Structural biology of the C1 complex of complement unveils the mechanisms of its activation and proteolytic activity

Molecular Immunology 39 (2002) 383–394

Review

Structural biology of the C1 complex of complement unveils the mechanisms of its activation and proteolytic activity Gérard J. Arlaud∗ , Christine Gaboriaud, Nicole M. Thielens, Monika Budayova-Spano, Véronique Rossi, Juan Carlos Fontecilla-Camps Laboratoire d’Enzymologie Moleculaire, Institut de Biologie Structurale Jean-Pierre Ebel, CEA-CNRS-Université Joseph Fourier, 41 Rue Jules Horowitz, Avenue des Martyrs, 38027 Grenoble Cedex 1, France Received 16 May 2002; received in revised form 25 June 2002; accepted 5 July 2002

Abstract C1 is the multimolecular protease that triggers activation of the classical pathway of complement, a major element of antimicrobial host defense also involved in immune tolerance and various pathologies. This 790 000 Da complex is formed from the association of a recognition protein, C1q, and a catalytic subunit, the Ca2+ -dependent tetramer C1s–C1r–C1r–C1s comprising two copies of each of the modular proteases C1r and C1s. Early studies mainly based on biochemical analysis and electron microscopy of C1 and its isolated components have allowed for characterization of their domain structure and led to a low-resolution model of the C1 complex in which the elongated C1s–C1r–C1r–C1s tetramer folds into a more compact, “8-shaped” conformation upon interaction with C1q. A major strategy used over the past years has been to dissect the C1 proteins into modular segments to characterize their function and solve their structure by either X-ray crystallography or nuclear magnetic resonance spectroscopy (NMR). The purpose of this review is to focus on this information, with particular emphasis on the architecture of the C1 complex and the mechanisms underlying its activation and proteolytic activity. © 2002 Elsevier Science Ltd. All rights reserved. Keywords: Innate immunity; Complement; C1; Molecular recognition; Serine protease; Protein modules

1. Introduction The classical pathway of complement is recognized as a major element of antimicrobial host defense due to its ability to bind pathogens either directly or after prior recognition by antibodies, and to trigger effector mechanisms aimed at limiting infection. A further role of complement activation is to instruct and stimulate the acquired immune response thereby providing a link between the innate and adaptive immune systems (Fearon and Locksley, 1996; Hoffmann et al., 1999). Besides its protective action against infection, the classical pathway of complement also plays a role in immune tolerance, through its ability to recognize and induce clearance of apoptotic cells (Fishelson et al., 2001), and is a major causative agent of rejection in xenograft transplantation (Dalmasso, 1992). A further, more paradoxical effect of the classical pathway of complement arises from the ability of the C1 complex to recognize abnormal structures from self, and thereby to trigger undesired effector mecha∗ Corresponding author. Tel.: +33-4-38-78-4981; fax: +33-4-38-78-5494. E-mail address: [email protected] (G.J. Arlaud).

nisms involved in various pathologies, such as Alzheimer’s (Rogers et al., 1992; Tacnet-Delorme et al., 2001) and prion diseases (Klein et al., 2001; Mabbott et al., 2001). The multiple facets of the physiological and pathological implications of the classical complement pathway have led to a renewed interest in the C1 complex, the multimolecular protease that triggers the pathway. C1 is a 790 000 Da complex formed from the association of a recognition protein, C1q, and a catalytic subunit, the Ca2+ -dependent tetramer C1s–C1r–C1r–C1s, comprising two copies of each of the C1r and C1s proteases (Cooper, 1985; Schumaker et al., 1987; Arlaud et al., 1987a; Arlaud et al., 2001). All activators of the classical pathway are recognized by the C1q moiety of C1, a process that is thought to generate a conformational signal that triggers self-activation of C1r, which in turn will convert proenzyme C1s into the active and highly specific protease that mediates cleavage of C4 and C2, the protein substrates of the C1 complex. Human C1q (Fig. 1) is assembled from 18 polypeptide chains of three different types (A, B, C) that are similar in length and exhibit homologous amino acid sequences (Reid, 1983; Kishore and Reid, 2000). Each chain comprises a short N-terminal region involved in the formation of A–B

0161-5890/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 6 1 - 5 8 9 0 ( 0 2 ) 0 0 1 4 3 - 8

384

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

Fig. 1. Modular structure of the C1 proteins and macroscopic model of the C1 complex. The nomenclature and symbols used for protein modules are those defined by Bork and Bairoch (1995). SP, serine protease domain (formerly designed B). The ␥ fragment corresponds to the CCP1 + CCP2 modules. The unlabeled segment in C1r and C1s corresponds to the activation peptide, and the arrow indicates the Arg–Ile bond cleaved upon activation. The only disulfide bond represented is that connecting the activation peptide to the serine protease domain. Close diamonds represent N-linked oligosaccharides. The C1 model shown is adapted from Arlaud et al. (1987a). C1q, C1r and C1s are shown in white, black and grey, respectively.

and C–C interchain disulfide bonds, that is followed by a repeating collagen-like sequence giving rise to the formation of six ABC heterotrimeric triple helices. These first associate to form a “stalk” and then, due to interruptions in the repeating Gly-Xaa-Yaa collagen-like sequence, diverge to form six “arms”. Each arm merges at its C-terminal end into a globular “head” region consisting of heterotrimers of protein domains known as “C1q” modules. A structural model of C1q was proposed in 1976 on the basis of amino acid sequence, limited proteolysis, and electron microscopy studies (Reid and Porter, 1976). Human C1r and C1s are glycoproteins comprising 688 and 673 amino acids, respectively (Fig. 1). They both exhibit a single-chain structure in the proenzyme form and are activated through cleavage of a single Arg–Ile bond, yielding two-chain active proteases. They are modular serine

proteases (SPs) exhibiting homologous structural organizations comprising, starting from the N-terminal end, a CUB module (Bork and Beckmann, 1993), an epidermal growth factor (EGF)-like module (Campbell and Bork, 1993), a second CUB module, two contiguous complement control protein (CCP) modules (Reid et al., 1986), and a C-terminal chymotrypsin-like serine protease domain. Early studies based on electron microscopy and neutron scattering revealed that the isolated C1s–C1r–C1r–C1s tetramer has an elongated structure (Tschopp et al., 1980; Boyd et al., 1983), whereas the same techniques provided evidence that the C1s–C1r–C1r–C1s subunit folds into a rather compact conformation upon interaction with C1q (Strang et al., 1982; Perkins et al., 1984). Then the key question was: how is C1s–C1r–C1r–C1s folded within the C1 complex? Answers to this question came initially from

