeric IgM is probably connected to the rnechanism by which uncomplexed IgM polymer is prevented from interacting fully with Clq. Our fmt approach was to study the Clq binding site of the monomeric subunit. The fmt possible explanation for the inactivity of monomenc IgM is that its Clq binding site, though comparable to the analogous site in IgG2b in its affinity for Clq, is rendered inaccessible by the bulk of its neighboring domains. In a complexed polymer, this inhibition would be relieved by the 'star to staple' conformational change. The second possibility is that the site, though accessible, is not inherently active and needs to be activated by a conformational change which can only be rnediated by antigen-binding coupled with polymerization. If only antigen binding was involved, then antigen-bound monomers should have activated complernent; similarly, if only polymerization were involved. free IgM pentamers should activate complement. The third possibility is a combination of the first and second scenarios - the site is neither inherently active nor exposed. An additional explanation is that the Clq binding site on each subunit is weak. Therefore only within the polymeric molecule when the sites are arranged so as to favour the binding of several globular heads of Clq at a time is the overali apparent affinity, enhanced by avidity, large enough to detect. This however, does not explain why the IgM polymer is not able to activate Cl in the absence of a hapten or antigen array.

To aid us in distinguishing among these possibilities, we took advantage of the fact that in IgG the Cy2 domain contains the Clq binding site (discussed in Chapter one) and that the isolated domain is as active as when it occurs within the context of the entire IgG molecule (Ellerson et al., 1972). This demonstrates that the C lq binding site in IgG is inherently active and that the IgG monomeric background within which it is found permits the expression of this binding site. Work reviewed in the introduction provided convincing evidence to suggest that the C lq binding site of IgM is located entirely within the Cp3 and Cp4 domains. Results from site-directed mutagenesis studies favour the CM location. By transplanting the putative Clq binding domain of IgM, Cp3, into the permissive IgG background, we can determine whether or not this domain has an inherently active and measurable C lq binding site. Conversely, by transplanting the active C@ domain of IgG into the monomenc IgM background and measuring its activity within this environment, we can gauge the effect of this environrnent on the CPdomain. Clackson and Winter (1989) were able to replace the Cy2 dornain of the inactive mouse IgGl subclass with the corresponding domain of the active IgG2b subclass and get complete lysis comparable to wildtype IgGZb. This suggests that the Cy2 domain cm be moved and still retain its inherent complement activity in a new immunoglobulin environment. FinaIly, by studying the activity of the polymeric species, we can also determine the role played by polyrnerization in the expression of the complement binding site. To this end, we created a panel of domain-switched mouse IgMIIgG2b hybrids with matched V-regions denved from the MOPC 315 plasmacytoma ce11 line. In both mouse IgG2b and mouse IgM genes, the hinge and constant domains are encoded by separate exons. We were therefore able to exchange domains by utilizing restriction sites in the flanking intronic sequences. The validity of domain shuffling between immunoglobulin classes has been fairly well-established. For example, Shopes et al. (1990) has used IgGlAgE hybnd constructs to identlfy the IgGl domains contributing to the interaction between IgGl and FcRI. The problem with this approach is that one cannot be certain that the domains assemble and are oriented in the hybrid molecule in the same way as within their native environments. Due to the high degree of sequence similarity between immunoglobulin domains, however, it is reasonable to assume that the domain structure remains, for the most part, intact. The hybrid mutants were expressed, purified and the monomeric species were separated from the polymeric species by ultracentrifugation through a sucrose density gradient. W5 then compared the monomers to the wildtype IgG2b positive control with respect to CML activity, using haptenated sheep red blood cells as targets and guinea pig serum as the source of cornplement, and Clq binding activity, using Clq purified from human serum. We also compared the polymers to the wildtype IgM polymer positive control with respect to CML and CLq binding activities. 4.2 ResuIts 4.2.1 Construction and expression of hybrid IgM/IgGZb immunoglobulins By exon shuffling, the hybrid genes of the corresponding panel of hybrid molecules were created and they are depicted schernatically in figure 4.1. The hybnd molecules are descnbed by four Greek symbols indicating the irnrnunoglobulin class origin of each constant domain where the following pairs are interchanged: Cyl and Cpl, gamma hinge and Cp2, Cy2 and Cp3, or Cy3 and Cp4. Using this designation, IgM is called pppp, IgG2b is cdied yrn,and a hybnd containing the following constant domain, Cl1- gamma hinge-Cy2-Cp4 would be cded pw. The constnicts were expressed by transforming the recipient cell line MOPC 3 15.26 with the appropriate hybrid heavy-chah encoding plasmids as in Chapter 3 wherein the resulting heavy chains combined with the endogenous lambda chah to produce a matched set of immunoglobulins with the sarne MOPC 3 15 TNP/DNP-specific binding site. The amount of irnrnunoglobulin secreted in most cases was greater than 1 pg/ml, however for ppyp and pyyp the yield was ten-fold less. The presence or absence of individual IgM constant domains in the various hybrids was confirmed by ELISA using rat antibodies specific for mouse Cp3 and Cy4 dornains. Also, the binding of S. aureus to the yyyy, kpyy and pyyy preparations confirmed the presence of complete Fcy in these mutants and the Iack of binding indicated its absence from all others. Imrnunoglobulins metabolicdly Iabeled with 3%-labeled L-methionine and L- cysteine were immunoprecipitated from the culture supematants of productive transfectants with a combination of rabbit anti-p-chah antibody and formalized Staphylococcus aureas cells and subjected to electrophoresis on a non-reducing SDS-PAGE gel. Representative electrophoretic patterns of the wildtype mouse IgM, IgGZb and the hybrid immunogfobulins are shown in figure 4.2. Our molecular weight controls are wildtype IgM, which migrates as a 950 kD pentarner, and wildtype IgGZb, which migrates as 150 kD monomer, and these are shown respectively in lanes 1 and 6 of the same figure. A faint Figure 4.1. Schematic diagram of the constant regions of wildtype IgG2b, wildtype IgM and their domain-switched hybrid constwcts. The y-domains are represented by open boxes and the p-domains are represented by shaded boxes. The short form notations of the constnicts are shown to the leR. P544G derivatives of wildtype IgM and hybrids ppw and pyyp were also employed in this study. Figure 4.2. Covalent species of IgG, IgM and the domain-switched hybrids. Non-reducing SDS-PAGEof metabolically labeled irnmunoglobulins shows the proportion of monomer to covalent polymer species in the various hybrid populations. The immunoglobulùis were metabolically labeled with 3%-~ethionineand 3%-~ysteine, immunoprecipitated and mn on a non-reducing 2.8%-3.5% acrylamide step gel reinforced with agarose as described in Materials and Methods. The position of the IgM polymer (P) and IgG2b monomer (M) controls are indicated beside the additional SDS-PAGE lanes (right). band that corresponds to the IgM rnonomer cm be seen in lane 1. The mobility of this band is expected to be less than that of the IgG2b monomer for two reasons. First the IgM has a 110 amino acid residue globular domain in place of the 22-residue gamma hinge of IgGZb. Secondly, the p-chah has more glycosylation sites than the y-chain (compare figures 1.4 and 1.7). Polymerization occurred to varying degrees in the different constructs (figure 4.2). Ail constructs lacking the Cp4 domain, e.g. the yypy, ppyy and ~yyychimeras, failed to form covalent polymers. These results are consistent with previous studies from Our laboratones which have shown that the Cp4 domain, which contains the y-tail, directs polymerization (Baker et al., 1986). The hybrid protein my migrates with the IgG2b control indicating that it is a monomer. The hybrid yypp migrates as several protein bands. The major component corresponds to a high molecular weight polymer. By comparison with the IgG2b wildtype. the small fast-migrating fraction of the

immunoglobulin appears to be H2L2 'monomers'. It is likely that the slow rnigrating species is a pentamer and that the species corresponding to the faint band that migrates slightly slower is the corresponding hexamer. The faster migration of these species by comparison with the corresponding IgM polyrners is likely due to the replacement of the large globular domain Cp2 with a comparatively smaller gamma hinge. Any additional mobility can be accounted for by the absence of the Cp2 glycosylation site. Hybrids ypycr and pyyp also migrate as mixtures of polymeric and monomeric species but additional bands corresponding to intermediate oligomeric species such as dimers, trimers and tetramers are also evident (figure 4.2). These two hybrïds were also secreted in low yieid. Together, these observations suggest that the Cp3 has a role both in polymerization and secretion of IgM. The high molecular weight polymers of ppyp correspond in mobility to the pentamers and hexamers of IgM whereas the high molecular weight polymers of pyyp migrate slightly faster. This can again be explained by the fact that the heavy chains of each H2L2 subunit of pyyj~are reduced in size by the substitution of a large Cpî domain with the srnaller gamma hinge. The chimeric proteins ppyy and pyyy migrate as H2L2 monomeric species. The slower migration of the ppv band by cornparison with the igG2b wildtype standard band

indicated by the arrow to the right of the panel cmbe accounted for by the replacement of the gamma huige by the comparatively larger Cp2. The other p-domain, Cp 1, is probably equal in size to the Cyl it replaces but it has a glycosylation site at position 17 1 which may explain the slight decrease in mobility. On reducing SDS-PAGE (data not shown) several heavy chah species were evident in ppyp, pyyp and also in the P544G derivatives of these molecules indicating that there may have been differences in glycosylation of the rnodified heavy chains. In all cases a single light chain band was observed on the reducing gels. For the hybrid immunoglobulins that were secreted as both monomers and polymers, the monomeric species had to be separated from the polymeric species. This was achieved by subjecting the affinity-purified immunoglobulin to size fractionation through a sucrose density gradient. Since the polymer has more hapten binding sites than does the monomer, the polymer would be expected to have a higher apparent affinity for a hapten array than would the monomer. To rernove the signal bias of the ELISA for polymers that would result from this difference in apparent affhities, small aliquots of the various fractions were subjected to reduction and alkylation before being quantified by ELISA. The gradient profiles for the wildtype and hybrid antibodies determined in this way are depicted in figures 4.3 to 4.4. The peak fractions containing the monomerk species, centered at about fraction number 19, and the high rnolecular weight polymers (centered at about fraction number 40) were saved for further analysis. The isolation of monomeric species of the ppy.~and pyyp hybrids proved difficult.

Fint, clones producing these hybrids were poor producers secreting no more than 200 ng of the protein per ml of culture supernatant. The problem of low yield was compounded by the fact that only a fraction of the secreted imrnunoglobulin was monomeric. The second problem was that in preliminary assays, it was found that the polymenc species were Figure 4.3. Sucrose density gradient pronie for IgG2b and Cy2-containing hybrids. Afinity purified immunoglobuiins were subjected to size fractionation by ultracentrifugation through a 5-20% sucrose density gradient. The monomeric species emerged at about fraction #19 whereas the polymeric species ernerged at about fraction #40. Fraction Number Polymer

0 10 20 30 40 50 60

Fraction Number

Figure 4.4. Sucrose density gradient profile for IgM and Cp3-COntaining hybrids. Affinity purified imrnunoglobulins were subjected to size fiactionation by ultracentrifugation through a 5-20% sucrose density gradient. The monomeric species emerged at about fiaction #19 whereas the polymeric species emerged at about fkaction #4O. highiy active in the CML assay. To assay the activity of the monomers, we had to be certain that the monomer preparations were free hmpolymenc contaminants. To reduce the proportion of the polymenc form, we decided to take advantage of the features of the P544G mutation described in Chapter 3. Since the proline to glycine mutation was able to increase the proportion of monomer to polymer in wildtype IgM, we surmised that this mutation would have the same effect when introduced into the ppyp and pyp hybrids. Figures 4.5 and 4.6 show the sucrose density gradient profies of these two hybrids and their corresponding P544G derivatives. The insets in these figures show that the electrophoretic patterns of the metabolicaiiy labeled and immunoprecipitated hybrids before and after the introduction of the P544G mutation are consistent with the sucrose density gradient results. In both cases, the proportion of monomer to polymer was higher in the P544G variant. Analysis of the monomers for these two hybnds was therefore performed with the P544G variant. The polymer forms were, however, isolated from the non-rnutated hybrid.

4.2.2 The Cp domain is not active in a monorneric IgM background Since the C$ domain is active in isolation, transplanting it into the IgM background should yield information as to whether or not this background has an inhibitory effect on the Clq binding site resident in Cy2. The complement binding properties of the Cy2- containhg hybrid monomen were assessed using both the CML assay and the more direct solid phase Clq binding assay. In the CML assay, we measured the ability of the immunoglobulin to initiate hemolysis of the target cells, an event that entails bath binding and activation of CI. In the C Iq binding assay, we measured the abili~of the monomer to bind to Clq under low ionic strength conditions. Figure 4.7 shows that when the C'y2 domain was placed in an IgM background, as in the mutant ppw, the resulting monomer was both unable to bind Clq or initiate CML by cornparison with wild-type IgG2b. This 1 PPW -P544G mutant

Fraction Number

Figure 4.5. Sucrose density gradient profile and a non-reducing SDS- PAGE gel of the ppyp hybrid and its P544G derivative. Affinity-purified immunoglobulin was separated on a 520% sucrose density gradient and fractions were collected and analyzed for immunoglobulin content by a reduced and alkylated ELISA as described in Matenals and Methods. Monomers came out in the early fractions centered about fraction 18 whereas the polymer emerged at fraction 40. Metabolically labeled immunoglobulin that was immunoprecipitated fkom ce11 culture supernatant was run on a 2.8% - 3.5% SDS polyacrylarnide gel (inset) under non-reducing conditions. As the lanes are fiom separate gels, the respective positions of controls, IgM poiymer (P) and IgG2b monomer 0,are indicated for each gel. The vertical scales for the hybrid (left) and its corresponding P544G denvative (nght) are different. PW -P544G mutant

Figure 4.6. Sucrose density gradient profile and a non-reducing SDS- PAGE gel of the pyyp hybrid and its P544G derivative. Affinity-purified immunoglobulin was separated on a 520% sucrose density gradient and fractions were collected and analyzed for immunoglobulin content by a reduced and alkylated ELISA as descnbed in Materials and Methods. Monomers came out in the early Fractions centered about fraction 16 whereas the polymer emerged at nïrction 40. Metabolically labeled immunoglobulin that was immunoprecipitated fkom cell culture supernatant was run on a 2.8% - 3.5% SDS polyacryla.de gel (inset) under non-reducing conditions. As the lanes are fiom separate gels, the respective positions of controls, IgM polymer (P) and IgG2b monomer @Q, are indicated for each gel. The vertical scales for the hybnd (left) and its corresponding P544G derivative (right) are different. Figure 4.7. CML and Clq binding activities of the Cy2-containing hybrid monomers vs. wildtype IgGZb monomers. A, CML activity of Cy2-con taining hybrid monomers. We tested the abilities of the monomer fractions of Cy2-containing hybrids to initiate CML under conditions that permit lysis by wildtype IgG2b (*P544G derivative). SRBC that were highly modified with TNP-hapten were used as CML targets as described in Materials and Methods.

