|Home | About | Journals | Submit | Contact Us | Français|
The human complement system is important in the immunological control of Staphylococcus aureus infection. We showed previously that S. aureus surface protein clumping factor A (ClfA), when expressed in recombinant form, bound complement control protein factor I and increased factor I cleavage of C3b to iC3b. In the present study, we show that, compared to the results for the wild type, when isogenic ClfA-deficient S. aureus mutants were incubated in serum, they bound less factor I, generated less iC3b on the bacterial surface, and bound fewer C3 fragments. It has been shown previously that two amino acids in ClfA (P336 and Y338) are essential for fibrinogen binding. However, S. aureus expressing ClfA(P336A Y338S) was less virulent than ClfA-deficient strains in animal models. This suggested that ClfA contributed to S. aureus virulence by a mechanism different than fibrinogen binding. In the present study, we showed that S. aureus expressing ClfA(P336A Y338S) was more susceptible to complement-mediated phagocytosis than a ClfA-null mutant or the wild type. Unlike ClfA, ClfA(P336A Y338S) did not enhance factor I cleavage of C3b to iC3b and inhibited the cofactor function of factor H. Fibrinogen enhanced factor I binding to ClfA and the S. aureus surface. Twenty clinical S. aureus strains all expressed ClfA and bound factor I. High levels of factor I binding by clinical strains correlated with poor phagocytosis. In summary, our results suggest that the interaction of ClfA with factor I contributes to S. aureus virulence by a complement-mediated mechanism.
Staphylococcus aureus is a significant cause of morbidity and mortality; methicillin-resistant S. aureus (MRSA) caused an estimated 18,650 deaths in the United States in 2005 (19). Antibiotic resistance continues to increase among S. aureus isolates, including isolates of community-associated MRSA (CA-MRSA) (7, 30), health care-associated MRSA (12), and S. aureus with reduced susceptibility to vancomycin (20). Understanding how this organism avoids host immune defenses is crucial for the development of new strategies to prevent and treat infections.
Complement is a major component of innate immunity and plays a vital role in the control of many bacterial pathogens (28), including S. aureus (15, 21, 33). Indeed, this organism secretes several small soluble proteins that interfere with normal complement host defense mechanisms, including SCIN and Efb (15, 32). We have previously shown that the human complement regulator factor I is captured on the S. aureus cell surface, where it is activated and cleaves the crucial opsonin C3b (22) to iC3b (3). This results in decreased phagocytosis by human neutrophils (2). We subsequently showed that the A domain of clumping factor A (ClfA), an important surface-located fibrinogen-binding protein, bound factor I and acted as a cofactor to trigger cleavage of C3b to iC3b (13).
The binding to fibrinogen by ClfA involves the C terminus of the γ-chain binding to a trench located between subdomains N2 and N3 by a “dock-lock-latch” mechanism (18). Residues Pro336 and Tyr338 are located in the trench and are crucial for ligand binding, and a P336S Y338A mutant (ClfAPYII) is completely defective in fibrinogen binding (23).
Clumping factor A is covalently anchored to the cell wall of S. aureus and promotes adhesion of the bacterium to fibrin clots and to thrombi created on heart valves in a rat model of endocarditis (25). In addition, ClfA is required for survival of bacteria following injection into the bloodstream of mice (16). This was attributed to the ability of the protein to promote bacterial resistance to phagocytosis by neutrophils. It was proposed that binding to fibrinogen prevented the deposition or recognition of opsonins. However, phagocytosis experiments performed in the absence of fibrinogen demonstrated that expression of ClfA still had an antiphagocytic effect, suggesting that there is another mechanism (14).
In mouse models of S. aureus bacteremia and septic arthritis, bacteria expressing the non-fibrinogen-binding mutant of ClfA were less virulent than a null mutant that was devoid of the surface protein (17). It was difficult to explain these effects by the loss of fibrinogen binding alone. In the present study, we analyzed the interaction of ClfA with factor I on the bacterial cell surface and the roles of these proteins in triggering cleavage of C3b to iC3b. In doing so, we developed a novel explanation for the role of ClfA in disrupting opsonophagocytosis.
