Binding of Complement Factor H (fH) to Neisseria meningitidis Is Specific for Human fH and Inhibits Complement Activation by Rat and Rabbit Sera

Abstract

Complement factor H (fH), a molecule that downregulates complement activation, binds to Neisseria meningitidis and increases resistance to serum bactericidal activity. We investigated the species specificity of fH binding and the effect of human fH on downregulating rat (relevant for animal models) and rabbit (relevant for vaccine evaluation) complement activation. Binding to N. meningitidis was specific for human fH (low for chimpanzee fH and not detected with fH from lower primates). The addition of human fH decreased rat and rabbit C3 deposition on the bacterial surface and decreased group C bactericidal titers measured with rabbit complement 10- to 60-fold in heat-inactivated sera from human vaccinees. Administration of human fH to infant rats challenged with group B strain H44/76 resulted in an fH dose-dependent increase in CFU/ml of bacteria in blood 8 h later (P < 0.02). At the highest fH dose, 50 μg/rat, the geometric mean number of CFU per ml was higher than that in control animals (1,050 versus 43 [P < 0.005]). The data underscore the importance of binding of human fH for survival of N. meningitidis in vitro and in vivo. The species specificity of binding of human fH adds another mechanism toward our understanding of why N. meningitidis is strictly a human pathogen.

Neisseria meningitidis is a commensal organism that is found frequently in the throats of healthy adolescents (21, 38). On rare occasions, the organism invades the bloodstream and causes meningitis or rapidly fatal sepsis. As far as is known, the organism is strictly a human pathogen. Reliable animal models of meningococcal disease have been difficult to develop (2, 19, 48).

Considerable data indicate that serum complement-mediated bactericidal antibody confers protection against meningococcal disease (3, 5, 7, 8). Standardized protocols for group A and C bactericidal assays that use infant rabbit serum as a complement source were described by an international consortium (14, 22). These assays have been widely used as a way to infer vaccine effectiveness and as a basis for licensure of new meningococcal vaccines (4, 15). Rabbit complement was used because of greater ease of standardization, although for many years it has been known that rabbit complement augments serum bactericidal titers compared with titers measured with human complement (29, 49). The mechanism for the higher titers measured with rabbit complement has not been identified.

Recently, N. meningitidis was reported to bind complement factor H (fH) (20, 30), a molecule that downregulates complement activation. Binding of fH to the bacterial surface increased resistance of the organism to complement-mediated bacterial killing and enhanced the ability of N. meningitidis to circumvent innate host defenses. With Neisseria gonorrhoeae, binding of fH was recently reported to be restricted to human fH, which may in part explain the species-specific restriction of natural gonococcal infection (23). Our hypothesis in the present study was that binding of fH to N. meningitidis may also be specific for human fH, which could contribute to the higher bactericidal titers measured with rabbit complement than with human complement. Species-specific fH binding may also contribute to the exclusive natural pathogenicity of this organism in humans.

MATERIALS AND METHODS

Bacterial strains.

We tested the ability of human fH to regulate rat C3 deposition on two group B N. meningitidis strains, H44/76 and NZ98/254 (25, 28, 31, 32, 41, 47) and to regulate rabbit C3 deposition on group A strain A2594, group B strain H44/76, and group C strain 4243. Strain H44/76 also was used to investigate the survival of organisms in the bloodstreams of infant rats upon administration of human fH. In previous studies, this strain has been used in an infant rat bacteremia model to measure the passive protective activity of antibodies elicited by outer membrane vesicle vaccines (33, 34, 36, 37). N. meningitidis group C strain 4243 was used to measure the bactericidal activity of immune human (vaccinee) sera by using infant rabbit complement. This strain has previously been characterized and used to measure group C bactericidal titers (9, 11, 43).

The serotype (PorB), serosubtype (PorA), and sequence type of strain 4243 were 2a, P1.5,2, and 11, respectively. The corresponding classifications for strain H44/76 were 15, P1.7,16, and 32; those for strain NZ98/254 were 4, P1.4, and 42; and those for strain A2594 were 4, P1.5,9, and 5.

