Roles of the Alternative Complement Pathway and C1q during Innate Immunity to Streptococcus pyogenes Free

The complement system is a major component of the early innate immune response and consists of over 30 plasma and surface proteins, forming three enzyme cascades (termed the classical, alternative, and mannan-binding lectin (MBL)³ pathways) and the terminal complement components C5 to C9 (1). The classical pathway is activated when the first component, C1q, binds to the Fc region of Ab complexed with Ag, and the MBL pathway is activated by binding of MBL to specific patterns of mannose or other sugar residues on the pathogen’s surface (1). In contrast, the alternative pathway is activated by spontaneous hydrolysis of C3 leading to covalent binding of C3b in association with factor B to the hydroxyl groups on cell-surface carbohydrates and proteins; pathogen specificity is provided by regulatory proteins, such as factor H, that prevent alternative pathway activity against host cells (1). All three pathways lead to opsonization of target pathogens with the C3 component derivative C3b, the release of the inflammatory mediators C3a and C5a, and the formation of the membrane attack complex on the target cell membrane, which results in the lysis of mainly Neisseria species.

The classical pathway is activated by Ab-Ag complexes and is therefore generally considered mainly an effector of the adaptive immune system. However, there is increasing evidence that the classical pathway is also important for innate immunity, mediated by serum components such as natural IgM or C-reactive protein or by direct binding of C1q to bacterial surfaces (2, 3, 4, 5, 6, 7, 8, 9). C3b deposition on the Gram-positive pathogen Streptococcus pneumoniae in serum from immune naive mice depends mainly on classical pathway activity (mediated at least partially by natural IgM), and S. pneumoniae infection in mice deficient in C1q results in increased replication of the bacteria, impaired activation of macrophages and CD4 cells at the site of infection, and a rapidly fatal illness (8). The classical pathway may also be important for innate immunity to group B Streptococcus (5), and C1q-deficient mice are more susceptible to Salmonella typhimurium infection (10). As well as mediating classical pathway activity, C1q can directly stimulate the phagocyte oxidative burst, chemotaxis, and phagocytosis (11, 12, 13, 14), and so could aid immunity to pathogens independently of complement. However, at present a role for the classical pathway and C1q has been demonstrated only in a small number of pathogens, and it is not known whether classical pathway- and C1q-mediated immunity are important for the innate immune responses to a wide range of microorganisms.

Streptococcus pyogenes is a major bacterial pathogen, commonly causing pharyngitis and soft tissue infections, as well as more severe infections such as necrotizing fasciitis and septicemia. Like S. pneumoniae, S. pyogenes is one of a limited number of pathogens that commonly cause septicemia in previously healthy individuals, and this pathogen has therefore evolved effective mechanisms for overcoming host immunity including the complement system. Several S. pyogenes proteins inhibit the complement system at multiple levels. In many S. pyogenes strains the hypervariable region of the important surface-exposed virulence factor M protein binds the complement regulator protein, C4b-binding protein (C4BP), and this is thought to result in inhibition of classical pathway activity. In addition some M protein types also bind factor H, a regulator of alternative pathway activity. Other S. pyogenes proteins that interact with complement include the streptococcal inhibitor of complement (SIC) that prevents effective formation of the membrane attack complex, a C5a peptidase, and Mac, a secreted protein that blocks phagocytosis mediated by C3b (15, 16, 17, 18, 19). The variety of S. pyogenes products that inhibit complement suggest that complement is essential for immunity to this bacterium, but the relative importance of the different complement pathways for innate immunity to S. pyogenes infection has not previously been investigated.

Using mice and human sera deficient in classical or alternative complement pathways in combination with flow cytometry assays of C3b deposition and opsonophagocytosis (OP) and mouse models of infection, we have investigated the role of both pathways for innate immunity to four different S. pyogenes strains representing important M protein serotypes. The results demonstrate that the alternative pathway is essential for opsonization by C3b and phagocytosis of S. pyogenes in both human and mouse sera, and as a consequence, factor B-deficient mice are very susceptible to S. pyogenes infection. In contrast, C3b deposition on S. pyogenes was not dependent on classical pathway activity in mouse sera for three of the strains and in human sera for two of the strains investigated, possibly due to the high degree of binding by these strains of the classical pathway inhibitor C4BP. However, despite the lack of effect of loss of C1q on C3b deposition, phagocytosis of these strains was impaired in serum deficient in C1q and C1q-deficient mice had increased susceptibility to S. pyogenes infection. These results suggest that the complement component C1q aids immunity to S. pyogenes by mechanisms that are independent of complement activation.