B, Cornparison of the Clq binding activities of the CV-containing hybrid monomers with wildtype IgG2b. A fixed concentration of radiolabeled Clq was incubated with increasing arnounts of TNP-bound monomer under low ionic strength conditions and the amount of specificaily bound Clq was determined after severai washes. Details of this assay are given in Matenuls and Methods. Results are presented as the mean of duplicate values. Amount of lmmunoglobulin Added (ng)

Amount of lrnmunoglobulin Added (ng) dernonsirates that the IgM background interferes with the expression of the Clq binding site of Cy2. To determine which of the Cp domains were responsible for this defect, we began by creating and testing for activity of the pyyy monomer, in which the Cyl domain of wildtype IgG2b was replaced by Cy 1. Because the monomer from this mutant was at least as active as wild-type IgGZb in both CML and Clq binding (figure 4.7), we concluded that the observed loss in Cy2 activity of the previous mutant is not due to the presence of the CpL domain. Next, we created mutants of py-yp in which either the Cp2 or Cp4 was replaced with their IgG2b counterparts, the gamma hinge and C'y3 respectively, and tested the monomer fractions for their abilities to bind C lq and to initiate hemolysis. The results of this expenment are also shown in figure 4.7. When Cp2 was replaced by the gamma hinge, as in p'yyp, the monomer still exhibited far less activity than pmboth in Clq binding and in CML. This suggested that the Cp4 done can inhibit the activity of the neighbouring Cy2 domain. Accordingly, when Cp4 was replaced with Cy3, the Cy2 of the resulting mutant rnonomer ppyy regained CIq binding activity equivalent to that in wildtype IgG2b though it was not klly active in the CM.assay. This demonstrates that Cp2 does not interfere with the ability of the adjacent Cy2 domain to bind C lq under the conditions of the Clq binding assay. The bais for the discrepancy between the CML and Clq binding data for ppyy monomer will be discussed in a subsequent section. We conclude that Cp4 cm inhibit an adjacent complement binding site and suggest that this same domain impairs the complement binding activity of monomeric IgM. In contrast, the inhibitory influence of the adjacent Cpî domain is decidedly less than that of Cy4. The trends seen in the Clq binding studies where the arnount of Clq was held constant (figure 4.7B) and the amount of monomeric immunoglobulin per well was varied were also rnirrored in a similar assay in which the antigen bound monomer was held constant and Clq was varied (see figure 4.8). From the data obtained in this latter assay, the apparent dissociation constants and the number of Clq binding sites per well could be II "*" denotes P544G mutants

1 2 3 [Free Cl q] x 10-8

Figure 4.8. Non-linear regression analysis of the binding data for pyyy, ppyy, pyyp, ppyp and IgGZb monomers. The hyperbolic curves are derived hma non-linear regression analysis of the binding data as described in Muterials and Methodr. calculated (summarized in table 4.1). The ppv and pyyy hybrids displayed similar apparent Kd values but the enhanced Clq binding seen in figures 4.7 and 4.8 for the pm hybrid relative to IgG2b and the p~~yyhybnd reflects an approximately two-fold increase in the number of Clq binding sites per pg bound.

4.2.3 The Cp3 domuin, alone or together with Cp4, is not active in a monomeric IgG2b background Because the monomeric IgG2b molecule allows the expression of the native complement binding site located within its C'y2 domain it should also permit the expression of a complement binding site transplanted from IgM if the binding site in the p-chain is sirnilarly organized. Since the literature indicates that the Cp3 and possibly the Cp4 domains are involved, we created mutants in which Cp3 alone or Cp3 together with Cp4 were substituted for Cy2 and CP-Cp, respectively. Figure 4.9 shows that the monomeric species from the resulting two mutants, yyjq and ypp, were both inactive by cornparison with the positive control, IgGZb, in terms of CML and Clq binding. This demonstrates that neither Cp3 nor Cp3-Cp4 are fully able to engage Clq and suggests that although the residues involved in complement binding may still be contained entirely within these two domains, other domains and additional factors such as polymerization and antigen binding are necessary for the full expression of the site. Altematively, since we have shown that Cp4 is capable of inhibiting the expression of an adjacent active C'y" domain, it is also possible that in the case of ypp Cp4 behaves sirnilarly toward such a site in Cp3.

4.2.4 CML activity of the polymer fractions To study the effect of polymerization on the expression of the Clq binding site in IgM we examined the hemolytic activities of the polymer fractions of these mutants. If the site is contained within Cm-Cp4 and polyrnerization produces the packing density needed to enable polyvalent binding of Cl by a group of weak sites, we would expect the polymer Monorner Polvmer

Hybrid Kd # sites Kd # sites

(X 109 M) (X 10-10) (X 109 M) (X 10-10)

Table 4.1. Summary of apparent Kd values and number of Clq binding sites per well. The values were deterrnined by non-linear analysis of the Clq binding curve for monomer and polymer as descnbed in Materials and Methods. The binding curves are shown in Figures 4.8 and 4.12. A dash appears for molecules that were not tested because they bound too poorly to Clq. The monomer fractions of the lpyp and pyyp hybnd immunoglobulins were isolated from their P544G derivatives. The errors indicated represent the confidence limits of the non-linear Ieast squares fitting routine. They do not represent means and standard deviations of multiple independent experiments. Figure 4.9. CML and Clq binding activities of the Cp3-containing hybrid monomers vs. wildtype Ig62b monomers. A, CML activity of Cp3-containing hybrid monomers. We measured the abilities of monomer fractions of Cp3-containing hybnds to initiate CML under conditions that permit lysis by wildtype IgG2b.

B, Cornparison of the Clq binding activities of CF~containing hybrid monomers with wildtype IgG2b. A fixed concentration of radiolabeled C lq was incubated with increasing amounts of TNP-bound monomer under iow ionic strength conditions and the amount of specifically bound Clq was determined after several washes. Details of this assay are given in Materials and Methods. Results are presented as the mean of duplicate values. Amount of lrnmunoglobulin Added (ng)

Amount of lmmunoglobulin Added (ng) form of the mutant yypp to be as active as the wild-type IgM. The CML and C 1q binding

assays shown in figure 4.10 demonstrate that this is not the case. We therefore conclude that the inactivity of the wppolymer is not simply due to the lack of a polyvalent myof Cp.3 and Cp4 within a single molecule and propose that either or both of the Cp2 and Cpl

domains may be necessary for the fidl expression of the site. When we compared the IgM and pp.~polymers with respect to their affiinities for

C lq (table 4. l), we found bat they both had Kd values of 1.1 x 10-9 M. This result Uidicates that the functional affinity of the C lq binding sites in the IgM polymzr is the sarne as the functional of the C lq binding sites in the C.(2 domains of the ppyp polymer. It then follows that the demonstrated Iack of C Lq binding by rnonomeric IgM is not simply due to a problem of avidity. A fact that suggests that the polymers behave differently in both Clq binding and C 1 activation is that the results of the CML assays are not reflected in the results of the Clq binding assays. Although the pyyp polymer was as much as five times more active than the IgM polymer in the CML assays (figure 4.1 lA), the apparent affinity of the pyyp polymer for Clq was lower than that of the IgM polymer. Indeed analysis of the Clq binding data shown in figure 4.12 and summarized in table 4.1 shows that the apparent Kd of the pwpolymer (2.1 x M) was higher than the apparent &j of the wildtype IgM polymer ( 1. l x M). Similarly the the-foid higher CM'activity of the ppyp polymer over the IgM polymer (figure 4. LIA) was not reflected in the results of the Clq binding assays in which pp.~and IgM were shown to be equally active (figure 4.1 1B and table 4.1). Interestingly, the total number of binding sites for Clq provided by pyyp is more than twice the number found for IgM or pp-yp (figure 4.12 and table 4.1) and may account for much of the enhanced CML activity observed for the p3cyc~polymer. Figure 4.10. CML and Clq binding activities of the Cp3-containing hybrid polymer yypp vs. that of wildtype IgM polymers. A, CML activity of a Cp3 - containing hybrid polymer, yypp. We tested the ability of the polymer fraction of the Cp3- containing polymer to initiate CML under conditions that permit lysis by wildtype IgM. SREK that were moderately modified with TNP-hapten were used as CML targets as described in Materials md Methods.

B, Cornparison of the C lq binding activity of polymeric yypp with that of wildtype IgM. A fixed concentration of radiolabeled C lq was incubated with increasing amounts of TNP- bound monomer under low ionic strength conditions and the amount of specifically bound C lq was detennined after several washes. Details of this assay are given in Materials and Methods. Results are presented as the mean of duplicate values. Amount of lmmunoglobulin Added (ng)

Amount of lmmunoglobulin Added (ng) Figure 4.11. CML and Clq binding activities of the CG-containing hybrid polymers, ppyp and pyyp, vs. that of wildtype IgM polymers. A, CML activity of Cy2-containing hybrid polymen ppyp and p.^. We tested the ability of the polymer fractions of Cfl-containing polymen to initiate CML under condition that permit lysis by wildtype IgM.

B, Cornparisons of the Clq binding activities of the ppw and pmpolymer with that of wildtype IgM. A fixed concentration of radiolabeled C lq was incubated with increasing amounts of TNP-bound monomer under low ionic strength conditions and the amount of specifically bound Clq was determined after several washes. Details of this assay are given in Materials and Methods. ResuIts are presented as the mean of duplicate values. O 2 4 6 8 10 Amount of lmmunoglobulin Added (ng)

- IgM - PPW

O 20 40 60 80 100 Amount of 1rnmunoglobuIin Added (ng) [Free Cl q] x 10-8

Figure 4.12. Non-linear regression analysis of the binding data for pyyl, ppw, yypp and IgM polymers. The hyperbolic curves are derived from a non-linear regression anaiysis of the binding data as described in Materials and Methodr. 4.3 Discussion Fundamentai to the interpretation of the observations reported in this chapter is an understanding of the mechanical nature of the activation of the Cl complex which is

thought to occur as a result of a distortional signal created when two or more globular heads of its C lq subunit are engaged by irnmunoglobulin (Circolo et al., 1985; Hanson et al., 1985; Schumaker et al., 1986; Arlaud et al., 1987; Hoekzema et al., 1988) In the case of IgG, two molecules brought together by interactions with polyvalent antigen or other physical or chernical means can bring about this event (Hyslop et al., 1970; Metzger, 1974; Metzger, 1978; Wright et al., 1980; Smith et al., 1995). Since IgM is a pentameric

assembly of IgG-like subunits, a single molecule has the potential to activate Cl and it has

been shown that, when suitably bound to antigen, a single molecule of IgM can initiate the classical complement cascade (Borsos and Rapp, 1965a). In the absence of antigen IgM is inactive. The sites that can react with Clq are only revealed as a consequence of antigen binding (Borsos and Rapp, 1965a; Borsos and Rapp, 1965b; Ishizaka et al., 19681, presumably when it undergoes the 'star to staple' conformational change visualized in electron rnicrographs by Feinstein (Feinstein and Munn, 1969; Feinstein et al., 1983). To simpliQ analysis of the Clq binding site in polymeric IgM, Taylor et al. (1994) decided to analyze the expression of this site in monomenc IgM. The study showed, rather

surprisingly, that there was no detectable interaction between Clq and hapten-arrayed IgM monomers. In Chapter 3, we have confirmed this observation using the MOPC 315- derived monomeric IgM. Therefore, to investigate the roles played by the various IgM constant domains in preventing the expression of the Clq binding activity in monomenc IgM, we studied the interaction between Clq and a variety of mouse IgMAgG2b domain- switched hybrids. The results of Our studies allowed us to determine whether the Clq binding site is inactive in the IgM monomer either because it is concealed in the IgM context, or because it is not fully formed, or because both these conditions prevail. Since the Cy2 domain is as active in isolation as it is in the context of the IgG molecule (Yasmeen et ai., 19761, the ppp hybrid was constmcted to see whether the IgM background permitted expression of the inminsic activity of the Cy2 domain. Conversely, we used the permissive nature of the IgG background to examine whether there is an intruisicaliy active Clq binding site in the Cp3 domain, as in wy, or whether adjacent domains are also required. Since no X-ray crystallographic data are currently available for the hybrids, we are limited to assuming that the transplanted domains are fonned and onented properly in their new environments thereby maintainhg their int~sicactivities. These assumptions are supported by the following observations - hybrids containing the Cp3 or Cp4 domains were recognized by anti-Cp.3 or anti-Cp4 antibodies respectively (data not shown), hybrids bearing a complete Fcy bound to S. aureus, and dl hybrids containing the Cp4 domain were able to polymerïze - which indicated that no gross conformational abnormalities had occurred. It might also be argued that the Clq binding site in monomer IgM is expressed but is not detected merely because its afinity is much lower than that in the IgG2b control. Since, however, the binding site of polymeric IgM has the same apparent Kd for Clq as does CyZ in the same polymenc IgM background (ypyp), about 1 x 10-9 M in both cases

(table 4. l), this suggests that, leaving conformational and steric arguments aside, monomer IgM should have an affinity for Clq similar to that of IgG2b.