Serum samples.

Frozen sera that had been obtained immediately before and 1 month after vaccination of children aged 4 to 5 years who were immunized with a quadrivalent meningococcal polysaccharide or sera from adults immunized with a group C meningococcal oligosaccharide-CRM₁₉₇ conjugate vaccine were available from previous studies (16, 39). For the present study, convenience samples of sera from 69 children and 11 adults were selected based on the availability of sufficient volumes of sera for performance of the assays.

Binding of primate fH to N. meningitidis.

Chimpanzee, rhesus macaque, and baboon sera were obtained from a commercial supplier (Bioreclamation, Hicksville, NY). Human sera (positive control for fH binding) were obtained from 10 healthy adult volunteers and pooled. All sera were heated at 56°C for 30 min to prevent complement activation and binding of C3b to bacteria, which could serve as an additional ligand for binding of fH. Binding of chimpanzee, rhesus macaque, and baboon fH to strain H44/76 was measured by Western blotting as described previously, using N. gonorrhoeae and polyclonal anti-human fH antiserum (Bethyl Laboratories, NY), which also binds to fH from other primates (23). Note that heat treatment of serum for inactivation of complement does not affect the binding properties of fH (13, 27).

Heterologous C3 deposition on N. meningitidis and the inhibitory effect of human fH.

We measured deposition of rat and rabbit C3 on live bacteria from N. meningitidis strains that had been incubated with infant rat or rabbit sera (final concentration of 20% [vol/vol]). Rat or rabbit C3 that bound to bacteria was measured by flow cytometry (23), using fluorescence isothiocyanate-conjugated anti-rat or anti-rabbit C3, respectively. As a control, we used zymosan, which is a potent activator of the alternative pathway of complement (6). Zymosan (Sigma) was suspended in phosphate-buffered saline (PBS) to give a concentration of 10 mg/ml. Ten microliters of this suspension was added to infant rat or rabbit sera that contained human fH as described above (final reaction volume of 50 μl).

Complement-mediated serum bactericidal activity.

Group C bactericidal titers were measured with log-phase, washed organisms that had been grown and resuspended in Dulbecco's PBS (DPBS) containing calcium, magnesium, and 1% bovine serum albumin (BSA) (DPBS-Ca²⁺-Mg²⁺-1% BSA) as previously described (45). The complement source was 20% pooled infant rabbit sera instead of nonimmune human sera. The rabbit serum pool (Pel-Freeze, Rogers, AR) showed no detectable bactericidal activity (the number of CFU/ml of the test strain in 20 or 40% rabbit sera increased by more than 150% during a 1-h incubation, compared with the number of CFU/ml at time zero).

Bacteremia in infant rats.

Animal experiments were conducted using protocols approved by the CHORI committee responsible for overview of animal experiments. Time-pregnant outbred Wistar rats (Charles River, Portage, MI) were obtained. Four to six days after birth, the pups were randomly redistributed to the nursing mothers to prevent confounding of the different treatment groups by maternal bias. Bacteria were grown and washed as described above for the serum bactericidal assay and resuspended in DPBS-Ca²⁺-Mg²⁺ containing 10% pooled heat-inactivated infant rat sera that had been passed over a 1-ml HiTrap protein G HP column (GE Healthcare, Piscataway, NJ) to remove immunoglobulin G (IgG). An enzyme-linked immunosorbent assay revealed no detectable IgG in the depleted rat sera (>99% of the IgG was removed) and no detectable bactericidal activity in the rat sera before heat inactivation. The bacterial suspension was divided into five aliquots and placed in tubes. To three aliquots, purified human fH (Complement Technologies, Inc., TX) was added to achieve concentrations of 20, 100, or 500 μg/ml. To a fourth aliquot, purified human C1 esterase inhibitor (Complement Technologies, Inc., TX) was added (500 μg/ml) as a negative control. A fifth aliquot, with bacteria alone, served as an additional negative control (0 μg of fH). At time zero, the pups were challenged intraperitoneally (i.p.) with 100 μl containing washed, log-phase N. meningitidis bacteria (∼7 × 10³ CFU/rat), together with doses of 0, 2, 10, or 50 μg human fH or 50 μg C1 esterase inhibitor. Eight hours after the bacterial challenge, blood specimens were obtained by cardiac puncture, and aliquots of 1, 10, and 100 μl of blood were plated onto chocolate agar (Remel, Lenexa, KS). The number of CFU/ml of blood was determined after overnight incubation of the plates at 37°C in 5% CO₂.