The S. pyogenes strains used in the present study were H372 (an opacity factor-positive M49 serotype, identical to CS101), H305 (an opacity factor-negative M1 serotype scarlet fever isolate), H330 (an opacity factor-negative M3 serotype invasive clinical isolate), and H378 (an opacity factor-positive M22 serotype invasive clinical isolate). Bacteria were cultured at 37°C 5% CO₂ on blood agar plates, or in Todd-Hewitt broth supplemented with 0.5% yeast extract (THYB) to an OD of 0.3 (corresponding to ∼10⁸ CFU/ml) and stored at −70°C in 10% glycerol as single-use aliquots.

The following immune-deficient mice backcrossed to the C57BL/6 background for 10 generations were used in this study: C1qa^−/− (C1q deficient, no classical pathway activity) (20), Bf^−/− (factor B deficient, no alternative pathway activity) (21), C3^−/− (C3 deficient, no effective complement activity) (5), C1qa^−/−.Bf^−/− (factor B and C1q deficient, no alternative or classical pathway activity) (8), and μ_s^−/− (deficient in secretory IgM and hence natural IgM) (22). All these mice were supplied by one of the investigators (MB) and the studies performed according to the institutional guidelines for animal use and care. Wild-type C57BL/6 mice were acquired from commercial suppliers. Blood samples for sera were obtained by terminal exsanguination, and the sera were stored at −70°C as single-use aliquots from individual mice. All mice had had no prior exposure to S. pyogenes, which is not a natural pathogen of mice (8). Human sera with specific complement deficiencies were supplied by Calbiochem (C9⁻, C1q⁻, and Bf⁻ sera). Alternative pathway activity was not affected in the C1q⁻ serum (manufacturer’s product specifications). Pooled normal human serum was obtained from healthy volunteers.

To detect C3b deposited on the surface of S. pyogenes, experiments were performed as previously described using 1/300 dilutions of FITC-conjugated polyclonal goat anti-mouse or anti-human C3 Ab (ICN) and 10⁷ S. pyogenes CFU incubated for 20 min, if not otherwise stated, in 10 μl of 20% sera (diluted in PBS) (8). Sera from mice with the same genetic background were pooled and experiments repeated on several occasions using different batches of serum. The bacteria were fixed in 3% paraformaldehyde and analyzed on a FACSCalibur flow cytometer (BD Biosciences) using forward and side scatter parameters to gate on at least 25,000 bacteria. The results were expressed as the proportion of bacteria showing fluorescence compared with controls incubated in PBS and the geometric mean fluorescence intensity (MFI) of the bacteria. The C3b deposition assay was adapted to assess S. pyogenes binding of factor H and C4BP in normal human serum, and 58 μg/ml purified C1q (Calbiochem) using polyclonal goat anti-human factor H (Calbiochem), polyclonal sheep anti-human C4BP (Abcam), polyclonal goat anti-human C1q (Calbiochem), and a FITC-labeled donkey anti-sheep/goat IgG (Serotec) secondary Ab.

OP of S. pyogenes in different sera was assessed by a flow cytometry assay that measures association of fluorescent bacteria with phagocytes and has previously been used for both S. pneumoniae and S. pyogenes. The general conditions of the assay were based on those described by other authors (23, 24, 25). Briefly, S. pyogenes strains were fluorescently labeled by incubation with FAM-succinimidyl ester (Molecular Probes) (10 mg/ml in DMSO; Sigma-Aldrich) in 0.1 M sodium bicarbonate buffer for 1 h at 37°C, then washed six times with HBSS-0.2% BSA and stored at −70°C in 10% glycerol as aliquots for further assays. Experiments were performed using the human cell line HL60 (promyelocytic leukemia cells, CCL240; American Type Culture Collection) differentiated into granulocytes as described previously (23, 24). Differentiated HL-60 cells were harvested by centrifugation, washed in HBSS and counted in a hematocytometer using trypan blue exclusion. FAM-succinimidyl ester bacteria were opsonized with 10 μl of dilutions of the test sera for 20 min at 37°C with horizontal shaking (150 rpm) using a 96-well plate, before addition of HL-60 differentiated cells for a bacteria to HL60 cell ratio of 20:1. After incubation for 30 min at 37°C with shaking, the suspension was fixed using 3% paraformaldehyde. For each sample a minimum of 6000 differentiated cells were analyzed using a FACSCalibur to assess the proportion of cells associated with bacteria.

