|Home | About | Journals | Submit | Contact Us | Français|
Streptococcus pneumoniae naturally colonizes the nasopharynx as a commensal organism and sometimes causes infections in remote tissue sites. This bacterium is highly capable of resisting host innate immunity during nasopharyngeal colonization and disseminating infections. The ability to recruit complement factor H (FH) by S. pneumoniae has been implicated as a bacterial immune evasion mechanism against complement-mediated bacterial clearance because FH is a complement alternative pathway inhibitor. S. pneumoniae recruits FH through a previously defined FH-binding domain of choline-binding protein A (CbpA), a major surface protein of S. pneumoniae. In this study, we show that CbpA binds to human FH but not to the FH proteins of mouse and other animal species tested thus far. Accordingly, deleting the FH-binding domain of CbpA in strain D39 did not result in obvious change in the levels of pneumococcal bacteremia or virulence in a bacteremia mouse model. Furthermore, this species-specific pneumococcal interaction with FH was shown to occur in multiple pneumococcal isolates from the blood and cerebrospinal fluid (CSF). Finally, our phagocytosis experiments with human- and mouse phagocytes and complement systems provide additional evidence to support our hypothesis that CbpA acts as a bacterial determinant for pneumococcal resistance to complement-mediated host defense in humans.
Streptococcus pneumoniae (the pneumococcus) is a gram-positive bacterium that causes a wide spectrum of infections, such as pneumonia, bacteremia, meningitis, otitis media and sinusitis (1). The nasopharynx of humans is the only natural reservoir for the pneumococci although other animal species can be experimentally infected with the bacterium (2). The bacterial and host determinants for the strict host tropism of S. pneumoniae have not been defined. S. pneumoniae can be frequently carried as a commensal organism in healthy adults, but causes severe infections in individuals without a fully functional immune system (1). Clinical surveys and experimental evidence in animal models have indicated the complement system is an essential element of host defense against the pneumococci (3–8). This is exemplified by the observations that patients deficient in complement proteins C2 and C3 have increased susceptibility to recurrent pneumococcal infections (9, 10).
Previous studies have also implicated several strategies used by S. pneumoniae to avoid complement attack. Pneumococcal surface protein A (PspA), a major surface protein, is able to interfere with activation of the alternative complement pathway by blocking the deposition of C3 on the pneumococcal surface (11–14). Pneumolysin, the only well-characterized pneumococcal toxin, is able to deplete complement by promoting activation of the classical complement pathway (15, 16). PspA- and pneumolysin-deficient strains of S. pneumoniae are significantly attenuated in terms of their virulence levels in mice (17, 18). A third complement evasion mechanism has been implicated in S. pneumoniae, which involves the recruitment of complement factor H (FH) by choline-binding protein A (CbpA) (19–25).
CbpA, also known as PspC (26), SpsA (27), Hic (19), or C3 binding protein (28), is a major surface-exposed protein of S. pneumoniae (29). The cbpA locus exists in all virulent strains tested thus far (30, 31). CbpA is considered a virulence factor because CbpA-deficient pneumococcal strains have attenuated capacity to colonize the nasopharynx and cause infections in the lungs and bloodstream in animal models (29, 32–34). The precise mechanisms of CbpA action in pneumococcal survival in vivo and pathogenesis are not completely understood. CbpA has been implicated as a pneumococcal adhesin based on in vitro investigations with epithelial cultures (29, 35, 36). In these studies, CbpA was shown to interact with sialic acid (29), human polymeric immunoglobulin receptor (pIgR) (35, 37), and complement C3 protein (36). In addition, CbpA has been shown to bind to free host factors, including FH (19, 20), C3 (28), secretory component (SC) (35, 37), and secretory IgA (SIgA) (27, 38). The findings from our previous studies (35, 38) and others (39) have demonstrated that CbpA only interacts with pIgR, SC, and SIgA of humans, but not the counterparts from common model animals including mouse, rat, and rabbit, suggesting CbpA as a bacterial determinant for the host tropism of S. pneumoniae. Finally, CbpA confers protective immunity against lethal challenge of virulent pneumococci in animal models (29, 30, 32, 40). CbpA is among a few pneumococcal proteins that can stimulate antibody production in humans (41, 42).
