Recent Progress in the Understanding of the Structure ... - Science Direct

Immunobiol. (2002) 205, pp. 355 – 364 © 2002 Urban & Fischer Verlag http://www.urbanfischer.de/journals/immunobiol

The Classical Pathway of Complement Activation 1

MRC Immunochemistry Unit, Department of Biochemistry, 2 Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, Oxford, United Kingdom, and 3 Department of Biochemistry, Sofia University, Bulgaria.

Recent Progress in the Understanding of the StructureFunction Relationships of the Globular Head Regions of C1q UDAY KISHORE1, 2, MIHAELA S. KOJOUHAROVA3, and KENNETH B. M. REID1

Abstract The first step in the activation of the classical pathway of complement cascade by immune complexes involves the binding of the C-terminal globular head regions of C1q to the Fc regions of IgG or IgM, each globular head being composed of the C-terminal halves of one A-, one B- and one C-chain. Recent studies using recombinant forms of globular region appear to suggest that each globular head of C1q may be composed of three, structurally and functionally, independent domains/modules. The heterotrimeric organisation thus could offer functional flexibility and versatility to the whole C1q molecule. The crystal structure of an adipocyte-specific serum protein, Acrp-30, has revealed the existence of a structural fold shared by members of a new C1q/tumor necrosis factor (TNF) superfamily, characterized by a distinctive globular domain. The protein members seem to be active as self-assembling noncovalent trimers, whose individual chains fold as compact ‘jellyroll’ b sandwiches. The recognition of a C1q/TNF superfamily, which has wide-ranging functions, highlights the possibility that the globular regions of C1q may fulfill more binding functions than previously envisaged.

Introduction The first subcomponent of the classical complement pathway (CCP), C1q (460 kDa), is composed of 18 polypeptide chains (6A, 6B, and 6C). The A chain (223 residues), B chain (226 residues), and C chain (217 residues) each have a short (3–9 residues) N-terminal region (containing a half-cysteine residue involved in interchain disulfide bond formation), followed by a collagen-like sequence of ~81 residues and a C-terminal globular region (gC1q domain) of ~135 residues (1). Out of four conserved cysteines in each chain (at positions 4, 135, 154, and 171, as per the B chain numbering), the cysteines at position 4 are involved in the interchain disulfide bridges, yielding the A-B and C-C subunits; the other three cysteines are considered to yield one intra-chain disulphide bond and one free thiol group per C-terminal globular region. The interchain disulfide bonding yields Abbreviations: TNF = tumour necrosis factor; CCP = classical complement pathway; MBP = Escherichia coli maltose-binding protein; ghA, ghB and ghC = carboxy-terminal, globular head region of the C1q A, B and C chains, respectively; gC1q = globular domain of the C1q/TNF superfamily of proteins; SLE = systemic lupus erythematosus; EMILIN = elastin microfibril interface-located protein; SP-D = lung surfactant protein D 0171-2985/02/205/04-05-355 $ 15.00/0