Mice aged 12–16 wk, weighing 20–25 g were used for experimental infections. Within each experiment, groups of animals were matched for age and sex. Mice were inoculated by the i.p. or i.v. routes with suspensions of S. pyogenes H372 appropriately diluted in PBS. For survival experiments, mice were killed when they exhibited signs of severe disease from which recovery was unlikely (26). For time point experiments, mice were killed 24 h after inoculation i.p or 4 h after mice inoculation i.v., and blood and spleens were harvested, serial diluted, and cultured to calculate bacterial CFU. Splenocytes from mice inoculated i.p. were analyzed by flow cytometry for lymphocyte subsets (anti-mouse CD4, CD8, and CD45RB antibodies) and the proportion of activated macrophages (anti-CD80 and I-A/I-E MHC class II antibodies) as previously described (8). Serum was stored at −70°C for subsequent cytokine analysis using the mouse inflammation cytometric bead array kit (BD Biosciences) according to manufacturer protocols, which quantitatively measures IL-6, IL-10, MCP-1, IFN-γ, TNF-α, and IL-12 protein levels using flow cytometry.

Data presented is representative of results obtained from at least two independent experiments. The results of C3b deposition experiments and OP assays were analyzed using two-tailed t tests. The proportion of activated lymphocytes and macrophages and bacterial CFU recovered from blood and spleen were analyzed using the Mann-Whitney U test for nonparametric data. Differences in survival curves between mouse strains were compared using the log rank method.

The role of the different complement pathways for C3b deposition on S. pyogenes was characterized using a flow cytometry assay and serum obtained from complement-deficient and wild-type mice with no prior exposure to S. pyogenes. To ensure the results were applicable to different S. pyogenes M serotypes, assays were repeated with four different isolates (H305, H330, H372, and H378) belonging to four different serotypes (M1, M3, M49, and M22, respectively). M1 and M3 serotypes are typical of the major “type 1” strains that cause both invasive and pharyngeal disease as well as rheumatic fever, and the M49 serotype represents a “type 2” strain that is an opacity factor-positive strain and associated with skin disease and glomerulonephritis. A serotype M22 strain was included because the interaction of the M protein of these strains with the complement regulatory protein C4BP has been well-described (17, 19).

For all four strains both the proportion of bacteria positive for C3b and intensity of C3b deposition were decreased when bacteria were incubated in serum from Bf^−/− mice (Fig. 1, A–C), demonstrating the alternative pathway is essential for maximal C3b deposition on S. pyogenes. However, when incubated in serum from C1qa^−/− mice, the proportion of bacteria positive for C3b was only reduced on strain H378 compared with the results for serum from wild-type mice, and the intensity of C3b deposition was unaffected for all strains (Fig. 1, A–C). To identify mild defects in C3b deposition in C1qa^−/− serum, the time course of C3b deposition on strain H372 in different complement-deficient sera was analyzed. The time course of C3b deposition on strain H372 in serum from C1qa^−/− mice had no discernible differences compared with the time course of C3b deposition in serum from wild-type mice (Fig. 1 D). Furthermore, C3b deposition on strain H372 was identical in serum from Bf^−/− mice and Bf^−/−.C1qa^−/− mice, showing that additional loss of classical pathway activity did not affect C3b deposition on S. pyogenes in serum lacking alternative pathway activity (data not shown). Overall these results demonstrate that in immune naive serum C3b deposition on S. pyogenes is mainly dependent on alternative pathway activity, and the classical pathway only has a significant role for strain H378.

Classical pathway-dependent complement deposition on S. pneumoniae is at least partially dependent on natural IgM Abs (8). Hence we investigated the effect of natural IgM on C3b deposition on S. pyogenes using mouse serum from wild-type and u_s^−/− mice (in which there is no circulating IgM). For all four S. pyogenes strains, there were no reductions in either the proportion of bacteria positive for C3b or the intensity of C3b deposition in serum from u_s^−/− mice compared with serum from wild-type mice (Fig. 1, E and F, and data not shown). These results demonstrate that there is no detectable natural IgM-dependent complement activity against S. pyogenes, and provide one explanation why deficiency of C1q has little effect on C3b deposition in mouse serum.

