|Home | About | Journals | Submit | Contact Us | Français|
Streptococcus sanguinis is a member of the viridans group of streptococci and a leading cause of the life-threatening endovascular disease infective endocarditis. Initial contact with the cardiac infection site is likely mediated by S. sanguinis surface proteins. In an attempt to identify the proteins required for this crucial step in pathogenesis, we searched for surface-exposed, cell wall-anchored proteins encoded by S. sanguinis and then used a targeted signature-tagged mutagenesis (STM) approach to evaluate their contributions to virulence. Thirty-three predicted cell wall-anchored proteins were identified—a number much larger than those found in related species. The requirement of each cell wall-anchored protein for infective endocarditis was assessed in the rabbit model. It was found that no single cell wall-anchored protein was essential for the development of early infective endocarditis. STM screening was also employed for the evaluation of three predicted sortase transpeptidase enzymes, which mediate the cell surface presentation of cell wall-anchored proteins. The sortase A mutant exhibited a modest (~2-fold) reduction in competitiveness, while the other two sortase mutants were indistinguishable from the parental strain. The combined results suggest that while cell wall-anchored proteins may play a role in S. sanguinis infective endocarditis, strategies designed to interfere with individual cell wall-anchored proteins or sortases would not be effective for disease prevention.
Viridans group streptococci are primary colonizers of the tooth surface and are instrumental in the development of the complex oral biofilm, dental plaque (1, 6). These mostly benign oral microbes may also act opportunistically to cause life-threatening bacterial infective endocarditis (IE) in persons with predisposing cardiac conditions. IE is thought to be preceded by endothelial damage inciting the deposition of a platelet and fibrin thrombus, referred to as a vegetation, which may then be colonized during transient bacteremia (14). Dental procedures are a known cause of bacteremia, and efforts to prevent IE have historically focused on antibiotic prophylaxis prior to dental visits. However, the value of this practice has been called into question on the basis of accumulating evidence that bacteremia resulting from daily activities is likely the cause of most IE cases (60). This observation, in addition to the risk of increased antimicrobial resistance among causative agents of IE, highlights the critical importance of developing new prophylactic measures against IE.
The first step in the development of IE, colonization, is thought to be mediated by bacterial surface-exposed adhesins, commonly referred to as MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) (45). MSCRAMMs are part of a larger, almost exclusively gram-positive surface protein class typified by covalent linkage to the peptidoglycan (PG) of the cell wall. A C-terminal cell wall-sorting signal (CWSS), typically containing an LPxTG pattern, a hydrophobic domain, and a C terminus rich in positively charged amino acids, directs extracytoplasmic processing of the precursor cell wall-anchored (Cwa) protein by a membrane-bound transpeptidase, sortase (33). The sortase responsible for anchoring the majority of Cwa proteins, sortase A (SrtA), catalyzes the formation of a covalent bond between the Thr of the LPxTG sequence and the free amino group of the cross bridge of the lipid II precursor of PG via a two-step transpeptidation reaction to link the mature protein to the cell wall (40, 46, 56).
The relevance of Cwa proteins for Streptococcus sanguinis IE has been suggested by previous studies. Experiments employing heterologous expression of the staphylococcal MSCRAMMs clumping factor A (ClfA) and fibronectin binding protein A (FnbA) in Lactococcus lactis suggest that these proteins mediate initial colonization and invasiveness, respectively, in staphylococcal IE (48, 49). Cwa protein-dependent platelet activation and aggregation functions have been suggested to be important for IE pathogenesis. Staphylococcus aureus protein A, ClpA, clumping factor B (ClpB), the serine/aspartate repeat protein SdrE, and the serine-rich protein SraP (43, 52) have been implicated in platelet interaction, and SraP has also been demonstrated to act as a virulence factor for staphylococcal IE (52). An S. sanguinis strain capable of inducing rabbit platelet aggregation has been shown to produce larger vegetations than a strain lacking this capability (22). An ortholog of SraP in S. sanguinis SK36, SrpA, was previously characterized as an accelerant of platelet activation via association with the platelet receptor GPIb (47). Taken together, these results suggest that Cwa protein-mediated platelet interactions may also play a role in S. sanguinis IE.
Among the viridans group streptococci, S. sanguinis is the most common causative agent of native-valve infection (12, 16, 51, 58), and virulence has been demonstrated for an oral isolate, SK36, in experimental models of IE (44). A random signature-tagged mutagenesis (STM) screen for SK36 virulence determinants identified genes encoding housekeeping functions, including amino acid and nucleic acid synthesis (44). No IE virulence factors that are ideal prophylactic candidates, such as MSCRAMMs, were identified in the STM screen. Therefore, we sought to determine the role of S. sanguinis Cwa proteins in the development of IE. This analysis was also extended to the predicted sortases of SK36 in order to determine the cumulative contribution of Cwa proteins. Virulence screening in a rabbit model of IE determined that individual Cwa proteins and sortases of S. sanguinis play a minimal role in competitive colonization at the onset of IE.