Based on extensive sequence variations in the CbpA locus, Iannelli et al. have divided the CbpA allelic variants into 11 PspC types (31). The typical CbpA alleles (types 1–6) in the majority of pneumococcal isolates consist of three N-terminal α-helical domains and are anchored to the cell wall choline via the C-terminal choline-binding domain (31, 43). In contrast, the CbpA alleles in PspC types 7–11 (referred to as nontypical CbpA hereafter) including the Hic protein possess non-typical N-terminal domains and are expressed at the cell surface via an LPXTG anchoring motif (31, 44). Consistently, the sequence homology levels between the typical and nontypical CbpA alleles are much lower (<30% sequence identity) than that among the CbpA alleles within the two CbpA groups. Our recent study has mapped the FH-binding activity to the N-terminal domain of CbpA in a type-2 strain D39 (PspC type 3) (25). Hic (PspC type 11) was the first CbpA allele shown to bind human FH, but our recent study showed CbpA of strain D39 has 23-fold or higher affinity to human FH as compared with Hic of a type-3 strain A66 (19, 25). The biological implications of these differences among the CbpA alleles on pneumococcal carriage and virulence remain unclear.
FH protects host cells from random deposition of C3b and nonspecific complement activation by inhibiting the activation of the alternative complement pathway (45). Many pathogenic bacteria including S. pneumoniae have been shown to evade complement-mediated host defense by recruiting FH to the bacterial surfaces (46–48). The FH proteins from the characterized animal species are all composed of 20 short consensus repeats (SCRs) and share a similar molecular size around 155 kDa (45). CbpA and its allelic variants bind to SCR 6–10 (23), 8–11 (24, 49), 13–15 (21), and 19–20 (49) of human FH. In addition to its antiphagocytic property (24, 50), recent studies also showed CbpA-FH interaction enhances pneumococcal adhesion to and invasion of host cells (49, 51). The FH proteins from different mammalian species have extensive sequence variations, which is exemplified by only 61% sequence identity between human (1,231 amino acids) and mouse FH (1,234 amino acids). Previous studies demonstrate that the OspE and BBA68 proteins of Borrelia burgdorferi selectively bind to human but not mouse FH (52, 53). In this study, we showed that CbpA binds only to human FH but not the FH variants from mouse and other animal species tested thus far. This species specific-interaction of S. pneumoniae with human FH is consistent with our observation that, in the mouse sepsis model, the FH-binding negative pneumococci had no significant reduction in virulence or survival defect in the bloodstream. Because humans are the only natural host for S. pneumoniae, our data suggest that the CbpA-mediated recruitment of complement FH may contribute to host tropism of this pathogen.
The capsular serotype 2 strain D39 and its isogenic derivatives ST588 and ST650 were previously described (25). Additional pneumococcal isolates from blood and cerebrospinal fluid (CSF) were provided by the Active Bacterial Core (ABC) Surveillance Program at the Center for Disease Control and Prevention (CDC, Atlanta, GA). The basic characteristics of these strains are provided in Table 1. The bacteria were routinely grown in Todd-Hewitt broth containing 0.5% yeast extract (THY) or on tryptic soy agar plates containing 3% (v/v) sheep blood.
Western blot was performed essentially as described previously (25). Briefly, recombinant CbpA2 protein with an N-terminal His tag (1 μg per lane) or bacterial cell lysates (5 μg total protein/lane) were boiled for 5 min in standard SDS-PAGE gel loading buffer (54), and subjected to electrophoresis in 10–20% Tris-HCl SDS-PAGE gels (BioRad, Hercules, CA). Protein concentrations of the fractions were determined by the BioRad protein assay reagent. CbpA2 represents the 254 amino acids of CbpA in strain D39 including the FH-binding domain (35). Serum samples were treated in a similar manner except that the final volumes of the serum-loading buffer mixtures were adjusted to equal volumes with deionized water before boiling. The proteins were blotted to polyvinylidene difluoride (PVDF) membranes (Millipore, Bedford, MA) with a semi-dry electro-transfer apparatus (BioRad) according to the manufacturer’s instructions. The blots were blocked with 5% milk (w/v) and washed three times in PBS before the detection step.