To assess whether the results of the C3b deposition assays in mouse serum were applicable to human serum, the assays were repeated using commercially available human serum deficient in C9 (an equivalent of wild-type serum because C9 deficiency should have no effect on C3b deposition), factor B or C1q. This serum may contain type-specific Ab to S. pyogenes and results of assays using this serum may not reflect innate immune responses. Nevertheless, for strains H372, H305, and H378 similar results were obtained with human serum and mouse serum, with all strains having a reduced proportion of bacteria positive for C3b after incubation in factor B-depleted serum but only strain H378 having a reduced proportion of bacteria positive for C3b after incubation in C1q-depleted serum (Fig. 2, A, B, and D). In contrast, the pattern of C3b deposition on strain H330 differed between mouse and human serum, with a reduced proportion of bacteria positive for C3b after incubation in C1q-deficient human serum (Fig. 2 C). For all strains, the intensity of C3b deposition was generally low with only small and statistically not significant differences between different sera (data not shown). These results demonstrate that there is little classical pathway-dependent C3b deposition on strains H372 and H305 in human serum.

S. pyogenes is thought to inhibit C3b deposition on its surface by binding to complement regulator proteins (18). The hypervariable region of many M protein serotypes binds to the natural inhibitor of classical pathway activity, C4BP and M proteins can also bind factor H, a regulator of the alternative pathway (17, 18). We therefore investigated whether the different patterns of C3b deposition in human sera deficient in C1q for the four S. pyogenes strains reflects variations in the binding of factor H or C4BP between strains. Flow cytometry assays were used to assess the degree of factor H and C4BP binding to strains H372, H305, H330, and H378 in human serum. Factor H bound to strains H372, H305, and H330 (Fig. 3), but as all these strains had significant alternative pathway-dependent C3b deposition in both human and mouse serum, this level of factor H binding was not enough to totally inhibit alternative pathway activity. Factor H binding to strain H378 did not reach statistically significant levels. Strains H372 and H305 had strikingly high levels of C4BP binding, markedly greater than the level for strains H330 and H378 (Fig. 4). This pattern of binding to C4BP mirrored the degree of classical pathway-dependent C3b deposition, with the strains showing high levels of binding to C4BP having no classical pathway-dependent C3b deposition. These results suggest that the differences in patterns of C3b deposition between the strains was at least partially due to the differences in binding of C4BP.

A murine model of septicemia was used to investigate whether C1q was required for innate immunity to the S. pyogenes H372 strain, despite the lack of C3b deposition mediated by the classical pathway on this strain. Wild-type, C3^−/−, C1qa^−/−, and Bf^−/− mice were inoculated i.p. with 5 × 10³ CFU of H372, and the course of disease progression monitored over time (Fig. 5,A). In the wild-type group, only 60% of mice developed fatal infection, and this was delayed until after 48 h, whereas all the C3^−/− mice developed fatal infection within 24 h, confirming the vital importance of complement for innate immunity against S. pyogenes in this model. As expected from the C3b deposition results, the infection also progressed rapidly in Bf^−/− mice, with all mice succumbing within 32 h. Surprisingly, fatal disease also developed in C1qa^−/− mice, with 90% of mice succumbing within 48 h (Fig. 5 A). Infection with S. pyogenes progressed more rapidly in Bf^−/− mice than C1qa^−/− mice (p = 0.026).

To investigate whether the differences in survival of C1qa^−/−, Bf^−/− and C3^−/− mice after inoculation with S. pyogenes were associated with higher bacterial CFU in target organs, mice were culled 24 h after inoculation with 3 × 10⁴ CFU of H372 and the bacterial CFU in target organs assessed by plating serial dilutions of blood and homogenized spleen (Fig. 5, B and C). Bacterial CFU from both blood and spleen were 4 to 6 logs greater in C3^−/−, C1qa^−/−, and Bf^−/− mice compared with wild-type mice, demonstrating that deficiency of either pathway results in markedly increased bacterial replication within target organs and rapidly developing septicemia. The slower progression of disease in C1qa^−/− mice was associated with over one log fewer bacterial CFU recovered from the spleen of C1qa^−/− mice compared with Bf^−/− mice (p = 0.016). These results confirm that the alternative pathway is the most important complement pathway for innate resistance to S. pyogenes, but also suggest that C1q has an important complement-independent role for immunity to H372.