The strains and plasmids used are described in Table Table1.1. S. sanguinis SK36, a human oral isolate, was obtained from Mogens Kilian (University of Aarhus, Aarhus, Denmark) (29). S. sanguinis strains were routinely cultured for 2 days on agar or for 1 day in broth at 37°C in jars under reduced-O2 conditions (7.2% H2, 7.2% CO2-, 79.6% N2, and 6% O2) generated using an Anoxomat system (Mart Microbiology, Lichtenvoorde, The Netherlands). All media used for bacterial culture were purchased from Difco Inc. (Detroit, MI). S. sanguinis strains were cultured in brain heart infusion (BHI) broth or on tryptic soy broth containing 1.5% agar (TSA). S. sanguinis was cultured as previously described for transformation experiments (44). When required for selective growth of S. sanguinis, erythromycin (Em), chloramphenicol (Cm), spectinomycin (Sc), and kanamycin (Kn) were used at 10, 5, 200, and 500 μg ml−1, respectively. ElectroMAX Escherichia coli DH10B (Invitrogen) was used as a bacterial host for plasmid construction. All E. coli strains were cultured in Luria-Bertani medium or on Luria-Bertani medium containing 1.5% agar. For selective culture of E. coli, Em was used at 300 μg ml−1.
Open reading frames (ORFs) predicted by Glimmer (11) from the compiled and contiguous S. sanguinis SK36 genome sequence were searched for the characteristic C-terminal CWSS. The EMBOSS (50) program fuzzpro was used for pattern searching of protein sequences. The two LPxTG query patterns used (Table (Table2)2) were modified from that published by Janulczyk and Rasmussen (27). The 70 N-terminal amino acids encoded by fuzzpro-positive ORFs were scanned for a gram-positive signal peptide using the SignalP 3.0 server (3). Proteins positive for both a signal peptide and a CWSS were then visually inspected to confirm the presence of an L-P-x-(T/A)-(G/N) pattern. An LPxTG hidden Markov model (HMM) (5) was secondarily used for the identification of Cwa proteins (E-score cutoff, 1). Identical searches were applied to the complete genome sequences of Streptococcus gordonii strain Challis substrain CH1 (GenBank accession number CP000725), Streptococcus mutans UA159 (GenBank accession number AE014133), and the incomplete genome of Streptococcus mitis NCTC 12261, version 11.0. A variation of the LPxTG prediction pattern, N-P-[SKTAQEHLDN]-[TA]-[GN]-[EDASTV]-X(0,8)-[VIFAGTSML](17)-[RK]-[RK](2)-X(0,5)> (where each set of brackets represents alternative possibilities for one residue; underlining indicates positions at which misses are not permitted; numbers in parentheses indicate repetition of the preceding residue; numbers separated by a comma indicate a range of acceptable repetitions; X is any amino acid; and > indicates the end of the protein), was used to search for potential SrtB substrates.
A previously described, prescreened signature-tagged plasmid library (44) was used for directed mutagenesis of SK36 Cwa protein and sortase ORFs by in vitro transposition, essentially as described previously (10). All primers used in this study are listed in Table S1 in the supplemental material. SK36 was transformed with 10 μl of the transposition product mixture (44). The transposon insertion site in Cm-resistant clones was located by PCR screening with the Mout primer (2), which anneals to the transposon inverted repeat, in conjunction with flanking primers of the target gene. The intragenic location of the transposon was confirmed by DNA sequencing (see Table S2 in the supplemental material). A single mutant for each gene targeted was selected for further study.
The srtA gene was mutagenized by an overlap extension PCR strategy (25) that inserted the aad9 Sc resistance cassette of pR412 (32) between the first 29 and last 23 codons of srtA. SK36 was transformed with the final, fused PCR product to create an Sc-resistant srtA mutant. A similar strategy was used for allelic exchange mutagenesis of srtB (codons 21 through 264 were replaced with a Kn resistance cassette ) and SSA_1635 (codons 72 to 659 were replaced with aad9).
The srtA gene was cloned downstream of the strong L. lactis promoter CP25 (28) in vector pCM18 (19) via an SphI restriction site. Primers CP25UP and srtAEcoRIDO were used for amplification of CP25 and srtA from pCM18, and the PCR product was then cloned into the EcoRI site of the E. coli-streptococcal shuttle vector pVA838 to create pJFP62. Transformation of the srtA mutant with pJFP62 produced the Emr Scr strain JFP62.
The virulence of S. sanguinis strains was assessed by a previously described rabbit model of IE (15, 44) using specific-pathogen-free male New Zealand White rabbits weighing 3 to 3.9 kg. Animal procedures received IACUC approval and complied with all applicable federal guidelines and institutional policies. Signature-tagged mutants were screened as described previously (44). The inocula for STM screening ranged from 1 × 107 CFU to 2 × 108 CFU. Each strain was tested in two or three independent experiments in triplicate.