Statistical analysis.

The respective geometric mean numbers of CFU/ml at 60 min were computed by exponentiating log₁₀ values. When the log₁₀ transformations were performed, samples below the lower limits of detection of bacteria were assigned a value of half of the lower limit (i.e., 5 CFU/ml). The significance of the differences in geometric mean numbers of CFU/ml for the four treatment groups given 50-, 10-, 2-, or 0-μg doses of fH/rat was determined by one-way analysis of variance (ANOVA). The significance of the differences in geometric mean numbers of CFU/ml obtained from animals treated with 50 μg of human fH versus 50 μg of the control human C1 esterase inhibitor was determined by Student's t test.

RESULTS

Species specificity of binding of primate fH to N. meningitidis.

Western blot analysis revealed that polyclonal anti-human fH antibody bound fH in sera from humans as well as nonhuman primates (Fig. 1A, lanes marked “serum only”). Therefore, the antibody could be used to detect binding of primate fH to the bacteria. Bacteria from strain H44/76 were incubated with heat-inactivated human or nonhuman primate sera. The cells were washed and lysed and the proteins separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. fH was detected by Western blot analysis (Fig. 1B, lanes marked “H44/76 + serum”). The cells incubated with the human sera showed the most fH binding, whereas binding of fH in chimpanzee sera was low, and binding was barely detectable with baboon or rhesus macaque sera.

FIG. 1.

Species specificity of binding of fH to N. meningitidis. (A) Binding of goat polyclonal anti-human fH to fH in primate sera (1:100 dilution) as measured by Western blot analysis. The antibody cross-reacts with fH in each of the primate sera. MW, molecular mass marker. (B) Binding of fH to N. meningitidis strain H44/76 after incubation of bacteria in different primate sera. After being washed, the pelleted bacteria were lysed in sample buffer, the lysate was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and binding of fH was detected by Western blot analysis using the antisera described for panel A.

Binding of rat C3 to meningococci is regulated by human fH.

We did not have an antibody specific for rat fH. Because rhesus and baboon fH did not bind to meningococci, we hypothesized that rat fH, which is evolutionarily more distant from human fH than fH from these other primate groups, also would not bind to the bacteria. We predicted that if this were true, there would be unregulated complement activation and deposition of rat C3 on the bacterial surface when N. meningitidis bacteria were incubated in infant rat sera. Further, the addition of human fH, which would bind to the surfaces of the bacteria, would decrease rat C3 deposition.

When N. meningitidis strains were incubated in infant rat sera, there was deposition of high levels of rat C3 (Fig. 2A). Adding 25 μg/ml of human fH to the reaction mixture resulted in >50-fold decreases in rat C3 binding. As little as 6 μg/ml of human fH gave strong inhibition of rat C3 deposition (Fig. 2B, left). To ensure that the inhibition of C3 binding by human fH did not result from nonspecific inhibition of complement activation, we performed parallel control experiments using zymosan, a potent activator of the alternative pathway of complement (6), which is not known to bind fH. At all tested concentrations of human fH up to 50 μg/ml, alternative pathway-specific complement activation and binding of rat C3 to zymosan occurred in an uninhibited manner (Fig. 2B, right). Thus, human fH was able to block rat C3 deposition on N. meningitidis, but not on zymosan, because the human fH bound directly to the bacteria.

FIG. 2.