S. aureus strains Newman and Reynolds were used in all experiments. Bacteria were grown in Columbia broth containing 2% NaCl at 37°C to mid-logarithmic phase, unless otherwise noted. Two ClfA-deficient strains that are isogenic mutants of strain Newman (9) were tested. ClfA-(2) was an isogenic mutant with the genotype clfA2::Tn917 (21). ClfA-(5) was an isogenic mutant with a frameshift mutation in clfA5 (11). A ClfAPYII-expressing strain expresses a non-fibrinogen-binding variant of ClfA [ClfA(P336A Y338S)] (17). Newman srtA::Tcr is a sortase A-deficient mutant (27).
In mid-logarithmic phase S. aureus strain Reynolds produces amounts of capsule that are not detectable by rocket immunoelectrophoresis (4). To evaluate the role of capsule, a capsule-deficient isogenic mutant of the Reynolds strain, JL022 (29), was tested.
Clinical S. aureus strains were obtained as discarded deidentified isolates from the Clinical Microbiology Laboratory of The Children's Hospital of The King's Daughters (Eastern Virginia Medical School IRB protocol 06-04-WC-0040). Twenty isolates were tested, including 5 CA-MRSA invasive isolates, 5 CA-MRSA noninvasive isolates, 5 methicillin-susceptible S. aureus (MSSA) invasive isolates, and 5 MSSA noninvasive isolates. CA-MRSA isolates were defined as MRSA isolates that were susceptible to clindamycin, while hospital-associated MRSA were defined as isolates that were resistant to clindamycin. Invasive isolates were recovered from blood, bone, or joint cultures. Noninvasive isolates were recovered from nasopharyngeal cultures, superficial wounds, or colonized tracheostomies of patients without systemic or deep-tissue S. aureus infection.
Recombinant ClfA (rClfA) was expressed from pQE30 in Escherichia coli. It comprised the full A region (residues 40 to 599) with an N-terminal His tag (26). Recombinant ClfAPYII(P336S Y338A) (rClfAPYII) is a variant that does not bind fibrinogen. It was also expressed in E. coli with a His tag (17). Recombinant protein expression and purification were performed as previously described (26).
S. aureus was incubated in serum diluted in 60% DGVBS++ buffer (60% Veronal-buffered saline [VBS] with 3% dextrose, 0.1% gelatin, 0.15 mM CaCl2, and 1.0 mM MgCl2), unless otherwise stated. Complement activation was inhibited with EDTA-GVBS−− (VBS with 0.1% gelatin and 0.01 M EDTA).
Normal human serum (NHS) was obtained from the blood of healthy human volunteers in accordance with an institutional review board-approved protocol (Eastern Virginia Medical School IRB 02-06-EX-0216). The sera from 5 persons were pooled, aliquoted, and frozen at −80°C, as previously described (4). Heat-inactivated serum was generated by warming NHS to 56°C for 30 min. Hirudin plasma was generated from the blood of human volunteers as previously described (31). Hirudin plasma was used because it does not alter the function of the complement system, unlike the plasma obtained by other methods of generating plasma, like the EDTA and heparin methods. Purified factor I was purchased commercially (CompTech, Tyler, TX) and tested for purity and functionality (3). C3-depleted serum was purchased commercially (CompTech).
Cell wall preparations were generated as previously described (13). Briefly, S. aureus cell walls were digested with lysostaphin in a 30% raffinose buffer (to stabilize the bacterial protoplasts) containing protease inhibitors and DNase. The protoplasts were sedimented, and the supernatant was recovered and used as the cell wall preparation. Detection of ClfA in cell wall preparations was performed by Western blotting as previously described (13).