The virulence of selected strains was also investigated by competitive index (CI) analyses employing a previously described protocol (44) with some modifications. Competition experiments were conducted with SK36 or with the virulent, Em-resistant competitor JFP36 as a control (57). For the preparation of a standard CI inoculum, overnight cultures of the competitor and mutant strains grown in BHI were combined and diluted 10-fold in 14 ml BHI for an additional 3 h of growth at 37°C. Cells were washed and suspended in phosphate-buffered saline (PBS) to an optical density at 660 nm (OD660) of 0.8, which corresponds to ~108 bacteria/0.5 ml. This inoculum was injected into catheterized rabbits and plated with or without selective antibiotics to determine the mutant-to-competitor ratio. Rabbits were sacrificed 20 h postinfection. Vegetations collected from the aortic valve during necropsy were homogenized with PBS, diluted, and plated. Cm-resistant strains were evaluated by a two-layer plating technique (44). Surface plating on TSA with or without selective antibiotics was used for CI evaluation of non-Cm-resistant strains. When specified in the text, the inoculum and homogenate were sonicated at 50% power for 1.5 min in a titanium cup adaptor (BioLogics Inc., Manassas, VA) prior to plating in order to eliminate differences in chain length. The resulting colonies were enumerated for determination of the CI value, which is defined as the mutant/wild-type CFU ratio of the homogenate divided by the mutant/wild-type CFU ratio of the inoculum. When the recovery of one of two strains competed was below the limit of detection in the most concentrated dilution plated, the value 0.5 was substituted for zero at that concentration for calculation of the CI value. Statistical significance was determined by a single-sample t test comparing log-transformed CI values to a hypothetical mean of zero, with a P value of <0.05 representing significance.
For individual strain virulence analyses, 0.5 ml of PBS-washed late-exponential-phase cells were inoculated into catheterized rabbits for 20-h infections. Serial dilutions of vegetation homogenates were spread on TSA. When noted in the text, the log10 CFU recovered was divided by the vegetation weight to control for differences in the colonization-prone surface area. Significance was determined by comparing the log10 CFU or log10 CFU/g recovered for each strain in a Welch-corrected, unpaired t test, with a P value of <0.05 indicating a significant difference in bacterial recovery.
Mutanolysin extracts of late-exponential-phase SK36 cells, prepared essentially as described previously (36), were used for the production of a rabbit polyclonal antiserum (Covance, Denver, PA). The anti-SrpA monoclonal antibody (MAb) was obtained from BALB/c mouse splenic hybridoma cells immunized with live S. sanguinis SK36. Supernatants from different hybridoma cell clones were used to screen for antibody specificity. A supernatant that reacted with wild-type S. sanguinis SK36 but failed to react with an srpA mutant of SK36 produced the SrpA-specific MAb used in this study.
S. sanguinis strains cultured overnight in 20 ml BHI were centrifuged at 5,000 × g for 10 min at 4°C. Cell wall-associated proteins were prepared by mutanolysin treatment of the harvested cell pellet in the presence of raffinose and protease inhibitor, as described previously (36). The supernatant containing released proteins was collected following centrifugation at 15,000 × g for 15 min at 4°C. The remaining spheroplast pellet was washed once in the same buffer, resuspended in 1 ml PBS plus protease inhibitor, and lysed by sonication at 80% for 10 min, followed by bead beating in Lysing Matrix B tubes (Bio 101). Ten micrograms of bacterial cell fractions was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in 5% precast Tris-HCl gels (Criterion; Bio-Rad, Hercules, CA). SYPRO Ruby (Invitrogen) or silver staining (Bio-Rad) was used per the manufacturer's instructions.
Late-exponential-phase S. sanguinis cells were fixed to the wells of a 96-well Immunolon 4 HBX plate (Thermo Electron) as described previously (24, 42) and were blocked with 100 μl of Tris-buffered saline containing 0.2% Tween 20 and 2.5% bovine serum albumin (TBST-B) overnight at 4°C. SrpA was detected by addition of 50 μl of a 1:50-diluted anti-SrpA MAb in TBST-B, followed by incubation at room temperature for 2.5 h and three washes with 100 μl of Tris-buffered saline containing 0.2% Tween 20. Fifty microliters of 1:1,000-diluted alkaline phosphatase-conjugated anti-mouse immunoglobulin G was added and incubated as described above. Bound antibody was detected by addition of 100 μl of p-nitrophenyl phosphate (Kirkegaard & Perry Laboratories, Gaithersburg, MD) according to the manufacturer's instructions. Significance was determined by a Kruskal-Wallis nonparametric analysis of variance (ANOVA) with a Dunn multiple-comparison post hoc test, with a P value of <0.05 representing a significant difference. For immunodetection of whole cells, 50 μl of the anti-SK36 polyclonal antiserum diluted 1:25,000 in TBST-B was added to each well, and plates were incubated and washed as described for SrpA detection. Fifty microliters of 1:2,000-diluted alkaline phosphatase-conjugated anti-rabbit immunoglobulin G as a secondary antibody was added, followed by the p-nitrophenyl phosphate substrate. Significance was determined by ANOVA with a Tukey-Kramer multiple-comparison post hoc test.
