Mutational Analyses of the Recombinant Globular Regions of Human C1q A, B, and C Chains Suggest an Essential Role for Arginine and Histidine Residues in the C1q-IgG Interaction¹

C1q is the target recognition protein of the classical pathway of complement activation and a major connecting link between innate immunity and IgG- or IgM-mediated acquired immunity (1). The binding of C1q to Ig-containing immune complex leads to the autoactivation of C1r, which, in turn, activates C1s. C1r and C1s, the two serine protease proenzymes, together with C1q, constitute C1, the first component of the classical pathway. The activation of the C1 complex (C1q + C1r₂ + C1s₂) subsequently leads to the activation of the C2-C9 components of the classical pathway. The human C1q molecule (460 kDa) is composed of 18 polypeptide chains: 6A, 6B, and 6C. Each of the chains (A chain, 223 residues; B chain, 226 residues; and C chain, 217 residues) has a short (3–9 residues) N-terminal region, containing a cysteine residue, followed by a collagen-like region of ∼81 residues and a C-terminal globular domain (gC1q)⁴ of ∼135 residues (2). Six of these chains associate via a combination of inter- and intrachain disulfide bonds, in the order ABC-CBA, to form a structural unit. Three of these structural units then associate via strong noncovalent bonds in the fibril-like central portion to yield the hexameric C1q molecule (3, 4).

C1q interacts with IgG via its gC1q domain, and this interaction is ionic in nature (5). A C1q-binding motif has been previously localized in murine IgG2b involving Glu³¹⁸, Lys³²⁰, and Lys³²², which is highly conserved in different IgG isotypes (6). However, studies by Idusogie et al. (7), using a chimeric mAb rituximab with a human IgG1 C region, have highlighted that Glu³¹⁸, Lys³²⁰, and Lys³²² are likely to be one of several possible C1q binding sites, and there are species-specific differences in the C1q binding site for IgG. Alanine substitutions at positions Glu³¹⁸ and Lys³²⁰ in rituximab have no serious effect on C1q binding or complement activation. Instead, alanine substitution at positions Asp²⁷⁰, Lys³²², Pro³²⁹, and Pro³³¹ significantly reduced the ability of chimera to bind C1q and activate complement. Site-directed mutagenesis studies to localize complementary IgG binding sites on gC1q have been hampered by difficulties in generating recombinant full-length C1q or gC1q because of its heterotrimeric organization.

Our recent studies using recombinant forms of C-terminal globular regions of human C1q A (ghA), B (ghB), and C (ghC) chains have suggested a modular organization of the gC1q domain (4, 8, 9, 10, 11, 12). It is becoming increasingly evident that each of the three modules within the heterotrimeric gC1q domain is likely to be functionally independent and capable of engaging ligands differentially (10, 11, 13). The availability of the expression constructs for ghA, ghB, and ghC has also allowed us to generate single residue mutations and identify residues on the gC1q that participate in the C1q-IgG interaction. Experiments involving chemical modification of arginine and histidine residues and subsequent cross-linking to heterologous IgG have implicated Arg¹¹⁴, Arg¹²⁹, and Arg¹⁶³ in the B chain; Arg¹⁶² in the A chain; and Arg¹⁵⁶ in the C chain in this interaction (14). We have examined the contributions of selected charged amino acid residues (Arg¹⁶² of ghA; Arg¹¹⁴, His¹¹⁷, Arg¹²⁹, and Arg¹⁶³ of ghB; and Arg¹⁵⁶ of ghC) to the interaction of C1q with IgG. Using site-directed mutagenesis, we engineered a series of single residue mutations, involving either arginine (R) to alanine (A), glutamate (E), or glutamine (Q) substitutions, or histidine (H) to alanine (A) or aspartate (D) substitution (Table I). Consistent with chemical modification (14) and molecular modeling studies (15), the functional characterization of point mutants suggests a dominant role for Arg¹¹⁴ of ghB and a complementary role for Arg¹²⁹, Arg¹⁶³, and His¹¹⁷ of ghB; Arg¹⁶² of ghA; and Arg¹⁵⁶ of ghC in the C1q-IgG interaction.