After mid-logarithmic-phase S. aureus cultures were washed, their concentrations were adjusted to 1 × 109 bacteria per ml. A staphylococcal suspension (0.125 ml) was added to 0.1 ml 60% DGVBS++ buffer and serum to obtain the final serum concentration described below and incubated for 5 min at 37°C. Bacteria were washed twice in GVBS−− buffer and then boiled in 30 μl of 2% SDS-0.05 M Tris buffer for 5 min to remove surface-bound proteins. Staphylococci were then sedimented, and the supernatant was recovered for analysis by a factor I enzyme-linked immunosorbent assay (ELISA) (below). The amount of factor I, expressed in molecules/CFU, was calculated as follows:
The amount of factor I was determined by an ELISA as previously described (13). Briefly, ELISA plates were coated with goat anti-human factor I antibody (CompTech, Tyler, TX). The plates were blocked with 3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS)-Tween. Test samples were then incubated for 1 h at room temperature. Factor I was detected with mouse anti-human factor I monoclonal antibody (Quidel, San Diego, CA), followed by goat anti-mouse horseradish peroxidase (HRP)-linked antibody. Western blot detection of factor I was performed as previously described (13). Briefly, a membrane was probed with an anti-factor I monoclonal antibody and a goat anti-mouse HRP-linked polyclonal antibody and then developed by using enhanced chemoluminescence (ECL). A modified ELISA was used to determine the amount of factor I from NHS binding to rClfA. Wells of an ELISA plate were coated with 20 μg/ml of either rClfA or rClfAPYII in carbonate buffer. The wells were washed, blocked, and then incubated with NHS in 3% BSA for 1 h. Washed wells were then incubated with an anti-factor I monoclonal antibody, followed by an anti-mouse HRP-linked antibody as described above.
In 100 μl of 60% DGVBS++, purified C3b (1 μg) and purified factor I (0.1 μg) were combined with rClfA or rClfAPYII (20, 100, or 200 μg) at 37°C for 16 h. A control containing C3b, factor I, and factor H was used in these experiments. The samples were examined to determine iC3b generation by ELISA, as described elsewhere (2).
S. aureus isolates were grown as described above, washed with GVBS++, and diluted to obtain a concentration of 1 × 109 CFU/ml based on the optical density at 600 nm. An aliquot containing 1 × 108 CFU was incubated in 1% NHS in 0.5 ml (total volume) of GVBS++ for 5 min at 37°C. The bacteria were then washed twice with GVBS−−.
Bacteria were also opsonized with C3b using purified complement components to activate the classical pathway as previously described (3). Briefly, antibody-sensitized S. aureus was incubated successively with purified complement components C1, C4, C2, and C3 to generate the classical pathway C3-convertase and bind C3b to the bacterial surface in the absence of other serum proteins. C3b-coated staphylococci were then incubated with factor I (4 μg/ml) with and without factor H (4 μg/ml) in GVBS−−.
Deposited C3 fragments were stripped from bacteria with 25 mM methylamine as previously described (4). iC3b was measured by an ELISA using anti-C3 polyclonal antibody for capture and anti-iC3b monoclonal antibody for detection, as previously described (2). Total C3 fragments were measured by an ELISA using different anti-C3 polyclonal antibodies for capture and detection, as previously described (2). C3 fragments were also analyzed by Western blot probing with anti-C3 polyclonal antibody that recognized the peptide chains of both C3b and iC3b, as previously described (5). Images were captured digitally using Versadoc (Bio-Rad).
Human neutrophils were prepared from heparinized human blood from healthy human volunteers by Hypaque-Ficoll step gradient centrifugation, dextran sedimentation, and hypotonic lysis. S. aureus isolates were grown, washed, and diluted as described above. An aliquot containing 1 × 108 S. aureus cells was incubated in 1% NHS or 1% heat-inactivated serum in 0.5 ml (total volume) of GVBS++ for 30 min at 37°C. These cells were not washed, but instead an aliquot containing 2 × 107 opsonized S. aureus cells was directly incubated with 1 × 106 neutrophils and tumbled for 45 min at 37°C. A 100-μl aliquot of the mixture was removed, stained with acridine orange (final concentration, 0.01%), and then quenched with crystal violet (final concentration, 0.03%) as previously described (2).