Overnight BHI cultures were diluted 10-fold and subcultured for 3 h. Cells were harvested and washed once in cold 150 mM NaCl prior to suspension in 8 ml of 150 mM NaCl. Cells were evenly divided among four tubes, two of which were supplemented with 333 μl of n-hexadecane, followed by immediate vortexing for 1 min. The OD660 of the aqueous phase was recorded 25 min later. Significance was determined by ANOVA with a Tukey-Kramer multiple-comparison post hoc test, with a P value of <0.05 representing a significant difference.
Three sortase-like proteins of S. sanguinis SK36 were identified and annotated as SrtA (SSA_1219), SrtB (SSA_0022), and SrtC (SSA_1631) on the basis of homology to publicly available sortase sequences, transpeptidase active-site residues, and the relative positioning of sortase substrates on the chromosome (9). Sortase nomenclature was used as suggested previously (13), with SrtA denoting the putative housekeeping sortase responsible for covalent attachment of the majority of substrates, SrtB denoting a sortase with limited and variant specificity in other species (31, 34), and SrtC denoting a potential mediator of surface pilus polymerization (37, 38, 55). Both SrtB and SrtC are encoded by a subset of gram-positive species.
Scanning of the SK36 genome sequence using two variations of a previously reported C-terminal CWSS predictive pattern (27) resulted in the identification of 32 Cwa proteins (see Table S2 in the supplemental material). Secondary screening with a genus-indiscriminate HMM, LPxTG-HMM (5), identified the same 32 proteins plus SSA_1635, bringing the total number predicted to 33. SSA_1635 was not identified by pattern searching due to mismatches exceeding the permitted threshold (see Table S2 in the supplemental material). SSA_1635 is clustered with additional predicted Cwa protein genes, SSA_1634, SSA_1633, and SSA_1632, and is proximal to srtC. Genome clustering of these genes with srtC, and the identification of pilin motifs in SSA_1632 to SSA_1635 (62) (our observation), suggests that this locus may encode a surface pilus. A previously characterized Cwa protein of SK36, SrpA (SSA_0829), was identified by pattern searching and LPxTG-HMM but was not predicted to have an N-terminal signal peptide (see Table S2 in the supplemental material). The N terminus of SrpA, however, has almost perfect identity with orthologs in Streptococcus gordonii (4) and Streptococcus parasanguinis (7), which contain noncanonical signal peptides directing alternative extracytoplasmic export. No protein with the conventional SrtB substrate CWSS of NPxTG was identified for SK36, despite the identification of an SrtB homolog (62).
In total, 33 proteins with characteristic features of the Cwa protein class were identified for SK36. Comparative sequence analysis or previous characterization indicates that 18 may be adhesins (see Table S3 in the supplemental material). Two predicted Cwa proteins are potentially dual-capacity proteins with both adhesin and enzymatic capabilities, while five additional Cwa proteins are predicted enzymes with nuclease/nucleotidase, sugar hydrolase, and aminopeptidase capacities. Further descriptions of predicted Cwa proteins are provided in Table S3 in the supplemental material. Pattern searching and the LPxTG-HMM identified 22, 6, or 13 Cwa proteins for S. gordonii, S. mutans, or S. mitis, respectively. These numbers were reduced to 12, 5, or 10, respectively, when the requirement for a signal peptide was included (data not shown). Thus, the predicted protein repertoire of SK36 far exceeds those of related oral streptococci.
A targeted STM approach was employed for mutagenesis of 31 of the 33 predicted Cwa protein ORFs and the 3 sortase genes. In vitro transposition of each ORF with a unique, signature-tagged magellan2 minitransposon, followed by the transformation of SK36 with the transposition product, resulted in a 34-strain mutant library. DNA sequencing of each Cwa protein mutant confirmed that transposon insertion occurred in the first half of each ORF, and commonly in the first third of the ORF. Transposon insertion in sortase mutants was 5′ to the predicted active site in all three cases (see Table S2 in the supplemental material). Strains were designated “CWA” for Cwa protein mutants or “SRT” for sortase mutants, followed by the unique signature tag identifier for that strain.
The mutant library was screened for virulence in a previously optimized rabbit model of IE (44). Results from two experiments examining the mutant library are presented in Fig. Fig.1.1. In STM, the presence or absence of a strain is indicated by the signal intensity of the hybridized tag of that strain (21). Strains exhibiting strong, reproducible signal intensity in the inoculum and recovery blot, or a signal intensity in the recovery blot greater than that in the inoculum, were judged to have retained virulence for IE. Strains with lower signal intensity in the recovery blot than in the inoculum were considered putatively avirulent. The virulence of strains with weak signal intensities in both the inoculum and the recovery blot was judged inconclusive. The STM screen differentiated (i) 29 mutants that retained virulence in the rabbit model of IE, encompassing 28 Cwa protein mutants and the srtC mutant (SRT18); (ii) a single putatively avirulent strain, CWA12; and (iii) 4 strains of inconclusive virulence, CWA15, CWA30, the srtA mutant (SRT21), and the srtB mutant (SRT24).