Regulation of rat C3 binding by human fH. Live bacteria were incubated with 20% rat sera. Rat C3 deposition was measured by flow cytometric detection as described in Materials and Methods. (A) Group B strains H44/76 and NZ98/254. Dotted histograms, control (no sera); open histograms (solid line), rat sera alone with 0 μg/ml of human fH; shaded histograms, rat sera plus human fH, 25 μg/ml. (B) Group B strain H44/76. Dotted histograms, control (no sera); shaded histograms, rat sera plus different concentrations of human fH incubated with bacteria (left) or, as a control, zyomsan (right), which does not bind fH.

Treatment with human fH increases meningococcal bacteremia in infant rats.

To determine whether human fH enhanced survival of N. meningitidis in vivo, we administered doses of 0, 2, 10, or 50 μg/rat of human fH and challenged the infant rats with ∼7 × 10³ CFU of group B strain H44/76. As an additional negative control, another group of rats was given a bacterial challenge with 50 μg/rat of human C1 esterase inhibitor, which does not bind to N. meningitidis (S. Ram, unpublished data). Blood cultures were obtained 8 h after the challenge. Increasing numbers of bacteria (CFU/ml) were isolated from the blood samples of animals that had been administered increasing amounts of human fH (Fig. 3) (P < 0.02; ANOVA). The geometric mean number of CFU/ml isolated from blood samples of animals given the highest human fH dose tested, 50 μg, was higher than that for negative-control animals given 50 μg of human C1 esterase inhibitor (1,050 versus 43 [P < 0.005; t test]).

FIG. 3.

Effect of coadministration of human fH on bacteremia after i.p. challenge of infant rats. Each group consisted of 15 rats except the group given 0 μg of fH (n = 10). Bacteremia in blood cultures obtained 8 h after bacterial challenge with ∼7 × 10³ CFU/rat of group B strain H44/76 was measured. The differences between the respective geometric (Geo.) means of the four treatment groups given 50, 10, 2, or 0 μg/rat of human fH were significant (P < 0.02; ANOVA). Rats treated with 50 μg of human fH had a higher geometric mean number of CFU/ml than controls treated with 50 μg of human C1 esterase inhibitor (Est. Inh.) (P < 0.005). CI, confidence interval.

Human fH decreases rabbit C3 deposition and immune human serum bactericidal titers measured with infant rabbit complement.

When group A, B, or C strains of N. meningitidis were incubated with infant rabbit sera, we observed downregulation of rabbit C3 deposition by the addition of 25 μg/ml of human fH (Fig. 4). The downregulation of complement was specific for binding of human fH to the bacteria since deposition of rabbit C3 to zymosan was not downregulated by the addition of 25 μg/ml of human fH (data not shown) or by a twofold-higher concentration of human fH, 50 μg/ml (Fig. 4).

FIG. 4.

Regulation of rabbit C3 binding by human fH. Live bacteria of group A (A2594), group B (H44/76), or group C (4243) strains of N. meningitidis were incubated with 20% infant rabbit sera alone (open histograms shown as solid lines) or rabbit sera plus human fH, 25 μg/ml (shaded histograms). Results for the control (no sera) are shown by the dotted lines. As an additional control, zymosan was incubated with 20% infant rabbit sera plus 50 μg/ml of human fH, and no downregulation of rabbit C3 deposition was observed. Rabbit C3 deposition was measured by flow cytometric detection as described in Materials and Methods.

We measured group C bactericidal titers by using rabbit complement in pre- and 1-month-postimmunization sera from 69 children, aged 4 to 5 years, who were immunized with meningococcal polysaccharide vaccine. Nineteen children had titers of <1:16 in preimmunization sera (the lowest dilution tested) and titers of 1:64 or greater in postimmunization sera. The titers of the 19 postimmunization sera were reassayed with rabbit complement alone, rabbit complement with 25 μg/ml of human fH added, or, as a negative control, 25 μg/ml of human complement C1 esterase inhibitor. Similar assays were performed on postimmunization sera from 11 adults immunized with a group C meningococcal conjugate vaccine. The results are summarized in Fig. 5. When human fH was added to the reaction mixtures, the serum geometric mean bactericidal titer of the adults decreased more than 8-fold (P < 0.001), and that of the children decreased 60-fold (P < 0.001). In these experiments, fH may have acted both in the fluid phase and on the bacterial surface to inhibit rabbit complement-mediated bacteriolysis. When the negative-control human C1 esterase inhibitor was added, the respective geometric mean titers were not significantly different from those measured using infant rabbit complement alone (P > 0.5).