The FH proteins in human and mouse serum samples were detected with a goat anti-human FH antibody (1:2,500) (Calbiochem, San Diego, CA) and a rat anti-mouse FH antibody (1:2,000) (55), respectively. The binding of serum FH to CbpA was assessed by incubating the protein blots of recombinant CbpA2 or pneumococcal lysates with normal human (1:1,000 dilution) or mouse serum (1:100 dilution) overnight at 4 °C at dilutions. The reactivity of pneumococcal lysates to purified human FH (Sigma, St. Louis, MO; 0.4 μg/ml) and purified human SIgA (Sigma; 0.4 μg/ml) was detected as described previously (25). The levels of reactivity were visualized with appropriate secondary antibody-peroxidase conjugates by the enhanced chemiluminescence (ECL) Western Blot Kit (Pierce, Rockford, IL) according to the supplier’s instructions. A peroxidase-conjugated rabbit anti-goat IgG antibody (1:5,000; BioRad) was used as secondary antibody to detect CbpA binding to human FH at a final dilution of 1:5000. CbpA binding to mouse FH was visualized with a peroxidase-conjugated rabbit anti-rat IgG antibody (Invitrogen) at a final dilution of 1:2,000.
For dot blot, the methanol-activated PVDF membranes were coated with purified proteins (1 kg/spot) or undiluted serum samples (2 kl/spot) from various species. After the membranes were air dried, they were rinsed with water and blocked by 5% milk (w/v). Following three washes with PBS, the membranes were incubated with biotin-labeled CbpA2 at a final concentration of 0.3 μg/ml for 1 hr at room temperature. Biotin labeling of CbpA was performed using the EZ-link biotinylation kit from Pierce according to the supplier’s instructions. The bound CbpA was detected using a streptavidin-peroxidase conjugate (1:2000; Pierce) and visualized by the ECL Western blot kit.
ELISA was performed essentially as described previously (38). Direct ELISA was performed by coating the wells of 96-well plates (Nalge Nunc International, Rochester, NY) with CbpA (0.5 kg/well) by incubating overnight at 4 °C. Serial dilutions of normal human and mouse sera were added as a source of FH. The bound FH proteins was reacted with a goat anti-human FH antibody (1:1,000) (Calbiochem) and a rat anti-mouse FH antibody (1:500) (55), respectively. The levels of reactivity were detected with peroxidase-conjugated rabbit anti-goat and anti-rat IgG antibodies (BioRad; 1:5,000). Absorbance was read on a microtiter plate reader at a wavelength of 490 nm (BioRad). Competitive ELISA was performed in a similar fashion with certain modifications. Purified human FH (0.5 kg/well) was used to coat to 96-well plates. After washing and blocking steps, the wells were treated with biotin-labeled CbpA at a final concentration of 0.3 μg/ml in the presence or absence of various concentrations of human or mouse serum samples. After 1 h incubation, the wells were washed and further incubated with streptavidin-peroxidase conjugate (1:1,000 dilution). The ELISA results for direct and competitive ELISA are presented as the means of the absorbance units from triplicate wells after subtraction of background readings. The background levels were determined by measuring the absorbance of wells without serum (direct ELISA) or biotin-labeled CbpA (competitive ELISA). Two-way ANOVA was used to perform statistic analysis using the GraphPad Prism 5.0 software for Mac OS X. A P value greater than 0.05 was considered significant.
The cbpA allelic variants were amplified by polymerase chain reaction (PCR) using primers Pr588 (5’-AATGAGAAACGAATCCTTAGCAATG-3’) and Pr589 (5’-AAGATGAAGATCGCCTACGAACAC-3’) representing the highly conserved flanking sequences of the cbpA locus as described previously (31). The high-fidelity DyNAzyme EXT DNA polymerase (New England Biolab, Beverly, MA) was used to minimize the amplification errors. DNA sequences were determined by automated sequencing of the cbpA-containing PCR fragments. Sequence analyses were carried out using the DNASTAR Lasergene version 7.1 (Madison, WI). The nucleotide sequences of the cbpA alleles in the pneumococcal isolates are contained in GenBank accession numbers as listed in Table 1.
Human PMNs were isolated from venous blood of healthy individuals as described previously (56). Studies were performed in accordance with a protocol approved by the Institutional Review Board for Human Subjects, National Institute of Allergy and Infectious Diseases. Purified PMNs were suspended in RPMI 1640 medium buffered with 10 mM Hepes, pH 7.2 (RPMI/H) and kept at ambient temperature until used. Purity of PMN preparations and viability were assessed by flow cytometry (FACSCalibur, BD Biosciences, San Jose, CA). Cell preparations contained 98–99% granulocytes, of which typically 94% were neutrophils and 5% were eosinophils. Murine PMNs were isolated from femurs and tibias of adult mice as described (57). Cells were suspended in RPMI/H and kept at ambient temperature until used. Cell purity was assessed by microscopy and ~60% of the cells were mature neutrophils.