A polyvinylidene difluoride (PVDF) membrane was blocked in Tris-buffered saline-Tween containing 3% BSA, which was also used for antibody diluents. The primary probe was a chicken anti-ClfA antibody (1:5,000), the secondary probe was a goat anti-chicken HRP-linked antibody (1:5,000; Genway Biotech Inc., San Diego, CA), and the blot was developed by using ECL.
Wells of ELISA plates were coated with a goat anti-factor I antibody at a 1:1,000 dilution in carbonate buffer for 4 h, washed with PBS containing 0.5% Tween (PBST) (which was also done between each of the following steps), and blocked with 3% BSA-PBST overnight at 4°C. The plates were then incubated with 10% NHS for 1 h at room temperature to capture serum factor I on the plates. Titrating amounts of rClfA or rClfAPYII were then applied to the plates, starting with 200 μg/ml diluted in 60% DGVBS++ for 3 h at 37°C. Bound rClfA was detected using a chicken anti-ClfA antibody (diluted 1:1,000 in 3% BSA-PBST) for 1 h at room temperature and a goat anti-chicken HRP-linked antibody (diluted 1:1,000 in 3% BSA-PBST) for 1 h at room temperature. Plates were developed with 3,3′,5,5′-tetramethylbenzidine (TMB), the reaction was stopped with 1 N H2SO4, and the results were read at 450 nm.
The values from replicate experiments were averaged, and means and standard errors were calculated (Microsoft Excel XP). Statistical comparisons were made using a Student t test, and P values of <0.05 were considered significant (GraphPad InStat).
Previously, we have shown that the soluble A domain of ClfA can bind complement factor I and act as a cofactor (13). In order to analyze the ability of ClfA expressed on the surface of S. aureus to promote binding of factor I, the wild-type ClfA+ strain Newman was compared with two null mutants lacking ClfA. Bacteria were incubated in human serum, and factor I bound to the cells was stripped off by boiling cells in 2% SDS and was measured by the ELISA (Fig. (Fig.1A).1A). In 0.5% serum, the ClfA+ strain bound 112% more factor I than the mutants (mean value for the two null mutants; P = 0.012), whereas in 1% serum, the wild-type bound 69% more factor I than the mutants (P = 0.012). These results show that expression of ClfA resulted in a higher level of factor I binding compared to the data obtained for cells lacking the protein but that some factor I still bound to the ClfA− mutants.
In order to evaluate the effect of ClfA on cleavage of C3b to iC3b on the S. aureus cell surface, the wild type and the ClfA-defective mutants were incubated in 1% NHS for 5 min, and the bound iC3b was stripped off using methylamine and measured by the ELISA (Fig. (Fig.1B).1B). The level of iC3b on the ClfA+ strain was 125% higher than the level on the mutants (mean value for the two mutants; P = 0.005). This showed that expression of ClfA increased the generation of iC3b but that some C3 degradation occurred in the absence of ClfA. A decrease in the level of iC3b was also detected by Western blotting of C3 fragments solubilized from the cell surface (Fig. (Fig.1C1C).
Cleavage of C3b to iC3b on the S. aureus surface should decrease the number of alternative pathway C3-convertases on the bacterial surface and inhibit the deposition of additional C3b. Therefore, we compared the total amount of C3 fragments deposited by serum on the surface of the wild-type strain to the total amount deposited by serum on the surface of a ClfA-null strain over 60 min (Fig. (Fig.1D).1D). The differences were most striking at 1 min, when the ClfA-null strain bound 9-fold more C3 fragments than the wild-type strain (P < 0.001). The difference in C3 fragment binding narrowed over time but remained significant. Thus, the presence of ClfA appears to correlate with an overall decrease in the deposition of C3 fragment opsonins.
In order to evaluate whether the residual factor I binding to the ClfA-null strains could be nonspecific, we tested a sortase-null mutant (27) of S. aureus strain Newman (Fig. (Fig.1E).1E). The sortase-null mutant is unable to anchor any LPXTG motif cell wall-associated proteins (24). No significant difference in factor I binding was evident between the ClfA-null strain [ClfA-(2)] and the sortase-null strain (Sortase−) (P = 0.17), suggesting that the non-ClfA-dependent factor I binding is not related to covalently anchored surface proteins.