4.3.1 Evidence that the Clq binding site in p-chain is concealed When Cy2 was placed in an IgM background (ppyp), the monomeric species (obtained as the P544G mutant) was without Clq binding activity, suggesting that the neighboring domains prevent the expression of its binding site. The lack of Clq binding by the pyp monorner (also obtained as the P544G mutant) shows that the presence of Cp4 adjacent to Cy2 is suficient to prevent binding of Clq. The absence of binding is aiso reflected in the lack of hemolytic activity of this monomer. These resdts demonstrate that Cp.4 cm hinder access to the Clq binding site in the Cy2 domain. The presence of Cp2 on the other hand appears to have no effect on the Clq binding activity of Cy2 in the ppyy monomer showing that Cy2, unlike Cp4, does not inhibit expression of the Clq binding site in Cy2. The ppv monomer is not as hemolyticaily active as wild-type IgG2b even though the apparent Kd values and the number of Clq binding sites are equal. One explanation is that the presence of the Cp1 and Cp2 domains may influence activation of whole Cl, as opposed to binding Clq, or it may influence events further down the cascade. An example of the latter has been found in human IgGI which was reported to be poorer at binding CIq than IgG3 but more efficient at CML due to some advantage at the stage of C4 activation (Bindon et al., 1988). Another explanation may be that although ppyy monomers bind simiiar numbers of Clq molecules, fewer clusters are available to bind multivalently to a single Clq molecule because rotation about their Cp2 domains is restricted. When both

Cp2 and Cp4 were replaced by their y-counterparts as in p"~yy,the monomer obtained displayed a Clq binding constant sirnilar to that of wildtype IgG2b but, surprisingly, had a higher number of binding sites. In this case it is possible that the C-terminal end of the Cp1 domain enhances the flexibility of the hinge and increases the nurnber of C lq binding pairs in pyyy.

4.3.2 Evidence that the Clq binding site is not formed in Cp3

When Cp3 was placed in the IgG background, the monomer was inactive throughout the forty-fold concentration range over which IgG2b showed dose-dependent binding to Clq. It would seem, therefore, that the Cp3 domain does not contain al1 the elements needed for a functional Clq binding site. It is further evident that the additional structures needed to complete the site are not dl available in the Cp4 domain because the monomer fraction of the constnict yypp was also without hernolytic activity under the same conditions. Altematively, the Cp4 domain in yypy might be blocking access to important residues in the Cp3. a hypothesis that is consistent with our findings regarding the inhibitory effects of the Cp4 domain in the Cy2-containing monorners. Thus the results of assays performed on the Cp3-containing monomers, myand yypp show that the C lq binding site putatively within this domain is not functional. This was confïrined with the CML assay. We conclude, therefore, that the Clq binding site on the Cp3 domain is neither formed nor exposed within the IgM monomer. It seems that nature has designed a double lock on the Clq binding site in IgM both by preventing access to the site and aiso by requiring additional structures and/or antigen-induced conformational changes to be present for its activity.

4.3.3 Properties of the hybrid polyrners The polymer form of yypp is inactive in a hemolytic assay (CML) even though it exhibits significant activity in the C 1q assay. The explanation may lie in the difference in the ionic strength used in the two assays. Low ionic strength is used in the Clq binding assay to enhance binding but in the CML assay, physiological ionic strength rnust be used and this may unduly weaken the binding by this hybrid wp. We have previously observed such a difference between Clq binding and CML activity in a study of an IgM mutant, P436S, where the difference was shown to be a result of a steeper dependency on ionic strength of the affinity of the mutant for Clq relative to wildtype (Wright et ai., 1988). It may well be the case that a similar situation prevails for the ypp polyrner in the present study. The diminished CML activity found with yypp suggests that even the combined Cp3 and Cp4 domains do not completely fumish the hl1 Clq binding site though the observed Clq binding indicates its partial completion. These data are also consistent with a mode1 in which a CIq binding site, hlly contained within Cp3, is oniy formed and exposed when the IgM polymer undergoes the antigen-induced 'star to staple' conformationd change observed by Feinstein and Munn (1969). In this model, the inability of the flexible y-hinge to support this conformational change could explain why the

Clq site in this mutant is not Myactive. A similar explanation was offered by Poon et al. (1995) to account for the weak hemolytic activity which they observed for the polymer yypp derived from human IgM and mouse IgG2b'. Altematively, the inability of the wy hybrid to initiate CML might be due to its strength of binding for C lq at physiological ionic strength king below the threshold required for C 1 activation. The CLq binding and CML results for the Cy2-containing polymers, ppyp and pyyp, show that the inhibitory effects of the Cp.4 domain can be partiaily or completely removed by polymerization and antigen binding. Both of these polymers are capable of binding Clq and activating the complement cascade at feast as well as wildtype IgM, aithough as monomers, ppyp and py~~both bind poorly to C lq. Interestingly, Poon et al. (1995) found that yp.~polymers are able to bind strongly to Cl in the absence of hapten, indicating that polyrnerization aione may 5e sufficient to remove most of the inhibitory effects of Cp4. Since wildtype IgM does not bind Clq under these conditions, this supports the view that the binding site of IgM is not inherently active but requires antigen binding to become active.

4.3.4 Possible reasons for the enhanced Clq binding and CML activity of the pmi polymer Analysis of the binding data for the pyyp polymers (figure 4.12) reveals that this molecule has a similar Kd for C lq as does ppyp and wildtype IgM but that it displays more than twice the number of Clq binding sites (table 4.1). A possible explanation for this phenomenon is that because of the flexibility of the hinge, the pyyp polymer does not assume the staple conformation when bound to antigen but remains roughly in the star conformation, oriented perpendicular to the antigenic surface, exposing both faces of iü Fc &SC. Since IgM apparently has a Clq binding site on each face of the Fc disc (Poon et al., 1985) molecules bound in this manner would be expected to bind twice as many molecules

* After a preliminary report of this work had been published in abstract fom, Poon et al. (1995) reported on the properties of IgG and IgM hybrid constnicts which were deriveci from mouse y2b and human p sequences. The properties of the hybrids that resembled ours yielded similar expenrnental results. of Clq as those bound in the staple conformation that can display only one face. Altematively, the flexibility of the hinge may permit more active Cl binding conformarionai possibilities. In IgM, rotational freedom about the Cp2 is believed to be restricted. This has been demonstrated by electron microscopy and synchrotron X-ray scattering and molecuiar graphics modeling (Feinstein and Munn, 1969; Perkins et al., 1991). Perkins reports that only a mode1 of IgM in which the plane of the Fab arms is pardel to the plane of the Fc disc is consistent with the X-ray data. Furthemore, in IgM pairs of adjacent Fab arms are uniformiy spaced due to restraints placed by the rigid Cp2 domain. If this equal spacing is maintained upon antigen-binding, a distance of 9 nm would separate C lq binding sites on adjacent subunits and 15 nm (for IgM pentarner with J chain) would separate sites which lie on the diagond of the pentagon projected by the IgM polymer, i.e. those sites on ncn-adjacent subunits (Perkins et ai., 1991). Since the reported distance between adjacent Clq globular heads is 14 nm (Perkins, 1989), it is most probable that only Clq sites on non-adjacent subunits could be engaged at the sarne time (Painter, 1993). If so, the IgM molecule would only be able to bind one molecule of Clq and the same should also be true of the ppyp polymer. A flexible hinge in place of the rigid Cp2 domain, as in pyyp, would, however, permit both side to side displacement of the Fab arms as well as rotation about the axis defined by the hinge. In this case, assuming that the Cp3 and Cp4 domains are not too bulky, this added flexibility would allow the polymer to assume the conformation depicted in figure 4.13, providing two pairs of CI activating binding sites as opposed to the one provided by pentameric IgM and pp*. In support of this argument it has been reported that hexamenc IgM activates complement six to twenty times better than pentameric IgM (Davis et al., 1988; Randall et al., 1990). Presumably, the enhanced activity occurs because a hexagonal Fcp6 disc can bind simultaneously to two Clq molecules using the binding sites on non-adjacent, non- opposing subunits whereas an Fcp5 disc cmooly bind one Clq (see figure 4.13). Figure 4.13 Schematic diagram of Clq subunits binding to the Fcs disc of wildtype IgM or ppyp pentamer, pyyp pentamer and IgM hexamer. The individual Fc portions of the Fc5 disc are shown as solid black lines whereas the Clq subunits are represented by shaded circles and sticks. a, Clq binding to IgM or ppyp pentamer. b, C lq binding to pyyp pentamer. c, Clq binding to IgM hexamer. In both b and c, two C lq subunits are bound by a single immunoglobulin. 4.3.5 Conclusion

The studies described in this thesis iead US to conclude that there is more thm one cause for the failure of IgM monomer to bind Clq. We canot exclude the possibility that the domains flanking the Cp3 domain contribute residues to the binding site, but if the binding site is contained solely within Cp3, it is neither intrinsicaiiy active nor exposed in the monomeric subunit. Polyrnerization can partially undo the masking effect that the Cp4 domain has on an adjacent Clq binding domain. Furthemore, the Cp.2 domain appears to have a role in completing this unmasking. This is consistent with the view that the Cp2 domain aids in the formation of the Clq binding site in IgM polymers by transmitting a distortional signal from the Fab to the Fcp upon antigen binding. CHAPTER 5

General Discussion and Future Perspectives 5.1 General Discussion A model for the interaction between IgM and Clq must be consistent with the existing data. A single antigen-bound IgM has been demonstrated to bind CL (Borsos and Rapp, 1965a) yet unbound IgM does not interact appreciably with Cl (Poon et al., 1991). Feinstein and Munn (1969) have hirthermore demonstrated that antigen-bound and free IgM assume different conformations (see figure 1.14). Together these observations suggest that the 'star' to 'staple' conformational change which IgM undergoes upon binding to antigen is closely linked with the expression of the C lq binding site. Studies conducted by several labs including our own have supported the view that the Clq binding site of IgM is located within the Fcp portion of the molecule. The Fcp5 fragment has been shown to contain most if not al1 of the CI binding activity of whole IgM (Piaut and

Tomasi, 1970; Bubb and Conradie, 1976). A study involving the use of site-directed mutagenesis hirther indicates that severai residues in the Cp3 dornain participate in the Clq binding site (Arya et al., 1994). This is consistent with an earlier study which showed that a Q3-specific antibody inhibits the hemolytic activity of IgM (Leptin et ai., 1983). The Cp4 domain may also be involved since Cl4 domains generated by proteolytic digestion of

IgM were found to fix Cl (Bubb and Conradie, 1976; Hurst et ai., 1976; Bubb and Conradie, 1977). One model for the interaction between IgM and C 1 q that is consistent with the EM, C 1 binding, fragment and site-directed mutagenesis studies is the distortional model (Meizger, 1974). According to this modei, antigen binding causes the IgM pentarner to becorne distorted and to expose or create a Clq binding site. In uncomplexed IgM, the Fab annç are hiily extended and in the same plane as the central Fcp disc. If the binding site is Iocated within the Fcp domains of the polymer it can be blocked by the neighbouring domain(s). When the IgM pentamer binds to antigen and undergoes the 'star' to 'staple' conformational change, the Fab arms are bent away from the plane of the Fcy disc thereby exposing potential sites on the Fc portion of the molecule. Additionaily, the gross conformational change involved in the transformation could create a binding site by rearranging othenvise incorrectly oriented functional groups. In Chapter 3 we showed that a derivative of mouse IgM which has a Pro to Gly substitution at position 544 (IgM P544G) is secreted predorninantly as a monomeric irnmunoglobulin. This result is consistent with the earlier finding of Taylor et al. (1994). Furthemore since the polymenc form of this mutant is as active as the wildtype polymer in

both Clq binding and CML activities, it is reasonable to assume that the activity of the monomeric form of this mutant reflects that of the wildtype monomer. As in the study by Taylor et al., monomeric IgM P544G did not bind Clq or activate complement even though monomeric 1gG did. The result suggests that the wildtype IgM monorner is inactive. We can explain this result using the distortional model. A single IgM monomeric subunit cannot undergo the gross conformational change required to express a distortional binding site because the activation energy for this conformational change would be too great. If the subunit is part of a polymer, the average energy required to distort each subunit could be reduced through the coordinated and concerted engagement of multiple Fab arms by antigen. Another explmation for why monomeric IgM does not bind Clq may be that its

inherent affiinity for C lq is too low. To effectively bind and activate C 1, multiple copies of the binding site must be arrayed in sufficient number and at the proper density and spacing, as in polymeric IgM, to be functional. If the spatial requirements are very strict then the

frequency at which an activating array is attained by the random distribution of monomers upon a haptenated surface may be too low. To understand why the IgM monomer is inactive, we created a set of mouse IgM/IgG2b hybrid molecules based upon the assumption that Cp3 and perhaps Cp4 are the Clq binding domains of IgM. The results of the CML and Clq binding assays for each hybrid immunoglobulin are summarized in table 5.1. Both monomers and polymen were studied where applicable. Monomer Polymer IgMngGb Hybrid CML ~lq CML clq activity binding activity binding