356 · U. KISHORE, M. S. KOJOUHAROVA and K. B. M. REID 6A-B dimer subunits and 3C-C dimer subunits. The triple-helical collagen-like sequences in the A and B chains of an A-B subunit, together with one of the C chains present in a C-C subunit, form a structural unit (ABC-CBA), which is held together by both covalent and noncovalent bonds. Three of these structural units associate, via strong noncovalent bonds in the fibril-like central portion, to yield the hexameric C1q molecule (2, 3). The first component of complement, C1, is a complex of three glycoproteins – C1q, C1r, and C1s. C1s and C1r interact to form a Ca2+-dependent tetrameric proenzyme complex, C1r2-C1s2 , which makes contacts with the C1q collagen domain. Binding of C1q to immune complexes (IgG or IgM) via the gC1q domain is considered to induce a conformational change in the collagen region of C1q, which leads to the autoactivation of C1r which, in turn, activates C1s. The activated C1 complex then cleaves components C4 and C2 in the CCP cascade. After C1 activation and removal of activated C1r2-C1s2 by C1 inhibitor, the collagen regions appear to interact with cell surface receptors. Many activating ligands for C1, including immune complexes, bind to the gC1q domains; however, a number of non-immunoglobulin substances, such as DNA, C-reactive protein (CRP), serum amyloid protein (SAP), decorin, and some putative C1q receptors are thought to bind C1q via the collagen domain. The non-immunoglobulin ligands which are considered to interact with gC1q regions include the outer membrane protein, OmpK36 porin from Klebsiella pneumoniae, non-immunoglobulin salivary agglutinins, histidine- rich glycoprotein, gp41 of HIV-I, HTLV-I, and b-amyloid peptide (reviewed in ref. 4). Interaction of globular head region of C1q with immunoglobulins Human C1q shows only weak binding to the Fc regions of non-aggregated IgG, however, upon the presentation of multiple, closely-spaced Fc regions, as are found in immune complexes, the strength of binding of the hexameric C1q to IgG increases a thousandfold. The precise binding region of the IgG molecule for C1q is considered to be located in the C-terminal half of the Cg2 domain of IgG and specifically to three amino acids, Glu 318, Lys 320, and Lys 322, which are highly conserved in different IgG isotypes (5). The charged residues Asp 417, Glu 418, and His 420 in the Cm3 region of IgM have been proposed to form the site to which the gC1q domain binds (6) and Asp-356 has also been implicated. Recent investigations on C1q binding site on IgG have indicated that this site is different in human IgG1 and murine IgG3 (7, 8, 9). Alanine substitution at position Glu 318 and Lys 320 in chimeric IgG3 rituximab had little or no effect on C1q binding, while Alanine substitutions (Asp 270, Lys 322, Pro 329 and Pro 331) significantly reduced the ability of IgG1 to bind C1q and activate complement. These results indicate that there are species-specific differences in C1q binding site on IgG. When antibody-dependent complement-mediated lysis of human IgG3 with mutated Lys 276, Tyr 278, Asp 280, Glu 318. Lys 320, Lys 322 and Glu 333, was examined, only the mutants that lacked the positively charged side chain of Lys 322 showed strong reduction in haemolysis. Therefore, C1q binding site of human IgG is different from that found in mouse IgG2b and possibly in human IgG1. Substitution of Glu 333 with Ser and Lys 326 with tryptophan in human IgG2 confers biological activity in the complement dependent cytotoxicity assay in which the wild type IgG2 is inactive (7).

The globular region of C1q · 357

The question, whether a specific class of immunoglobulin is bound selectively by either of the A or B or C chains of C1q, or by a combination of all three, has been addressed recently (10, 11). In order to dissect the structural and functional autonomy of ghA, ghB and ghC, we have expressed the C-terminal globular head regions of human C1q A (ghA), B (ghB) and C (ghC) chains, in E. coli as soluble proteins linked to maltose-binding protein (MBP). The affinity-purified fusion proteins (MBP-ghA, ghB and ghC) bound differentially to heat-aggregated IgG and IgM, and three known C1q-binding peptides, derived from HIV-I, HTLV-I and b-amyloid. In ELISA assays, the MBP-ghA bound heataggregated IgG and IgM, in addition to binding specifically to HIV-I gp41-derived peptide; the MBP-ghB bound preferentially to aggregated IgG rather than IgM, in addition to binding b-amyloid peptide; whereas the MBP-ghC showed preference for IgM as well as HTLV-I peptide. Both MBP-ghA and MBP-ghB could inhibit C1q-dependent hemolysis of IgG- and IgM-sensitized sheep erythrocytes. However, for IgM-coated erythrocytes, MBP-ghC was a better inhibitor than MBP-ghB. These results appear to suggest that the C1q globular head region is likely to have a modular organization, being composed of three, structurally and functionally, independent domains/modules which retain multivalency in the form of a heterotrimer. Such heterotrimeric organisation thus could offer functional flexibility and versatility to the whole C1q molecule (KISHORE et al., submitted). The availability of these recombinant constructs has also allowed us to localise complementary binding sites and identify residues on the gC1q domains, which are involved in the interaction of C1q with immunoglobulins. We have investigated the contributions of the arginine residues (A162, B114, B129, B163 and C156) in the IgG binding by engineering a number of single-residue mutants (A162-Arg/Ala/Glu, B114-Arg/Ala/Glu, B129-Arg/Ala/Glu, B163-Arg/Ala/Glu and C156-Arg/Ala/Glu) via site-directed mutagenesis (KOJOUHAROVA et al., unpublished). The examination of the mutants for their ability to bind IgG and IgM confirms the importance of arginine residues of gC1q domain and suggests a central role of B114 in C1q-IgG interaction. Interaction of gC1q modules with apoptotic cells Programmed cell death via apoptosis is the physiological mode of cell death. Apoptotic cells are characterised by cell shrinkage, collapse of the nucleus and cytoplasmic blebbing. Cells undergoing death by apoptosis generate discrete subcellular structures, called apoptotic blebs, that contain either nuclear or cytoplasmic constituents, many of which are targeted by autoantibodies in systemic lupus erythematosus (SLE) patients. The finding that C1q can bind specifically to the surface blebs of apoptotic keratinocytes (12) and that the common autoantigens targeted in SLE can be found in high concentrations on the surface of the apoptotic cells (13), have led to the view that C1q deficiency may cause SLE as a result of an impaired clearance of apoptotic cells (14). In addition, immunisation with apoptotic cells has been shown to stimulate autoantibody production (15), suggesting that defects in the pathways by which apoptotic cells are processed could play an important role in driving an autoimmune response. In C1q knock-out mice, which have glomerulonephritis with immune deposits, a large number of apoptotic bodies are also present in diseased glomeruli. C1q-deficient mice that also lack C2 and factor B develop glomerulonephritis without glomerular C3 deposition (16). However, mice lacking C2 and factor B only do not develop either glomerulonephritis or autoantibodies. Therefore,