ClfAPYII is a ClfA mutant with substitutions at positions 336 and 338 [ClfA(P336A Y338S)] that does not bind fibrinogen (6), the primary function described for ClfA. In order to test the role of ClfAPYII in factor I cleavage of C3b to iC3b on the S. aureus surface, a mutant of strain Newman expressing ClfAPYII was incubated in 1% normal human serum and stripped of C3 fragments, and iC3b was measured by the ELISA (Fig. (Fig.2A).2A). The ClfAPYII-expressing strain generated 57% less iC3b than the wild-type strain (P = 0.02, based on absolute values). This suggested that the presence of ClfAPYII altered the ability of factor I in serum to cleave C3b to iC3b on the bacterial surface.
In order to further elucidate how the expression of ClfAPYII affects factor I cleavage of C3b to iC3b on the S. aureus surface, experiments were conducted using purified complement components. S. aureus strains expressing ClfAPYII or ClfA (wild type) were incubated with purified components to activate the classical pathway and bind C3b to the bacterial surface (3). The C3b-coated bacteria were then incubated with purified factor I or both purified factor I and purified factor H, C3 fragments were stripped off using methylamine, and the iC3b and total C3 fragments were measured by the ELISA (Fig. (Fig.2B).2B). The ClfA-expressing (wild-type) strain incubated with factor I alone showed increased C3b cleavage as the iC3b/C3 ratio increased (P = 0.05) compared with the control. However, the ClfAPYII-expressing strain incubated with factor I alone showed no increase in C3b cleavage compared with the control, suggesting that ClfAPYII could not act as a cofactor for factor I on the S. aureus surface. Curiously, when the C3b-coated ClfAPYII-expressing strain was exposed to both factor I and a strong cofactor, factor H, the C3b cleavage was similar to that of the control. This suggests that ClfAPYII interaction with factor I on the S. aureus surface may inhibit the normal cofactor activity of factor H.
We then tested whether recombinant ClfAPYII and recombinant ClfA differ in cofactor activity for factor I. Recombinant ClfA and ClfAPYII were incubated with purified C3b and purified factor I, and generation of iC3b was measured by the ELISA (Fig. (Fig.2C).2C). In the presence of 200 μg of recombinant ClfAPYII factor I produced 57% less iC3b than it produced in the presence of the same amount of recombinant ClfA (P = 0.01). These findings again suggest that ClfAPYII has diminished cofactor activity for factor I compared with wild-type ClfA.
In order to test factor I binding on the bacterial surface, serum factor I binding to S. aureus expressing ClfAPYII was compared with serum factor I binding to the wild-type and ClfA-null strains. The bacteria were incubated in serum, the surface proteins were stripped off by boiling the bacteria in 2% SDS, and factor I was measured by the ELISA (Fig. (Fig.3A).3A). A minimal amount of factor I was detected in the stripped surface protein supernatant for the ClfAPYII-expressing strain; the amount was 86% less than the amount for the ClfA-deficient strains (P < 0.03). This suggested either that factor I did not bind to the bacterial surface of the ClfAPYII-expressing strain or that factor I could not be stripped off the ClfAPYII strain.
In order to test these possibilities, we incubated both the ClfAPYII-expressing and wild-type S. aureus strains in purified factor I and then solubilized cell wall proteins with lysostaphin (13). The cell wall preparations were analyzed by Western blotting and probed for factor I with anti-factor I antibody (Fig. (Fig.3B).3B). The factor I content of the cell wall preparations was also measured by the ELISA (Fig. (Fig.3C).3C). The Western blot analysis showed that factor I was present in the cell wall preparations of both organisms. The ELISA results showed that there were similar amounts of factor I in the cell wall preparations of the ClfAPYII-expressing and wild-type S. aureus strains. This demonstrated that a similar amount of factor I bound to the cell wall of the ClfAPYII-expressing strain but that factor I could not be removed from the cell wall by boiling in 2% SDS, which readily removed factor I from the surface of the wild-type strain. These results suggested that ClfAPYII bound factor I but the binding properties were altered compared to those of wild-type ClfA.