C1q was purified from pooled human serum using IgG-Sepharose (16). The purity of C1q was assessed by SDS-PAGE (15% w/v) under reducing conditions, in which it appeared as three bands of 34, 32, and 27 kDa, corresponding to the A, B, and C chains, respectively.

The codon mutations within the DNA sequences of ghA, ghB, and ghC were generated by site-directed mutagenesis using the overlapping PCR approach (17). The PCR products incorporating the mutations were digested with XbaI and HindIII (in the case of the ghB mutants) or with KpnI and HindIII (in the case of the ghA and the ghC mutants) and subcloned into the pMal-c2 expression vector (New England Biolabs, Beverly, MA). The primers, listed in Table I, were custom made (Invitrogen, Paisley, U.K.). The expression constructs containing mutant sequences were verified by DNA sequencing (ABI Prism 3100 analyzer; Applied Biosystems, Foster City, CA) using bacteriophage M13 and maltose-binding protein (MBP)-specific malE primers.

The globular head regions of the A chain (ghA, residues 88–223), the B chain (ghB, 90–226), and the C chain (ghC, 87–217), and their respective mutants were expressed in Escherichia coli BL21 as fusion proteins linked to MBP and purified, as described recently (11).

Proteins were biotinylated using EZ-link succinimidyl-6-(biotinamido) hexanoate (Pierce, Rockford, IL). The concentration of N-hydroxysuccinimido-biotin used was 25-fold molar excess to the protein solution in PBS, pH 7.2. The reaction was conducted at room temperature for 3 h, followed by extensive dialysis against appropriate buffers to remove free biotin. The biotinylation efficiency was determined using the 2-hydroxyazobenzene-4′-carboxylic acid assay (Pierce). The biotinylation of recombinant proteins did not alter their ability to compete with unlabeled native human C1q in the hemolytic assays (see below).

The wild-type and mutant forms of ghA, ghB, or ghC modules were coated in 0.2 M carbonate buffer (pH 9.6) at different concentrations (0.125, 0.25, 0.5, and 1 μg/well) on microtiter wells overnight at 4°C, washed, and then blocked with PBS containing 2% (w/v) BSA for 2 h. After three rounds of washing, the wells were incubated with heat-aggregated human IgG (10 μg/well) in PBS containing 0.05% Tween 20 (PBST; 2 h at 37°C). Following washing, bound IgG was detected using goat anti-human IgG-HRP conjugate and o-phenylenediamine.

The inhibitory effects of mutant proteins on the interaction between ghA, ghB, or ghC and heat-aggregated IgG were examined by coating heat-aggregated human IgG (2 μg/well) on the microtiter wells. Following blocking and washing, biotinylated ghA, ghB, or ghC (2 μg/well) and different amounts of nonbiotinylated mutant proteins (1.25, 2.5, 5, 10 μg/well) were added to the wells. The amount of bound ghA, ghB, or ghC was detected using extravidin-alkaline phosphatase (AP) conjugate.

SRBC (E), sensitized with rabbit anti-sheep hemolysin (A; EA_IgG), were prepared in DGVB²⁺ (dextrose gelatin veronal buffer containing 2.5 mM sodium barbital, 71 mM NaCl, 0.15 mM CaCl₂, 0.5 mM MgCl₂, 2.5% w/v glucose, 0.1% w/v gelatin, pH 7.4). The addition of human C1q (1 μg) to C1q-deficient serum (Sigma-Aldrich, St. Louis, MO; 1/40 dilution in DGVB²⁺) was sufficient to lyse >95% EA_IgG in the assay described below. This concentration of human C1q was subsequently used to determine whether the pretreatment of EA_IgG with ghA, ghB, ghC, or mutant proteins protected EA_IgG from C1q-mediated hemolysis.