Table 5.1 Summary of CML and Clq binding assay results for mouse wildtype IgM, wildtype IgG2b and their hybrids. Represented are the assays shown in figures 3.2, 3.4, 4.7, 4.9, 4.10 and 4.1 1. For a cornparison of Kd values see table 4.1. 'NIA' denotes species that do not occur. The inabiiity of the Cp3 domain to function in the context of the permissive monomeric IgG2b environment (wy) suggests that the binding site in Cp3 is not inherently active. If the site was merely blocked by its fl anking domains, Cp.2 and Cp4, in the IgM monomer, then it should be active in the IgG2b background. We were not able to measure the hemolytic and Clq binding activities of a Cp3 domain in a polyrnenc IgG2b environment because the myhybrid did not secrete polymers. We were however able to measure the hemolytic and Clq binding activities of the yypp polymer. This polymer was not active in the CML assays and was not as active as wildtype IgM in the Clq binding assay. Assuming that the binding site is contained entirely within the Fcp portion of IgM and that the correct orientation and spacing of the resident Clq binding sites were maintained in this hybrid polymer, then the Clq binding site must not be fully fonned or exposed. Since the yypp polymer has a flexible hinge between its Fab and Fc regions, one can argue that this molecule would not be able to sustain the strain required to bend the CjG domain and cause the distortional change necessary to reveal or shape the existing site. Our results therefore suggest that the site within the Cp3 domain of monomeric IgM is not hnctional because it is not inherently active. In this way IgM differs from IgG because the Clq binding site of IgG is an intrinsic property of an isolated domain. To determine what effect the flanking Cp2 and Cp4 domains have on an inherently functiond domain, we created a hybrid in which the Cy2 domain was substituted for the Cp3 domain of IgM. The hybrid monomer with this substitution (ppyp) was inactive in both CML and C lq binding assays. This suggests that the IgM environment can exert an inhibitory effect on the expression of an inherently active binding site. By systernatically replacing the Cp2, Cp4 and both domains of this hybrid rnonomer with their corresponding IgG2b counterparts, we were able to detemine which of the two domains is directly involved. The results of this sîudy show that only Cp4 must be replaced to restore full Clq binding activity to the Cy2-containing monomer. To restore full CML activity however, both Cp2 and Cp4 must be replaced. This suggests that the Cp4 domain conceds the Clq binding site of C@ in the ppp monomer and that the Cp2 domain interferes with the CML activity of this monomer at a later stage of the complement pathway. Similar des may be played by these domains in the IgM monomer. Since the ppyp polymer is at least as active as the wildtype IgM polymer in both CML and Clq binding assays, polymerization and antigen binding must reiieve the inhibitory effect of the Cp2 and Cp4 domains. Poon et ai. (1995) found that the pp'yp polymer was able to bind

Clq in the absence of antigen. In contrast, wildtype IgM exhibited virtuaily no Clq binding activity under the same conditions. This suggests that the inhibitory effect of the Cp2 and Cp4 domains on an intrinsically active Cl binding domain are at least partially if not completely relieved by polymerization alone. 5.2 Future Perspectives The work presented in this thesis raises many questions. For example, is the ypp polymer inactive because it lacks the Cp2 domain? If so, then what is the role played by this domain? Does the ngid Cp2 domain simply transmit the distortional signai from the Fab regions of the IgM molecule to the distaliy located Fcp domains or does it contribute residues to the Clq binding site? By creating and testing a 'ppp polymer for C 1q binding activity, one can deterrnine whether or not the Cp2 domain is necessary for the expression of the Clq binding site in IgM. If the Cp2 domain of this mutant is replaced with the similarly rigid CE^ domain from IgE and the resulting ~pphybrid is tested for Clq binding activity, one can determine whether or not amino acid residues in the Cp.2 domain participate directly in the binding of Clq to IgM. If the ypp polymer is active, one can conclude that the Cp2 domain does not participate directly in this interaction but oniy mediates the conformational change necessary to form the site in a neighbounng domain. On the other hand if the y~pppolymer is not active, this would suggest that part of the C lq binding site is located in the Cp2 domain. By substituting the CE^ domain for the Cp2 domain, we can ensure that the interheavy chain disulfide bonds equivalent to the p-chah 337 disulfide bond is maintained (the Ce2 domain of IgE is analogous to the Cp2 domain of IgM). If the results suggest that part of the binding site is found in the Cp2 domain one cm attempt to identiQ the critical residues by site-directed mutagenesis of residues in this domain. To determine whether the lack of CML activity of the yypp polymer is really due to the lack of an appropriate conformational change it would be useful to obtain EM images of this molecule to see if it can undergo a 'star to staple' conformation in the presence of an antigen. The same would be tme of the ppyp and pyyp polymers. Polymers containing a hinge would be expected to have a different conformation than those containing a rigid Cp.2 domain. Does the 'star' to 'staple' conformational change alter the accessibility of the residues identified by site-directed mutagenesis as Clq binding site residues? To see if the accessibility of the aspartic acid residues in the putative binding site (Arya et ai., 1994) is affected by conformational changes, one can compare the extent to which these residues can be chemically modified in the two conformations of IgM. Proteolytic fragments of the modifed IgM could be isolated and the frequency at which the Cp3-containing fragment is modified could be measured. Other studies could focus on the IgM binding site within Clq. Several arginine residues in the A-, B- and C-chahs are proposed to interact with IgG (Marques et al., 1993). The interaction between C lq and both IgM and IgG could be studied using site- directed mutagenesis of these and other surrounding residues. Furthemore one could obtain convincing evidence that the aspartic acid residues in Cp3 interact direcdy with the arginine residues of Clq if the aspartic acid residues of Cp3 were mutated to arginine residues and the arginine residues of Clq were mutated to aspartic acid residues to give complementary change-in-specificity mutations.

A less risky venture would be to use the panel of hybnd IgGAgM mutants to study the structural requirements for other biological relevant interactions. Both polymeric IgM and IgA are known to associate with secretory component, for exarnple. By studying the interaction between the polymeric hybrids and secretory component, we couid determine which domains were necessary for this interaction. Once the location of the binding site is narrowed to a region of IgM, site-directed mutagenesis could be used to identify critical residues. Our lab is also presently attempting to determine the domain requirements for the association between IgM and the Fcp receptor. An important part of interpreting site-directed mutagenesis data is knowing where the important residues are located within the three-dimensional structure of the molecule being studied. The lack of a crystal structure for IgM is one of the major hurdles to studying IgM functions. Presently Our understanding of the IgM structure is limited to models based on the known structure for IgG. X-ray crystallography studies should therefore be a top pnority in the study of IgM. These proposed studies are important to understanding the nature of IgM and its interactions with various effector functions. An understanding of the mechanisms by which these immunoregulatory proteins interact would provide researchers with a potent tool for devising strategies to combat illnesses due to the malfunction of these important immunologicd systems. REFERENCES

Alcolea, J. M., Anton, L. C., Marques, G., Sanchez-Corral, P. and Vivanco, F. (1986) The interaction of 1-aniline-8-naphthalene sulphonate with human C 1q. Molec. Immunol. 23, 39-44,

Amzel, L. M., Poljak, R. J., Saul, F., Varga, J. M. and Richards, F. F. ( 1974) The three dimensional structure of a combining region-ligand complex of irnmunoglobulin NEW at 3.5-A resohtion. Proc. Natl. Acad. Sci. USA 71, 1427-1430.

Anderson, D. R., Atkinsoo, P. H. and Grimes, W. J. (1985) Major carbohydrate structures at five glycosylation sites on murine IgM determined by high resolution 'H- NMR spectroscopy. Arch. Biochem. Biophys. 243, 605-618.

Anderson, D. R. and Grimes, W. R. (1982) Heterogeneity of asparagine-linked oligosaccharides of five glycosylation sites on immunoglobu1in M heavy chain from minerai oil plasmacytoma 1O4E. J. Biol. Chem. 257, 14858-14864.

Appella, E., Weber, 1. T. and Blasi, F. (1988) Structure and function of epidermal growth factor-like regions in proteins. FEBS Letrers 23 1, 1-4.

Arlaud, G. J., Chesne, S., Villiers, C. L. and Colomb, M. G. (1980a) A study on the structure and interactions of the Cl sub-components and C~Sin the fluid phase. Biochim. Biophys. Acta 616, 105- 1 15.

Arlaud, G. J., Colomb, M. G. and Gagnon, J. (1987) A functional mode1 of the human C 1 complex. Immunol. Today 8, 106- 1 11.

Arlaud, G. J. and Gagnon, J. (1983) Complete amui0 acid sequence of the catalytic chah of human complement subcomponent . Biochemis~22, 1758- 1764.

Arlaud, G. J. and Gagnon, J. (1985) Identification of the peptide bond cleaved during activation of human Clr. FEBS Letters 180, 234-238. ArIaud, G. J., Meyer, C. M. and Colomb, M. G. (1976) Use of an IgG fragment prepared with particulate plasmin to study the Cl binding and activation. FEBS Leîters 66, 132- 136,

Arlaud, G. J., Thielens, N. M. and IIly, C. (1993) Assembly of the Cl complex. Behring Inst. Mit?. 93, 189-195.

Arlaud, G. J., Villiers, C. L., Chesne, S. and Colomb, M. G. (1980b) Purified proenzyme Clr - Some characteristics of its activation and subsequent proteolytic cleavage. Biochim. Biophys. Acta 616, 116- 129.

Arlaud, G. J., Gagnon, I. and Porter, R. R. (1982) The catalytic chah of human complement subcomponent C lr. Purification and N-terminal amino acid sequences of the major cyanogen bromide-cleavage fragments. Biochem. J.201,49-59.

Arya, S. K., Chen, F., Spycher, S., Isenman, D. E., Shulman, M. J. and Painter, R. H. (1994) Mapping of amino acid residues in the Cp3 domain of mouse IgM important in macrornolecular assernbly and complement-dependent cytolysis. J. Irnrnunol. 152, 1206- 12.

Augener, W., Grey, H. M., Cooper, N. R. and Müller-Eberhard, H. J. (1971) The reaction of monomeric and aggregated immunoglobuIins with C 1. Imrnunochem. 8, 10 11 - 1020.

Baker, M. D., Wu, G. E., Toonem, W. M., Murialdo, H., Davis, A. and Shulman, M. J. (1986) A region of the imrnunoglobulin p- heavy chah necessary for forming pentameric IgM. J. Immunol. 137, 1724-1728.

Barlow, P. N., Baron, M., Norman, D. G., Day, A. J., Willis, A. C., Sim, R. B. and Campbell, I. D. (1991) Secondary structure of a complement control protein module by two-dimensional 1H-NMR. Biochemistry 30, 997- 1004.

Bartholomew, R. M. and Esser, A. F. (1977) The fxst complement component: Evidence for an equilibrium between C 1s free in serum and C 1s bound in the Cl complex. J. Immunol. 119, 19 16-1922. Baumann, M. A. and Anderson, B. E. (1990) An immune complex selective afFinity matrix utilking a synthetic peptide. hi. Chem. 265, 18414-18422.

Bede, D. and Feinstein, A. (1969) Studies on the reduction of a human 19s irnmunoglobului M. Biochem. J. 112, 187- 194.

Beale, D. and Fazakerley, J. K. (198 1) A cornparison of the actions of trypsin and pepsin on porcine imrnunoglobulin M and their effects on biological activity . Biochim. Biophys. Acta 670, 230-235.

Beale, D. and Feinstein, A. (1976) Structure and function of the constant regions of irnrnunoglobulins. Quart. Rev. Biophys. 9, 18 1-2 15.

Bianchino, A. C., Poon, P. H. and Schumaker, V. N. (1988) A mechanism for the spontaneous activation of the frst component of complement, Cl, and its regulation by Cl- Inhibitor- J. Immunol. 141, 3930-3936.

Bindon, C. L, Haie, G., Bruggemann, M. and Waldmann, H. (1988) Human monoclonal IgG isotypes differ in complement activating hinction at the level of C4 as well as Clq. J. Exp. Med. 168, 127-142.

Bindon, C. L, Hale, G. and Waldmann, H. (1990) Complement activation by immunoglobulin does not depend solely on C 1 q binding. Eur. J. Immunol. 20, 277-28 1.

Bork, P. (1 99 1) Complement components C lr/C ls, bone morphogenic protein 1 and Xenopus laevis developmentally regulated protein WS.2 share cornmon repeats. FEBS Let fers 282, 9- 12.

Borsos, T. (1989) Immune complex mediated activation of the classical complement pathway. Behring Inst. Mift. 84, 93- 10 1.

Borsos, T. Chapuis, R. M. and Langone, J. J. (1981) Distinction between fixation of C1 and the activation of cornplement by naturd IgM anti-hapten antibody: Effect of cell surface hapten density . Molec. Immunol. 18, 863-868. Borsos, T. and Circolo, A. (1983) Binding and activation of Cl by ce11 bound IgG: Activation depends on ce11 surface hapten density. Molec. Immunol. 20,433-438.

Borsos, T., Colten, H. R., Spalter, J. S., Rogentine, N. and Rapp, H. J. (1968) The C'la fixation and transfer test: examples of its applicability to the detection and enurneration of antigens and antibodies at cell surfaces. J. Irnmunol. 101,392-398.

Borsos, T. and Rapp, H. 1. (1965a) Hemolysin titration based on fixation of the activated fint component of complement: Evidence that one molecule of hemolysin suffices to sensitize an erythrocyte. J. Imrnunol. 95,559-566.

Borsos, T. and Rapp, H. 1. (1965b) Complement fixation on ce11 surfaces by 19s and 7s antibody. Science 150, 505-506.

Brandtzaeg, P. (1985) Role of J chah and secretory component in receptor-mediated glandular and hepatic transport of irnmunoglobulins in man. Scand. J. Irnmzmol. 22, 11 1- 146.

Brass, A., Kadlar, K. E-,Thomas, J. T., Grant, M. E. and Boot-Handford, R. P. (1992) The fibrillar collagens, coilagen WII, collagen X and the C lq complement proteins share a sirnilar domain in their C-terminal non-collagenous regions. FEBS Letiers 303, 126- 128.