In order to evaluate further the binding of factor I to ClfAPYII, we tested the binding of serum factor I to recombinant ClfA (rClfA) or recombinant ClfAPYII (rClfAPYII) in an ELISA-type assay (Fig. (Fig.3D).3D). rClfA and rClfAPYII bound serum factor I with similar half-maximal binding values (50 nM), showing that both forms bind serum factor I effectively.
In order to understand better the impact of ClfA on complement-mediated phagocytosis, ClfAPYII-expressing S. aureus was compared with the wild-type and ClfA-null strains. The bacteria were incubated with 1% normal human serum (NHS) or heated serum without complement activity and then added to purified human neutrophils at a 20:1 ratio. Bacteria were stained with acridine orange, and extracellular bacteria were quenched with crystal violet. With heat-inactivated serum minimal phagocytosis of each strain occurred, whether the absolute number of bacteria phagocytized per 100 neutrophils (Fig. (Fig.4A)4A) or the percentage of neutrophils phagocytizing bacteria (Fig. (Fig.4B)4B) was determined. In normal human serum, neutrophils phagocytized 3-fold more ClfAPYII-expressing S. aureus cells than wild-type cells (P < 0.01) and phagocytized 2-fold more ClfAPYII-expressing S. aureus cells than ClfA-null strain cells (P = 0.01). More neutrophils ingested ClfAPYII-expressing bacteria than ingested either wild-type bacteria (P = 0.02) or ClfA-null strain bacteria (P = 0.03). Although a trend toward increased phagocytosis was noticed for the ClfA-null strain compared with the wild type, the difference was not statistically significant. These findings show that ClfAPYII-expressing S. aureus is much more susceptible to complement-mediated phagocytosis than the wild-type strain or the ClfA-null mutant, suggesting that the substitutions at positions 336 and 338 of ClfA are important in modulating S. aureus susceptibility to phagocytosis.
In order to determine if complement activation and C3b deposition had any effect on the binding of factor I to the S. aureus surface, ClfA+ wild-type strain Newman was incubated in normal human serum or in serum in which complement was inactivated by heat treatment or with EDTA-GVBS−− buffer. Complement proteins were stripped, and factor I was measured by the ELISA (Fig. (Fig.5A).5A). No difference in the level of factor I was detected when samples incubated in NHS and samples incubated in complement-inactivated sera were compared (P > 0.11).
In order to test if C3 or C3b was necessary for factor I binding to ClfA, recombinant ClfA was incubated with normal human serum and with C3-depleted serum, and factor I binding was measured by the ELISA (Fig. (Fig.5B).5B). Serum factor I binding to rClfA was not significantly different in the presence and in the absence of C3 or C3b with 25% serum (P = 0.07), but with 50% serum rClfA bound 16% less factor I in the presence of C3-containing serum than in the presence of C3-deficient serum (P = 0.02). These findings suggest that C3 is not required for serum factor I binding to ClfA.
In order to determine if expression of capsular polysaccharide interfered with factor I binding to S. aureus, the well-characterized capsule type 5-expressing strain Reynolds and an isogenic capsule-deficient mutant were compared. Bacteria were grown under conditions in which capsule was expressed strongly (stationary phase) or poorly (mid-logarithmic phase) (4). Similar levels of factor I bound to the two strains under both growth conditions (Fig. (Fig.6A)6A) (P = 0.17), suggesting that neither capsule expression nor the phase of growth significantly affected factor I binding. Control experiments showed that ClfA was expressed by bacteria in both phases (Fig. (Fig.6B6B).