CI assays were used for virulence quantitation of Cwa protein mutants with inconclusive or avirulent phenotypes by STM screening. Each mutant strain was coadministered to catheterized rabbits with a competitive Em-resistant derivative of SK36, JFP36 (57). The CI value for each strain was calculated as the ratio of mutant to JFP36 cells recovered from the infected vegetation divided by the ratio of mutant to JFP36 cells in the inoculum. CI results are summarized in Table Table3.3. In contrast to the preliminary dot blot hybridization results, CWA12 exhibited wild-type competitiveness by the CI assay. The reduced output signal in the STM study may be attributed to a reduced plating efficiency of this strain; CWA12 formed ~0.5 colony for every colony of JFP36 at an equal optical density (data not shown). Differences in plating efficiency would not be expected to have affected the results of the CI assay, because both inoculated and recovered cells were plated. CI analyses determined that the levels of competitiveness of CWA15 and CWA30 were slightly greater than or equal to that of SK36, respectively (Table (Table33).
CI assays were also used for analysis of two Cwa proteins not targeted by STM: SSA_1635 and SrpA. The mildly hypercompetitive phenotype observed for the SSA_1635 mutant (Table (Table3)3) indicates that expression of this protein does not promote colonization in the development of IE, in agreement with our findings for the strains possessing mutations in the other genes in the putative pilus locus (Fig. (Fig.1).1). SrpA was evaluated by CI comparison of a previously described mutant (VT1614) to the parental strain, SK36 (47). The mean CI value derived for the srpA mutant, 1.6, which was not significantly different from 1, demonstrates that mutation of SrpA did not reduce virulence in vivo (Table (Table33).
Reduced virulence for IE was previously reported for mutants of the srpA homologs S. gordonii DL1 hsa and S. gordonii M99 gspB in rat endocarditis models when lower coinoculum levels were tested (107 CFU for the hsa mutant; 105 CFU for the gspB mutant) (54, 61). To determine whether the inoculum size might also influence the virulence outcome for srpA, the mutant was competed against JFP36 in total inocula of 107 or 106 CFU. Testing of the 107-CFU inoculum in six rabbits in two independent studies produced a statistically insignificant mean CI of 2.1 (P = 0.23) (Fig. (Fig.2A).2A). The 106-CFU coinoculum resulted in a mean CI of 5.4 (P = 0.31) (Fig. (Fig.2A).2A). The ranges of CI values obtained with lower inocula were much greater than those observed for the 108-CFU inoculum (Fig. (Fig.2A)2A) and correspond with highly divergent bacterial recoveries from infected rabbits. The lower recovery and increased variability suggest a bottleneck effect (20) in this experiment, as will be discussed below.
The finding that no single SK36 Cwa protein mutant was defective in IE prompted us to consider the possibility that coinoculated strains might compensate for virulence defects in some mutants. This possibility can be tested by inoculating mutant and parent strains into separate animals. It was not feasible to undertake such a comparison for every mutant; therefore, we chose to focus on the srpA mutant, since mutation of srpA homologs has been reported to reduce IE virulence in other species (52, 54, 61). Five rabbits each were inoculated with 108 CFU of SK36 or the srpA mutant. Similar numbers of colonies were recovered for the two strains (means ± standard deviations [SD], 7.3 ± 0.6 log10 CFU for the srpA mutant versus 7.6 ± 0.6 log10 CFU for SK36; P = 0.43), suggesting that the competitiveness observed for the srpA mutant by CI analysis did not result from a beneficial association with SK36.
We next focused on the overall contribution of S. sanguinis Cwa proteins by analyzing SrtA, the putative major housekeeping sortase, and SrtB, the substrate(s) of which is unknown. Variation in signal intensity by STM screening (Fig. (Fig.1)1) prompted our development of new srtA and srtB mutant strains. In both cases, allelic replacement mutants were constructed by direct transformation with PCR amplicons containing antibiotic resistance cassettes fused to abbreviated 5′ and 3′ sequences of the srtA or srtB gene. The competitiveness of the mutants relative to JFP36 was determined as described for the Cwa mutants above.