358 · U. KISHORE, M. S. KOJOUHAROVA and K. B. M. REID C1q may protect against autoimmunity by serving as an opsonin in the efficient recognition and physiological clearance of apoptotic cells. Although C1q is required to maintain immune tolerance, the molecular mechanism is not fully understood. Recently, C1q has been shown to bind directly to apoptotic blebs of peripheral blood mononuclear cells (PBMCs) and vascular endothelial cells via gC1q domain (17), suggesting that surface blebs may be capable of directly activating the CCP. Thus, appropriate recognition of apoptotic cells by gC1q and targeted clearance of the molecular contents of surface blebs to complement receptors via exposed collagen region of C1q (for example, CR1 on the surface of erythrocytes) may be critical for the maintenance of immune tolerance. The recombinant forms of globular heads (ghA, ghB and ghC) also show specific and dose-dependent binding to the apoptotic blebs of PBMCs (KISHORE et al., submitted). We are currently trying to ascertain if individual globular heads bind differentially to different autoantigens clustered within discrete populations of blebs. Proteins containing C1q-like globular domains Modules of the same type as the gC1q domains are also found in a variety of noncomplement proteins, which include both collagenous and noncollagenous molecules (reviewed in ref. 4). It is a growing family of novel proteins which include human type VIII (18) and type X collagens (19), hibernation proteins- HP-20, -25, and -27 (20), adiposespecific mouse Acrp-30/AdipoQ and human adiponectin (21), saccular collagen (22), elastin microfibril interface-located protein (EMILIN) (23), precerebellins (24), and multimerin (25). A sequence comparison between the gC1q domains reveals a conserved framework of aromatic and other hydrophobic residues found in a region of ~130 amino acids. The polypeptide chains containing these gC1q modules form either a homotrimeric structure (type X collagen, multimerin, Acrp-30, and possibly saccular collagen), or a heterotrimeric structure (C1q, hibernation proteins, type VIII collagen, and possibly precerebellins). Although the short chain collagen variant, type VIII, is considered to be composed of two distinct gene products, a1 and a2 , it has been recently demonstrated that both chains can also preferentially form pepsin-resistant, homotrimeric molecules and exist as two distinct proteins (26). The type X collagen molecule consists of three a1 (X) chains, each having a C-terminal gC1q domain. Interestingly, in the disease called Schimid’s metaphyseal chondrodysplasia which is a mild autosomal disorder associated with growth plate abnormalities, all the mutations have been located in the gC1q domain and this affects the folding of the domain and the trimerization of the collagen region (27). Multimerin is localised in platelets and endothelium of blood vessels as complexed with factor V. It is composed of homotrimeric subunits, which are linked by interchain disulfide bonds to form large homomultimers. In addition to having a gC1q domain, which is a likely site for protein-protein interactions, multimerin also has an RGDS (Arg-Gly-Asp-Ser) motif, epidermal growth factor (EGF) domain, and a domain containing coiled-coil structures. EMILIN has a region containing two leucine zippers and at least four heptad repeats with a high potential for forming amphipathic coiled-coil a-helices, and at the N-terminus, a partial EGF-like motif as found in multimerin (23). The gC1q domain is a likely site for initial interchain association and the coiled-coil a-helices may be involved in associating homotrimeric subunits of EMILIN into larger aggregates. The gC1q domain in