To evaluate if factor I could bind to clinical isolates of S. aureus, five strains in each of four categories (invasive CA-MRSA, noninvasive CA-MRSA, invasive MSSA, and noninvasive MSSA) were incubated with serum, surface proteins were stripped off, and factor I was measured by the ELISA (Fig. (Fig.7A).7A). All 20 strains bound factor I. A 3-fold difference was found when the strains which bound the most factor I were compared with the strains which bound the least factor I. We calculated that, on average, approximately 60,000 molecules of factor I were bound to each cell of the S. aureus isolates. These experiments demonstrate that all clinical strains of S. aureus bind factor I. Control experiments showed that ClfA was expressed by all clinical strains; a representative blot is shown in Fig. Fig.7B7B.
The clinical strains that bound the most factor I would be expected to cleave C3b to iC3b more effectively and to limit opsonic C3 fragment deposition on the S. aureus surface by the alternative pathway, as shown in Fig. Fig.1D.1D. Neutrophil phagocytosis experiments were performed with the two clinical strains that bound the most factor I and the two clinical strains which bound the least factor I (Fig. (Fig.7C).7C). The phagocytosis efficiency of the clinical isolates that poorly bound factor I was 1.6-fold greater than that of the strains that bound the most factor I (P = 0.003). These findings support the hypothesis that there is probably a clinically relevant association between factor I binding and virulence.
In order to test whether the presence of fibrinogen interferes with factor I binding to ClfA, we compared factor I binding in serum, where there is minimal fibrinogen, to factor I binding in plasma, which has a physiological concentration of fibrinogen (Fig. (Fig.8A).8A). Use of the anticoagulant hirudin does not affect activation of complement, which occurs in plasma stabilized with EDTA or heparin (1). The factor I binding to wild-type (ClfA+) S. aureus was greater in both plasma preparations than in serum (P ≤ 0.01). There was no difference between the factor I binding to the ClfA-null strain [ClfA-(2)] in the plasma and the factor I binding to the ClfA-null strain [ClfA-(2)] in the serum (P ≥ 0.1). These results suggest that the presence of fibrinogen increases factor I binding to wild-type S. aureus.
In order to test whether factor I bound to the S. aureus continued to cleave C3b to iC3b in the presence of a physiological concentration of fibrinogen (2 to 4 mg/ml), we compared the generation of iC3b on the surface of bacteria incubated in NHS to the generation of iC3b on the surface of bacteria incubated in hirudin plasma (Fig. (Fig.8B).8B). For the wild-type strain (ClfA+), there was no difference between the iC3b generated on the bacterial surface in NHS and the iC3b generated on the bacterial surface in hirudin plasma (P = 0.63). This suggests that factor I functions normally on the S. aureus surface in the presence of physiological concentrations of fibrinogen.
The influence of fibrinogen was also tested using purified factor I and purified fibrinogen in a solid-phase binding assay with immobilized rClfA (Fig. (Fig.8C).8C). Purified fibrinogen increased purified factor I binding to rClfA in a dose-dependent manner. This result is consistent with the results of the experiments whose results are shown in Fig. Fig.5A,5A, where factor I binding to S. aureus was increased in plasma, suggesting that fibrinogen increases the association of factor I with ClfA.
We then tested if fibrinogen in plasma was associated detectably with factor I in a solid-phase assay in which factor I was captured from plasma with an immobilized anti-factor I monoclonal antibody (Fig. (Fig.8D).8D). Factor I was readily captured from plasma, but no bound plasma fibrinogen was detected when the preparation was probed with an anti-fibrinogen antibody. We also investigated whether fixing fibrinogen to a surface caused binding by factor I (Fig. (Fig.8E).8E). Here an immobilized anti-fibrinogen antibody was used to capture fibrinogen from plasma, and bound plasma factor I was measured with an anti-factor I antibody. No plasma factor I bound to the solid-phase plasma fibrinogen.
In order to clarify whether fibrinogen is required for ClfA binding to factor I, we performed a solid-phase assay to capture factor I from serum, followed by incubation with different concentrations of rClfA or rClfAPYII (Fig. (Fig.8F).8F). Wells were coated with anti-factor I antibody to capture factor I from serum. After washing, rClfA or rClfAPYII was added, and binding was detected with anti-ClfA antibody. The rClfA and rClfAPYII bound to serum factor I in a dose-dependent manner with similar affinities, suggesting that ClfA binding to serum factor I does not require fibrinogen.