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

the identification of specialized functional regions in C1r and C1s, indicating that the N-terminal part of each protease mediates the Ca2+ -dependent interactions involved in the assembly of C1s–C1r–C1r–C1s, whereas the regions responsible for catalytic activity are derived from their C-terminal parts (Villiers et al., 1985). These and other studies (Tschopp et al., 1980; Weiss et al., 1986) led to a macroscopic model of the C1s–C1r–C1r–C1s assembly where the catalytic domains of C1r are located at the center and those of C1s at both ends of the complex. From this location, and given that each of these domains contains both the serine protease site and the Arg–Ile bond cleaved upon activation, followed the concept that, within the C1 complex, C1s–C1r–C1r–C1s folds into a compact, “8-shaped” conformation allowing for contact between the catalytic domains of C1r and C1s, a prerequisite for C1s activation by C1r (Arlaud et al., 1987a; Fig. 1). This concept represents the basis of most of the low-resolution C1 models proposed so far (Colomb et al., 1984; Weiss et al., 1986; Arlaud et al., 1987a; Schumaker et al., 1986). Our major objective over the past years has been to generate more detailed information about the structure of C1 at the atomic level. This task is rendered difficult by the fact that C1 is a non-covalent, multimolecular complex of three modular proteins, each exhibiting areas of flexibility. These features preclude a high-resolution study of the C1 complex as a whole by standard structural biology techniques. As a consequence, the strategy used over the past years has been based on a molecular dissection of the individual C1 proteins into modular segments, with a view to precisely characterize their function and solve their three-dimensional structure by nuclear magnetic resonance spectroscopy (NMR) or X-ray crystallography. This approach has proven quite successful, and now allows deep insights into the structure–function relationships of C1. The purpose of this article is to review this information, with particular emphasis on the architecture of the complex and the mechanisms underlying its activation and proteolytic activity. 2. The protein modules of C1 2.1. The gC1q modules Although there are several reports that certain non-immune activators bind C1 through the collagen-like moiety of C1q (Gewurz et al., 1993), it seems now well established that most of the activating ligands of C1 are recognized by the C1q globular “heads” at the C-terminal end of the C1q arms. This is well documented for immune complexes (Cooper, 1985), as well as for several non-immune activators, including HIV-1 and ␤-amyloid fibrils (Tacnet-Delorme et al., 1999, 2001). Each globular head is thought to be a non-covalent, heterotrimeric association of protein modules known as gC1q domains (Kishore and Reid, 2000). Modules of the gC1q family are about 140 amino acid residues

385

long and occur at the C-terminal end of various proteins, including types VIII and X collagens, the adipocyte complement-related protein (ACRP)-30, precerebellin, the brain C1q-related factor (CRF), and multimerin (Bérubé et al., 1999; Kishore and Reid, 2000). With the exception of precerebellin and multimerin, the gC1q modules are always located at the carboxyl terminus of a collagen-like sequence. Whereas the gC1q module self-associates into heterotrimeric structures in C1q and possibly in type VIII collagen, homotrimeric structures are likely formed in most other known related proteins. Comparative sequence analysis of known gC1q modules shows conservation of a large number of aromatic, hydrophobic, and neutral residues, most of which are very likely involved in the conserved framework of this type of module (Smith et al., 1994). The X-ray structure of the gC1q module has been solved for ACRP-30 (Shapiro and Scherer, 1998) and type X collagen (Bogin et al., 2002), both of which form homotrimeric structures. In both these proteins, the individual gC1q module shows a 10-stranded ␤-sandwich structure with a jelly-roll folding topology. Both trimers are bell-shaped, with a broad base, and are stabilized by a hydrophobic interface near the base, where the polypeptide chain N- and C-termini emerge, suggesting a role in the assembly of the adjacent collagen-like triple-helical structure. On the opposite side of the trimer, the collagen X structure also shows a buried cluster of Ca2+ ions, likely to contribute to the remarkable stability of the trimer. Most of the structural features seen in the C1q modules are observed in proteins of the tumor necrosis factor (TNF) family (Jones et al., 1989), suggesting that these form a superfamily of protein modules. As judged from Fourier-transform infrared (FTIR) spectroscopy and secondary structure predictions, it is likely that the globular heads of C1q also exhibit a ␤-sheet structure (Smith et al., 1994). A fragment corresponding to this domain was generated by collagenase digestion of C1q and crystallized, allowing collection of high-resolution X-ray diffraction data (Gruez et al., 1998). Although the ACRP-30 structure was a suitable model for molecular replacement, it did not allow for resolution of the structure. Nevertheless, it provided strong indication that C1q and ACRP-30 exhibit similar ␤-strand topologies and trimer assemblies, but display significant differences in their loop structures. 2.2. The CUB modules The N-terminal, non-catalytic A chains of C1r and C1s have homologous modular structures each comprising two CUB modules surrounding a single EGF module and a pair of contiguous CCP modules (Fig. 1). CUB modules were first recognized in C1r/C1s, the sea urchin protein Uegf, and the human bone morphogenetic protein-1, hence their name (Bork, 1991; Bork and Beckmann, 1993), and have since been identified in a number of extracellular proteins, many of which are involved in developmental processes. Most of the CUB modules contain four cysteine residues

386

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

that form two disulfide bridges in a Cys1–Cys2, Cys3–Cys4 pattern (Romero et al., 1997), but the N-terminal CUB1 modules in C1r and C1s feature only the Cys3–Cys4 pair. The CUB modules of C1r and C1s contain 110–120 amino acids, and show the characteristic CUB sequence pattern with conserved hydrophobic and aromatic residues. Whereas the C1s CUB1 module is not glycosylated, one N-linked oligosaccharide is present in both C1r CUB modules (at Asn108 and Asn204) and in the second C1s CUB module (at Asn159). As shown by mass spectrometry, Asn159 of C1s bears a complex-type biantennary bisialylated oligosaccharide NeuAc2 Gal2 GlcNAc4 Man3 (Pétillot et al., 1995). Although no structure is available yet for the C1r and C1s CUB modules, the CUB domain fold has been revealed by the crystal structure of two spermadhesins, and shown to consist of a compact ellipsoidal ␤-sandwich structure organized in two five-stranded ␤-sheets (Romero et al., 1997; Varela et al., 1997; Romao et al., 1997). 2.3. The EGF modules EGF-like modules have been identified in a number of soluble and membrane-bound proteins involved in diverse biological functions, such as blood coagulation, cell adhesion, or neural development (Campbell and Bork, 1993). They contain six conserved cysteine residues separated by segments of varying lengths, that form three disulfide bonds (Cys1–Cys3, Cys2–Cys4, Cys5–Cys6). The EGF modules of C1r and C1s contain 53 and 44 amino acids, respectively, and belong to a particular subset involved in Ca2+ binding, that features a characteristic consensus sequence pattern Asp/Asn, Gln/Glu, Asp∗ /Asn∗ , Tyr/Phe, where asterisk (∗) indicates a ␤-hydroxylated residue. Although in C1r, the corresponding asparagine at position 150 was shown to be fully hydroxylated, only 50% of the equivalent Asn134 in C1s was found to be converted to erythro-␤-hydroxyasparagine (Arlaud et al., 1987b; Thielens et al., 1990a). This modification does not appear to be required for Ca2+ -binding, since recombinant C1s expressed in insect cells lacks ␤-hydroxylation but nevertheless retains its Ca2+ -dependent interaction properties (Luo et al., 1992). A quite unusual feature of the C1r EGF module is the large size (14 residues) of the loop between Cys1 and Cys3 (Cys129 and Cys144), which contains the single polymorphic site (Ser135/Leu) identified in human C1r (Journet and Tosi, 1986; Leytus et al., 1986; Arlaud et al., 1987c). The human C1r EGF module was synthesized chemically (Hernandez et al., 1997) and its solution structure was determined by 1 H NMR spectroscopy (Bersch et al., 1998). The C-terminal part exhibits the typical EGF fold, with major and minor anti-parallel double-stranded ␤-sheets, whereas the N-terminal end of the module, and the unusually large loop between Cys129 and Cys144, are disordered. NMR spectroscopy also provided evidence of the ability of the C1r EGF module to bind Ca2+ , although with a KD of about 10 mM, i.e. with a very low affinity compared to the larger