The EA_IgG (10⁷ cells/100 μl) were pretreated with a range of concentrations of ghA, ghB, ghC, or mutant proteins (1.25, 2.5, 5, and 10 μg each) for 1 h at 37°C. Cells were centrifuged, and the pellet was washed and resuspended in 100 μl of DGVB²⁺. Each aliquot of pretreated EA_IgG cells was added to a mixture composed of 1 μg of C1q in 10 μl of DGVB²⁺ buffer, 2.5 μl of C1q-deficient serum, and 87.5 μl of DGVB²⁺. Following 1-h incubation at 37°C, the reaction was stopped by transferring the tubes to ice and adding 600 μl of ice-cold DGVB²⁺. The unlysed cells were pelleted by centrifugation, and the A₄₁₂ of the supernatants was read. Total hemolysis (100%) was taken as the amount of hemoglobin released upon cell lysis with water. The C1q-dependent hemolytic activity was expressed as a percentage of total hemolysis. MBP was used as a negative control protein.

Using PCR-based site-directed mutagenesis, the alanine and glutamate variants of key arginine residues (Arg¹⁶² of ghA; Arg¹¹⁴, Arg¹²⁹, and Arg¹⁶³ of ghB; Arg¹⁵⁶ of ghC) were generated. Another three mutants were also engineered, which included substitution of Arg¹¹⁴ of ghB to glutamine (ghB-R114Q) and of His¹¹⁷ of ghB to either alanine (ghB-H117A) or aspartate (ghB-H117D; Table I). The incorporation of each mutation was confirmed by automated DNA sequencing. The PCR products were subcloned in the pMal-c2 vector and expressed in the soluble fraction as MBP fusion proteins in E. coli BL21 cells, as described recently for the wild-type ghA, ghB, and ghC (11). Following induction with 0.4 mM isopropyl β-d-thiogalactoside (IPTG) for 3 h, each fusion protein accumulated intracellularly as an overexpressed protein of ∼60 kDa, as judged by SDS-PAGE under reducing conditions (Fig. 1, a–c). The majority of the MBP fusion proteins, extracted in the soluble fraction after cell lysis and sonication, bound to amylose resin and eluted as >95% pure soluble fractions (Fig. 1, d–f). The mutants were expressed to levels comparable to their wild-type counterparts. When the fusion proteins were passed through a Q-Sepharose anion-exchange column to remove contaminating DNA, the fusion proteins bound at 0.1 M NaCl and eluted as a sharp peak at ∼0.6 M NaCl, with the mutants behaving in a very similar fashion to the wild-type proteins. The factor Xa cleavage, used to separate the globular domain from the MBP, caused aggregation of the wild-type as well as the mutants; therefore, the MBP fusions of each globular region were used for functional assays.

The point mutants of the ghA, ghB, and ghC modules were recognized by rabbit anti-human C1q antisera. All the point mutants retained the antigenic characteristic of their wild-type forms because they were well recognized by module-specific Abs using ELISA (data not included). The fusion proteins containing mutant modules, when loaded onto a Superose 12 gel filtration column (Pharmacia) in the buffer containing 20 mM Tris-HCl, 100 mM NaCl, and 1 mM EDTA, pH 7.5, eluted between ∼150 kDa (alcohol dehydrogenase) and ∼68 kDa (BSA) molecular size markers, suggesting that the fusion proteins were either dimers, or monomers probably being retarded during gel permeation.

Using ELISA, we examined the IgG-binding ability of recombinant ghA, ghB, and ghC and their single residue mutants (ghA-R162A, ghA-R162E, ghB-R114A, ghB-R114E, ghB-R114Q, ghB-R129A, ghB-R129E, ghB-R163A, ghB-R163E, ghC-R156A, ghC-R156E). Different amounts of ghA, ghB, ghC, and their corresponding mutants were coated onto the microtiter wells, incubated with heat-aggregated IgG, washed, and then probed with anti-human IgG-HRP conjugate.