saccular collagen may self-associate to form a supramolecular organisation together with hydrating glycoproteins in the otolithic membrane (28). In the chipmunk, a mammalian hibernator, four proteins (HP-20, -25, -27, and -55) disappear from blood prior to the onset of hibernation (29). Of these proteins, HP-20, -25, and -27 are members of the C1q family, whereas HP-55 is similar to a1-antitrypsin, a member of the serpin (serine protease inhibitor) superfamily. In the nonhibernating state, HP-20, -25, and -27 bind via their collagen regions to HP-55 in the plasma to form a 140-kDa complex and may thus regulate the functions of HP-55. Precerebellin has a remarkable similarity to the globular region of the C1q B chain. The sequence alignment between precerebellin and human C1q B chain is only shifted by four amino acids relative to the C-terminal ends of both proteins (24). Crystal structure of murine homologue, Acrp-30/AdipoQ The crystal structure of the recombinant form of homotrimeric Acrp-30 gC1q domain has revealed an asymmetrical trimer of b-sandwich protomers, each of which has a tenstrand jelly-roll folding topology. Such folding is also seen in the TNF family proteins (30), which control many aspects of inflammation, adaptive immunity, apoptosis, energy homeostasis, and organogenesis. This fold would be expected to be common to all the members of C1q family proteins. The trimer is bell-shaped, with a wide base. The trimer contacts take place through a cluster of hydrophobic interactions near the base. The trimer interface near the apex is largely hydrophilic: these features are in common with TNF family trimers (31). The Acrp-30 structure shows that the globular region forms strong trimers, stabilised by a central hydrophobic interface, thus suggesting a structural basis for their role in collagen assembly. Each of the ten b-strands of Acrp-30 can be superimposed onto the ten-strands of TNF-a, TNF-b, and CD40 ligand (CD40L). The relative positions and lengths of these b-strands are almost identical between Acrp-30 and TNFs. The TNF and C1q proteins also have similar gene structures – their gC1q domains are each encoded within one exon. It is likely that these two groups of proteins arose by divergence from a common precursor molecule of the innate immune system, and thus establish a C1q/TNF molecular superfamily. Curiously, the three-fold jelly-roll structure is remarkably similar to the capsid proteins of small RNA viruses (several avian virus receptors are TNF receptors). Given similar intracellular signalling domains of some TNF receptors and toll-like receptors, which function as germline-encoded receptors for surface proteins of pathgens, C1q/TNF may well have descended via horizontal capture of a gene encoded by an ancient viral pathogen. There are several aspects of immune mechanisms and energy homeostasis where members of C1q/TNF superfamily cross-over. CD40L-deficiency impairs CD4+ T cell priming, follicular dendritic cell differentiation, germinal centre formation, and class switching. The IgM expressing cells can not undergo isotype conversion to IgG expression, leading to hyper IgM syndrome (32). Administration of neutralising anti-CD40L in mature mice causes the abrupt disappearance of existing germinal centres, revealing the need for CD40-CD40L interaction (33). C1q knock-out mice are also defective in antibody class switching in the T cell dependent humoral response (34). Like C1q, TNF-a is produced in response to infection and brings about inflammation, cell proliferation, and apoptosis. Adiponectin can inhibit proliferation of myelomonocytic lineage cells, possibly via induc-