Mutation of srtB had no effect on competitiveness (Table (Table3).3). In contrast, three independent CI experiments suggested a 10-fold reduction in competitiveness for the srtA mutant (P < 0.0001) (Table (Table3).3). An in vitro coculture comparison of the two strains demonstrated that the attenuation observed in vivo was not due to a deficiency in competitive growth in vitro (data not shown). However, mutation of srtA resulted in extraordinarily long chains relative to the control, JFP36, which exhibits a chain length comparable to that of SK36. The longer-chain phenotype was unique to the srtA mutant among the mutant strains examined. Concern that this difference in chain morphology might affect the outcome of in vivo comparisons prompted our reevaluation of the competitiveness of the srtA mutant with the added step of sonicating cells to create homogenous chain lengths prior to plating. The sonication parameters used did not reduce streptococcal viability in trial experiments (our unpublished observations). This experiment was also designed to include complementation analysis to determine whether any phenotype observed for the srtA mutant was attributable to loss of SrtA expression. The wild-type srtA gene under the control of a constitutive promoter was cloned into the shuttle vector pVA838 and introduced into the srtA mutant. Other strains were transformed with pVA838 to control for potential plasmid effects. An insignificant (P = 0.14) CI value of 0.7 was obtained when an srtA vector control strain was compared to SK36(pVA838) (Table (Table3).3). The complemented mutant, JFP62, possessed near-wild-type competitiveness, with a mean CI of 0.95 (Table (Table33).
The experiments described above indicated that chain length was indeed an important factor and that srtA mutation had no more than a minor effect on competitiveness when this variable was removed. Given these results, we again considered the possibility that a reduced inoculum size might be more discriminating for slight changes in virulence. The srtA mutant was competed against JFP36 using inocula of 108, 107, and 106 CFU, with sonication prior to plating as described above. All three inocula produced CI values suggestive of slightly reduced competitiveness (Fig. (Fig.2B).2B). The 108-CFU inoculum produced a CI value of 0.70, which borders on significant difference from an equal competitiveness value of 1 (P = 0.057) (Fig. (Fig.2B).2B). As with the srpA reduced-inoculum experiment, total bacterial recovery levels were generally lower with the two lower inocula, and variability was much higher. The mean CI of 0.4 resulting from the 107-CFU inoculum borders on significant difference from equal competitiveness (P = 0.05) (Fig. (Fig.2B).2B). The CI value of 0.018 resulting from the 106-CFU inoculum was not significant (P = 0.15), due to the highly varied CI values observed (Fig. (Fig.2B2B).
A final consideration was whether the wild-type competitor strain might be compensating for the srtA mutant in trans, as investigated for the srpA mutant. A single-strain infection experiment was therefore performed. Four rabbits each were inoculated with 2 × 108 CFU of the srtA mutant or 7 × 107 CFU of SK36. The mean recovery for rabbits inoculated with the srtA mutant (7.5 ± 0.2 log10 CFU) was not significantly different from that for rabbits receiving SK36 (7.8 ± 0.1 log10 CFU) (P = 0.098). These data were also analyzed to determine whether the size of the vegetation influenced the result. No difference between the srtA mutant and SK36 was observed when the CFU recovered from each rabbit was normalized by the weight of the vegetation (8.7 ± 0.3 log10 CFU/g vegetation for the srtA mutant versus 9.0 ± 0.1 log10 CFU/g vegetation for SK36; P = 0.104).
Given the relatively minor defect in virulence observed for the srtA mutant, we wanted to confirm that this strain possessed in vitro phenotypes that would be expected of a mutant lacking a “housekeeping” sortase. First, cells were examined for an altered cell wall-linked protein profile. Fractions containing mutanolysin-extracted proteins or proteins interior to the cell wall (spheroplast proteins) were collected and examined by SDS-PAGE. Spheroplast proteins were stained with SYPRO Ruby, whereas silver staining was required to visualize the mutanolysin-released proteins. Figure Figure33 shows the protein banding profiles of SK36, the srtA mutant, the complemented strain (JFP62), and the srpA Cwa protein mutant. Similar banding pattern were observed in the spheroplast fractions of all strains tested. As anticipated, mutation of srtA caused an overall reduction in the amount of cell wall-associated proteins. In contrast, mutation of the srpA gene, which would be expected to eliminate only a single Cwa protein, had no discernible effect on the overall profile. The apparent loss of multiple protein bands in the srtA mutant is the expected phenotype following loss of expression of the major sortase.
To further demonstrate altered Cwa protein cell surface localization as a result of srtA mutation, a whole-cell ELISA for SrpA was performed. Poor association of the srtA and srpA mutants with polystyrene plates was observed following overnight incubation. This reduced-adhesion phenotype was also previously documented for an S. sanguinis ATCC 10566 srtA mutant (63). Since the whole-cell ELISA requires equal cell numbers of each strain, all strains tested were fixed to the wells with a 0.25% glutaraldehyde treatment, as previously described (24, 42). Equivalent reactivity to an anti-SK36 antiserum demonstrated the fixation of similar numbers of cells of each strain (Fig. (Fig.4A).4A). The presence of SrpA on polystyrene-fixed whole cells was determined for SK36 and its isogenic derivatives by reactivity with an anti-SrpA MAb. As expected, the srpA mutant exhibited only background levels of reactivity (Fig. (Fig.4B).4B). The srtB and srtC mutants showed SrpA levels similar to those of the wild type, confirming that SrpA cell surface localization is not dependent on these sortases. The srtA mutant exhibited a significant reduction in cell surface-exposed SrpA levels, which were restored in the complemented strain. Thus, mutation of the putative major housekeeping sortase in S. sanguinis, SrtA, resulted in loss of detectable SrpA from the surfaces of whole cells.