Phagocytosis assays were performed using a previously described method (58), but with modifications. S. pneumoniae strains were cultured to mid-exponential phase of growth (optical density of 0.35–0.45 at 600nm) as described above, washed, and resuspended in Dulbecco’s phosphate-buffered saline at 1.25 × 108/ml. Bacteria were labeled with 5.0 μg/ml FITC (Sigma) for 20 min at 37°C and unbound label was removed by 2 washes in DPBS. Labeled S. pneumoniae were resuspended to in RPMI/H and chilled on ice until used. PMNs (106) were combined with 107 S. pneumoniae and pooled normal human or mouse serum (10% final concentration) in wells of a 96-well microtiter plate on ice. Assays with human or mouse PMNs were performed with bacteria opsonized with serum from the same species (e.g., human PMNs were combined with bacteria opsonized with human serum). Samples were mixed gently with a pipette and rotated at 37°C for 30 or 60 min. At the desired times, samples were placed on ice and analyzed by flow cytometry. Samples were analyzed to determine the total number of PMNs with bound/ingested bacteria and then re-analyzed immediately in the presence of an equal volume of Trypan Blue (2 mg/ml in 0.15 M NaCl/0.02 M citrate buffer, pH 4.4) to measure the number of PMNs with ingested bacteria (phagocytosis). 10,000 events were collected per sample and a single gate was used to exclude debris and free bacteria (Cell Quest Pro Software, BD Biosciences). Percent phagocytosis was determined by the percentage of FL1-H (FITC) positive PMNs after quenching with Trypan Blue.
Groups of five female BALB/c mice (Taconic, Germantown, NY) were infected with S. pneumoniae strains by intravenous inoculation as described previously (32). The pneumococci were grown to middle log phase in THY (OD620 = 0.4). Bacterial cultures were centrifuged and the pellets were resuspended in PBS to estimated density according to the pre-established correlation between optical absorbance and bacterial density (CFU/ml). The bacterial suspensions were further diluted in PBS and plated on blood agar plates to confirm the concentrations of viable bacteria. Mice were intravenously infected with 0.2 ml of pneumococcal suspensions (5×106 CFU/ml in PBS) and bled retroorbitally at 0, 12, 24, and 36 h post infection. The blood samples were diluted in PBS and spread on blood agar plates to enumerate recovered bacteria. Mortality was monitored daily for 10 days post infection. The levels of bacterial density in the bloodstream are presented as CFU per ml of blood after the diluting factors are considered. All animal infection procedures were in compliance with the guidelines of the Institutional Animal Care and Use Committee.
Our previous study revealed a sequence motif at the N-terminus of CbpA in strain D39 that is responsible for the FH binding activity (25). In this study, we first determined whether the pneumococcal recruitment of FH enhances pneumococcal resistance to the complement-mediated bacterial clearance in vivo using a common mouse bacteremia model. BALB/c mice were intravenously infected with similar CFUs of strain D39 or its isogenic mutant strains ST588 and ST650. ST588 lacked the entire coding region of cbpA, whereas ST650 contains a deletion only in the FH-binding domain and retains the rest of CbpA (25). The mice infected with all three strains displayed similar bacteremia levels at five time points (0, 12, 24, 36, and 48 h) (Fig 1A). Similarly, no significant difference in survival rate was observed among the groups of mice infected with strains D39, ST588, and ST650 (Fig 1B). This is in agreement with a previous report indicating that D39 and an isogenic mutant strain lacking CbpA did not show significant differences in the mouse bacteremia model (32). We thus conclude that the FH-binding activity of CbpA does not confer a survival advantage for the pneumococci in the bloodstream of mice.
The above mouse infection experiments raised the question whether the pneumococci are able to interact with mouse FH. All previous studies on the FH-pneumococcus interaction have been carried out with purified human FH protein or normal human serum (19, 20, 22–25, 59) with one exception (60). Quin et al. have recently shown that the pneumococci recovered from the blood of mice retained mouse FH in a CbpA-dependent manner as detected by flow cytometry using an anti-human FH antibody (60). It should be noted that virtually all the FH-related available reagents are based on human FH. We first attempted to use this antibody to detect binding interaction between CbpA and mouse FH by Western blot. The antibody readily detected FH in human serum but not FH in mouse serum (data not shown). Since the normal plasma contains high concentration of FH (300 to 450 μg/ml) (45), our data indicated that this antibody does not significantly cross-react with mouse FH.