Brekke, O. H., Bremnes, B., Sandin, R., Aase, A., Michaelsen, T. E. and Sandlie, 1. (1993a) Human IgG3 cm adopt the disulfide bond pattern characteristic for IgGl without resembling it in cornplement mediated cell lysis. Molec. Immunol. 30, 1419- 1425.

Brekke, O. H., Michaelsen, T. E., Sandin, R. and Sandlie, 1. (1993b) Activation of complement by an IgG molecule without a genetic hinge. Nature 363.628-630 (Published erratum appears in Nature 383, 103).

Brewer, J. W., Randall. T. D., Parkhouse, R. M. and Corley, R. B. (1994) IgM hexamers? Immunol. Today 15, 165- 168.

Brown, J. C. and KoshIand, M. E. (1975) Activation of antibody Fc function by antigen- induced conformational changes. Pruc. Nat. Acad. Sci. USA 72,s1 1 1-5 115. Brown, J. C. and Koshiand, M. E. (1977) Evidence for a long-range conformational change induced by antigen binding to IgM antibody. Proc. NatL Acad. Sei. USA 74, 5682-5686.

Bubb, M. 0. and Conradie, J. D. (1976) The importance of quaternary structure in the expression of the C 1-binding site of IgM. Immunology 31,893-903.

Bubb, M. 0. and Conradie, J. D. (1977) The Cp3-domain of IgM: Isolation and identification of the intact fragment. Biochem. Biophys. Cornm. 77,613-620.

Burton, D. R. (1985) Immunoglobulin G: Functional sites. Molec. Imrntcnol. 22, 161- 206.

Burton, D. R., Boyd, J.. Brampton, A. D., Easterbrook-Smith, S. B., Emanuel, E. J., Novotny, J., Rademacher, T. W., van Schravendijk, M. R., Sternberg, M. J. E. and Dwek, R. A. (1980) The Clq receptor site on immunoglobulin G. Nature 288,338-344.

Calcott, M. A. and Müller-Eberhard, H. J. (1972) Clq protein of human complement. Biochem. 11, 3443-3450.

Cals, LM.M., Guenzi, S., Carelli, S., Simmen, T., Sparvoli, A. and Sitia, R. (1996) IgM polymerization inhibits the Golgi-mediated processing of the mu-chah carboxy-terminai glycans. Molec. Immunol. 33, 15-24.

Cam, G. M., Zaritsky, A. and Koshland, M. E. (1982) Primary structure of the imrnunoglobulin J chain from the mouse. Proc. Natl. Acad. Sci. USA 79, 6656-6660.

Capra, J. D. and Kehoe, J. M. (1975) Hypervariable regions, idiotypy, and the antibody- combining site. Adv. Immunol. 20, 1-40.

Carter, P. E., Dunbar, B. and Fothergill, J. E. (1983) The serine proteinase chain of human complement component C~S: Cyanogen bromide cleavage and N-terminai sequences of the fragments. Biochem. J. 215, 565-57 1. Cathou, R. E. and Domngton, K. J. (1974) Stniciure and function of immunoglobulins. from Biological Macrornolecules - Subunits in Biological Systems. Timasheff, S. N. and Fasman, G. D. Dekker: New York.

Cattaneo, A. and Neuberger, M. S. (1987) Polymeric immunoglobulin M is secreted by transfectants of non-lyrnphoid cells in the absence of immunoglobulin J chain. EMBO J. 6, 2753-2758.

Cazenave, J. P., Assimeh, S. N., fainter, R. H., Packham, M. A. and Mustard, J. F. (1976) Clq inhibition of the interaction of coliagen with human platelets. J. Immunol. 116, 162-163.

Chan, J. C. (1996) Purification and characterization of recombinant human basic proline- nch protein precursor. M.Sc. Thesis, Department of Biochemistry, University of Toronto. P27.

Chen, F. H., Arya, S. K., Rinfret, A., Isenman, D. E., Shulman, M. J. and Painter, R. H. (1997) Domain-switched mouse IgMIIgG2b hybrids indicate individual roles for Cp2, Cp3 and Cp4 domains in the regulation of the interaction of IgM with cornplement C lq. Immunol. 159, 3354-3363.

Circolo, A., Battista, P. and Borsos, T. (1985) Efficiency of activation of complement by antihapten antibodies at the red ce11 surface: Effect of patchy vs random distribution of hapten. Mol. Irnmunol. 22, 207-2 14.

Clackson, T. and Winter, G. (1989) 'Sticky feet' -directed mutagenesis and its application to swapping antibody domains. Nucleic Acids Research 17, 10 163-10 170.

Colomb, M. and Porter, R. R. (1975) Characterization of a plasmin-digest fragment of rabbit irnmunoglobulin gamma that binds antigen and complement. Biochem J. 145, 177- 183.

Colten, H. R., Borsos, T. and Rapp, H. J. (1969) Titration of the first component of complement on a molecular basis: suitability of IgM and unsuitability of IgG hemolysins as sensitizer. Irnrnunochern. 6, 46 1-467. Cooper, N. R. (1985) The classical complement pathway: activation and regdation of the fitcomplement component Adv. Immunol. 37, 151-2 16.

Cunnif, R. V. and Stollar, B. D. (1968) Properties of 19s antibodies in complement fixation. 1.Temperature dependence and role of antigen structure. J. Immunol. 100,7- 14.

Davis, A. C., Collins, C., Yoshimura, M. L, D'Agostaro, G. and Shulman, M. J. (1989a) Mutations of the mouse mu H chain which prevent polymer assembly. J. Immunol. 143, 1352- 1357.

Davis, A. C., Collins, C. and Shulman, M. 1. (1989b) Differential glycosylation of polymeïic and monorneric IgM. Molec. Immunol. 26, 147- 152.

Davis, A. C., Roux, K. H., Pursey, J. and Shulman, M. 1. (1989~)Intermolecular disulfide bonding in IgM: effects of replacing cysteine residues in the mu heavy chain. EMBO J. 8, 25 19-2526.

Davis, A. C., Roux, K. H. and Shulman, M. 1. (1988) On the structure of polymenc IgM. Errr. J. Immunol. 18, 1001-1008.

Davis, A. C. and Shulman, M. J. (1989d) IgM-Molecular requirements for its assembly and function. Immunology Today 1 18- 123.

Davis, A. E. d. (1988) Cl inhibitor and hereditary angioneurotic edema. Ann. Rev. Immunol. 6, 595-628.

Deisenhofer, J. (1981) Crystallographic refinement and atomic models of a human Fc fragment and its complex with fragment B of Protein A from Staphylococcus aureus at 2.9 - and 2.8-A resolution. Biochernistry 20, 2361-2370.

Deisenhofer, J., Colman, P. M., Epp, 0. and Huber, R. (1976) Crystallographic structural studies of a human Fc fragment. II. A complete mode1 based on a Fourier map at 3.5 A resolution. Huppe-Seylers Zeitschrifrfür Physiologische C'hernie 357, 1421 - 1434. Dierich, M. P., Schulz, T. F., Eigentler, A., Huemer, H. and Schwable, W. (1988) Structural and functional relationships among receptors and regulators of the complement system. Molec. Immunol. 25, 1043- 1051.

Dodds, A. W., Sim, R. B., Porter, R. R. and Kerr, M. A. (1978) Activation of the first component of human complement (C 1) by antibody-antigen aggregates. Biochem. J. 175, 383-390.

Duncan, A. R. and Winter, G. (1988) The binding site for C lq on IgG. Nature 332,738- 740.

Dwek, R. A., Sutton, B. J., Perkins, S. J. and Rademacher, T. W. (1984) Structure- function relationships in immunoglobulins from Molecular Variants of Protein- Biosynthesis and Clinical Relevance Ed. Campbell, P. S. and Phelps, C. Biochem. Soc. Symp. 49, 123-136.

Edelman, G. M. and Gall, W. E. (1969) The Antibody Problem. Ann. Rev. Biochem. 38, 4 15-466.

Ellerson, J. R., Yasmeen, D., Painter, R. H. and Dorrington, K. (1972) A fragment corresponding to the CH^ region of immunoglobulin G (IgG) with complement fixing activity. FEBS Letters 24, 3 18-323.

Ely, K. R., Wood, M. K., Rajan, S. S., Hodsdon, J. M., Abola, E. E., Deutsch, H. F. and Edmundson, A. B. (1985) Unexpected similarities in the crystal structures of the Mcg light-chah dimer and its hybnd with the Weir protein. Molec. ImrnunoL 22, 93-100.

Emanuel, E. J., Brampton, A. D., Gagnon, J. and Dwek, R. A. (1982) Chernical verification of the Clq receptor site on IgG. FEBS Letters 137,298-302.

Endo, S. and Arata, Y. (1985) Proton nuclear magnetic resonance study of human immunoglobulin G1 and its proteolytic fragments: structure of the hinge region and effects of a hinge-region deletion. Biochemistry 24, 156 1 - 1568.

Fazel, S., Wiersma, E. I. and Shulman, M. J. (1997) Interplay of J chah and disulfide bonding in assembly of polymenc IgM. International Imrnwiol. 9, 1 149- 1 158. Feinstein, A. and Munn, E. A. (1969) Conformation of the fiee and antigen-bound IgM antibody molecules. Nature 224, 1307- 1309.

Feinstein, A., Munn, E. A. and Richardson, N. E. (1971) The three-dimensional conformation of yM and yA globulin molecules. Annals New York Academy of Sciences 190, 104-121.

Feinstein, A. and Richardson, N. E. (1981) Tertiary structure of the constant regions of immunoglobulins in relation to their function. Monogr. Allergy 17,2847.

Feinstein, A., Richardson, N. E., Gorick, B. D. and Hughes-Jones, N. C. (1983) Immunoglobulin M conformational change is a signal for complement activation. Protein Conformation as an ïmmunological Signal. Plenum Press: New York. P47-58.

Folkerd, E. J., Gardner, B. and Hughes-Jones, N. C. (1980) The relationship between the binding ability and the rate of activation of the complement component C 1. Immunology 41, 179-185.

Fust, G., Medgyesi, G. A., Bazin, H. and Gergely, J. (1980) Differences in the ability of rat IgG subclasses to consume cornplement in homologous and heterologous serum. Immunology Let fers 1, 249-253.

Fust, G., Medgyesi, G. A. and Franek, F. (1977) Complement consumption by immune complexes containing various pig anti-DNP antibodies and DNP-serurn alburnin. immunochem. 14, 259-262.

GA, P., Cseh, S., Schumaker, V. N. and Zavodszky, P. (1994) The structure and function of the first component of complement: genetic engineering approach. Acta Microbiologica et Immunologica Hungarica 4l,361-380.

Ga, P., Sarvari, M., Szilagyi, K., Zavodszky, P. and Schumaker, V. N. (1989) Expression hemolyticaliy active human complement component Clr proenzyme in insect cells using a baculovirus vector. Complement Inflamrn. 6,433-441. Gettins, P., Boyd, I., Glaudemans, C. P., Potter, M. and Dwek, R. A. (1981) Interaction of saccharide haptens with myelorna proteins. A 270-MHz proton nuclear magnetic resonance study. Biochemistry 20,7463-7469.

Goto, Y. and Hamaguchi, K. (1979) The role of the intrachain disulfide bond in the conformation and stability of the constant fragment of the immunoglobulin light chain. J. Biochern. 86, 1433- 1441.

Greenwood, J., Clark, M. and Waldmann, H. (1993) Structural motifs involved in human IgG antibody effector functions. Eur. J. IrnmunoL 23, 1098-1 104.

Haines, A. L. and Lepow. 1. H. (1964) Studies on human C'l-esterase. 1. Purification and enzymatic properties. J. Immunol. 92,456-467.

Haines, A. L. and Lepow, 1. H. (1964) Studies on human C'l-esterase. II. Function of purified Cl-esterase in the human complement system. J. Imrnunol. 92,465-478.

Hanson, D. C., Siegel, R. C. and Schumaker, V. N. (1985) Segmenta1 flexibility of the Clq subcomponent of human complement and its possible role in the immune response. J. Biol. Cham. 260, 3576-3583.

Harris, L. J., Larson, S. S., Hasel, K. W., Day, J., Greenwood, A. and McPherson, A. (1992) The three-dimensional structure of an intact monoclonal antibody for canine lymphoma. Nature 360, 369-372.

Heinz, H.-P., Burger, R., Golan, M. and Loos, M. (1984) Activation of the first component of complement, Cl, by a monoclonal antibody recognizing the C Chain of Clq. J. Imrnunol. 132, 804-808.

Heusser, C., Boesman, M., Nordin, J. H. and Isliker, H. (1973) Effect of chemical and enzymatic radioiodination on in vitro human C lq activities. J. Immunol. 110,820-828.

Hill, R. L., Delaney, R., Lebovitz, H. E. and Fellows, R. E. J. (1966) Studies on the amino acid sequence of heavy chahs from rabbit immunoglobulin G. Proc. Royal Soc. . London - Series B 166, 159-175. Hoekzema, R., Martens, M., Brouwer, M. C. and Hack, C. E. (1988) The distortive mechanism for the activation of complernent component C1 supported by studies with a monoclonal antibody against the "anns" of Clq. Molec. Immunol. 25,485-494.

Holowka, D. A. and Cathou, R. E. (1976a) Conformation of immunoglobulin M. 2. Nanosecond fluorescence depolarization anaiysis of segmental flexibility in anti-epsilon-1- dimethylamino-5-naphthalenesulfonyl-L-lysineanti-immunoglobulin from horse. pig, and shark. Biochernistry 15, 3379-3390.

Holowka, D. A. and Cathou, R. E. (1976b) Conformation of immunoglobulin M. 1. Characterization of anti-epsilon- 1-dimethylarnino-5-naphthalenesulfonyl-L-lysine immunoglobulin M antibodies from horse, pig, and shark. Biochemisby 15, 3373-79.