Cell surface hydrophobicity, a property conferred by certain Cwa proteins (35), has also been linked to SrtA expression in S. sanguinis ATCC 10556 (63). To investigate this global cell surface phenotype in SK36, n-hexadecane phase partitioning was employed. As expected, the srtA mutant exhibited a significantly decreased association with hexadecane, indicating that SrtA-dependent Cwa proteins confer hydrophobicity in strain SK36. Mutation of srtB or srtC had no effect on association with hexadecane (Fig. (Fig.5).5). To the best of our knowledge, no srtB or srtC mutant of S. sanguinis or any other bacterium has been examined previously by this assay.
This is the first study we are aware of to adopt a systematic approach for comprehensive virulence analysis of Cwa proteins in a single species. This is significant because the more common approach of virulence screening by random mutagenesis does not ensure the identification of every virulence factor. Although we failed to identify any SK36 Cwa proteins as virulence factors in a prior random STM study (44), it was not clear whether this was because (i) their mutation was lethal, (ii) their expression was not required for virulence, or (iii) their genes were not mutagenized.
To elucidate the roles of S. sanguinis Cwa proteins and sortases in early IE, a combination of STM, CI, and individual inoculation analyses was applied to 33 predicted Cwa protein ORFs and 3 sortase ORFs. The STM screening strategy was the most efficient in terms of animal use, allowing for the testing of 34 mutants in a single pool and the elimination of 29 from further consideration. The remaining mutants were examined by CI analysis. The CI approach was less efficient but generated fewer artifacts and allowed for the quantitation of virulence. We have used both STM and CI assays extensively for SK36 virulence analysis (10, 18, 44, 57).
The most notable finding to emerge from these studies is that no single sortase or Cwa protein is essential for the development of IE in the rabbit model. This finding is particularly intriguing given the numerous adhesin-like Cwa proteins predicted for SK36. Of course, it is possible that S. sanguinis possesses adhesins specific for human receptors and that such interactions were missed. Nevertheless, previous studies have implicated various Cwa proteins (38, 48, 49, 52, 54, 61) and sortase A (59) as virulence factors for endocarditis in animal models. One explanation for our disparate findings might be differences in study design, including the animal model used, the inoculum size, and the duration of infection. Several prior studies have used catheterized rats to model IE (38, 48, 49, 54, 61). In our previous random STM analysis, both rat and rabbit models were assessed (44). In a trial experiment, a similar set of avirulent strains was identified in both models. However, most rats were infected with only a few strains, the identities of which varied from animal to animal. This suggested that a bottleneck effect had transpired. This phenomenon occurs when the number of cells establishing infection is too small to ensure that differences in the proportion of each strain recovered reflect differences in competitiveness rather than a sampling error (20). We chose the rabbit model of IE for the current study because bottleneck effects were rare in our prior use of this model for repeat STM analyses, resulting in reproducible outcomes (44), and because of the proven efficacy of this model for the identification of virulence factors (10, 18, 44, 52). The similarity of clinical signs and extracardiac pathologies in the rabbit model to those observed in infected humans has also been noted (14, 15).
In several previous studies, deficiencies in Cwa protein mutants were apparent only after the bacterial inocula were decreased (54, 61). We therefore used lower inocula to test the SK36 genes most likely to encode virulence factors, srtA and srpA. We found that lowering the inoculum apparently produced the same bottleneck effect seen with the rats in our earlier study. Fewer bacteria were recovered from vegetations, and these lower recoveries were associated with dramatic variations in CI values. This result is also consistent with our earlier random STM study, in which we noted that the mean recovery of bacteria from all rabbits was 1.2 × 108 CFU/animal but that six of the seven rabbits with unusually low recoveries of 105 to 106 CFU also produced highly variable results (44). This suggests that both the animal model and the standard 108-CFU inoculum used to evaluate the mutants in this study were ideal. Despite the variability resulting from the lower inocula, the results for the srtA mutant are in agreement with those obtained from the larger inoculum in suggesting the possibility of a slight decrease in the competitiveness of this strain. It is possible that the use of larger numbers of animals or a different statistical treatment for the animals in which no srtA mutant cells were recovered would have produced statistically significant results for all the inocula. Nevertheless, it is clear that if srtA mutation has an effect on virulence, it is minor.