The substitution of Arg¹⁶² to alanine in the ghA (ghA-R162A) resulted in up to 16% reduction in IgG binding (Fig. 2,a). When Arg¹⁶² was substituted with the negatively charged residue glutamate, the ghA-R162E mutant bound even less IgG (∼35% reduction). For the ghB module, the Arg¹¹⁴ substitution had the most significant effect on IgG binding (Fig. 2,b) The mutants ghB-R114A and ghB-R114E showed ∼50% reduction in IgG binding, compared with the wild-type ghB. The substitution of Arg¹²⁹ and Arg¹⁶³ with either alanine or glutamine also caused up to 30–40% reduction in IgG binding (Fig. 2, c and d), suggesting that the contributions of Arg¹²⁹ and Arg¹⁶³ to the ghB-IgG interaction were comparable. A smaller reduction in IgG binding (∼18–22%) was observed for ghC mutants involving Arg¹⁵⁶ (Fig. 2 e). In each case, the glutamate mutants bound less IgG than the alanine mutants.

The finding that only the substitution of Arg¹¹⁴ of human C1q B chain with alanine or glutamate reduced the IgG binding >40% (up to 50% of the wild-type ghB) appears to suggest that Arg¹¹⁴ is crucial for IgG binding. To understand whether the reduced IgG binding was due to the hydrophobicity of the alanine residue, an additional ghB mutant (ghB-R114Q) was constructed and the influence of a more polar amino acid, glutamine, was examined (Fig. 2 b). However, the IgG-binding ability of ghB-R114Q was found to be comparable to ghB-R114A.

In a competitive ELISA, the mutant proteins were allowed to compete with their respective wild-type globular head modules for binding solid-phase heat-aggregated IgG. All three mutants involving Arg¹¹⁴ of ghB (ghB-R114A, ghB-R114E, and ghB-R114Q) failed to compete with ghB in binding IgG, indicating that their interaction with IgG is of much lower affinity than that of their wild-type counterpart (Figs. 2 and 3,b). The inhibitory activities of the other mutant proteins were ∼30% of the wild-type modules (Fig. 3).

We also generated two mutants involving His¹¹⁷ of the ghB module (ghB-H117A and ghB-H117D), which retained ∼80% IgG-binding properties of the ghB. Thus, the His¹¹⁷ may contribute ∼20% of the IgG-binding ability of the wild-type ghB (Fig. 4,a). The inhibitory experiments for these mutants confirmed this observation (Fig. 4 b).

To examine the inhibitory effects of ghA, ghB, ghC, and their respective mutants on C1q-dependent hemolysis, SRBC (E) were sensitized with the IgG fraction of anti-sheep E hemolysin (A) to yield EA_IgG cells. Reconstitution of C1q-deficient serum with 1 μg of C1q was found to completely lyse (>95%) the sensitized EA cells. EA_IgG cells were pretreated with various concentrations of ghA, ghB, ghC, or mutant proteins (1.25, 2.5, 5, and 10 μg) before adding C1q-deficient serum reconstituted with exogenous human C1q. Compared with ∼55% inhibition by the ghA at the maximum concentration tested (10 μg), the mutants ghA-R162A and ghA-R162E had reduced inhibitory activities (∼30% at 10 μg; Fig. 5,a). The inhibitory activity of the ghB at 10 μg concentration was found to be ∼70%. However, all three Arg¹¹⁴ mutants, ghB-R114A (∼10%), ghB-R114E (∼28%), and ghB-R114Q (∼22%), competed poorly with native C1q in the hemolytic assay (Fig. 5,b). The mutants ghB-R129A and ghB-R129E were weak inhibitors (∼28% inhibition at 10 μg; Fig. 5,c), whereas ghB-R163A and ghB-R163E retained about one-half of the potency of the wild-type ghB as inhibitors of C1q-dependent hemolysis (∼38% inhibition at 10 μg; Fig. 5,d). At 10 μg concentration, the wild-type ghC failed to achieve 50% inhibition, which was slightly better than its mutants ghC-R156A and ghC-R156E (Fig. 5,f). For the His¹¹⁷ mutants tested, ghB-H117A and ghB-H117D showed ∼40 and ∼50% inhibition at 10 μg concentration, compared with ∼70% inhibition by ghB at the same concentration of protein (Fig. 5 e).