360 · U. KISHORE, M. S. KOJOUHAROVA and K. B. M. REID tion of apoptosis. Adiponectin also suppresses mature macrophage function by significantly inhibiting their phagocytic activity and their LPS-induced production of TNF-a. It appears to act as a negative regulator in hematopoiesis and immune systems, and therefore, may be involved in ending inflammatory responses through its inhibitory functions. Both C1q and adiponectin inhibit proliferation of myelomonocytic progenitors, which is a novel function of this family (35). Adiponectin has been shown to reverse insulin resistance associated with obesity (36, 37), by decreasing triglyceride content in muscle and liver of obese mice. Decreased adiponectin has been implicated in the development of insulin resistance in the mouse models of obesity and type 2 diabetes. Interestingly, the gC1q domain of adiponectin can ameliorate hyperglycemia and hyperinsulinemia much more potently than full-length adiponectin. A collagen-free form of adiponectin/Acrp-30 has been shown to be present in serum, suggesting that full-length adiponectin can undergo proteolytic processing (36). Like adiponectin, TNF-a is a major secretory product of adipocytes and plays a role in energy homeostasis where it is implicated in obesity and insulin resistance. Similarly, the HP-27 from Siberian chipmunk has also been implicated in energy homeostasis. Trimerization of gC1q domains Proteins belonging to the C1q/TNF superfamily are characterized by a distinctive globular domain (gC1q domain), which is situated at the C-terminus of a collagen stalk. They often form a characteristic superstructure in which three protomers trimerize to form a collagen triple-helix, and these trimers multimerize to form a bouquet. In the structure of Acrp-30 and the members of the TNF family, the N- and the C-termini within one module are directly adjacent to one another (30), suggesting that gC1q domains might assemble as either N-terminal or C-terminal appendages to the collagen stalks, as observed in the C-type lectin domains (38). However, the C-terminal globular region containing carbohydrate recognition domains (CRDs) in the C-type lectins have a very different protein fold. This includes a neck region composed of a coiled-coil of three a-helices that accounts for the major part of the trimer interface. The C-type lectin domains lacking this neck region are not formed as trimers (39). In the proteins belonging to the C1q/TNF superfamily, the gC1q domain leads directly into the Gly-X-Y repeats of the collagen region (except in multimerin), with no intervening neck region, and thus, the folding of collagen triple-helices is thought to nucleate within the gC1q domain. Multimerin and EMILIN are unique in having a-helical coiledcoil structures (EMILIN also has two leucine zippers) but so far, their involvement in trimerisation or multimerisation has not been formally demonstrated. It appears likely that the specific hydrophobic bonds within the sequence of the gC1q domains facilitate the interchain recognition and alignment of the three chains (whether homotrimeric or heterotrimeric) at their C-terminal ends and act as a nucleation centre for collagen triple-helix to trimerize (40). However, a single globular head module of C1q does not appear to homotrimerise on its own, unlike other members of the C1q family (41), and engineering a trimerising, a-helical coiled-coil, neck region of human surfactant protein D (SP-D), upstream to single-chain globular head can yield a stable and functional homotrimer. Although precerebellin (Cbln1) is the precursor of the brain-specific hexadecapeptide, and has properties of a conventional neuropeptide, its function is controversial


because of its structural similarity to gC1q domain. Recently, a new member of this family, Cbln3, has been cloned (42) having gC1q domain. The members of Cbln family are capable of formimg homomeric and heteromeric complexes. Cbln3 binds avidly to Cbln1, but its homomeric interaction is weak. It appears that the function of Cbln3 is dependent on its incorporation into a complex with Cbln1. Since this complex is likely to be a heterotrimer, there may yet be another member of the precerebellin family to be identified. The Cbln1, Cbln2 and Cbln3 are independent genes that map to the mouse chromosome 8, 18 and 14 respectively. Given their levels of homology, these genes have probably arisen by gene duplication, as also reflected by their identical intron-exon organisation (42). Concluding remarks The expression of the individual globular heads of each of the A, B, and C chains has allowed examination of the modular organisation of the C-terminal regions of C1q. It is also possible to produce a recombinant homotrimer of C1q ghB module by making use of the trimerising, a-helical coiled-coil, neck region of human SP-D (41). The homotrimeric globular head of C1q B chain is an inhibitor of the CCP as it blocks haemolysis of erythrocytes by intact C1q. Although complement is an important line of defense against pathogens, its upregulated activation may cause host tissue damage, leading to autoimmune diseases, adult respiratory distress syndrome, stroke, burn injuries, and complications of cardiopulmonary bypass and xenotransplantation. The generation of homotrimers of ghB (and of ghA and ghC) would open up the possibility of blocking activation of the CCP at an early stage. Based on the Acrp-30 crystal structure, four residues are conserved throughout C1q and TNF families – Tyr 161, Gly 159, Phe 237, and Leu 242 (based on Acrp-30 numbering). Each of these residues is important in the packing of the protomer’s hydrophobic core. Chemical modification studies have implicated two regions of the C1q globular domain in IgG binding (43): these are in the C1q B chain (site 1, localised to residues 114–129) and in the A and C chains (site 2, both around residue 160). Each of these sites maps to the exterior of the trimer and would be expected to be on two separate loops, although site 1 appears more attractive as a candidate binding surface. The availability of recombinant globular heads of C1q should facilitate identification of interaction sites via mutagenesis. The Acrp-30 crystal structure has also given plenty of clues regarding the overall fold of proteins belonging to newly designated C1q/TNF superfamily. The protein members seem to be active as self-assembling noncovalent trimers, whose individual chains fold as compact ‘jellyroll’ b sandwiches and interact at hydrophobic interfaces. Trimers would require more contacts and may cause an exponential increase in the avidity with cognate receptor/protein complexes. Therefore, future discovery of ligand-receptor interaction of many members of this new family is likely to involve stoichiometrically defined protein complexes with three-fold symmetry. Trimers should also provide a unity of design and function for the members of this supefamily. A cumulative account of C1q/TNF superfamily necessitates new lines of investigation into their roles in the processes as disparate as host defense, inflammation, apoptosis, autoimmunity, cell differentiation, organogenesis, hibernation and insulin-resistant obesity. Although these proteins have gC1q domain as one unique