During the course of our investigation, a rat anti-mouse FH monoclonal antibody became available (55). We initially assessed the reactivity of the mouse FH antibody in recognizing FH in mouse serum. Different amounts (0.125–2 kl) of normal mouse or human serum were subjected to SDS-PAGE. As shown in Fig 2A (right panel), the antibody was able to recognize the mouse FH in mouse serum at a 1:2,000 dilution although its potency was lower than the polyclonal antibody against human FH (left panel). Four-fold more mouse serum was needed for the mouse FH antibody to yield a similar level of signal intensity to that generated with human serum and the cognate antibody. Consistent with the amino acid sequence diversity between mouse and human FH proteins (61% amino acid sequence identity), neither antibody showed detectable cross-reactivity to the non-cognate FH proteins under the same conditions (data not shown). This monoclonal antibody was subsequently used to determine the binding between CbpA and mouse FH.
In agreement with our previous study (25), the FH protein in normal human serum was capable of binding to CbpA2, a recombinant CbpA containing the intact FH-binding domain (Fig 2B, left panel). In contrast, a similar procedure failed to detect reactivity with the normal mouse serum even with the highest amount of CbpA (4 kg) used (Fig 2B, right panel). We confirmed this finding with a more sensitive and quantitative ELISA method using normal serum samples as a source of FH. While human serum FH showed a dose-dependent binding to CbpA, the same reactions with mouse serum did not result in any detectable binding activity (Fig 2C). Our separate ELISA experiments showed that the anti-mouse FH antibody was able to detect mouse serum FH with up to 5,000-fold diluted serum samples (data not shown). These initial findings suggested that CbpA has very weak binding, if any, to mouse FH. This result is reminiscent of a previous report by McDowell et al. that the BBA68 protein of B. burgdorferi binds specifically to human FH but not the mouse counterpart (53). However, the Western blot and direct ELISA data obtained with two different anti-FH antibodies were not absolutely conclusive. We next verified this finding by competitive ELISA. Human FH was coated onto 96-well plates and incubated with biotin-labeled CbpA2 in the presence or absence of human or mouse serum. Consistent with the Western blot experiment (Fig 2B), the CbpA-FH binding was blocked by human serum in a dose-dependent manner (there was ~50% inhibition by 2% human serum) (Fig 2D). In sharp contrast, mouse serum produced a marginal blocking effect on the CbpA-FH binding even at its highest concentration (100%). Together, the data obtained with complementary experimental methods demonstrated that pneumococcal CbpA allele of strain D39 binds to human FH but not mouse FH under these experimental conditions.
We further determined whether CbpA binds to the FH proteins of other animal species by dot blot. Equal volumes (2 μl) of undiluted serum samples from human, mouse, rat, rabbit, horse, and cattle bovine were each spotted in duplicate onto a PVDF membrane and probed with biotinylated CbpA. Multiple proteins were also included as positive (human FH and SIgA) and negative (BSA) controls. While the positive controls from the human origin (FH, SIgA, and serum) showed strong binding activity to CbpA, no CbpA binding signal was detected with the serum samples from mouse, rat, rabbit, horse, and bovine serum along with BSA (negative control). This reactivity pattern was reproducible in our additional experiments (data not shown). These observations have convinced us that the pneumococci interact with FH in a human-specific fashion. This finding is consistent with the fact that the pneumococci are naturally carried only by humans (2).
All virulent strains of S. pneumoniae tested thus far contain the cbpA locus (30, 31). However, there is extensive sequence variations among CbpA allelic variants from pneumococcal strains (30, 31). Since our previous experiments regarding CbpA-FH binding activity were all conducted with CbpA alleles from strains D39 and TIGR4 (25)(Figs 2 and and3),3), we were interested in whether this species-specific host-pathogen interaction occurs in additional S. pneumoniae isolates. We determined the FH-binding activity of 11 pneumococcal isolates from invasive (blood and CSF) infections by Western blot. The rationale was that FH-mediated evasion of the complement system by S. pneumoniae is most likely to operate in the invasive infections due to high concentrations of complement proteins in the bloodstream. These isolates represent 11 different capsular serotypes (types 3, 6A, 6B, 7F, 10A, 15C, 19A, 19F, 22F, 23F, and 38). None of the strains showed detectable binding to mouse FH when mouse serum was used as a source of FH at a 1:100 dilution (data not shown). When the same blot was stripped and re-probed with human serum at a 1:1,000 dilution as described for Fig 2B, all but one (ST865) of the isolates reacted with FH (Fig 4A). These results have further verified the species-specific binding between CbpA and human FH.