The studies described in this paper show that ClfA expression on the surface of S. aureus is associated with increased binding by factor I from serum, increased cleavage of surface-bound C3b to iC3b, and decreased C3 fragment binding to the S. aureus surface. We have shown previously that factor I-mediated cleavage of C3b to iC3b on the S. aureus surface is associated with decreased phagocytosis by human neutrophils (2). Taken together, these findings suggest that ClfA binding of serum factor I and the resultant C3b cleavage to iC3b on the S. aureus surface is a plausible mechanism contributing to S. aureus evasion of complement host defenses.
Previous studies have shown that ClfA(P336A Y338S) does not bind fibrinogen (23) and that in mouse models of S. aureus bacteremia and septic arthritis there is a consistent virulence pattern (the ClfAPYII mutant is less virulent than the ClfA-null mutant) (17). It is quite striking that complement-mediated phagocytosis of the strains shows a similar progression since ClfAPYII-expressing S. aureus is more readily phagocytized than the ClfA-null or wild-type strain. These differences in complement-mediated phagocytosis can be explained at least in part by the failure of ClfAPYII to act as a cofactor for factor I-mediated cleavage of C3b to iC3b. ClfAPYII also appears to inhibit the cofactor activity of factor H. Thus, ClfA has an important interaction with the complement system that likely contributes to the differences in virulence found in vivo that could not be attributed to ClfA interaction with fibrinogen. These findings support the hypothesis that factor I binding to ClfA results in C3b cleavage to iC3b on the S. aureus surface and contributes to immune evasion and virulence.
The presence of fibrinogen at physiological concentrations increases factor I binding to the S. aureus surface, increases factor I binding to ClfA, and does not adversely affect the cleavage of C3b on the bacterial surface. Factor I in serum does not appear to require fibrinogen for binding to ClfA or ClfAPYII (the latter is a molecule that cannot bind fibrinogen).
Previous studies have suggested that upon binding the factor H-C3b complex, factor I undergoes a conformational change that enables it to cleave C3b (8, 10). We speculate that factor I may undergo a similar conformational change upon binding of the fibrinogen-ClfA complex that enables it to cleave C3b. We also propose that the “activating” conformational change for factor I does not occur upon binding to ClfAPYII.
Based on these data, we propose the model shown in Fig. Fig.9.9. The apo form of ClfA is shown with the latching peptide emanating from the C terminus of the N3 domain unbound to the N2 domain. When the D domain of fibrinogen contacts ClfA and the gamma-chain peptide inserts into the ligand-binding trench, a conformational change occurs which results in the fibrinogen peptide being locked in place by the latching peptide undergoing beta-strand complementation with two beta strands in N2. Factor I binds the fibrinogen-ClfA complex and undergoes a conformational change to an “active” form that can cleave C3b to iC3b. ClfAPYII, in contrast, is unable to bind fibrinogen and remains in the apo form with the latching peptide free. Factor I is also able to bind ClfAPYII, but it does not undergo a conformational change to an “active” form. Thus, factor I complexed with ClfAPYII is unable to cleave C3b to iC3b. The stronger binding of factor I to ClfAPYII could be due to local changes in conformation due to the amino acid substitutions.
All 20 clinical isolates expressed ClfA and bound factor I. Clinical isolates with increased factor I binding were poorly phagocytized compared to isolates with decreased factor I binding. These data support the likely physiological relevance of ClfA-mediated factor I cleavage of C3b.
In future studies we will continue to investigate the binding and functional interaction between factor I, fibrinogen, ClfA, and ClfAPYII in order to further elucidate this relationship.
This work was supported by the Children's Hospital of the King's Daughters Research Endowment. T.J.F. acknowledges the support of a Science Foundation Ireland Principal Investigator Programme grant.
Strain JL022 was kindly provided by Jean Lee, Channing Laboratory, Harvard Medical School.
Editor: J. B. Bliska
Published ahead of print on 25 January 2010.