The human C1q has a characteristic heterotrimeric gC1q domain (3, 4, 8), a module that is also found in a variety of noncomplement proteins, including collagen VIII and X, precerebellin, hibernation proteins, multimerin, adipocyte complement-related protein of 30 kDa (ACRP-30)/adiponectin, saccular collagen, and EMILIN (1, 4, 10). The crystal structure of a recombinant gC1q homotrimer of mouse ACRP-30 has revealed a structural and evolutionary link between TNF and gC1q-containing proteins, and hence recognition of a C1q/TNF molecular superfamily (19, 20). A recently published crystal structure of the gC1q domain of human C1q at 1.9 Å resolution (15) has revealed a compact, spherical, heterotrimeric assembly, held together predominantly by nonpolar interactions, with 3-fold symmetry. Like ACRP-30, the N- and C-terminal ends of the ghA, ghB, and ghC emerge at the base of the trimer. It has previously been suggested that the heterotrimeric organization of the gC1q is maintained due to the presence of structural patches within ghA, ghB, and ghC, which probably interact specifically during C1q biosynthesis (3). Approximately 27% of the residues, which are conserved between the A, B, and C chains of human gC1q, including three cysteine and several hydrophobic and neutral residues, form the gC1q scaffold and impart upon it a largely β-sheet structure (21). Consistent with this view, the crystal structure of gC1q has identified a series of interactions along the 3-fold axis of the heterotrimer that include hydrogen bonds, a well-exposed Ca²⁺ ion located near the apex, and main-chain polar interactions. Additional lateral interactions, which are hydrophobic at the base and polar and hydrophilic toward the apex, further stabilize the heterotrimeric assembly.

Given the heterotrimeric organization of the gC1q, it has been debated whether the C-terminal globular head regions of human C1q A, B, and C chains are functionally autonomous modules (with ghA, ghB, and ghC having distinct binding properties), or whether the ability of C1q to bind its ligands is dependent upon a combined, globular structure (10, 11). To address these questions, we generated the recombinant forms of ghA, ghB, and ghC and examined their interaction with IgG and IgM. In ELISA, the ghA module appeared the most effective region for interacting with both heat-aggregated IgG and IgM, whereas ghB and ghC showed individual preference for aggregated IgG and IgM, respectively. In C1q-dependent hemolytic assays, both ghA and ghB competed with C1q and protected IgG- and IgM-sensitized SRBC against lysis. However, for IgM-coated SRBC, the ghC was a better inhibitor of C1q than the ghB (11).

The expression and functional characterization of ghA, ghB, and ghC thus allowed assessment of the modular nature of the gC1q domain. It also provided an opportunity to mutate single amino acids in the gC1q domain and identify complementary binding sites involved in IgG binding. The roles of arginine residues of C1q have previously been considered important during IgG interaction (5). Experiments involving chemical modification of specific amino acid residues and subsequent cross-linking to heterologous IgG have implicated Arg¹¹⁴ and Arg¹²⁹ of the B chain as being the important residues in IgG binding, while highlighting participation of Arg¹⁶² in the A chain, Arg¹⁶³ in the B chain, and Arg¹⁵⁶ in the C chain in the C1q-IgG interaction (14).