Previously we have shown that the proportion of activated macrophages and CD4 T cells are reduced in C1qa^−/− mice compared with wild-type mice after infection with S. pneumoniae (8), and impaired inflammatory and cellular immune responses may be one mechanism decreasing the immunity of C1qa^−/− mice to S. pyogenes. To investigate this possibility, splenocytes were isolated from wild-type, Bf^−/− and C1qa^−/− mice 24 h after infection i.p. with strain H372, and the numbers of CD4, CD8, and B cell lymphocyte subsets, activated CD4 and CD8 cells, and activated macrophages analyzed using Abs to cell surface Ags and flow cytometry. For both Bf^−/− and C1qa^−/− mice, infection with H372 resulted in a nearly 10-fold expansion of the activated (CD80 positive) macrophage population within the spleen compared with wild-type mice (Fig. 6, A and B). There were no consistent differences in the size of the spleen CD8 T cell or B cell populations between the mouse strains (data not shown), but the proportion of splenocytes consisting of CD4 T cells was consistently reduced in Bf^−/− mice compared with wild-type mice, resulting in a marked depletion in the total numbers of activated CD4 T cells (Fig. 6 C).

To further characterize the inflammatory response to infection with S. pyogenes in wild-type, Bf^{−/ −}and C1qa^−/− mice, the serum levels of important inflammatory cytokines and chemokines were measured using a flow cytometry bead assay. In serum obtained from mice 24 h after i.p. inoculation with strain H372, the patterns of the cytokine responses in Bf^−/− and C1qa^−/− mice were similar, with both strains having marked increases in TNF-α, IFN-γ, IL-6, MCP-1, IL-10, and IL-12 compared with wild-type mice (Table I). The absolute levels of these cytokines were generally higher in Bf^−/− mice. These differences in cytokine response to S. pyogenes infection between the three mouse strains matched the levels of bacterial CFU in the blood and the clinical severity of infection (all Bf^−/− mice were moribund at this time point, C1qa^−/− mice generally exhibited signs of severe infection, and wild-type mice were well or exhibited signs of mild infection only). We therefore analyzed whether early cytokine responses were impaired in C1qa^−/− mice before severe infection had been established by measuring cytokine levels 4 h after i.v. infection. There were only small increases in the levels of TNF-α, IL-6, and MCP-1 with no significant differences between mouse strains (data not shown). Levels of IFN-γ, IL-10, and IL-12 remained undetectable. These results suggest that the cytokine response to infection is not impaired in Bf^−/− or C1qa^−/− mice.

C1q has been shown to bind directly to S. pyogenes (9), and this may explain how C1q can affect immunity to H372 in immune naive mice. We therefore assessed the degree of direct binding of C1q to the four S. pyogenes strains by incubating the bacteria in a physiologically relevant concentration (58 μg/ml) of purified human C1q and measuring C1q bound to the bacterial surface using a flow cytometry assay. All four strains had significant binding to C1q (Fig. 7), providing a mechanism by which C1q may be able to influence immunity to S. pyogenes independent of Ab recognition.

To assess whether the increased CFU of strain H372 in target organs 24 h after i.p. inoculation reflects an impaired ability to clear S. pyogenes from the blood, C3^−/−, C1qa^−/−, Bf^−/−, and wild-type mice were inoculated i.v. with 2 × 10⁶ CFU of strain H372 and bacterial CFU measured in the blood and spleen after 4 h. Compared with wild-type mice, C3^−/−-, C1qa^−/−-, and Bf^−/−-deficient mice failed to clear the bacteria and had over 2 logs greater numbers of bacteria in the blood and over 1/2 log greater bacterial CFU recovered from the spleen (Fig. 8). There were no detectable differences in bacterial CFU recovered from the different complement-deficient mice backgrounds. Hence, complement components including C1q are required for the control of bacterial CFU within the blood at very early stages during infection.