N-terminal C1r␣ fragment (Hernandez et al., 1997). Analysis of the chemical shift variations induced by Ca2+ and modeling studies (Bersch et al., 1998) were consistent with the hypothesis that Ca2+ binding by the C1r EGF module involves residues Asp125, Leu126, Glu128, Asn150, and Tyr151, i.e. ligands homologous to those identified in the EGF modules of blood coagulation factors IX and X and of fibrillin (Rao et al., 1995; Sunnerhagen et al., 1996; Downing et al., 1996). 2.4. The CCP modules The CCP (complement control protein) modules of C1r and C1s vary in length from 60 to 70 amino acids, and belong to a family of protein motifs occurring, often as contiguous arrays, in various complement receptors and regulatory proteins (Reid et al., 1986). They exhibit a characteristic consensus sequence comprising a few conserved aromatic and hydrophobic residues and four cysteines that form two disulfide bonds (Cys1–Cys3, Cys2–Cys4). Whereas the C1r CCP modules are not glycosylated, the CCP2 module of C1s bears a complex-type N-linked oligosaccharide at Asn391. As shown by mass spectrometry, this oligosaccharide is heterogeneous, with the occurrence of a biantennary form (NeuAc3 Gal3 GlcNAc4 Man3), a triantennary form (NeuAc3 Gal3 GlcNAc5 Man3), and a fucosylated triantennary form (NeuAc3 Gal3 GlcNAc5 Man3 Fuc1), in approximately 1/1/1 relative proportions. This heterogeneity yields three major C1s species in serum, with molecular masses of 79 318, 79 971, and 80 131 (Pétillot et al., 1995). The structures of both C1r CCP modules, and of the second C1s CCP module are now available from the X-ray structure analysis of the larger fragments CCP1–CCP2–SP (C1r) and CCP2–SP (C1s) (Gaboriaud et al., 2000; Budayova-Spano et al., 2002a,b). The fold of these modules is similar to that described for other members of the CCP family (Bork et al., 1996), with six ␤-strands enveloping a compact hydrophobic core. The N- and C-termini lie at opposite ends of the long axis of the ellipsoidal modules, and the ␤-strands are approximately aligned with this axis. Consistent with the observed amino acid sequence homologies, the C1r CCP2 module structure is closer to that of its counterpart in C1s than to that of its contiguous CCP1 module. However, the C1r CCP2 module features two insertions not present in the homologous C1s module: one in the so-called “hypervariable” loop (Wiles et al., 1997) between strands B1 and B2, the other in the loop connecting strands B3 and B4. 2.5. The serine protease domains The serine protease domains of C1r and C1s contain 242 and 251 amino acids, respectively, and belong to the chymotrypsin-like family. Each exhibits at the S1 substratebinding sub-site an Asp residue indicative of trypsin-like specificity, consistent with the known ability of C1r and C1s to cleave arginyl bonds in their natural protein substrates

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

387

Fig. 2. Three-dimensional structures of the active SP domains of C1r (A) and C1s (B). Loops are labeled according to Perona and Craik (1997). The ␤-strand are represented as arrows, and ␣-helices are shown as spirals. Residues of the active sites (a.s.) are shown as ball and stick. Dots represent residues not defined in the structures. (*): N-terminal Ile residue resulting from cleavage of the Arg–Ile bond upon activation. (**): C-terminal residue of the activation peptide. C-ter: C-terminal residue of the SP domain.

(Arlaud et al., 1998). Thus, both C1r autoactivation and C1r-mediated cleavage of C1s involve cleavage of Arg–Ile bonds, and active C1s cleaves C4 and C2 at Arg–Ala and Arg–Lys bonds, respectively. Two of the disulfide bridges conserved in other chymotrypsin-like SPs, namely the one contained in the “methionine loop” and the disulfide bridge connecting the primary and secondary substrate-binding sites, are conserved in C1r and C1s. However, unlike MASP-1 (mannan-binding lectin-associated protease-1) but like MASP-2 and MASP-3, C1r and C1s lack the “histidine loop”, a disulfide loop present in most other known mammalian SPs (Arlaud and Gagnon, 1981). Although in human, the SP domain of C1s is not glycosylated, the C1r SP domain bears two N-linked oligosaccharides at Asn497 and Asn564, and each position is occupied by heterogeneous complex-type biantennary species containing either one or two sialic acids and one or no fucose, with a major species NeuAc2 Gal2 GlcNAc4 Man3 being present in both cases (Lacroix et al., 1997). The three-dimensional structures of the active C1s SP domain and of the zymogen and active forms of the C1r SP domain have been solved by X-ray crystallography analysis of fragments also containing one or two of the preceding CCP modules (Gaboriaud et al., 2000; Budayova-Spano et al., 2002a,b). In both proteases, the core of the SP domain has the typical fold of chymotrypsin-like enzymes, with two six-stranded ␤-barrels connected by three trans-segments and a C-terminal ␣-helix (Fig. 2). The catalytic triad residues (Ser637, His485, Asp540 in C1r; Ser617, His460, Asp514 in C1s) are located at the junction between the two ␤-barrels in a conformation virtually identical to that observed in chymotrypsin. In addition, the active sites exhibit other structural features characteristic of the active conformation of chymotrypsin-like SPs (Perona and Craik, 1997).

On the other hand, C1r and C1s display unique structural features in most of the surface loops (1–3 and A–E, as described by Perona and Craik, 1997) that together define fine substrate specificity in mammalian serine proteases. Thus, loop E (the Ca2+ -binding loop in other SPs such as trypsin) exhibits in C1r an ␣-helical fold that has not been seen in any of the chymotrypsin-like SPs of known structure (Budayova-Spano et al., 2002a,b). Another particular feature of C1r is loop B, that represents a major insertion when compared to other SPs, and is partly disordered, suggesting that it may undergo structural changes upon substrate and/or inhibitor binding, as observed in thrombin (Malkowski et al., 1997). In the same way, C1s exhibits major insertions in loops 3 and C on the same side of the active site cleft, and deletions in loops 1, 2 and A on the opposite side. Disordered conformations are also observed in loops 3 and E at both ends of the substrate-binding region, and in the five-residue long segment that prolongs the canonical C-terminal ␣-helix (Gaboriaud et al., 2000). 3. Assembly of the C1 complex 3.1. Assembly of the C1s–C1r–C1r–C1s tetramer A fundamental characteristic of C1r and C1s is that they exert their catalytic activities in the context of the multimolecular C1 complex, a property that results from the ability of each protease to mediate various protein–protein interactions, both within the C1s–C1r–C1r–C1s tetramer and between the tetramer and C1q. The production and characterization of a series of modular fragments from the N-terminal regions of C1r and C1s has allowed for identification of the domains responsible for these interactions, and