We further attempted to address why these CbpA alleles have different levels of FH binding activity and reasoned that sequence polymorphisms and differential expression of the CbpA alleles are two major possibilities. Variable sizes of the FH-binding bands suggested sequence variations in these CbpA alleles (Fig 4A), which could in turn affect the FH-binding. We first assessed the sequence variations of the CbpA allelic variants by determining the full cbpA-coding sequences of 10 strains. Repeated attempts to amplify the cbpA locus of strain ST873 by PCR were unsuccessful. Consistent with variable sizes of the FH-binding bands (Fig 4A), sequence analysis showed extensive sequence variations among the CbpA alleles (data not shown). The sequence variations may in part explain different FH-binding levels of the CbpA alleles. As reflected from the order of the CbpA alleles in Fig 4B, the highly reactive strains (D39, ST866, ST869, ST860, ST863, ST872, and ST861) shared higher levels of sequence homology in the FH-binding domain of CbpA (25). Consistent with this notion, the FH-binding domains of the weakly reactive strains (ST858 and ST862) have reduced sequence homology compared with that of the highly reactive strains. The CbpA allele in ST864 is an exception, which showed relative strong FH-binding but possesses weak sequence homology with the FH-binding motif of D39. Finally, the sequence information also revealed the lack of the FH binding domain in strain ST865, a type-3 CSF isolate (Fig 4B). ST865 possesses a PspC type-8 allele, which is highly similar to the CbpA allele in another type-3 strain G396. Our previous study could not detect FH-binding activity in G396 (25). This result thus explains the absence of the FH binding activity in ST865 (Fig 4A).
The variable FH-binding intensities among the pneumococcal isolates could also reflect different expression levels of the CbpA variants although an approximately equivalent amount of total proteins (5 μg) was used for each strain (Fig 4A). We further assess the CbpA variants with human SIgA since the SIgA binding site is highly conserved among many pneumococcal isolates (31, 39). All isolates except for ST865 showed positive binding to purified human SIgA. The levels of the reactivity were also somewhat variable, which was likely due to uneven expression levels of different CbpA alleles and/or sequence variations in the SIgA binding motif (Fig 4C). The lack of SIgA binding in strain ST865 is consistent with the lack of the SIgA binding site in this strain as reported for all CbpA variants from type-3 isolates tested thus far (31). Multiple reactive bands detected with human FH and SIgA in certain strains were likely to represent degradation products of CbpA because this phenomenon appeared to occur only when the reactivity was strong. Taken together, our data demonstrate that the species-specific binding of pneumococcal CbpA to human FH occurs in the majority of clinical isolates, but levels of the binding interaction may vary from strain to strain.
The human-specific pneumococcal recruitment of FH suggested that this interaction may enhance bacterial adaptation in the human host, thus contributing to the tropism of this pathogen to humans. Since our previous experiments indicated that mouse models are not appropriate to address this possibility, we approached this issue by assessing the impact of the pneumococcus-FH interaction on phagocytosis in vitro. Strains D39 and ST650 (isogenic mutant lacking the FH-binding domain of CbpA) were tested in both human- and mouse phagocytosis systems (human serum + human PMNs vs. mouse serum + mouse PMNs). Serum was used as a source of FH and other proteins of the complement system. There was in general equivalent association of D39 with human and mouse PMNs, indicating that PMN binding in these assays is mediated primarily by molecules not impacted by CbpA FH-binding activity (Fig. 5A). Although on average ST650 were associated (surface bound and ingested combined) with more PMNs than D39 in either phagocytosis system, this difference was greater in the human system and significant only with the human system (Fig 5A). There was also comparable, albeit relatively limited, ingestion of D39 in the human and mouse phagocytosis systems, suggesting that phagocytosis of S. pneumoniae occurs in part in the presence of CbpA (Fig. 5B). By comparison, phagocytosis of S. pneumoniae was more pronounced in the absence of CbpA FH-binding activity, as there were significantly more human PMNs containing ingested ST650 at the two time points tested (P < 0.01) (Fig 5B). Thus, absence of the CbpA FH-binding domain enhanced uptake of S. pneumoniae by PMNs in the human phagocytosis system, whereas levels of D39 and ST650 phagocytosis were not significantly different in the mouse system at 30 min (43.3±6.5% versus 33.5±8.2%, respectively) (Fig 5B). There were significantly more mouse PMNs with ingested ST650 than D39 by 60 min, but the difference between these two strains was greater (15.5%) in assays with human PMNs and serum (Fig. 5B). Taken together, the lack of CbpA FH-binding activity led to a more profound impact in the human phagocytosis system compared with the mouse system.