FIG. 5.

Effect of addition of human fH (25 μg/ml) on group C serum bactericidal titers measured with infant rabbit complement. The postimmunization sera were from 11 adults given a group C meningococcal conjugate vaccine or 19 children, aged 4 to 5 years, immunized with quadrivalent meningococcal polysaccharide vaccine (see Materials and Methods). The differences between the respective geometric mean titers (GMT) in human vaccinee sera measured using rabbit complement in the absence or presence of human fH were significant (P < 0.01). CI, confidence interval.

DISCUSSION

The most important new findings of this study were that (i) binding of fH to N. meningitidis was specific for human fH, (ii) the addition of human fH to infant rat or rabbit sera decreased rat or rabbit C3 deposition, (iii) administration of human fH to infant rats challenged with group B N. meningitidis increased the survival of organisms by more than 1 log₁₀, and (iv) group C human serum bactericidal titers measured with infant rabbit complement decreased 8- to 60-fold when human fH was added to the test reaction mixtures. Collectively, these observations imply a cross-species interaction of human fH with rat or rabbit C3b, which is consistent with previous observations that human fH could regulate rabbit C3b on gonococci (23). The importance of serum fH in susceptibility of humans to meningococcal disease has also been underscored by recent epidemiological observations that a single-nucleotide polymorphism (C-496T) within a presumed NF-κB-responsive element in the promoter region of the cfH gene was associated with higher serum fH levels (C/C homozygous genotype) and an increased risk of acquiring meningococcal disease (10). The present data provide further support for the hypothesis that the ability of N. meningitidis to bind human complement fH is an important mechanism that enables this organism to evade innate host defenses. Several other species-specific mechanisms also undoubtedly contribute to the ability of N. meningitidis to invade the bloodstreams of humans and cause disease. These include the ability of N. meningitidis to scavenge essential elements for growth, such as iron complexed with human transferrin (48), and the requirement of human CD46 to cross the blood-brain barrier and cause meningitis (14a).

The bactericidal antibodies measured in the human vaccinee sera (Fig. 5) were directed against the capsular polysaccharide while fH was bound to the outer membrane. Complement-mediated bacteriolysis of gram-negative bacteria requires formation of C3/C5 convertases and assembly and insertion of the membrane attack complex in the bacterial membrane. Lipooligosaccharide and opacity proteins appear to be the major acceptors for C3 and C4 on meningococci (18), even when anti-capsular monoclonal antibodies were used to activate complement (unpublished data). The present results indicated that fH bound to meningococcal fHbp was well oriented to downregulate complement activation elicited by antibodies to the group C capsule.

Meningococcal strains vary in their abilities to cause bacteremia in infant rats. In our previous studies, i.p. challenge by 10³ to 10⁴ CFU of group B strain NZ98/254 resulted in high levels of bacteremia (12, 46), while challenges with even higher inocula of strain H44/76 or Cu385 resulted in rapid clearance of the bacteria from the bloodstream (our unpublished observations). In contrast, Toropainen et al. reported high levels of bacteremia in infant rats 6 h after i.p. challenge with strain H44/76 or Cu385 (34-37). Toropainen et al. used challenge doses as high as 10⁶ CFU/rat of bacteria grown in brain heart infusion broth that were either inoculated directly i.p. into the infant rats or given after a 1:10 dilution in PBS (33). In contrast, we used lower challenge doses of 10³ to 10⁵ CFU of bacteria that had been grown in Mueller-Hinton broth and washed and resuspended in Dulbecco's salt solution containing either 1% BSA (in our previous studies) (12, 43, 44, 46) or, in the present study, 10% pooled, IgG-depleted, heat-inactivated infant rat sera. The coadministration of bacteria and nutrients from the culture broth by Toropainen et al. may have permitted replication of the bacteria in the peritoneal cavity and contributed to the ability of the organism to maintain high levels of bacteremia for the 6-h duration of their experiments in the absence of added human fH.