388

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

has provided precise insights into their structure–function relationships. Limited proteolysis with trypsin or plasmin was initially used to generate fragments C1r␣ and C1s␣ encompassing the N-terminal CUB1 and EGF modules of each protein, plus a short segment from the following CUB2 module (Fig. 1) (Busby and Ingham, 1987; Thielens et al., 1990b; Busby and Ingham, 1990). Differential scanning calorimetry revealed that in the absence of Ca2+ ions, fragment C1r␣ displays a low-temperature transition, with a mid-point of 26–40 ◦ C, that is shifted upwards by more than 20 ◦ C upon addition of Ca2+ ions (Busby and Ingham, 1987). Seemingly, the recombinant CUB1–EGF segment of C1s was shown to exhibit a temperature transition at 44–52 ◦ C, that is shifted to 61 ◦ C in the presence of Ca2+ ions (Thielens and Kardos, unpublished data). Further characterization of the C1r␣ fragment indicated that it retains the ability of intact C1r to associate to C1s in the presence of Ca2+ ions, and contains one high-affinity Ca2+ binding site (KD = 32 ␮M) (Thielens et al., 1990b). Conversely, the homologous C1s␣ fragment has the ability to form Ca2+ -dependent C1s␣–C1r␣ heterodimers, with the concomitant binding of two Ca2+ atoms per dimer, i.e. one atom per ␣ fragment. Further fragmentation of the C1r␣ region provided evidence that its ability to bind Ca2+ and to associate to C1s in the presence of Ca2+ ions involves residues contributed by both its CUB1 and EGF modules. First, as mentioned earlier, the synthetic C1r EGF module did bind Ca2+ , but with an affinity about 300-fold lower than that of the larger C1r␣ fragment (Hernandez et al., 1997), suggesting that, as observed for other Ca2+ -binding EGF domains (Downing et al., 1996; Sunnerhagen et al., 1996), residues located outside this module either provide additional Ca2+ ligands or stabilize the Ca2+ binding site conformation. In support of this hypothesis, a mutant C1r molecule lacking most of the N-terminal CUB1 module was found to lose the ability to bind C1s in the presence of Ca2+ ions (Zavodszky et al., 1993; Cseh et al., 1996). Surface plasmon resonance spectroscopy allowed for detailed analysis of the interaction properties of the CUB1, EGF, and CUB1–EGF fragments of C1r (Thielens et al., 1999). Neither the isolated CUB1 and EGF modules, nor a CUB1 + EGF mixture had the ability to bind to immobilized C1s in the presence of Ca2+ . In contrast, the CUB1–EGF pair bound C1s under these conditions, with a KD of 1.5–1.8 ␮M, that decreased to 15–20 nM when CUB1–EGF was used as the immobilized ligand and C1s was free, likely due to an increased stability of the CUB1–EGF fragment upon attachment to the surface of the sensor chip. No protein–protein interaction was observed in the absence of Ca2+ ions, and half-maximal binding was achieved at comparable Ca2+ concentrations (5–16 ␮M) for intact C1r and its CUB1–EGF and C1r␣ fragments. Gel filtration analysis and measurement of intrinsic fluorescence of tyrosine residues provided evidence that Ca2+ binding induces a more compact conformation of the CUB1–EGF module pair (Thielens et al., 1999). Taken together, the data currently available are consistent with the model depicted

Fig. 3. Model of the Ca2+ -induced conformational changes in the N-terminal CUB1–EGF moiety of C1r and its interaction with the homologous CUB1–EGF moiety of C1s. Adapted from Thielens et al. (1999).

in Fig. 3, in which Ca2+ binds primarily to ligands in the EGF module, thereby allowing the CUB1 and EGF module to move closer to each other, and resulting in the formation of a compact CUB1–EGF assembly. This conformation is expected to stabilize the Ca2+ binding site and to provide the appropriate ligands for interaction with C1s within the C1s–C1r–C1r–C1s tetramer. The homologous CUB1–EGF fragment of C1s was produced in insect cells, and was shown to retain the ability to bind C1r in the presence of Ca2+ (Tsai et al., 1997). Although it has not been determined whether this fragment binds Ca2+ with an affinity comparable to that of intact C1s, the model proposed for C1r very likely also applies in the case of C1s. Thus, C1r–C1s interactions within the C1s–C1r–C1r–C1s tetramer are likely to be mediated by the CUB1–EGF moiety of each protease (Fig. 3). 3.2. Assembly of the C1 complex Information currently available indicates that the N-terminal regions of C1r and C1s also play a major role in the interaction between the C1s–C1r–C1r–C1s tetramer and C1q. The hypothesis that the C1r moiety of the tetramer contributes to that interaction is supported by several experiments. Thus, evidence that C1r alone binds with high affinity to C1q was obtained by ultracentrifugation analysis (Lakatos, 1987), an observation that is consistent with the finding that the activation rate of C1r is significantly

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

increased by C1q in the presence of Ca2+ ions (Thielens et al., 1994). Further experiments have shown that treatment of C1s–C1r–C1r–C1s by a water-soluble carbodiimide modifies acidic amino acids of C1r, thereby preventing interaction of the tetramer with C1q (Illy et al., 1993). On the other hand, as shown initially by Busby and Ingham (1990), the truncated tetramer C1s␣–C1r–C1r–C1s␣ associates to C1q to form a pseudo-C1 complex with a stability comparable to that of intact C1, demonstrating the ability of C1s␣ to promote the interaction of C1r with C1q. This pseudo-C1 complex also retains its C1r activation properties (Thielens et al., 1994), demonstrating that the missing part of the C1s molecule, i.e. its C-terminal catalytic region, is not required to achieve C1r activation in the C1 complex. Further studies by Tsai et al. (1997) have shown that the N-terminal CUB1–EGF fragment of C1s is sufficient to promote interaction of C1r with C1q, yielding a complex in which C1r, again, retains its activation properties. A plausible hypothesis is that the CUB1–EGF moiety of C1s and the neighboring, homologous domain of C1r, each provide ligands for the interaction with C1q and act in synergy to achieve formation of a stable C1 complex. Thus, the CUB1–EGF pairs of C1r and C1s would represent key elements of the C1 architecture, at both the interfaces between the proteases, and the proteases and C1q. Less information is available about the sites of C1q responsible for the interaction with C1s–C1r–C1r–C1s. Ultracentrifugation analysis has provided strong evidence that these sites are located in the collagenous moiety of the molecule (Siegel and Schumaker, 1983). Furthermore, electron microscopy studies on the chemically cross-linked C1 complex (Strang et al., 1982) suggest that the collagen-like “arms” of C1q, and not the central bundle, are involved in the interaction, as featured in most of the low-resolution C1 models proposed thus far (Weiss et al., 1986; Schumaker et al., 1987; Arlaud et al., 1987a; see Fig. 1).

4. Activation of C1: structure and function of the C1r catalytic domain It is well established that the catalytic activities of C1r (i.e. autolytic activation and subsequent cleavage of C1s) are mediated by its C-terminal region, comprising the SP domain (also known as the B chain) and the preceding CCP1 and CCP2 modules, which altogether form the ␥ segment (Villiers et al., 1985; Weiss et al., 1986). This region associates as a non-covalent homodimer that forms the core of the C1s–C1r–C1r–C1s tetramer and the corresponding fragment (␥-B)2 has been generated by autolytic cleavage of active C1r and by limited proteolysis (Arlaud et al., 1986) and shown to retain the ability to activate proenzyme C1s. The proenzyme form of C1r (␥-B)2 was also obtained by limited proteolysis with thermolysin, and again was found to retain the autoactivation properties of native zymogen C1r (Lacroix et al., 1989).

389

Information on the structure of the catalytic domain of C1r was provided initially by chemical cross-linking and homology modeling studies performed on the active (␥-B)2 fragment, which indicated a strong interaction between the CCP2 module and the SP domain of each monomer, and provided evidence for an inter-monomer cross-link between the N-terminal end of the ␥ segment and the SP domain (Lacroix et al., 1997). These data led to a three-dimensional model of the active (␥-B)2 homodimer, featuring a “head-to-tail” interaction of the monomers, with their active sites facing opposite directions towards the outside of the dimer. The production of recombinant fragments of varying sizes, using both bacterial and baculovirus-mediated expression systems, allowed for further insights into the role of individual modules in the assembly and catalytic properties of the C1r catalytic domain (Lacroix et al., 2001; Kardos et al., 2001). First, the CCP1–CCP2–SP fragments associated as homodimers, whereas the shorter CCP2–SP and SP fragments were monomeric, underlying the critical role of the CCP1 module in the assembly of the dimer. In contrast, the CCP2–SP fragment cleaved proenzyme C1s more efficiently than did CCP1–CCP2–SP, indicating that CCP1 is not involved in C1s recognition. Each C1r construct was expressed with the wild-type sequence and with a point mutation, either at the Arg–Ile activation site (Arg446Gln) or at the active serine residue (Ser637Ala). The Arg446Gln and Ser637Ala mutants were all recovered in a totally uncleaved, zymogen form, and did not undergo activation either during the purification process or upon subsequent incubation. In contrast, the wild-type species, whatever their size, were all recovered as two-chain, active proteases, indicating that activation had occurred in each case either during biosynthesis or after secretion. These data provide the first experimental evidence that C1r activation, a subject that for several decades has been a controversial issue (Cooper, 1985), is indeed a self-activation process. The observed ability of the monomeric CCP2–SP and SP molecules to undergo activation in their wild-type form is more unexpected, as this implies that formation of a stable homodimer is not a prerequisite for self-activation. The above observations are consistent with the threedimensional structure of the human C1r catalytic domain, recently solved by X-ray crystallographic analysis of a CCP1–CCP2–SP fragment stabilized in the zymogen form by means of the Arg446Gln mutation (Budayova-Spano et al., 2002a). The protein exhibits a homodimeric structure, featuring a head-to-tail interaction of the molecules through contacts between the CCP1 module of one monomer and the SP domain of its counterpart (Fig. 4). The structure is rather elongated and, unexpectedly, shows a large opening in the center of the dimer. The CCP1–SP contacts at the monomer–monomer interfaces reveal extensive shape complementarity, and feature a major interaction framework consisting of hydrogen bonds and hydrophobic interactions. The most striking characteristic of this head-to-tail assembly lies in the fact that the catalytic site of one monomer