After i.v. inoculation, clearance of streptococcal pathogens from the blood is dependent on phagocytosis by professional phagocytes (27, 28). We therefore investigated the efficiency of serum from wild-type or complement-deficient mice to opsonize S. pyogenes using a flow cytometry assay that detects association (either adherence or internalization) of phagocytes with bacteria (25, 29). FAM-succinimidyl ester-labeled H372, H330, H305, and H378 S. pyogenes strains were incubated in mouse serum and then the human neutrophil cell line HL60, and the proportion of neutrophils associated with bacteria assessed using flow cytometry. The level of OP for all strains was reduced when the bacteria were incubated with serum from Bf^−/− mice compared with serum from wild-type mice (Fig. 9, A–D), consistent with the results for the C3b deposition assays in serum from Bf^−/− mice. Although in mouse serum C1q deficiency did not affect C3b deposition for three of the strains, there were significant reductions in OP of all four S. pyogenes strains in serum from C1qa^−/− mice compared with serum from wild-type mice. For strains H372 and H305, the strains for which there was no detectable classical pathway-mediated C3b deposition in human serum, OP was also reduced in C1q-deficient human serum (Fig. 9 E). Hence, C1q assists OP of S. pyogenes in human and mouse serum independent of the role of C1q for initiating classical pathway complement activation.

Clinical data suggest the importance of complement for preventing infections with pyogenic bacteria (30), and experiments using animal models of disease and complement deficient mice has confirmed the essential role for complement for innate immunity to S. pneumoniae and other Gram-positive pathogens such as group B Streptococcus (5, 8, 31, 32). Surprisingly, the classical pathway has been shown to be important for innate immunity to both S. pneumoniae and group B Streptococcus (5, 8), suggesting that perhaps classical pathway-mediated complement activity is generally an important component of the innate immune response to streptococci and other extracellular pathogens. We have therefore investigated the role of the alternative and classical pathways during innate immunity to four clinically important strains of the other major streptococcal pathogen, S. pyogenes, using mice genetically engineered to have specific complement defects. Mouse strains with genetically engineered deficiencies of specific complement components and which have not had prior exposure to S. pyogenes are very valuable tools for dissecting out the roles of different complement pathways during innate immunity. However, S. pyogenes is not a natural pathogen of mice and at least some interactions between S. pyogenes proteins and complement are thought to be specific to primates (33). The experiments were therefore repeated using human serum in which specific complement components had been removed, but which could contain specific Abs to some S. pyogenes strains and results for which therefore may not reflect innate immune responses. Assays for C3b deposition and OP in human and mouse serum gave broadly similar results for three of the strains investigated, suggesting that the results obtained using complement-deficient mice are likely to represent the interaction of complement with S. pyogenes during infection in humans.

In both mouse and human sera C3b deposition on S. pyogenes was mainly dependent on the alternative pathway, and this correlated with impaired phagocytosis of the bacteria in mouse or human serum-deficient in factor B and reduced clearance of S. pyogenes from the blood of Bf^−/− mice after i.v. inoculation. As a consequence, after i.p. inoculation with S. pyogenes Bf^−/− mice develop a rapidly progressive infection and high numbers of bacterial CFU in the blood and spleen. These data demonstrate the vital role of the alternative pathway for innate immunity to S. pyogenes, and support previous data showing the importance of the alternative pathway for immunity to other streptococcal pathogens (8, 32, 34). The absence of alternative pathway dependent C3b deposition on a S. pyogenes serotype M22 strain in the reports by Carlsson et al. (17, 19) contrasts with our results and is difficult to fully explain at present. Type-specific Ab within the source of serum used by Carlsson et al. could result in mainly classical pathway-dependent complement deposition, and the chelating agents used to selectively inhibit classical or all complement pathways activity may not give as clear results as serum with specific deficiencies of individual complement components. In addition, Carlsson et al. measured C3d levels whereas we have measured all forms of C3 bound to the bacterial surface, and this may have affected direct comparison of the results.