Huber, R., Deisenhofer, J., Colman, P. M. and Matsushima, M. (1976) Crystdlographic structure studies of an IgG molecule and an Fc fragment. Nature 264,415-420.

Hughes, G. J., Frutiger, S., Paquet, N. and Jaton, J. C. (1990) The amino acid sequence of rabbit J chah in secretory immunoglobulin A. Biochem. J. 271, 641-647.

Hughes-Jones, N. C. (1977) Functional affinity constants of the reaction between i251 labelled Clq and Clq binders and their use in the measurement of plasma Clq concentrations. Immunology 32, 19 1- 198.

Hughes-Jones, N. C. and Gardner, B. (1978) The reaction between the complement subcomponent C lq, IgG complexes and polyionic molecules. Immunology 34,459-463.

Hughes-Jones, N. C. and Gardner, B. (1979) Reaction between the isolated globular subunits of the complement component Clq and IgG complexes. Molec. Immunol. 16, 697-70 1.

Hurst, M. M., Voianakis, J. E., Stroud, R. M. and Bennett, J. C. (1976) A comparative analysis of the Cl-binding ability of fragments derived from complement-fixing and noncomplement-fing IgM proteins. J. Clin. Invest. 58, 16-2 1. Hyslop, N. E. J., Dourmashkin, R. R., Green, N. M. and Porter, R. R. (1970) The fixation of complement and the activated first component (Cl) of complement by complexes formed between antibody and divalent hapten. J. of Ep. Med. 131,783-802.

Illy, C., Thielens, N. M. and Arlaud, G. J. (1993) Chexnical characterizaton and location of ionic interactions involved in the assembly of the Cl complex of human complement J. Protein Chem. 12,771-780.

Irimajiri, S., Franklin, E. C. and Frangione, B. (1968) Some chernical and biological properties of the Fc' fragment of human 6 globulins. Immunochem. 5,383-397.

Isenman, D. E., Dorrington, K. J. and Painter, R. H. (1975) The structure and function of immunoglubulin domains II. The importance of interchain disulfide bonds and the possible role of molecular flexibility in the interaction between immunoglobulin G and complement. J. Immunol. 114, 1726- 1729.

Ishizaka, K., Ishizaka, T. and Sugahara, T. (1962) Biologicd activity of soluble antigen- antibody complexes: VIL Role of an antibody fragment in the induction of biological activities. J. Imrnirnol. 88, 690-70 1.

Ishizaka, T., Ishizaka, K., Salmon, S. and Fudenberg, H. (1967) Biologic activites of aggregated y-Globulin: VIII. Aggregated immunoglobulins of different classes. J. Immunol. 99, 82-9 1.

Ishizaka, T., Tada, T. and Ishizaka, K. (1968) Fixation of C' and C'la by rabbit gamma- G- and gamma-M-antibodies with particulate and soluble antigens. J. Imrnunol. 100, 1145-53.

Jeske, D. J. and Capra, J. D. (1984) Immunoglobulins: Structure and Function from Fundamental Imrnunology Ed. Paul, W.E. Raven Press: New York. 13 1- 165.

Jiang, H., Robey F. A., and Gewurz, H. (1992) Localization of sites through which C- reactive protein binds and activates complement to residues 14-26 and 76-92 of the human Clq A chah J. Exp. Med. 175, 1373-1379. Johnson, B. I. and Thames, K. E. (1976) Investigations of the Complement-fixing sites of immunoglobulins. J. Immwzol. 117, 149 1- 1494.

Johnston, M. F. M., Barisas, B. G. and Sturtevant, J. M. (1974) Thermodynamics of hapten binding to MOPC 3 15 and MOPC 460 mouse myeloma proteins. Biochemistry 13, 390-396. -

Journet, A. and Tosi, M. (1986) Cloning and sequencing of full-length cDNA encoding the precursor of human complement component C lr. Biochem. J. 240,783-787.

Karush, F., Chua, M. M. and Rodwell, J. D. (1979) Interaction of a bivalent ligand with IgM anti-lactose antibody. Biochemists, 18,2226-2232.

Kehoe, J. M., Bourgois, A., Capra, J. D. and Fougereau, M. (1974) Amino acid sequence of a murine immunoglobulin fragment that possesses complement fixing activity. Biochemistry 13, 2499-2504.

Kehoe, I. M. and Fougereau, M. (1969) Immunoglobulin peptide with cornplement fixing activity. Nature 224, 12 12- 12 13.

Kehry, M. R., Fuhrman, I. S., W., S. J., Rogers, J., Sibley, C. H. and Hood, L. E. (1982) Complete arnino acid sequence of a rnouse mu chain: homology among heavy chah constant region domains. Biochemistry 21,54 15-5424.

Kilchherr, E., Schumaker, V. N., Bianchino, A. C. and Curtiss, L. K. (1987) Kinetics of Cl activation by a monoclonal anti-Clq antibody and its fragments. J. Imrnunol. 138, 849-855.

Kilchherr, E., Schumaker, V. N., Phillips, M. L. and Curtiss, L. K. (1986) Activation of the first component of human complement, C 1, by monoclonal antibodies directed against different domains of subcomponent Clq. J. Immunol. 137,255-262.

Kirschfink, M., Blase, L., Engelmann, S. and Schwartz-Albiez, R. (1997) Secreted chondroitin sulfate proteoglycan of human B ceii lines binds to the complement protein Clq and inhibits cornplex formation of C 1. J. Imrnunol. 158, 1324- 133 1. Knobel, H. R., Heusser, C., Rodrkk, M. L. and Lsliker, H. (1974) Enzymatic digestion of the fmt component of human complement (C lq). J. Immunol. 112,2094-2 101.

Knobel, H. R., Villiger, W. and Isliker, H. (1975) Chernical analysis and electron microscopy studies of human Clq prepared by different methods. Eur. J. Immunol. 5,78- 82.

Koide, N., Nose, M. and Muramatsu, T. (1977) Recognition of IgG by Fc receptor and complement: effects of glycosidase digestion. Biuchem. Biophys. Res. Cornm. 75, 838- 844.

Koshland, M. E. (1985) The corning of age of the irnmunoglobulin J chain. Ann. Rev. Irnmunol. 3, 425-453.

Law, S. K. A. and Reid, K. B. M. (1995) Complement., 2nd Ed. (Ed. D. Male). Oxford University Press Inc.: New York.

Leatherbarrow, R. J. and Dwek, R. A. (1984) Binding of complement subcomponent Clq to mouse IgG1, IgG2b and IgG2b: A novel Clq binding assay. Molec. Immunul. 21, 32 1-327.

Leatherbarrow, R. J., Rademacher, T. W., Divek, R. A., Woof, J. M., Clark, A., Burton, D. R., Richardson, N. and Feinstein, A. (1985) Effector functions of a monoclonal aglycosylated mouse IgGZa: binding and activation of complement C 1 and interaction with human monocyte Fc receptor. Mol. Immrcnol. 22,407-415.

Lepow, 1. H., Naff, G. B., Todd, E. W., Pensky, J. and Hinz, C. F. (1963) Chromatographie resolution of the first component of human complement into three activities. J. Exp. Med. 177,

Leptin, M. and Melchers, F. (1983) A monoclonal antibody with specificity for murine mu heavy chain which inhibits the formation of antigen-specific direct IgM plaques. J. Imrnrtnol, Meth. 59, 53-6 1.

Leytus, S. P., Kurachi, K., Sakariassen, K. S. and Davie, E. W. (1986) Nucleotide sequence of the cDNA coding for human cornplement Clr. Biochemistry 25,48554863. Lin, T. Y. and Fletcher, D. S. (1978) Interaction of human Clq with insoluble immunoglobulin aggregates. Immunochemistry 15, 107- 1 11.

Lintin, S. J. and Reid, K. B. (1986) Studies on the structure of the human C4b-binding protein gene. FEBS Letters 204-77-81.

Loos, M., Martin, H. and Petry, F. (1989) The biosynthesis of Clq, the collagen-like and Fc-recognizing molecule of the complement system. Behring Insr. Min. 84, 32-4 1.

Lukas, T. J., Munoz, H. and Erikson, B. W. (1981) Inhibition of Cl-mediated immne hernolysis by monomenc and dheric peptides from the second constant domain of human imrnunoglobulin G. J. Immrtnol. 127, 2555-2560.

Luo, C., Thielens, N. M., Gagnon, J., Gal, P., Sarvari, M., Tseng, Y., Tosi, M., Zavodszky, P., Arlaud, G. J. and Schumaker, V. N. (1992) Recombinant human complement subcomponent Cls lacking B-hydroxyasparagine, sialic acid, and one of its two carbohydrate chains still reassembles with Clq and Clr to form a functional Cl complex. Biochemistry 31, 4254-4262.

Mackinnon, C. M., Carter, P. E., Smyth, S. J., Dunbar, B. and Fothergill, J. E. (1987) Molecular cloning of cDNA for human complement component Cls. The complete arnino acid sequence. Eur. J. Biochern. 169,547-553.

Makela, 0. and Litman, G. W. (1980) Lack of heterogeneity in antihapten antibodies of a phylogeneticaily primitive shark. Nature 287,639-640.

Marquart, M., Deisenhofer, J., Huber, R. and Palm, W. (1980) Crystallographic refinement and atornic models of the intact immunoglobulin molecule Kol and its antigen- binding fragment at 3.0 A and 1.0 A resolution. J. Molec. Biol. 141,369-391.

Marques, G., Anton, L.C., Barrio, E., Sanchez, A., Ruiz, S., Gavilanes, F. and Vivanco, F. (1993) Arginine residues of the globular regions of human Clq involved in the interaction with immunoglobulin G. J. Biol. Chem. 268, 10393- lO4Oî. Matsuuchi, L., Cam, G. M. and Koshland, M. E. (1986) Immunoglobulin J chain gene from the mouse. Proc. Natl. Acad. Sci. USA 83,456460,

Max, E. E. and Korsmeyer, S. J. (1985) Human J chain gene. Structure and expression in B lymphoid cells. J. Erp. Med 161, 832-849.

Metzger, H. (1974) Effect of antigen binding on the properties of antibody. Adv. in Imrnunol. 18, 169-207.

Metzger, H. (1978) The effects of antigen on antibodies: recent studies. Contemp. Top. Mol. Imrnunol. 7, 119- 152.

Michaelsen, T. E., Aase, A., Norderhaug, L. and Sandlie, 1. (1992) Antibody dependent cell-mediated cytotoxicity induced by chimeric mouse-human IgG subclasses and IgG3 antibodies with aitered hinge region. Molec. Immunol. 29, 3 19-326.

Michaelsen, T. E., Aase, A., Westby, C. and Sandlie, 1. (1990) Enhancement of complement activation and cytolysis of human IgG3 by deletion of hinge exons. Scand. J. hmttnol. 32, 5 17-528-

Milstein, C. P., Richardson, N. E., Deverson, E. V. and Feinstein, A. (1975) Interchain disulphide bridges of mouse immunoglobulin M. Biochem. J. 151,615-624.

Mole, J. E., Bhown, A. S. and Bennett, J. C. (1977) Primary structure of human J chain: alignrnent of peptides from chernical and enzymatic hydrolyses. Biochemistr-y 16, 3507- 3513.

Mosmann, T. R., Baumal, R. and Williwamson, A. R. (1979) Mutations affecting immunoglobulin light chah secretion by myeloma cells. 1. Functional analysis by cell fusion. Eur. J. Immunol. 9, 5 11-5 16.

Müller-Eberhard, H. (1975) Cornplement. Ann. Rev. Biochem. 44, 697-724.

Müller-Eberhard, H. J. (1968) Chemistry and reaction mechanisms of complement. Adv. Immunol. 8, 1-80. Müller-Eberhard, H. 1. (1988) Molecular organization and function of the complement System. Ann. Rev. Biochem. 57, 32 1-347.

Muraoka, S. and Shulman, M. J. (1989) Structural requirements for IgM assembly and cytolytic activity: Effect of mutations in the oligosaccharide acceptor site at Asn402 J. Imrnunol. 142, 695-70 1.

Nagaki, K. and Stroud, R. M. (1969) The relationship of the hemolytic activity of active C'ls to its TAMe esterase action: a new method of purification and assay. J. Imrnunol. 102, 42 1-430.

Nardella, F. A., Teller, D. C. and Mannik, M. (198 1) Studies on the antigenic determinants in the self-association of IgG rheumatoid factor. J. Exp. Med. 154, 112- 125.

Nardella, F. A., Teller, D. C., Barber, C. V. and Mannik, M. (1985) IgG rheumatoid factors and staphylococcal protein A bind to a comrnon molecular site on IgG. J. Exp. Med. 162, 18 11-1824.

Nezlin, R. S., Zayansky, Y. A., Kaivarainen, A. 1. and Stefani, D. V. (1973) Properties of myeloma immunoglobulin E(Yu). Chernical, fluorescence polarisation and spin-labeled s tudies. Immunochem. 10, 68 1-688.

Norderhaug, L., Brekke, O. H., Brernnes, B., Randi, S., Aase, A., Michaelsen, T. E. and Sandlie, 1. (199 1) Chimeric mouse human IgG3 antibodies with an IgG4-Like hinge region induce complement-mediated lysis more eficiently than IgG3 with normal hinge. Eut-. J. Immunol. 21, 2379-2384.

Nose, M. and Wigzell, H. (1983) Biological significance of carbohydrate chahs on monoclonal antibodies. Proc. Natl. Acad. Sei. USA 80, 6632-6636.

Odermatt, E., Berger, H. and Sano, Y. (1981) Size and shape of human Cl-inhibitor. FEBS Letters 131, 283-285.

Ohishi, K., Kanoh, M., Shinomiya, H., Hitsumoto, Y. and Utsurni, S. (1995) Complement activation by cross-linked B cell-membrane IgM. J. Imrnunol. 154, 3 173- 3 179. Oi, V. T., Vuong, T. M., Hardy, R., Reidler, I., Dangl, J., Henenberg, L. A. and Styer, L. (1984) Correlation between segmental flexibility and effector function of antibodies. Nature 307, 136-140.