Previous in vitro studies have implicated the CbpA-mediated pneumococcal interaction with FH as an important strategy for bacterial evasion of complement-mediated host defense (24, 50, 60). However, it is unclear whether the pneumococcal recruitment of FH affects bacterial survival and virulence during pneumococcal infections. By utilizing the pneumococcal strains with deletion only in the FH-binding domain (25), we have shown that the FH-binding domain is critical for pneumococcal resistance to complement-mediated phagocytosis in a well-defined phagocytosis cell culture model (61, 62). However, the deletion of the FH-binding domain in CbpA or the entire CbpA did not significantly affect pneumococcal bacteremia or virulence in mice. This finding fully agrees with the lack of detectable binding interaction between S. pneumoniae and mouse FH in our molecular analyses. Together with the species-specific interaction between CbpA and human pIgR/SC/SIgA (35, 38, 39), our data further suggest that CbpA is a bacterial determinant for natural host tropism of S. pneumoniae to humans.
The results of the phagocytosis experiments using a mouse system are intriguing, since the lack of the CbpA FH-binding domain had an impact on pneumococcal phagocytosis (but not binding) with mouse PMNs in the presence of mouse serum (complement system). The CbpA FH-binding domain may possess another uncharacterized activity that is involved in pneumococcal interaction with mouse phagocytes. This agrees with the previous observations indicating that CbpA-deficient pneumococci are attenuated in nasal colonization and lung infection in mouse infection models (29, 32, 33). It is also possible that this domain directly or indirectly influences pneumococcal interaction with serum factors including complement C3 and other members of the FH family (or FH-like proteins). This notion is consistent with the previous reports that CbpA binds to C3 (28), which promotes pneumococcal adhesion to host cells (36). Many pathogenic bacteria have been shown to bind to FH-like proteins (45, 46). Lastly, we cannot completely exclude the possibility of a low-affinity binding between CbpA and mouse FH due to the detection limit of Western blot and ELISA, although the mouse infection data argue against this possibility. Such a low-affinity binding of CbpA to mouse FH would inhibit complement-mediated phagocytosis by mouse phagocytes. This would resolve the discrepancy between our results and observations reported by Quin et al. (60). The low-affinity binding between TCR and MHC cannot be detected by conventional biochemical binding analysis, but can be detected when both of TCR and MHC are focally enriched in cellular context (63).
Nasopharyngeal colonization and subsequent dissemination to other tissue sites including the bloodstream are the key aspects of pneumococcal pathogenesis. CbpA has been shown to contribute to nasopharyngeal colonization of S. pneumoniae in rat and mouse infection models (29, 32, 64). However, the contribution of CbpA to pneumococcal sepsis is not entirely clear. Rosenow et al. first showed that CbpA is not required for pneumococcal sepsis in an infant rat model (29). Other investigators have reported marginal effects of CbpA mutations on bacterial survival in the bloodstream (32) and virulence in mice when the animals were intravenously infected (32, 65). Conversely, other studies have recently reported that CbpA significantly contributes to pneumococcal sepsis in mice (34, 64). Our mouse infection experiments have confirmed the previous observations that CbpA is not essential for pneumococcal infection in the bloodstream of mice (29, 32). The pneumococcal mutant strain lacking the FH-binding domain or the entire CbpA protein produced a phenotype comparable to the wild-type strain. The mice infected with CbpA mutant or wild-type strains had similar levels of bacteremia at various time points post intravenous infection. Consistent with that observation, survival rates were similar for the mice intravenously infected with the wild type and isogenic CbpA mutant strains. The discrepancy among different studies regarding the contribution of CbpA to pneumococcal sepsis in mice may be explained in part by technical variations including the use of different bacterial strains, infection doses, and mouse strains.