390

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

Fig. 4. Homodimeric structure of the zymogen catalytic domain of C1r. (A) Overall view of the structure. Molecules A and B of the homodimer are shown in red and green, respectively. Residues at the catalytic sites (a.s.) and at the cleavage sites are shown. NA , NB and CA , CB indicate the N- and C-terminal ends of molecules A and B. (B) Surface view of the structure. Modified from Budayova-Spano et al. (2002a).

and the cleavage site of its counterpart are located at opposite ends of the dimer, some 92 Å away from each other. This configuration has functional implications, since it cannot account for C1r self-activation, which requires cleavage of the susceptible Arg–Ile bond in each monomer by the catalytic serine residue of the other (Lacroix et al., 2001). Thus, C1r activation within the C1 complex is expected to involve transient conformational states allowing cleavage of the SP domain of molecule B by that of molecule A, and conversely, a process that requires dissociation of the homodimeric structure through disruption of the CCP1–SP interfaces. We have proposed that this is achieved by means of a mechanical stress that is transmitted from C1q to C1r when C1 binds to an activator (Budayova-Spano et al., 2002a). This hypothesis is consistent with the occurrence of a semi-flexible hinge in C1q (Schumaker et al., 1981; Poon et al., 1983), and with the fact that each C1r catalytic domain is connected to the CUB1–EGF interaction domain, which itself is expected to be bound to the C1q collagen arms (see above). Thus, multivalent binding of the C1q globular “heads” to a pattern of sites at the surface of a target may be expected to increase the angle between the C1q arms and the central part of the protein, thereby generating a stress transmitted to the C1r catalytic domains and

leading to the disruption of the CCP1–SP interfaces. Other specific characteristics, such as the large central opening of the dimeric structure, and the expected flexibility at the interface between the CCP1 and CCP2 modules (Wiles et al., 1997) are probably key features of the activation mechanism, since they are expected to facilitate the approaching of the SP domains. Further information relevant to C1r activation was derived from the X-ray structure of shorter CCP2–SP fragments from the catalytic domain of C1r (Budayova-Spano et al., 2002b). Thus, the structure of the zymogen Ser637Ala mutant confirms that, compared to other known zymogens, the SP domain of C1r exhibits a high degree of flexibility, with several loops displaying weak or no electron density. Comparison of the zymogen structure with that of the wild-type, active CCP2–SP fragment allows precise analysis of the multiple conformational changes that take place in the SP domain upon activation, and reveals that these changes are often important, with amplitudes as high as 8 Å, similar to those observed in other SPs (Budayova-Spano et al., 2002b). Analysis of the substrate-binding sub-sites reveals restrained access to most of them, indicating that, although C1r catalytic activity is confined to the interior of the C1 complex, its restricted specificity likely arises, at least partially, from

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

structural constraints that limit access to its active site. Comparison of the CCP2–SP assembly in the three C1r structures now available (Budayova-Spano et al., 2002a,b) reveals that the interface between the CCP2 module and the SP domain is not completely rigid, but exhibits some flexibility, due to the occurrence of a restricted “hinge” at the level of the intermediate residue Val433. This flexibility may be relevant to the mechanism of C1r activation in C1, since it may help achieve appropriate positioning of one C1r molecule with respect to its partner in the restrained space of the C1 macromolecule.

5. C1 activity: structure and function of the C1s catalytic domain The enzymatic activity of C1s is mediated by its C-terminal catalytic domain, homologous to that of C1r (see above). The corresponding segment ␥-B, which forms the outer portion of the C1s–C1r–C1r–C1s tetramer, can be obtained by limited proteolysis with plasmin and retains the ability to mediate specific cleavage of C4 and C2, the protein substrates of C1 (Villiers et al., 1985; Weiss et al., 1986). Insights into the role of individual domains of this region were made possible by expression in a baculovirus-insect cells system of fragments lacking either the first CCP module (CCP2–SP) or both CCP modules (SP) (Rossi et al., 1998). Both could be activated by C1r and were in turn able to mediate specific cleavage of C2 with similar efficiencies.

391

In contrast, compared to intact C1s and the whole ␥-B fragment, the C4-cleaving activity of CCP2–SP was dramatically reduced, and that of SP was abolished. Thus, whereas proteolytic activity on C2 only requires structural determinants located in the SP domain, efficient C4 cleavage clearly involves accessory substrate-binding sites contributed by both CCP modules, in keeping with previous studies based on the use of monoclonal antibodies (Matsumoto et al., 1989). Structural characterization of the C1s ␥-B region was initially achieved by differential scanning calorimetry, providing evidence for three independently folded domains: the CCP1 and CCP2 modules, which each unfold reversibly at about 60 ◦ C, and the less stable SP domain, which melts at 49 ◦ C (Medved et al., 1989). Further insights came from combined chemical cross-linking and homology modeling studies, which indicated that the CCP2 module closely interacts with the SP domain on a side opposite to both the active site and the susceptible Arg–Ile bond cleaved upon activation (Rossi et al., 1995). Based on multiple sequence alignments, it was proposed that all proteins of the “CCP–SP” family, including C1r, the mannan-binding lectin-associated serine proteases (MASPs), and Limulus factor C, exhibit a similar CCP module/SP domain assembly at their C-terminal end (Gaboriaud et al., 1998). The three-dimensional structure of the CCP2–SP fragment of human C1s was solved by X-ray crystallography and refined to 1.7 Å resolution (Gaboriaud et al., 2000). The overall shape of the structure is that of a bludgeon, with the ellipsoidal CCP2 module tightly anchored on the

Fig. 5. The putative role of the CCP modules in the displacement of the C1s SP domain upon C1 activation. (A) Inside conformation of the proenzyme C1s catalytic domain required for activation. The thin line indicates the C1r catalytic domain in the appropriate position for C1s activation. (B) Outside conformation of the active C1s catalytic domain allowing its access to C4 and C2.