Okada, M. and Utsumi, S. (1989) Role for the third constant domain of the IgG H chah in activation of complement in the presence of C 1-Inhibitor. J. Immunol. 142, 195-20 1.

Ovary, Z., Saluk, P. H., Quijada, J. and Lamm, M. E. (1976) Biologic acîivities of rabbit immunoglobulin G in relation to domains of the Fc region. J. Imrnunol. 116, 1265- 127 1.

Painter, R. H. (1993) The binding of Clq to irnrnunoglobulins. Behring Inst. Mit?. 93, 131-137.

Parekh, R. B., Dwek, R. A., Sutton, B. J., Fernandes, D. L., Leung, A., Stanworth, D., Rademacher, T. W., Mizuochi, T., Taniguchi, T., Matsuta, K. et al. (1985) Association of rheumatoid arthritis and primary osteoarthritis with changes in the glycosylation pattern of total serum IgG. Nature 316,45247.

Patri, S. and Nau, F. (1992) Isolation and sequence of a cDNA coding for the immunoglobulin mu chah of the sheep. Molec. Imrnunol. 29,829-836.

Perkins, S. J., Villiers, C. L., Arlaud, G. J., Boyd, J., Burton, D. R., Colomb, M. G. and Dwek, R. A. (1984) Neutron scattering studies of subcornponent Clq of first component C 1 of human complement and its association with subunit C lrzClsz within C 1. J. Mol. Biol. 179, 547-557.

Perkins, S. J. (1989) Models for the Cl cornplex determined by physical techniques. Behring Inst. Mitt. 84, 129- 14 1.

Perkins, S. J., Smith, K. F., Amatayakul, S., Ashford, D., Rademacher, T. W., Dwek, R. A., Lachmann, P. J. and Harrison, R. A. (1990) Two-domain structure of the native and reactive centre cleaved forms of CI inhibitor of huma. complement by neutron scattering. J. Mol. Biol. 214, 751-763. Perhs, S. J., Nealis, A. S., Sutton, B. J. and Feinstein, A. (1991) The solution structure of hurnan and mouse immunoglobulin M by synchrotron X-ray scattering and molecular graphics modelling. A possible mechanism for complement activation. J. Mol. Biol. 221, 1345-1366.

Perkins, S. J. ( 1993) Three-dimensional structure and molecular modelling of C 1- inhibitor. Behring Inst. Mitt. 93, 63-80.

Petillot, Y., Thibault, P., Thielens, N. M., Rossi, V., Lacroix, M., Coddeville, B., Spik, G., Schumaker, V. N., Gagnon, J. and Arlaud, G. 5. (1995) Analysis of the N-linked oligosaccharïdes of human C 1s using electrospray ionisation mass spectrometry. FEBS Letters 358, 323-8.

Petry, F., Reid, K. B. and Loos. M. (1991) Gene expression of the A- and B-chain of mouse Clq in different tissues and the charactenzation of the recombinant A-chain. J. Imrnunol. 147, 3988-3993.

Petry, F., Reid, K. B. and Loos, M. (1992) Isolation, sequence analysis and characterization of cDNA clones coding for the C chah of mouse Clq. Sequence sirnilarity of complement subcomponent C lq, collagen type VEI and type X and precerebellin. Eur. J. Biochem. 209, 129- 134.

Piez, K. A. and Tm, B. L. (1978) Sequence regularities and packing of collagen molecules. J. Molec. Biol. 122, 419-432.

Plaut, A. G., Cohen, S. and Tomasi, T. B., Jr. (1972) Immunoglobulin M: Fixation of hurnan complement by the Fc fragment. Science 176,55-56.

Plaut, A. G. and Tomasi, T. B. J. (1970) Immunoglobulin M: pentameric Fcmu fragments released by trypsin at higher temperahires. Proc. Natl. Acad. Sci. USA 65, 3 18-322.

Poljak, R. J. ( 1973) X-ray crystallographic studies of irnmunoglobulins. Contemp. Top. Molec. Immunol. 2, 1-26. Poon, P. H., Momson, S. L. and Schumaker, V. N. (1995) Structure and function of several anti-dansyl chimeric antibodies formed by domain interchanges between human IgM and mouse IgG2b. J. Biol. Chem. 270, 8571-8577.

Poon, P. H., Phillips, M. L. and Schumaker, V. N. (1985) lmmunoglobulin M possesses two binding sites for complement subcomponent C lq, and soluble 1: 1 and 2: 1 complexes are formed in solution at reduced ionic strength. L Bioi. Chem. 260,9357-9365.

Poon, P. H. and Schumaker, V. N. (1991) Measurement of macromolecular interactions between complement subcomponents Clq, Clr, Cls, and immunoglobulin IgM by sedimentation analysis using the analytical ultracentrifuge. J. Biol. Chem. 266, 5723- 5727.

Poon, P. H., Schumaker, V. N., Phillips, M. L. and Strang, C. J. (1983) Conformation and restricted segmental flexibility of Cl, the first component of human complement. J. Mol. Biol. 168, 563-577.

Porter, R. R. ( 1973) Structural studies of immunoglobulins. Science 180,713-7 16.

Porter, R. R. (1977) Structure and activation of the early components of complement. Federation Proceedings. 36,219 1-2 196.

Prystowsky, M. B., Kehoe, J. M. and Erickson, B. W. (1981) Inhibition of the classicaI complement pathway by synthetic peptides from the second constant domain of the heavy chah of human immunoglobulin G. Biochemistry 20,6349-6356.

Putnam, F. W., Florent, G., Paul, C., Shinoda, T. and Shimizu, A. (1973) Complete amino acid sequnce of the mu heavy chain of a human IgM immunoglobulin. Science 182, 287-290.

Rademacher, T. W., Homans, S. W., Fernandes, D. L., Dwek, R. A., Mizuochi, T., Taniguchi, T. and Kobata, A. (1983) Structural and conformational analysis of irnmunoglobulin-derived N-linked oligosaccharides. Biochem. Soc. Trans. 11, 132- 134. Randall, T. D., Brewer, J. W. and Corley, R. B. (1992) Direct evidence that J chah regulates the polymenc structure of IgM in antibody-secreting B Ceus. J. Biol. Chem 267, 18002-18007.

Randall, T. D., King, L. B. and Corley, R. B. (1990) The biologicd effects of IgM hexamer formation. Eur. J. immunol. 20, 1971 - 1979.

Rees, D. J., Jones, 1. M., Handford, P. A., Walter, S. J., Esnouf, M. P., Smith, K. J. and Brownlee, G. G. (1988) The role of beta-hydroxyaspartate and adjacent carboxylate residues in the fmt EGF domain of human factor IX. EMBO Journal 7,2053-206 1.

Reid, K. B. (1977) Amino acid sequence of the N-terminal forty-two amino acid residues of the C chain of subcomponent Clq of the first component of human cornplement. Biochem. J. 161, 247-25 1.

Reid, K. B. (1979) Complete amino acid sequences of the three collagen-like regions present in subcomponent Clq of the first component of human complement. Biochem. J. 179, 367-37 1.

Reid, K. B. (1983) Proteins involved in the activation and control of the two pathways of human complement. Biochem. Soc. Tram 11, 1- 12.

Reid, K. B., Lowe, D. M. and Porter, R. R. (1972) Isolation and characterization of Clq, a subcomponent of the first component of complement, from human and rabbit sera. Biochem. J. 130, 749-763.

Reid, K. B. and Porter, R. R. (1975) The structure and mechanism of activation of the first component of complement. Contemp. Topics Molec. Immunol. 4, 1-22.

Reid, K. B. and Thompson, E. 0. (1978) Amino acid sequence of the N-terminal 108 amino acid residues of the B chah of subcomponent Clq of the frst component of human complement. Biochem. J. 173, 863-868.

Reid, K. B. M. and Porter, R. R. (1976) Subunit composition and structure of subcomponent Clq of the first component of human complement. Biochem. J. 155, 19- 23. Rinfret, A., Home, C., Boux, H-,Marks, A., Domngton, K. J. and Klein, M. (1990) Isotype modulation of idiotypic expression in recombinant isotypic variants of MOPC 3 15. J. Immunol. 145, 925-93 1-

Romans, D. G., Tilley, C. A., Crookston, M. C., Falk, R. E. and Dorrington, K. J. ( 1977) Conversion of incomplete antibodies to direct agglutinins by mild reduction: evidence for segmental flexibility within the Fc fragment of immunogiobuiin G. Pruc. Natl. Acad. Sci. USA 74, 253 1-2535.

Rook, G. A., Steele, J., Brealey, R., Whyte, A., Isenberg, D., Sumar, N., Nelson, J. L., Bodman, K. B., Young, A., Roitt, 1. M. et al. (1991) Changes in IgG glycoform leveis are associated with remission of arthritis dunng pregnancy. Journal of Autoimmunity 4, 779-794.

Rudd, P. M., Leatherbarrow, R. J., Rademacher, T. W. and Dwek, R. A. (1991) Diversification of the IgG molecule by oligosaccharides. Molec. Imrnunol. 28, 1369- 1378.

Sakai, K. and Stroud, R. M. (1974) The activation of Cis with purified Clr. Immunochem. 11, 191-196.

Sdvesen, G. S., Catanese, J. J., Kress, L. F. and Travis, J. (1985) Primary structure of the reactive site of human Cl-inhibitor. J. Biol. Cham. 260, 2432-2436.

Sarnbrook, J., Fritsch, E. F. and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual., 2nd Ed. Cold Spring Harbor Laboratory Press: Plainview.

Sandberg, A. L. and Stollar, B. D. (1966) Cornparisons of antibodies reacting with DNA. II. Rabbit antibodies induced by nucleoside-protein conjugates. J. Immunol. 96,764-77 1.

Sasaki, T. and Yonemasu, K. (1984) Comparative studies on biological activities of subcomponents Clq of the first component of human, bovine, mouse and guinea-pig complement. Biochim. Biophys. Acta 785, 118- 122. Schiffer, M., Ainsworth, C., Xu, 2. B., Carperos, W., Olsen, K., Solomon, A., Stevens, F. J. and Chang, C. H. (1989) Structure of a second crystal form of Bence-Jones protein Loc: stnkingly different domain associations in two crystd forms of a single protein. Biochemistry 28,4066472.

Schiffer, M., Girling, R. L., Ely, K. R. and Edmundson, A. B. (1973) Structure of a lambda-type Bence-Jones protein at 3.5-A resolution. Biochemistry 12,46204631.

Schneider, W. P., Wensel, T. G., Stryer, L. and Oi, V. T. (1988) Geneticaily engineered irnmunoglobulins reveal structural features controlling segmenta1 flexibility. Proc. Natl. Acad, Sci. USA 85, 2509-25 13.

Schumaker, V. N., Calcott, M. A., Spiegelberg, H. L. and Mulier-Eberhard, H. J. (1976) Ultracentrifuge studies of the binding of IgG of different subclasses to the C lq subunit of the frst component of complement Biochemistry 15,5175-5 18 1.

Schurnaker, V. N., Hanson, D. C., Kilchherr, E., Phillips, M. L. and Poon, P. H. (1986) A molecular mechanism for the activation of the fmt component of complement by immune complexes. Mol. Imrn~inol.23, 557-565.

Schumaker, V. N. and Phillips, M. L. ( 1993) Electron microscope studies of C ls, C ls2, C 1r2, C 1r2C 1s2 and C 1-inhibitor. Behring Inst. Mitt. 93, 17-24.

Sellar, G. C., Blake, D. J. and Reid, K. B. (199 1) Characterization and organization of the genes encoding the A-, B- and C-chains of human complement subcornponent Clq. The complete derived amino acid sequence of human Clq. Biochem. J. 274,48 1-490.

Shelton, E., Yonemasu, K. and Stroud, R. M. (1972) Ultrastructure of the human complement component, Clq (negative staining-glutamine synthetase-biologically active Clq). Proc. Natl. Acad. Sci. USA 69, 65-68.

Shinkai, H. and Yonemasu, K. (1979) Hydroxylysine-linked glycosides of human complement subcomponent Clq and various collagens. Biochem. J. 177,847-852. Shopes, B., Weetail, M., Holowka, D. and Baird, B. (1990) Recombinant human IgGl- murine IgE chimeric Ig. Construction, expression, and binding to human Fc gamma receptors. J. Imrnunol. 145, 3842-3848.

Shopes, B. (1993) A genetically engineered human IgG with limited flexibility fully initiates cytolysis via complement. Molec. Immmol. 30,603-609.

Shulman, M. J., Heusser, C., Filkin, C. and Kohier, G. (1982) Mutations affecting the structure and hnction of irnmunoglobulin M. Molec. Cell. Biol. 2, 1033-1043.

Shulman, M. J., Pennell, N., Collins, C. and Hozumi, N. (1986) Activation of complement by irnmunoglobulin M is impaired by the substitution serine406 to asparagine in the irnmunoglobuLin y heavy chain. Proc. Natl. Acad. Sci. USA 83,7678-7682.

Siegel, R. C. and Schumaker, V. N. (1983) Measurement of the association constants of the complexes formed between intact C lq or pepsin-treated Clq stalks and the unactivated or activated C lr2C1 s2 tetramers. Molec. Imrnunol. 20, 53-66.

Sim, R. B., Arlaud, G. J. and Colomb, M. G. (1979) CI inhibitor-dependent dissociation of human complement component CL bound to immune complexes. Biochern. J. 179, 449-457.

Sim, R. B., Porter, R. R., Reid, K. B. and Gigli, 1. (1977) The structure and enzymic activities of the Clr and Cls subcomponents of Cl, the fiist component of human semm cornplement. Biochem. J. 163,219-227.