Another variable in virulence studies is the duration of infection. In several previous studies, the impact of srtA mutagenesis was apparent only at later time points of multiple-day infections (8, 17, 59). In this regard, it should be noted that in humans, streptococcal IE presents as a chronic, subacute infection, often lasting for weeks. A longer period of virulence assessment might reveal Cwa proteins required for chronic infection, potentially involved in reseeding of the vegetation following embolization, as well as in long-term bacterial survival. The 20-h infections used in this study are intended to model early IE, because we are interested primarily in targeting the establishment rather than the progression of the disease.
We also considered that retention of virulence by Cwa protein and sortase mutants in STM and CI experiments might have been attributed to complementation by other strains in the inoculum. This possibility seemed particularly relevant, since our study focused on surface proteins in an organism capable of interbacterial interaction (26). However, individual inoculation experiments with both the srpA and srtA mutants revealed no change in virulence.
The virulence of all Cwa protein mutants presents two possibilities: either (i) Cwa proteins are not required for S. sanguinis infection or (ii) functional redundancy masks the individual contributions of Cwa proteins. As an example of the latter possibility, seven Cwa proteins with putative collagen adhesion capability have been identified in S. sanguinis SK36 (see Table S3 in the supplemental material). In STM screening, no single collagen adhesin was identified as being required for the development of IE, and a putative collagen adhesin mutant, CWA30, was determined to be of wild-type virulence by CI comparison. If collagen binding promotes S. sanguinis infection, as has been suggested for other IE pathogens (39), it might be missed in our study, given the number of Cwa proteins that potentially confer this function. Our finding that srpA was not required for IE appears inconsistent with the reduced IE pathogenesis observed for orthologous mutants in S. gordonii (54, 61) and S. aureus (52). The fact that srpA was not required for virulence in S. sanguinis IE but possesses in vitro properties similar to those of orthologous proteins (47) suggests that additional virulence factors may eclipse the need for srpA expression in S. sanguinis. It is not clear whether this is because platelet binding is not required for S. sanguinis disease or because this function is being performed by other proteins. In this regard, two additional Cwa proteins have been reported to mediate platelet adherence in S. sanguinis (23). The larger number of Cwa proteins predicted for S. sanguinis than for S. gordonii would be consistent with a greater level of functional redundancy in S. sanguinis.
It is less clear how the hypothesis of functional redundancy among Cwa proteins fits with the modest decrease in competitiveness observed for the srtA mutant, given that this mutant should be defective in the processing of almost all Cwa proteins. We confirmed that mutation of srtA resulted in the expected cellular phenotypes, including a general reduction in levels of cell wall-associated proteins, loss of a SrtA substrate from the surfaces of whole cells, and decreased cell surface hydrophobicity. However, the fate of Cwa proteins in the srtA mutant was not investigated further. Unanchored protein substrates are likely held transiently in the cell membrane in the absence of covalent attachment to PG, prior to release from the cell surface into the extracellular milieu (41). It is possible that Cwa proteins retain some function while transiently associated with the cell membrane or can reassociate with the cell surface after release, thus explaining the minor effect of srtA mutation. This possibility requires further investigation. Regardless of the explanation for our results, they suggest that determination of Cwa protein function by analysis of mutants may be problematic and that gain-of-function studies employing heterologous expression in a carefully selected host, as previously described for S. aureus adhesins (48, 49), may be a suitable alternative.
The initiation of S. sanguinis IE undoubtedly requires multiple factors in addition to Cwa protein and sortase expression. It is interesting that all of the S. sanguinis mutants we have identified thus far with severe virulence reductions (100- to 10,000-fold attenuation in CI assays) have mutations in genes predicted to encode proteins required for nutrient uptake or synthesis (10, 18, 44). This suggests that prevention of S. sanguinis IE by inhibition of nutrient acquisition, if this is possible, may be more effective than targeting adhesion. This situation may not be unique to S. sanguinis, given that the effect of Cwa protein mutagenesis on IE virulence reported for other bacteria, though statistically significant, has generally been relatively modest (38, 52, 54, 61). Regardless of their necessity for IE causation in S. sanguinis or other bacteria, several features of Cwa proteins make them attractive prophylactic targets, including the fact that these proteins are highly surface exposed. Our data indicate that a Cwa protein- and/or sortase-based strategy for the prevention of S. sanguinis IE should emphasize opsonophagocytic killing of the infecting bacteria over interference with the function of Cwa proteins. The findings presented here, combined with those of our previous random STM screen for SK36 virulence factors (44), underscore the versatility of S. sanguinis in causing IE and the challenge of designing prophylactics to prevent this disease.
We thank Nicai Zollar for excellent technical assistance. We are grateful to Ping Xu and Sankar Das for stimulating discussions. We thank Don Morrison, Bernard Martin, and Jean-Pierre Claverys for generously providing antibiotic resistance cassettes.
This work was supported by grants R01AI47841 (to T.K.), K02AI05490 (to T.K.), R21DE016891 (to H.W.), and DE16891 (to H.W.) from the National Institutes of Health.
Editor: A. Camilli
Published ahead of print on 24 August 2009.
†Supplemental material for this article may be found at http://iai.asm.org/.