362 · U. KISHORE, M. S. KOJOUHAROVA and K. B. M. REID structural attribute, it is tempting to envisage certain common themes that may unite their actions in different tissues, therefore justifying the evolutionary success of this superfamily. Acknowledgements

The authors are funded by the European Commission (CEC_2LK2-2000-00325) and the Medical Research Council, UK. We thank P. WATERS for critically reading the manuscript.

References 1. SELLAR, G. C., D. J. BLAKE, AND K. B. M. REID. 1991. Characterisation and organisation of the genes encoding the A, B and C chains of human complement subcomponent C1q. The complete derived amino acid sequence of human C1q. Biochem. J. 274: 481–490. 2. REID, K. B. M., and R. R. PORTER. 1976. Subunit composition and structure of subcomponent C1q of the first component of human complement. Biochem. J. 155: 19–23. 3. KISHORE, U., and K. B. M. REID. 1999. Modular organization of proteins containing C1q-like globular domains. Immunopharmacol. 41: 15–21. 4. KISHORE, U., and K. B. M. REID. 2000. C1q: structure, function, and receptors. Immunopharmacol. 49: 159–170. 5. DUNCAN, A. R., and G. WINTER. 1988. The binding site for C1q on IgG. Nature 332: 738–740. 6. PERKINS, S. J., A. S. NEALIS, B. J. SUTTON, and A. FEINSTEIN. 1991. Solution structure of human and mouse immunoglobulin M by synchrotron X-ray scattering and molecular modelling: a possible mechanism for complement activation. J. Mol. Biol. 221: 1345–1366. 7. IDUSOGIE, E. E., P. Y. WONG, L. G. PRESTA, H. GAZZANO-SANTORO, K. TOTPAL, M. ULTSH and M. G. MULKERRIN. 2001. Engineered antibodies with increased activity to recruit complement. J. Immunol. 166: 2571. 8. IDUSOGIE, E. E., L. G. PRESTA, H. GAZZANO-SANTORO, K. TOTPAL, P. Y. WONG, M. ULTSH, Y. G. MENG and M. G. MULKERRIN. 2000. Mapping of the C1q binding site on rituxan, a chimeric antibody with a human IgG1 Fc. J. Immunol. 164: 4178. 9. THOMMESEN, J. E., T. E. MICHAELSEN, G. A. LOSET, I. SANDLIE and O. H. BREKKE. 2000. Lys322 in the human IgG3 CH2 domain is crucial for antibody-dependent complement activation. Mol. Immunol. 37: 995. 10. KISHORE, U., L. E. A. LEIGH, P. EGGLETON, P. STRONG, M. V. PERDIKOULIS, A. C. WILLIS, and K. B. M. REID. 1998. Functional characterisation of a recombinant form of the carboxy-terminal, globular head region of the B chain of human serum protein, C1q. Biochem. J. 333: 27–32. 11. KOJOUHAROVA, M. S., I. D. PANCHEV, M. I. TCHORBADJIEVA, K. B. M. REID, and H. J. HOPPE. 1998. Differential binding of IgG and of a HIV gp41 peptide by the B chain and A chain globular head sequences of C1q, respectively. J. Immunol. 161: 4325. 12. KORB, L. C., and J. M. AHEARN. 1997. C1q binds directly and specifically to surface blebs of apoptotic human keratinocytes: complement deficiency and systemic lupus erythematosus revisited. J. Immunol. 158: 4525–4528. 13. CASCIOLA-ROSEN, L. A., G. ANHALT, and A. ROSEN. 1994. Autoantigens targeted in systemic lupus erythematosus are clustered in two populations of surface structures on apoptotic keratinocytes. J. Exp. Med. 179: 1317–1330. 14. BOTTO, M., C. DELL’AGNOLA, A. E. BYGRAVE, E. M. THOMPSON, H. T. COOK, F. PETRY, M. LOOS, P. P. PANDOLFI, and M. J. WALPORT. 1998. Homozygous C1q deficiency causes glomerulonephritis associated with multiple apoptotic bodies. Nat. Genet. 19: 56–59. 15. MEVORACH, D., J. L. ZHOU, X. SONG, and K. B. ELKON. 1998. Systemic exposure to irradiated apoptotic cells induces autoantibody production. J. Exp. Med. 188: 387–392. 16. MITCHELL , D. A., P. R. TAYLOR, H. T. COOK, J. MOSS, A. E. BYGRAVE, M. J. WALPORT, and M. BOTTO. 1999. C1q protects against the development of glomerulonephritis independently of C3 activation. J. Immunol. 162: 5676–5679.