392

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

more globular SP domain, on the side opposite to the active site (Fig. 5). The CCP2 module is held perpendicularly to the SP domain through a rigid interface arising from multiple non-covalent interactions involving residues from three proline- and tyrosine-rich sequence stretches contributed by the CCP2 module, the SP domain, and the short (15-residue) segment that connects them. Whether this interface is fully rigid in C1s, or slightly flexible as observed in C1r (Budayova-Spano et al., 2002b) will await resolution of further structures of the CCP2–SP assembly in other C1s fragments. However, compared to the other members of the “CCP–SP” family, the CCP2–SP interaction appears significantly stronger in C1s, and the interface is therefore possibly fully rigid. Analysis of the substrate-binding sub-sites of the C1s SP domain has revealed that the access to most of them (particularly S4, S2, S1, and S2 ) is severely restricted, a feature also observed, although to a lesser extent, in C1r (see above) and some of the blood coagulation proteases (Bode et al., 1997). These steric constraints are likely to be a major determinant of the highly restricted specificity of C1s, which in plasma cleaves only two protein substrates and is sensitive to a single protease inhibitor, C1-inhibitor (Arlaud and Thielens, 1993). Thus, the occurrence of a cluster of bulky aromatic residues (Phe511, Tyr595, Trp640) in the vicinity of the S2–S4 sub-sites may explain the specificity of C1s for C2, C4, and C1-inhibitor, since these contain hydrophobic P3 residues (Leu, Val) and small P4 residues (Gly, Ser). On the other hand, the CCP2 module probably provides an extension of the substrate-binding region and thereby contributes additional recognition sites, as those needed for efficient cleavage of C4 (Rossi et al., 1998). Thus, the finely tuned proteolytic specificity likely results from both steric constraints in the SP domain and accessory recognition sites in the CCP modules. The C1s CCP modules are likely to have further implications relevant to the function of C1s in the context of the C1 complex (Fig. 5). Indeed, based on our current view of the C1 complex, the SP domain of C1s is thought to be initially positioned inside the C1 complex (a location that allows its activation by C1r), and then is expected to gain access to its C4 and C2 protein substrates, which implies a movement towards the outside of the complex. We have proposed that this shift is achieved by the conjunction of two structural features: (i) the occurrence of a flexible hinge at the interface between modules CCP1 and CCP2, as observed in other pairs of CCP modules (Barlow et al., 1993; Wiles et al., 1997); (ii) the fact that the CCP2 module is oriented perpendicularly to the SP domain, which makes it both a spacer and a handle, and allows it to amplify the shift of the SP domain (Fig. 5). Although this mechanism remains to be demonstrated experimentally, it appears likely that, due to their particular mechanic and interaction properties, the CCP modules of C1s, like those of C1r, play essential roles in the activation and activity of the C1 complex.

References Arlaud, G.J., Gagnon, J., 1981. C1r and C1s subcomponents of human complement: two serine proteases lacking the “histidine loop” disulfide bridge. Biosci. Rep. 1, 779–784. Arlaud, G.J., Thielens, N.M., 1993. Human complement serine proteases C1r and C1s and their proenzymes. Methods Enzymol. 223, 61–82. Arlaud, G.J., Gagnon, J., Villiers, C.L., Colomb, M.G., 1986. Molecular characterization of the catalytic domains of human complement serine protease C1r. Biochemistry 25, 5177–5182. Arlaud, G.J., Colomb, M.G., Gagnon, J., 1987a. A functional model of the human C1 complex. Immunol. Today 8, 106–111. Arlaud, G.J., van Dorsselaer, A., Bell, M., Mancini, A., Aude, C., Gagnon, J., 1987b. Identification of erythro-␤-hydroxyasparagine in the EGF-like domain of human C1r. FEBS Lett. 222, 129–134. Arlaud, G.J., Willis, A.C., Gagnon, J., 1987c. Complete amino acid sequence of the A chain of human complement-classical-pathway enzyme C1r. Biochem. J. 241, 711–720. Arlaud, G.J., Volanakis, J.E., Thielens, N.M., Narayana, S.V.L., Rossi, V., Xu, Y., 1998. The atypical serine proteases of the complement system. Adv. Immunol. 69, 249–307. Arlaud, G.J., Gaboriaud, C., Thielens, N.M., Rossi, V., Bersch, B., Hernandez, J.-F., Fontecilla-Camps, J.C., 2001. Structural biology of C1: dissection of a complex molecular machinery. Immunol. Rev. 180, 136–145. Barlow, P.N., Steinkasserer, A., Norman, D.G., Kieffer, B., Wiles, A.P., Sim, R.B., Campbell, I.D., 1993. Solution structure of a pair of complement modules by nuclear magnetic resonance. J. Mol. Biol. 232, 268–284. Bersch, B., Hernandez, J.-F., Marion, D., Arlaud, G.J., 1998. Solution structure of the epidermal growth factor-like module of human complement protease C1r, an atypical member of the EGF family. Biochemistry 37, 1204–1214. Bérubé, N.G., Swanson, X.H., Bertram, M.J., Kittle, J.D., Didenko, V., Baskin, D.S., Pereira-Smith, O.M., 1999. Cloning and expression of CRF, a novel C1q-related factor expressed in areas of the brain involved in motor functions. Mol. Brain Res. 63, 233–240. Bode, W., Brandstetter, H., Mather, T., Stubbs, M.T., 1997. Comparative analysis of haemostatic proteinases: structural aspects of thrombin, factor Xa, factor IXa, and protein C. Thromb. Haemost. 78, 501–511. Bogin, O., Kvansakul, M., Rom, E., Singer, J., Yayon, A., Hohenester, E., 2002. Insights into Schmid metaphyseal chondrodysplasia from the crystal structure of the collagen X NC1 domain trimer. Structure 10, 165–173. Bork, P., 1991. Complement components C1r/C1s, bone morphogenetic protein-1 and Xenopus laevis developmentally regulated protein UVS.2 share common repeats. FEBS Lett. 282, 9–12. Bork, P., Bairoch, A., 1995. Extracellular modules. Trends Biochem. Sci. 20 (Suppl.), C03. Bork, P., Beckmann, G., 1993. The CUB domain. A widespread module in developmentally regulated proteins. J. Mol. Biol. 231, 539–545. Bork, P., Downing, A.K., Kieffer, B., Campbell, I.D., 1996. Structure and distribution of modules in extracellular proteins. Q. Rev. Biophys. 29, 119–167. Boyd, J., Burton, D.R., Perkins, S.J., Villiers, C.L., Dwek, R.A., Arlaud, G.J., 1983. Neutron scattering studies of the isolated C1r2 C1s2 subunit of first component of human complement in solution. Proc. Natl. Acad. Sci. U.S.A. 80, 3769–3773. Budayova-Spano, M., Lacroix, M.B., Thielens, N.M., Arlaud, G.J., Fontecilla-Camps, J.C., Gaboriaud, C., 2002a. The crystal structure of the zymogen catalytic domain of complement protease C1r reveals that a disruptive mechanical stress is required to trigger activation of the C1 complex. EMBO J. 21, 231–239. Budayova-Spano, M., Grabarse, W., Thielens, N.M., Hillen, H., Lacroix, M., Schmidt, M., Fontecilla-Camps, J.C., Arlaud, G.J., Gaboriaud, C., 2002b. Monomeric structures of the zymogen and active catalytic domain of complement protease C1r: a further step towards elucidation of the C1 activation mechanism, Structure, in press.