Sitia, R., Neuberger, M., Alberini, C., Bet, P., Fra, A., Valetti, C., Williams, G. and Milstein, C. (1990) Developmental regulation of IgM secretion: The role of the carboxy- terminal cysteine. Ce11 60, 78 1-790.

Sledge, C. R. and Bing, D. H. (1973) Binding properties of the human complement protein C lq. J. Biol. Chem. 248, 28 18-2823.

Smith, R. 1. F., Coloma, M. J. and Morrison, S. L. (1995) Addition of a p-tailpiece to IgG results in polymeric antibodies with enhanced effector functions including complement-mediated cytolysis by IgG4. J. Immunol. 154,2226-2236. S~rensen,V., Rasmussen, 1. B., Norderhaug, L., Natvig, I., Michaelsen, T. E, and Sandlie, 1. (1996) Effect of the IgM and IgA secretory tailpieces on polyrnerization and secretion of IgM and IgG. J. Immunol. 156, 2858-2865.

Spiegelberg, H. L. (1974)Biological activities of immunoglobulins of different classes and subclasses. Adv. Imrnunol. 19, 259-294.

Spycher, S. E., Nick, H. and Rickli, E. E. (1986) Human complement component C~S. Partial sequence determination of the heavy chah and identification of the peptide bond cleaved during activation. Eur. J. Biochem. 156,49-57.

Strang, C. J., Siegel, R. C., Phillips, M. L., Poon, P. H. and Schumaker, V. N. (1982) Ultrastructure of the first component of human complement: Electron rnicroscopy of the crosslinked cornplex. Proc. Natl. Acad. Sci. USA 79, 586-90.

Swanson, S. M., Dombrink-Kurtzman, M. A. and Voss, J. E. W. (1988)Clq binding by a high affinity anti-fluorescein murine monoclonal IgM antibody and monomeric subunits. Molec. Immunol. 25, 545-554.

Suba, E. A. and Csako, G. (1976) Clq (cl) receptor on human platelets: inhibition of collagen-induced platelet aggregation by C lq (C1) molecules. J. Immunol. 117, 304-409.

Sutton, B. J. and Phillips, D. C. (1983) The three-dimensional structure of the carbohydrate within the Fc fragment of immunoglobulin G. Biochem. Soc. Tram 11, 130- 132.

Takada, A., Shirahama, S. and Takada, Y. (1985) Inhibition by various peptides of the activation of Cl, the first component of complement, and the interaction of Cy2 domain of IgG with Clq. Imrnunophannacology 9, 87-95.

Tan, L. K., Shopes, R. J., Oi, V. T. and Momson, S. L. (1990) Influence of the hinge region on complement activation, C lq binding, and segmentai flexibility in chimeric human irnrnunoglobulins. Proc. Natl. Acad. Sci USA 87, 162- 166 (Published erratum appears in Proc. AM. Acad. Sci. USA ( 199 1) 88, 5066). Tao, H.-H. and Momson, S. L. (1989) Studies of aglycosylated chimenc mouse-human IgG: Role of carbohydrate in the structure and effector functions mediated by the human IgG constant region. J. Immunol. 143, 2595-260 1.

Tao, M.-H., Canfield, S. M. and Momson, S. L. (1991) The differential ability of human IgGl and IgG4 to activate complement is determined by the COOH-terminal sequence of the CH2 domain. J. Exp. Med. 173, 1025- 1028.

Tao, M. H., Smith, R. 1. F. and Morrison, S. L. (1993) Structural features of human irnrnunoglobulin G that determine isotype-specific differences in complement activation. J. Exp. Med. 178, 661-667.

Taylor, B., Wright, J. F., Arya, S., Isenman, D. E., Shulman, M. J. and Painter, R. H. (1994) Clq binding properties of monomer and polymer forms of mouse IgM p chah variants: Pro544Gly, and Pro434Ala. J. Immunol. 153, 5303-53 13.

Tenner, A. I., Lesavre, P. H. and Cooper, N. R. (198 1) Purification and radiolabeling of huma. C lq. J. Immunol. 127, 648-653.

Thielens, N. M., Aude, C. A., Lacroix, M. B., Gagnon, J. and Arlaud, G. J. (1990) ca2+ binding properties and ca2+-dependent interactions of the isolated NH,-terminal- a fragments of hurnan complement proteases cT~and C~S.J. Biol. Chem. 265, 14469- 14475.

Thielens, N. M., Illy, C., Bally, 1. M. and Arlaud, G. J. (1994) Activation of human complement serine-proteinase C lr is dom-regulated by a ca2+-dependent intramolecular control that is released in the C 1 complex through a signal transrnitted by Clq. Biochem J. 301,

Tosi, M., Duponchel, C., Meo, T. and lulier, C. (1987) Complete cDNA sequence of human complement Cls and close physical linkage of the homologous genes Cls and Ch. Biochem. 26, 85 16-8524.

Tseng, Y., Zavodszky, P. and Schumaker, V. N. (1997) The huma. complement Cl complex has a picomolar dissociation constant at room temperature. J. Immunol. 158, 937-944. Udaka, K., Okada, M. and Utsumi, S. (1986) Co-operation between the pair of Cy2 domains in C 1q-binding by rabbit IgG. Molec. Immunol. 23, 1 103- 11 10.

Utsumi, S. (1969) Stepwise cleavage of rabbit immunoglobulin G by papain and isolation of four types of biologicdy active Fc fragments. Biochem. J. 112,343-355.

Utsumi, S., Okada, M., Udaka, K. and Amano, T. (1985) Preparation and biologic characterization of hgments containing dimeric and monomeric Cfl domain of rabbit IgG. Molec. Imrnunol, 22, 8 1 1-8 19-

Vaientine, R. C. and Green, N. M. (1967) Electron microscopy of an antibody-hapten complex. J. Mol. Biol. 27, 615-617.

Valet, G. and Cooper, N. R. (1974a) Isolation and characterization of the proenzyme form of the C lr subunit of the first complement component. J. Imrnunol. 112, 1667-1673.

Valet, G. and Cooper, N. R. (1974b) Isolation and characterization of the proenzyme fom of the Cls subunit of the first complement component. J. Immunol. 112,339-350.

Van Der Zee, J. S., Beuvery, E. C., Van Ree, R. and Aaiberse, R. C. (1986) Human IgM antibodies do not activate guinea-pig complernent after interaction with soluble antigen. Molec. Immcrnol. 23, 669-673.

Van Schravendijk, M.-R. and Easterbrook-Smith, S. B. (1985) Analysis of the inhibition of Clq binding to aggregated imrnunoglobulin. Molec. Immunol. 22,503-504.

Van Schravendijk, M. R. and Dwek, R. A. (1982) Interaction of Clq with DNA. Molec. Immunol. 19, 1179-1 187.

Viiliers, C. L., Arlaud, G. J. and Colomb, M. G. (1985) Domain structure and associated functions of subcomponents CIr and Cls of the first component of human cornpiement. froc. Natl. Acad. Sci. USA 82, 4477-448 1.

Wautier, J.-L., Souchon, H., Reid, K. B. M., Peltier, A. P. and Caen, J. P. (1977) Studies on the mode of reaction of the fint component of complement with platelets: Interaction between the collagen-like portion of Clq and platelets. Imrnunochemistry 14, 763-766.

Wawrzynczak, E. J., Cumber, A. J., Parnell, G. D., Jones, P. T. and Winter, G. (1992) Blood clearance in the rat of a recombinant mouse monoclonal antibody lacking the N- Iinked oligosaccharide side chahs of the c~2domains. Molec. Immwtol. 29,213-220.

Weiss, V., Fauser, C. and Engel. J. (1986) Functional mode1 of subcomponent Cl of human complement. J. Mol. Biol. 189, 573-58 1.

White, K. D., Cummings, R. D. and Waxman, F. J. (1997) Ig N-glycan orientation can influence interactions with the complement system. J. Immunol. 158,426435.

Wiersrna, E. J., Chen, F., Bazin, R., Collins, C., Painter, R. H., Lemieux, R. and Shulman, M. J. (1997) Analysis of IgM structures involved in J chah incorporation. J. Immunol. 158, 17 19- 1726.

Wiersma, E. J. and Shulman, M. J. (1995) Assembly of IgM: Role of disulfide bonding and noncovalent interactions. J. Imrnrtnol. 154, 5265-5272.

Wiersrna, E. J., Collins, C., Fazel, S. and Shulman, M. J. (Subrnitted) Structural and functional analysis of J chain-deficient IgM.

Wiger, D. and Natvig, J. B. ( 1972) The effect of enzyme digestion on the antigenicity and biologic activity of Clq. J. Immunol. 109, 933-939.

Winchester, R. J., Kunkel, H. G. and Agnello, V. (1971) Occurrence of -globulin complexes in serum and joint fluid of rheumatoid arthritis patients: use of monoclonal rheumatoid factors as reagents for their demonstration. 3. Exp. Med. 134,286s+.

Wines, B. D. and Easterbrook-Smith, S. B. (1988) Enhancement of the binding of Clq to immune complexes by polyethylene glycol. Molec. Immunol. 25,263-266.

Wines, B. D. and Easterbrook-Smith, S. B. (1990) Carbodiirnide crosslinking of human Clq and rabbit IgG. Molec. Immunol. 27, 221-226. Winkelhake, J. L., Kunicki, T. J., Elcombe, B. M. and Aster, R. H. (1980) Effects of pH treatments and deglycosylation of rabbit immunoglobulin G on the binding of Clq. J. Biol. Chem. 255, 2822-2828.

Wright, A. and Momson, S. L. (1994) Effect of altered CH2-associated carbohydrate structure on the functional properties and in vivo fate of chimeric mouse-human immunoglobuh G1. J. Exp. Med. 180, 1087-96.

Wright, J. F., Shulman, M. J., Isenman, D. E. and Painter, R. H. (1988) Cl binding by murine IgM: the effect of a Pro-to-Ser exchange at residue 436 of the p-chain. J. BioZ. Chem. 263, 1122 1- 11226.

Wright, J. K., Tschopp, J., Jaton, i.C. and Engel, J. (1980) Dimeric, trimeric and tetrarneric complexes of irnmunoglobulin G fix complement. Biochem 187,775-780.

Wu, T. T. and Kabat, E. A. (1970) An analysis of the sequences of the variable regions of Bence-Jones proteins and myeloma light chains and their implications for antibody complementarity. J. Exp. Med. 132, 21 1-250.

Xu, Y., Oomen, R. and Kiein, M. H. (1994) Residue at position 331 in the IgGl and IgG4 CH2 domains contributes to their differentiai ability to bind and activate complement. J. BioZ. Chern. 269, 3469-3474.

Yarnawaki-Kataoka, Y., Miyata, T. and Honjo, T. (1981) The complete nucleotide sequence of mouse immunoglobin gamma 2a gene and evolution of heavy chain genes: further evidence for intervening sequence-mediated dornain transfer. Nucleic Acids Res. 9, 1365-1381.

Yasmeen, D., Ellerson, J. R., Domngton, K. J. and Painter, R. H. (1976) The structure and function of immunoglobulin domains. IV. The distribution of some effector functions among the Cy2 and Cy3 homology regions of human immunoglobulin G. J. Immunol. 116, 5 18-526.

Yonemasu, K. and Stroud, R. M. (1972) Structural studies on human Clq: non-covalent and covalent subunits. Imrnunochem. 9,545-554. Yonemasu, K., Stroud, R. M., Niedenneier, W. and Butler, W. T. (197 1) Chernical snidies on Clq: A modulator of immunoglobulin biology. Biochem. Biophys. Res. Cornm. 43, 1388-1395.

Ziccardi, R. J. (1981) Activation of the early components of the classicd complement pathway under physiologie conditions. J. Imrnunol. 126, 1769-1773.

Ziccardi, R. J. (1982a) A new role for chhibitor in homeostasis: Control of activation of the frst component of human complement. J. Immunol. 128,2505-2508.

Ziccardi, R. 1. (1982b) Spontaneous activation of the first component of human complement (Cl) by an intrarnolecdar autocatalytic mechanism. J. Imrnunol. 128,2500- 2504.

Ziccardi, R. J. (1985) Demonstration of the interaction of native Cl with monomeric imrnunoglobulins and C 1 inhibitor. J. ImrnunoL 134, 2559-2563.

Ziccardi, R. J. (1984) The role of immune complexes in the activation of the first component of human complement. J. Immunol. 132,283-288.

Ziccardi, R. 1. (1985) Nature of the interaction between the Clq and Clr2s2 subunits of the first component of human complement. Molec. Imrnunol. 22,489-494.

Ziccardi, R. J. and Cooper, N. R. (1979) Active disassembly of the fxst complement component, C- 1, by C- 1 inactivator. J. Imrnunol. 123,788-792.

Ziccardi, R. J. and Cooper, N. R. (1980) Development of an immunochemical test to assess Cl inactivator function in human semm and its use for the diagnosis of hereditary angioedema. CINI. Imunol. Immunopth. 15,465-47 1.

Ziccardi, R. J. and Tschopp, J. (1982~)The dissociation properties of native Cl. Biochem. Biophys. Res. Comrn. 107, 6 18-623.

Zikan, J., Novotny, I., Trapane, T. L., Koshland, M. E., Urry, D. W., Bennett, J. C. and Mesteclq, J. (1985) Secondary structure of the immunoglobulin J chah. Proc. Natl. Acad. Sci. USA 82, 5905-5909. Zollinger, W. D. and Mandrell, R. E. (1983) Importance of complement source in bactericidal activity of human antibody and murine monoclonal antibody to Meningococcal group B polysaccharide. Infection and Immunity 40,257-264. IMAGE EVALUATION TEST TARGET (QA-3)

APPLIED IMAGE . lnc -S 1653 East Main Street --2 ----2 Rochester, NY 14609 USA ---- Phone: 716f482-0300 ------Fa 71 ô/28û-5989