The globular region of C1q · 363 17. NAVRATIL J. S., S. C. WATKINS, J. J. WISNIESKI, and J. M. AHEARN. 2001. The globular heads of C1q specifically recognize surface blebs of apoptotic vascular endothelial cells. J. Immunol. 166: 3231–3239. 18. YAMAGUCHI, N., P. D. BENYA, M. VAN DER REST, and Y. NIMOMIYA. 1989. The cloning and sequencing of a1 (VIII) collagen cDNAs demonstrate that the type VIII collagen is a short chain collagen and contains triple-helical and carboxy-terminal non-triple helical domains similar to those of type X collagen. J. Biol. Chem. 270: 16022–16029. 19. NINOMIYA, Y., M. GORDON, M. VAN DER REST, T. SCHMID, T. LINSENMAYER, and B. R. OSLEN. 1986. The developmentally regulated type X collagen gene contains a long open reading frame without introns. J. Biol. Chem. 261: 5041–5050. 20. TAKAMATSU, N., K. I. OHBA, J. KONDO, N. KONDO, and T. SHIBA. 1993. Hibernation associated gene regulation of plasma proteins with a collagen-like domain in mammalian hibernators. Mol. Cell Biol. 13: 1516–1521. 21. SCHERER, P. E., S. WILLIAMS, M. FOGLIANO, G. BALDINI, and H. F. LODISH. 1995. A novel serum protein similar to C1q, produced exclusively in adipocytes. J. Biol. Chem. 270: 26746–26749. 22. DAVIS, J. G., J. C. OBERHOLTZER, F. R. BURNS, and M. I. GREENE. 1995. Molecular cloning and characterisation of an inner ear-specific structural protein. Science 267: 1031–1034. 23. DOLIANA, R., M. MONGIAT, F. BUCCIOTTI, E. GIACOMELLO, R. DEUTZMANN, D. VOLPIN, G. M. BRESSAN, and A. COLOMBATTI. 1999. EMILIN, a component of the elastic fibre and a new member of the C1q/tumour necrosis factor superfamily of proteins. J. Biol. Chem. 274: 16773–16781. 24. URADE, Y., J. OBERDICK, R. MOLINAR-RODE, and J. I. MORGAN. 1991. Precerebellin is a cerebellum-specific protein with similarity to the globular domains of complement protein C1q B chain. Proc. Natl. Acad. Sci. 88: 1069–1073. 25. HAYWARD, C. P. M., J. A. HASSELL, G. A. DENOMME, R. A. RACHUBINSKI, C. BROWN, and J. G. KELTON. 1995. The cDNA sequence of human endothelial cell multimerin. A unique protein with RGDs, coiled-coil and epidermal growth factor-like domains and a carboxy-terminus similar to the globular domain of component C1q and collagen type VIII and X. J. Biol. Chem. 270: 18246–18251. 26. GREENFIELD, N. S., B. M. RUGER, Q. HASAN, and P. F. DAVIS. 2000. The a1 (VIII) and a2 (VIII) collagen chains form two distinct homotrimeric proteins in vivo. Matrix Biol. 19: 19–28. 27. DUBLET, B., T. VERNET, and M. VAN DER RESET. 1999. Schimid’s metaphyseal chondrodysplasia mutations interfere with folding of the C-terminal domain of human collagen X expressed in Escherichia coli. J. Biol. Chem. 274: 18909–18915. 28. DAVIS, J. G., F. R. BURNS, D. NAVARATNAM, A. M. LEE, S. ICHIMIYA, J. C. OBERHOLTZER, and M. I. GREENE. 1997. Identification of a structural constituent and one possible site of postembryonic formation of teleost otolithic membrane. Proc. Natl. Acad. Sci. U.S.A. 94: 709–712. 29. KONDO, N., and J. KONDO. 1992. Identification of novel blood proteins specific for mammalian hibernation. J. Biol. Chem. 267: 474–478. 30. SHAPIRO, L., and P. E. SCHERER. 1998. The crystal structure of a complement-1q family protein suggests an evolutionary link to tumour necrosis factor. Curr. Biol. 8: 335–338. 31. JONES, E. Y., D. I. STUART, and N. P. WALKER. 1989. Structure of tumour necrosis factor. Nature 338: 225–228. 32. GREWAL, I. S., and R. A. FLAVELL. 1988. CD40 and CD154 in cell-mediated immunity. Annu. Rev. Immunol. 16: 111–135. 33. RENSHAW, B. R., W. C. R. FANSLOW, R. J. ARMITAGE, K. A. CAMPBELL, D. LIGGIT, B. WRIGHT, B. L. DAVISON, and C. R. MALISZEWSKI. 1994. Humoral response in CD40 ligand deficient mice. J. Exp. Med. 180: 1889–1900. 34. CUTLER, A. J., M. BOTTO, D. VAN ESSEN, R. RIVI, K. A. DAVIES, D. GRAY, and M. J. WALPORT. 1998. T cell-dependent immune response in C1q-deficient mice: defective interferon gamma production by antigen specific T cells. J. Exp. Med. 187: 1789–1797. 35. YOKOTA, T., K. ORITANI, I. TAKAHASHI, J. ISHIKAWA, A. MATSUYAMA, N. OUCHI, S. KIHARA, T. FUNAHASHI, A. J. TENNER, Y. TOMIYAMA, and Y. MATSUZAWA. 2000. Adiponectin, a new member of the family of soluble defense collagens, negatively regulates the growth of myelomonocytic progenitors and the functions of macrophages. Blood 96: 1723–1732.