We examined the contributions of Arg¹⁶² (of ghA), Arg¹¹⁴, His¹¹⁷, Arg¹²⁹, Arg¹⁶³ (of ghB), and Arg¹⁵⁶ (of ghC) to the C1q-IgG interaction by substituting each of them with either neutral or negatively charged residues. Substitutions with negatively charged glutamate residues led to a larger decrease in the IgG binding by the mutants compared with their corresponding alanine substitutions. The lower IgG-binding capacity of the glutamate or aspartate mutants compared with alanine mutants indicates that an ionic interaction is required at those specific positions for correct C1q-IgG interaction. Substitution of Arg¹¹⁴ of ghB with either alanine (ghB-R114A) or glutamate (ghB-R114E) caused maximum reduction in IgG binding (∼50%), highlighting a major role for Arg¹¹⁴ of human C1q B chain in the C1q-IgG interaction. We generated another mutant, ghB-R114Q, to examine whether the reduction in IgG binding by ghB-R114A originated due to the hydrophobic nature of alanine residue. The mutant ghB-R114Q showed ∼40% IgG binding compared with the wild-type ghB. The substitution of Arg¹⁶² of ghA and Arg¹⁵⁶ of ghC with alanine (ghA-R162A, ghC-R156A) or glutamate (ghA-R162E, ghC-R156E) led to ∼20% reduction in IgG binding. The results of ELISA and hemolytic assays suggest that Arg¹¹⁴ of ghB plays a central role in the C1q-IgG interaction, with Arg¹²⁹ and Arg¹⁶³ of the ghB making important contributions.

The crystal structure of the gC1q heterotrimer of human C1q (15) has supported the general view of the modular nature of the C1q globular region (4, 8, 11). It has also revealed that Arg¹⁶² of ghA and Arg¹⁵⁶ of ghC are engaged in internal salt bridges with Asp¹⁹¹ of ghA and Glu¹⁸⁷ of ghC, respectively, and hence, are unlikely contributors to the C1q-IgG interaction within a heterotrimer. But, using individual modules in solution, we found that both ghA and ghB bound IgG equally well, highlighting an interesting inconsistency between the modeling studies and these solution studies. The surface of the ghB is predominantly positively charged, characterized by a stretch of three basic residues: Arg¹⁰¹, Arg¹¹⁴, and Arg¹²⁹. These three arginine residues, which are considered important for IgG binding (14), appear on the outside edge of the ghB module. In the crystal structure, Arg¹¹⁴ and Arg¹²⁹ are shown to have ordered structures; however, this may be due to the stabilizing effects of crystal contacts and, like Arg¹⁶³, they are likely to be available for ligand binding in solution. Our previous studies (18, 22, 23) are quite consistent with the structural modeling, suggesting a predominant role of the ghB module in C1q-IgG interaction. The biochemical data from our present study using mutants of Arg¹¹⁴, Arg¹²⁹, and Arg¹⁶³ also support a pivotal role of Arg¹¹⁴ of the ghB module in binding IgG. The most attractive model that attempts to describe ghB-IgG interaction (15) positions the two molecules in such a way that Asp²⁷⁰ and Lys³²² of IgG form salt bridges with Arg¹²⁹ and Glu¹⁶² of the ghB, respectively. In this orientation, the Arg¹²⁹ appears to act like a wedge between the C_H2 and L chain C domains. Gaboriaud et al. (15) have contemplated that the gC1q may bind to Fab region as well through interaction with L chains. Thus, Fab/Fc orientation may be a critical factor in dictating access of the ghB module to C_H2 domain, with Arg¹²⁹ probably having a significant role. Chemical modification of the histidine residues has been shown to reduce the ability of C1q to activate complement (14). The IgG-binding properties of two of the histidine mutants, ghB-H117A and ghB-H117Q, have confirmed the importance of His¹¹⁷ of the C1q B chain.

In conclusion, the availability of the recombinant forms of the wild-type and mutant globular head regions of C1q should make it possible to examine their interactions with a range of ligands known to bind via gC1q. It may be possible to generate a soluble and stable heterotrimer using MBP-free ghA, ghB, and ghC via denaturation-renaturation procedures. If successfully generated, such a recombinant heterotrimeric gC1q molecule should allow inclusion of the point mutants of ghA, ghB, or ghC modules individually to study the effects of single residue mutations on the structure-function relationship in a heterotrimeric context.