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394 Busby, T.F., Ingham, K.C., 1987. Calcium-sensitive thermal transitions and domain structure of human complement subcomponent C1r. Biochemistry 26, 5564–5571. Busby, T.F., Ingham, K.C., 1990. NH2 -terminal calcium-binding domain of human complement C1s mediates the interaction of C1r with C1q. Biochemistry 29, 4613–4618. Campbell, I.D., Bork, P., 1993. Epidermal growth factor-like modules. Curr. Opin. Struct. Biol. 3, 385–392. Colomb, M.G., Arlaud, G.J., Villiers, C.L., 1984. Activation of C1. Philos. Trans. R. Soc. Lond. B 306, 283–292. Cooper, N.R., 1985. The classical complement pathway: activation and regulation of the first complement component. Adv. Immunol. 37, 151–216. Cseh, S., Gal, P., Sarvari, M., Dobo, J., Lorincz, Z., Schumaker, V.N., Zavodszky, P., 1996. Functional effects of domain deletions in a multidomain serine protease, C1r. Mol. Immunol. 33, 351–359. Dalmasso, A.P., 1992. The complement system in xenotransplantation. Immunopharmacology 24, 149–160. Downing, A.K., Knott, V., Werner, J.M., Cardy, C.M., Campbell, I.D., Handford, P.A., 1996. Solution structure of a pair of calcium-binding epidermal growth factor-like domains: implications for the Marfan syndrome and other genetic disorders. Cell 85, 597–605. Fearon, D.T., Locksley, R.M., 1996. The instructive role of innate immunity in the acquired immune response. Science 272, 50–53. Fishelson, Z., Attali, G., Mevorach, D., 2001. Complement and apoptosis. Mol. Immunol. 38, 207–219. Gaboriaud, C., Rossi, V., Fontecilla-Camps, J.C., Arlaud, G.J., 1998. Evolutionary conserved rigid module-domain interactions can be detected at the sequence level: the examples of complement and blood coagulation proteases. J. Mol. Biol. 282, 459–470. Gaboriaud, C., Rossi, V., Bally, V., Arlaud, G.J., Fontecilla-Camps, J.C., 2000. Crystal structure of the catalytic domain of human complement C1s: a serine protease with a handle. EMBO J. 19, 1755–1765. Gewurz, H., Ying, S.C., Jiang, H., Lint, T.F., 1993. Non immune activators of the classical complement pathway. Behring Inst. Mitt. 93, 138–147. Gruez, A., Piras, C., Pignol, D., Lacroix, M., Arlaud, G.J., Fontecilla-Camps, J.C., 1998. X-ray crystallographic analysis of the heterotrimeric globular heads of human C1q. Mol. Immunol. 35, 331 (Abstract). Hernandez, J.-F., Bersch, B., Pétillot, Y., Gagnon, J., Arlaud, G.J., 1997. Chemical synthesis and characterization of the epidermal growth factor-like module of human complement protease C1r. J. Peptide Res. 49, 221–231. Hoffmann, J.A., Kafatos, F.C., Janeway, C.A., Ezekowitz, R.A., 1999. Phylogenetic perspectives in innate immunity. Science 284, 1313–1318. Illy, C., Thielens, N.M., Arlaud, G.J., 1993. Chemical characterization and location of ionic interactions involved in the assembly of the C1 complex of human complement. J. Protein Chem. 12, 771–781. Jones, E.Y., Stuart, D.I., Walker, N.P., 1989. Structure of tumour necrosis factor. Nature 338, 225–228. Journet, A., Tosi, M., 1986. Cloning and sequencing of full-length cDNA encoding the precursor of human complement component C1r. Biochem. J. 240, 783–787. Kardos, J., Gal, P., Szilagyi, L., Thielens, N.M., Szilagyi, K., Lorincz, Z., Kulcsar, P., Arlaud, G.J., Zavodszky, P., 2001. The role of individual domains in the structure and function of the catalytic region of a modular serine protease, C1r. J. Immunol. 167, 5202–5208. Kishore, U., Reid, K.B.M., 2000. C1q: structure, function, and receptors. Immunopharmacology 49, 159–170. Klein, M.A., Kaeser, P.S., Schwarz, P., Weyd, H., Xenarios, I., Zinkernagel, R.M., Carroll, M.C., Verbeek, J.S., Botto, M., Walport, M.J., Molina, H., Kalinke, U., Acha-Orbea, H., Aguzzi, A., 2001. Complement facilitates early prion pathogenesis. Nat. Med. 7, 488–492. Lacroix, M.B., Aude, C.A., Arlaud, G.J., Colomb, M.G., 1989. Isolation and functional characterization of the proenzyme form of the catalytic domains of human C1r. Biochem. J. 257, 885–891.

393

Lacroix, M., Rossi, V., Gaboriaud, C., Chevallier, S., Jaquinod, M., Thielens, N.M., Gagnon, J., Arlaud, G.J., 1997. Structure and assembly of the catalytic region of human complement protease C1r: a three-dimensional model based on chemical cross-linking and homology modeling. Biochemistry 36, 6270–6282. Lacroix, M.B., Ebel, C., Kardos, J., Dobo, J., Gal, P., Zavodszky, P., Arlaud, G.J., Thielens, N.M., 2001. Assembly and enzymatic properties of the catalytic domain of human complement protease C1r. J. Biol. Chem. 276, 36233–36240. Lakatos, S., 1987. Subunit interactions in the first component of complement, C1. Biochem. Biophys. Res. Commun. 149, 378–384. Leytus, S.P., Kurachi, K., Sakariassen, K.S., Davie, E.W., 1986. Nucleotide sequence of the cDNA coding for human complement C1r. Biochemistry 25, 4855–4863. Luo, C., Thielens, N.M., Gagnon, J., Gal, P., Sarvari, M., Tseng, Y., Tosi, M., Zavodszky, P., Arlaud, G.J., Schumaker, V.N., 1992. Recombinant human complement subcomponent C1s lacking ␤-hydroxyasparagine, sialic acid, and one of its two carbohydrate chains still reassembles with C1q and C1r to form a functional C1 complex. Biochemistry 31, 4254–4262. Mabbott, N.A., Bruce, M.E., Botto, M., Walport, M.J., Pepys, M.B., 2001. Temporary depletion of complement component C3 or genetic deficiency of C1q significantly delays onset of scrapie. Nat. Med. 7, 485–487. Malkowski, M.G., Martin, P.D., Guzik, J.C., Edwards, B.F., 1997. The co-crystal structure of unliganded bovine alpha-thrombin and pre-thrombin-2: movement of the Tyr-Pro-Pro-Trp segment and active site residues upon ligand binding. Protein Sci. 6, 1438–1448. Matsumoto, M., Nagaki, K., Kitamura, H., Kuramitsu, S., Nagasawa, S., Seya, T., 1989. Probing the C4 binding site on C1s with monoclonal antibodies. Evidence for a C4/C4b binding site on the ␥ domain. J. Immunol. 142, 2743–2750. Medved, L.V., Busby, T.F., Ingham, K.C., 1989. Calorimetric investigation of the domain structure of human complement C1s: reversible unfolding of the short consensus repeat units. Biochemistry 28, 5408–5414. Perkins, S.J., Villiers, C.L., Arlaud, G.J., Boyd, J., Burton, D.R., Colomb, M.G., Dwek, R.A., 1984. Neutron scattering studies of subcomponent C1q of first component C1 of human complement and its association with subunit C1r2 C1s2 within C1. J. Mol. Biol. 179, 547–557. Perona, J.J., Craik, C.S., 1997. Evolutionary divergence of substrate specificity within the chymotrypsin-like serine protease fold. J. Biol. Chem. 272, 29987–29990. Pétillot, Y., Thibault, P., Thielens, N.M., Rossi, V., Lacroix, M., Coddeville, B., Spik, B., Schumaker, V.N., Gagnon, J., Arlaud, G.J., 1995. Analysis of the N-linked oligosaccharides of human C1s using electrospray ionisation mass spectrometry. FEBS Lett. 358, 323–328. Poon, P.H., Schumaker, V.N., Phillips, M.L., Strang, C.J., 1983. Conformation and restricted segmental flexibility of C1, the first component of human complement. J. Mol. Biol. 168, 563–577. Rao, Z., Hanford, P., Mayhew, M., Knott, V., Brownlee, G.G., Stuart, D., 1995. The structure of a Ca2+ -binding epidermal growth factor-like domain: its role in protein–protein interactions. Cell 82, 141–141. Reid, K.B.M., 1983. Proteins involved in the activation and control of the two pathways of human complement. Biochem. Soc. Trans. 11, 1–12. Reid, K.B.M., Porter, R.R., 1976. Subunit composition and structure of subcomponent C1q of the first component of human complement. Biochem. J. 155, 19–23. Reid, B.F., Bentley, D.R., Campbell, R.D., Chung, L.P., Sim, R.B., Kristensen, T., Tack, 1986. Complement system proteins that interact with C3b or C4b. Immunol. Today 7, 230–234. Rogers, J., Cooper, N.R., Webster, S., Schultz, J., McGeer, P.L., Styren, S.D., Civin, W.H., Brachova, L., Bradt, B., Ward, P., Lieberburg, I., 1992. Complement activation by beta-amyloid in Alzheimer disease. Proc. Natl. Acad. Sci. U.S.A. 89, 10016–10020. Romao, M.J., Kölln, I., Dias, J.M., Carvalho, A.L., Romero, A., Varela, P.F., Sanz, L., Töpfer-Petersen, E., Calvete, J.J., 1997. Crystal structure of acidic seminal fluid protein (aSFP) at 1.9 A resolution: a bovine polypeptide of the spermadhesin family. J. Mol. Biol. 274, 650–660.