The infant rat bacteremia model has been used to measure passive protective activity conferred by purified antibodies or human sera from clinical vaccine trials (9, 12, 33-37, 40, 43, 44, 46). While protective activity by both purified antibodies and immune human sera that did not activate human complement-mediated bactericidal activity was observed in the model, interpretation of the in vivo passive protective activity must now take into consideration that prevention of the bloodstream infection in the rats occurred in the absence of bound fH. As a consequence, rat C3 deposition on the bacteria was not downregulated, which would have rendered the organisms more susceptible to clearance by the antibodies and rat complement than would have been the case in human blood with fH bound to the bacterial surface.

The essential role of complement in host defense against N. meningitidis was first reported nearly 30 years ago (24). Twenty-five years ago, Zollinger and Mandrell reported that serum bactericidal titers measured with rabbit complement were much higher than the respective titers measured with human complement (49), an observation confirmed by subsequent studies (29). Our data (Fig. 5) showed that the addition of human fH decreased group C bactericidal titers measured with rabbit complement in sera from immunized children or adults. Thus, the ability of meningococci to selectively bind to human fH may be one reason for the higher serum bactericidal titers measured with nonhuman complement. Note that at the high dilutions of human sera tested in our study (≥1:64), human fH in the test sera would be expected to be limited. However, at lower dilutions, concentrations of human fH in the heat-inactivated human sera would be expected to be sufficient to downregulate rabbit C3, which may contribute to observed “prozones” with rabbit complement (i.e., lack of bactericidal activity at low dilutions of human test sera but presence of bactericidal activity at higher serum dilutions when rabbit C3 activation is no longer downregulated by human fH).

Serum bactericidal titers measured with rabbit complement have been correlated with the effectiveness of meningococcal conjugate vaccination introduced to large populations (1, 3, 5). Indeed, a titer of 1:8 or greater has been correlated with long-term protection against disease (1). However, many of these sera would lack bactericidal activity if tested with human complement (29). Thus, the correlations observed between vaccine effectiveness and serum bactericidal titers measured with rabbit complement may not totally reflect the actual mechanisms by which the vaccine-induced antibodies conferred protection. For example, the positive titers measured with rabbit complement could be a surrogate for alternative mechanisms of clearing of N. meningitidis when human complement is present and serum antibody concentrations or quality is insufficient to elicit bactericidal activity but is sufficient to support opsonophagocytosis (26, 42).

For the infant rat challenge study (Fig. 3) and the human vaccinee serum bactericidal titers measured with rabbit complement, we chose human C1 esterase inhibitor as a negative control because we could not detect binding of C1 esterase inhibitor to meningococci. Also, adding 50 μg/ml of human C1 esterase inhibitor to the sera would only modestly increase overall C1 esterase concentrations in the fluid phase. The lack of an effect of the C1 esterase treatment of the infant rats on the level of bacteremia or on rabbit complement-mediated bactericidal titers highlighted the importance of binding of complement inhibitors to the bacterial surface to prevent bacterial killing. Alternatively, it is possible that human C1 esterase inhibitor does not regulate rabbit complement, but the human inhibitor has been shown to inhibit rat and mouse classical pathways (50), and pig C1 esterase inhibitor showed broad species specificity (17).

In conclusion, our results demonstrating species specificity of binding of human fH to N. meningitidis and the ability of human fH, but not rat or rabbit fH, to downregulate complement activation and bactericidal activity underscore the importance of binding of human fH on survival of N. meningitidis in vitro and in vivo. The species specificity of binding of human fH adds another mechanism toward our understanding of why N. meningitidis is strictly a human pathogen.