364 · U. KISHORE, M. S. KOJOUHAROVA and K. B. M. REID 36. YAMAUCHI, T., J. KAMON, H. WAKI, Y. TERAUCHI, N. KUBOTA, K. HARA, Y. MORI, T. IDE, K. MURAKAMI, N. TSUBOYAMA-KASAOKA, O. EZAKI, Y. AKANUMA, O. GAVRILOVA, C. VINSON, M. L. REITMAN, H. KAGECHIKA, K. SHUDO, M. YODA, Y. NAKANO, K. TOBE, R. NAGAI, S. KIMURA, M. TOMITA, P. FROGUEL, and T. KADOWAKI. 2001. The fat-derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat. Med. 7: 941–946. 37. BERG, A. H., T. P. COMBS, X. DU, M. BROWNLEE, and P. E. SCHERER. 2001. The adipocytesecreted protein Acrp-30 enhances hepatic insulin action. Nat. Med. 7: 947–953. 38. KISHORE, U., and K. B. M. REID. 2001. Structures and functions of mammalian collectins. Mammalian carbohydrate recognition systems. In: Crocker, P. C., Editor. In the Series Results and Problems in Cell Differentiation, Springer-Verlag, Heidelberg. 39. KISHORE, U., J. Y. WANG, H. J. HOPPE, and K. B. M. REID. 1996. The a-helical neck region of human lung surfactant protein D is essential for the binding of the carbohydrate recognition domains to lipopolysaccharides and phospholipids. Biochem. J. 318: 501–511. 40. ZHANG, Y., and Q. CHEN. 1999. The noncollagenous domain a1 of Type X collagen. A novel motif for trimer and higher order multiple formation without a triple helix. J. Biol. Chem. 274: 22409–22413. 41. KISHORE, U., P. STRONG, M. V. PERDIKOULIS, and K. B. M. REID. 2001. A recombinant homotrimer, composed of the a-helical neck region of human surfactant protein D and C1q B chain globular domain, is an inhibitor of the classical complement pathway. J. Immunol. 166: 559–565. 42. PANG, Z., J. ZUO, and J. I. MORGAN. 2000. Cbln3, a member of the pecerebellin family that binds specifically to Cbln1. J. Neurosci. 17: 6333–6339. 43. MARQUES, G., L. C. ANTON, E. BARRIO, A. SANCHEZ, S. RUIZ, F. GAVILANES, and F. VIVANCO. 1993. Arginine residues of the globular regions of human C1q involved in the interaction with immunoglobulin G. J. Biol. Chem. 268: 10393–10402. Note added in proof: The crystal structure of human collagen × NC1 domain at 2 Å resolution has recently been published: BOGIN, O., M. KVANSAKUL, E. ROM, J. SINGER, A. YAYON, and E. HOHENESTER. 2002. Insight into Schmid Methaphyseal Chondrodysplasia from the crystal structure of the collagen × NC1 Domain trimer. Structure 10: 165–173. Dr. U. KISHORE, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, Oxford OX3 9DS, UK (Fax: +44 1865 222402; e-mail: [email protected])