394

G.J. Arlaud et al. / Molecular Immunology 39 (2002) 383–394

Romero, A., Romao, M.J., Varela, P.F., Kölln, I., Dias, J.M., Carvalho, A.L., Sanz, L., Töpfer-Petersen, E., Calvete, J.J., 1997. The crystal structure of two spermadhesins reveal the CUB domain fold. Nat. Struct. Biol. 4, 783–788. Rossi, V., Gaboriaud, C., Lacroix, M.B., Ulrich, J., Fontecilla-Camps, J.C., Gagnon, J., Arlaud, G.J., 1995. Structure of the catalytic region of human complement protease C1r: study by chemical cross-linking and three-dimensional homology modeling. Biochemistry 34, 7311–7321. Rossi, V., Bally, I., Thielens, N.M., Esser, A.F., Arlaud, G.J., 1998. Baculovirus-mediated expression of truncated modular fragments from the catalytic region of human complement serine protease C1s. Evidence for the involvement of both complement control protein modules in the recognition of the C4 protein substrate. J. Biol. Chem. 273, 1232–1239. Schumaker, V.N., Poon, P.H., Seegan, G.W., Smith, C.A., 1981. Semi-flexible joint in the C1q subunit of the first component of human complement. J. Mol. Biol. 148, 191–197. Schumaker, V.N., Hanson, D.C., Kilchherr, E., Phillips, M.L., Poon, P.H., 1986. A molecular mechanism for the activation of the first component of complement by immune complexes. Mol. Immunol. 23, 557–565. Schumaker, V.N., Zavodszky, P., Poon, P.H., 1987. Activation of the first component of complement. Annu. Rev. Immunol. 5, 21–42. Shapiro, L., Scherer, P.E., 1998. The crystal structure of a complement-1q family protein suggests an evolutionary link to tumour necrosis factor. Curr. Biol. 8, 335–338. Siegel, R.C., Schumaker, V.N., 1983. Measurement of the association constants of the complexes formed between intact C1q or pepsin-treated C1q stalks and the unactivated or activated C1r2 C1s2 tetramer. Mol. Immunol. 20, 53–66. Smith, K.F., Haris, P.I., Chapman, D., Reid, K.B.M., Perkins, S.J., 1994. ␤-Sheet secondary structure of the trimeric globular domains of C1q of complement and collagen types VIII and X by Fourier-transform infrared spectroscopy and averaged structure predictions. Biochem. J. 301, 249–256. Strang, C.J., Siegel, R.C., Philipps, M.L., Poon, P.H., Schumaker, V.N., 1982. Ultrastructure of the first component of human complement: electron microscopy of the cross-linked complex. Proc. Natl. Acad. Sci. U.S.A. 79, 586–590. Sunnerhagen, M., Olah, G.A., Stenflo, J., Forsen, S., Drakenberg, T., Trewhella, J., 1996. The relative orientation of Gla and EGF domains in coagulation factor X is altered by Ca2+ binding to the first EGF domain. A combined NMR-small angle X-ray scattering study. Biochemistry 35, 11547–11559. Tacnet-Delorme, P., Boyer, V., Thielens, N.M., Hernandez, J.-F., Bally, I., Sim, R.B., Desgranges, C., Arlaud, G.J., 1999. In vitro analysis of complement-dependent HIV-1 cell infection using a model system. J. Immunol. 162, 4088–4093.

Tacnet-Delorme, P., Chevallier, S., Arlaud, G.J., 2001. Beta-amyloid fibrils activate the C1 complex of complement under physiological conditions. Evidence for a binding site for A beta on the C1q globular regions. J. Immunol. 167, 6374–6381. Thielens, N.M., van Dorsselaer, A., Gagnon, J., Arlaud, G.J., 1990a. Chemical and functional characterization of a fragment of C1s containing the epidermal growth factor-like homology region. Biochemistry 29, 3570–3578. Thielens, N.M., Aude, C.A., Lacroix, M.B., Gagnon, J., Arlaud, G.J., 1990b. Ca2+ binding properties and Ca2+ -dependent interactions of the NH2 -terminal ␣ fragments of human complement proteases C1r and C1s. J. Biol. Chem. 265, 14469–14475. Thielens, N.M., Illy, C., Bally, I.M., Arlaud, G.J., 1994. Activation of human complement serine proteinase C1r is down-regulated by a calcium-dependent intramolecular control that is released in the C1 complex through a signal transmitted by C1q. Biochem. J. 301, 378– 384. Thielens, N.M., Enrié, K., Lacroix, M., Jaquinod, M., Hernandez, J.F., Esser, A.F., Arlaud, G.J., 1999. The N-terminal CUB-epidermal growth factor module pair of human complement C1r binds Ca2+ with high affinity and mediates Ca2+ -dependent interaction with C1s. J. Biol. Chem. 274, 9149–9159. Tsai, S.W., Poon, P.H., Schumaker, V.N., 1997. Expression and characterization of a 159 amino acid, N-terminal fragment of human complement component C1s. Mol. Immunol. 34, 1273–1280. Tschopp, J., Villiger, W., Fuchs, H., Kilchherr, E., Engel, J., 1980. Assembly of subcomponents C1r and C1s of first component of complement: electron microscopy and ultracentrifugal studies. Proc. Natl. Acad. Sci. U.S.A. 77, 7014–7018. Varela, P.F., Romero, A., Sanz, L., Romao, M.J., Töpfer-Petersen, E., Calvete, J.J., 1997. The 2.4 Å resolution crystal structure of boar seminal plasma PSP-I/PSP-II: a zona pellucida-binding glycoprotein heterodimer of the spermadhesin family built by a CUB domain architecture. J. Mol. Biol. 274, 635–649. Villiers, C.L., Arlaud, G.J., Colomb, M.G., 1985. Domain structure and associated functions of subcomponents C1r and C1s of the first component of human complement. Proc. Natl. Acad. Sci. U.S.A. 82, 4477–4481. Weiss, V., Fauser, C., Engel, J., 1986. Functional model of subcomponent C1 of complement. J. Mol. Biol. 189, 573–581. Wiles, A.P., Shaw, G., Bright, J., Perczel, A., Campbell, I.D., Barlow, P.N., 1997. NMR studies of a viral protein that mimics the regulators of complement activation. J. Mol. Biol. 272, 253–265. Zavodszky, P., Gal, P., Cseh, S., Schumaker, V.N., 1993. Protein engineering studies on C1r and C1s. Behring Inst. Mitt. 93, 103–114.