In contrast, in mouse serum there was no detectable classical pathway-dependent C3b deposition on three of the S. pyogenes strains and only a low level of classical pathway-dependent C3b deposition on the fourth strain. In human serum, there was no classical pathway-dependent C3b deposition on two of the strains. Classical pathway-dependent C3b deposition on S. pneumoniae and other pathogens is partially mediated by natural IgM, a spontaneous occurring repertoire of Abs that recognize Ags on the surface of bacterial pathogens without prior exposure of the host to that pathogen (6, 8, 35, 36). However, there was no natural IgM-dependent C3b deposition on the four S. pyogenes strains investigated, perhaps because mouse natural Abs do not recognize S. pyogenes ligands, and this may be one reason why loss of C1q has little effect on C3b deposition on some S. pyogenes strains. Alternatively, S. pyogenes proteins interact with different aspects of complement activity including C4BP, a complement regulatory protein that binds to the hypervariable region of the important surface-expressed S. pyogenes virulence factor, M protein, and may inhibit classical pathway-mediated C3b deposition (16, 17, 18). We have found that the two strains with no C1q-dependent C3b deposition bind very strongly to C4BP, and this may explain the lack of classical pathway-mediated C3b deposition on these strains. In addition, the extracellular S. pyogenes protein H might divert classical pathway complement activity away from the bacteria (37), and two S. pyogenes enzymes, EndoS and SpeB (the streptococcal cysteine protease), can modify host IgG and potentially prevent host IgG bound to bacterial Ags activating the classical pathway (38). The combination of these different proteins could dramatically reduce classical pathway-dependent C3b deposition on S. pyogenes, and this is supported by the observation that in human serum complement deposition on an M22 strain deficient in M protein was mainly classical pathway-dependent (17, 19). Variations in the expression and structure of S. pyogenes proteins that interact with different aspects of the complement system (including C4BP) could explain the differences in the pattern of C3b deposition on the four strains investigated.

Although C1q deficiency had little effect on complement deposition on strain H372, C1qa^−/− mice were highly susceptible to infection with this strain. After i.p. inoculation, fatal disease developed rapidly in C1qa^−/− mice and was associated with large numbers of bacteria from target organs, and after i.v. inoculation, C1qa^−/− mice failed to clear S. pyogenes from the blood. A similar pattern of inflammation was seen in C1qa^−/− mice and Bf^−/− mice, with an influx of macrophages into the spleen and massively increased serum inflammatory cytokine levels. Surprisingly, phagocytosis of S. pyogenes was reduced in both mouse and human serum deficient in C1q, even for strains in which there was no effect on complement deposition in these sera. These data suggest C1qa^−/− mice did not have an impaired inflammatory response to S. pyogenes but were more susceptible to S. pyogenes infection due to the reduced efficiency of phagocytosis, leading to a failure to control bacterial replication. The results of the C3b deposition assays demonstrate that the impaired phagocytosis of some strains of S. pyogenes strains in C1q-deficient serum is not due to reduced opsonization with C3b. C1q bound directly to the S. pyogenes strains used in these studies, and the collagen-like domain of the C1q molecule can bind to a range of cell types, including neutrophils and monocytes, via poorly defined receptors (14, 39, 40, 41, 42, 43). These interactions enhance phagocytosis of erythrocytes and trypanosomes, trigger the oxidative burst and stimulate chemotaxis (11, 12, 13, 14, 42), and could potentially aid OP of S. pyogenes and explain the susceptibility of C1qa^−/− mice to strain H372. As far as we are aware, no role has previously been described for complement-independent C1q-mediated immunity to S. pyogenes, and in general the importance of directly mediated C1q immunity to microbial pathogens has not been clearly defined. Further investigation is required to characterize the mechanisms underlying C1q-dependent immunity to S. pyogenes and other bacterial pathogens.

To conclude, we have shown that the alternative pathway is the dominant complement pathway required for innate immunity to S. pyogenes. The effect of C1q on complement activity varies between S. pyogenes strains, perhaps due to variation in the degree of binding of C4BP. Despite the differences in the effects of C1q deficiency on C3b deposition, phagocytosis was reduced in C1q-deficient serum for all the strains we have investigated, and C1qa^−/− mice were very susceptible to S. pyogenes infection. These results indicate that the balance of immune mechanisms underlying innate immunity to bacterial pathogens vary between related bacterial species and even within a species. This is likely to reflect complex interactions between bacterial proteins and host immunity, and a full understanding of the pathogenesis of important bacterial infections will require careful evaluation of these factors for the major human pathogens.

We are grateful to Aaron Rae (Imperial College) for assistance in cytokine analysis.

The authors have no financial conflict of interest.