It is well documented that S. pneumoniae is able to bind to human FH (19–25). Our molecular analysis has shown that pneumococcal CbpA specifically interacts with human FH, but the same approach did not detect obvious binding between CbpA and mouse FH. This human-specific interaction was initially identified with purified CbpA of strain D39 by Western blot analyses and ELISA. The species-specific binding between CbpA and human FH is not limited to a single pneumococcal strain because the native CbpA variants of multiple pneumococcal isolates were able to bind to human FH but not the mouse counterpart. Similarly, no detectable binding was observed between CbpA and the FH proteins of rat, rabbit, horse, and bovine. The lack of detectable binding interaction between pneumococcal CbpA and mouse FH is fully consistent with the mouse infection data of this study and previous studies (29, 32). Deleting the entire CbpA protein in strain D39 did not have any significant effect on pneumococcal survival in the bloodstream or virulence post intravenous infection, which was also observed previously by other investigators (29, 32). The molecular basis for this human-specific binding is not clear. It appears that multiple domains of the FH protein are involved in CbpA binding (21, 23, 49). It is likely that amino acid sequence difference(s) in these regions between the human and mouse FH proteins determines the binding specificity to CbpA.
Our observations in this study are not in full agreement with a recent report by Quin et al. (60), despite that both studies used the same D39 strain. These investigators reported that the pneumococci recovered from the blood after intravenous or intraperitoneal infection had significant binding to mouse FH as detected by flow cytometry with an anti-human FH antibody (60). In this study, this antibody readily recognized human FH but could not detect mouse FH in serum samples by Western blot (data not shown). A plausible explanation is that the flow cytometry method used by Quin et al. (60) may be able detect a potentially weak interaction between CbpA and mouse FH due to its higher detection sensitivity. This possibility is consistent with our observation that high concentrations of mouse serum had a marginal blocking effect on CbpA binding to human FH in our ELISA experiments although the inhibition was not statistically significant.
The noted mouse-human difference in the context of interaction with S. pneumoniae is probably a reflection of human-specific interaction between CbpA and other mucosal factors in humans. The biochemical analyses from our previous studies and others have revealed that CbpA also binds to human pIgR/SC/SIgA, but not to the counterparts of common model animals including mouse, rat, and rabbit (35, 38, 39). Because SIgA and SC represent the extracellular portion of pIgR, CbpA apparently binds to these host factors by the same mechanism (37, 38). pIgR is predominantly expressed by the mucosal epithelial cell to transport IgA to mucosal surfaces membrane (66). We (35) and others (37) have shown that CbpA-pIgR interaction promotes pneumococcal adhesion to and invasion of respiratory epithelial cells, thus suggesting this human-specific pneumococcal-host interaction enhances pneumococcal adhesion and thereby colonization in humans, although their actual biological impact has not been defined due to the lack of appropriate animal models. Likewise, it is reasonable to postulate that the species-specific pneumococcal interaction with human FH promote bacterial evasion of complement-mediated host defense in humans.
This study and others (35, 38, 39) have suggested that CbpA may be one of the bacterial factors that define pneumococcal host specificity. Humans are the only natural host for S. pneumoniae, but the underlying mechanisms for this strict host tropism remain to be elucidated. Identifying human-specific binding interactions of S. pneumoniae with FH and pIgR/SC/SIgA has provided the first line of molecular evidence that the bacterium prefers humans to other animal species. A challenging but important issue in future studies is to develop appropriate model systems to assess the in vivo impact of these human-specific pneumococcal interactions in human host. It should be noted that the CbpA-negative pneumococcal mutants are deficient in the nasopharyngeal colonization in mice and rats (29, 32, 64). It is thus possible that CbpA interacts with other unknown host factor(s), which promotes pneumococcal colonization and/or infection in mice. However, the information from this study and others (35, 38, 39) have highlighted the clear differences between humans and mice in terms of interacting with the pneumococci at the molecular level. Accordingly, cautions are well justified when future experimental findings from mice are extrapolated to human infections of S. pneumoniae.
We thank Jun Yang, Daimin Zhao, Yueyun Ma for technical assistance, and James R. Drake for valuable advice on the phagocytosis experiments. We are also grateful for the pneumococcal strains provided by the Active Bacterial Core surveillance (ABCs)/Emerging Infections Programs (EIP) Network at CDC.
This work was supported in part by the Intramural Research Program of the NIH, NIAID, and by a research grant from the National Institutes of Deafness and Other Communication Disorders (NIH/NIDCD) (DC006917).