Role of σ^B in the Expression of Staphylococcus aureus Cell Wall Adhesins ClfA and FnbA and Contribution to Infectivity in a Rat Model of Experimental Endocarditis

Abstract

Isogenic Staphylococcus aureus strains with different capacities to produce σ^B activity were analyzed for their ability to attach to fibrinogen- or fibronectin-coated surfaces or platelet-fibrin clots and to cause endocarditis in rats. In comparison to the σ^B-deficient strain, BB255, which harbors an rsbU mutation, both rsbU-complemented and σ^B-overproducing derivatives exhibited at least five times greater attachment to fibrinogen- and fibronectin-coated surfaces and showed increased adherence to platelet-fibrin clots. No differences in adherence were seen between BB255 and a ΔrsbUVWsigB isogen. Northern blotting analyses revealed that transcription of clfA, encoding fibrinogen-binding protein clumping factor A, and fnbA, encoding fibronectin-binding protein A, were positively influenced by σ^B. σ^B overproduction resulted in a statistically significant increase in positive spleen cultures and enhanced bacterial densities in both the aortic vegetations and spleens at 16 h postinoculation. In contrast, at 72 h postinoculation, tissues infected with the σ^B overproducer had lower bacterial densities than did those infected with BB255. These results suggest that although σ^B appears to increase the adhesion of S. aureus to various host cell-matrix proteins in vitro, it has limited effect on pathogenesis in the rat endocarditis model. σ^B appears to have a transient enhancing effect on bacterial density in the early stages of infection that is lost during progression.

Staphylococcus aureus is a major human pathogen with the capacity to cause a broad spectrum of diseases including native and prosthetic valve endocarditis (34). Its ability to colonize vascular tissue is thought to occur via ligand-adhesin interactions between the organism's surface determinants and host proteins at sites of endovascular injuries and prosthetic materials (8, 11, 14, 40). S. aureus expresses a variety of distinct surface proteins, termed MSCRAMMs (for “microbial surface components recognizing adhesive matrix molecules”), which allow the pathogen to bind to host extracellular matrix proteins and establish a focus of infection (17, 21, 38, 43). Among the MSCRAMMs are the clumping factors ClfA and ClfB (35, 36, 37, 42), which modulate bacterial adhesion to fibrinogen, and the fibronectin-binding proteins FnbA and FnbB, which mediate binding to fibronectin (15, 16, 17, 26) and to a lesser extent fibrinogen (53). Fibrinogen is a large (340-kDa) protein found predominantly in blood. It is also the most abundant host protein in endothelial lesions (13), as well as the major blood protein deposited on newly implanted biomaterials (52). Fibronectin is a dimeric glycoprotein, which forms a significant component of the conditioning layer that coats many implanted medical devices such as heart valves and catheters. Both the clumping factors and the fibronectin-binding proteins influence the pathogenicity of S. aureus in various infection models (14, 32, 40, 44, 46, 48, 52). It has been shown previously that σ^B significantly influences the clumping of S. aureus in the presence of soluble fibrinogen or fibronectin in vitro (3, 41).

In the present study, we used a series of isogenic strains to characterize the effects of σ^B activity on the capacity of S. aureus to adhere to immobilized fibrinogen, fibronectin, and platelet-fibrin clots. Additionally, we examined the effects of σ^B on S. aureus infectivity in a rat model of experimental endocarditis (14, 22, 40).

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The bacterial strains and relevant phenotypes are listed in Table 1. Microorganisms were grown in tryptic soy broth (Difco Laboratories, Detroit, Mich.) or Luria-Bertani medium, with aeration at 37°C and 200 rpm. For both in vitro adhesion experiments and in vivo inoculum preparation, overnight cultures of bacteria were diluted 1:100 in fresh medium and grown to an optical density at 600 nm of 2.0, where σ^B was shown to be highly active (2). Where indicated, the media were supplemented with 5 μg of erythromycin per ml or 10 μg of tetracycline per ml. The presence of antibiotics did not alter the adherence phenotype in vitro, and the mutants grew as well as the parental strain did.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant genotype and phenotypea	Source or reference
Strains
E. coli
XL1Blue	recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F+proAB lacldZΔM15 Tn10 (Tcr)]	Stratagene
S. aureus
BB255	rsbU; low σB activity	1
IK181	BB255 ΔrsbUVWsigB; Emr	30
GP268	BB255 rsbU+; Tcr	19
MB138	BB255 rsbW7 (Am); Emr (high level of σB activity)	2
DU5880	8325-4 clfA1::Tn917; Emr	35
Plasmids
pAC7	Cmr, expression plasmid containing the PBAD promoter and the araC gene	45
pAC7-sigB	Cmr, 767-bp PCR fragment of the sigB ORF from strain COL into pAC7	23
pSB40N	Apr, promoter probe plasmid	29
pSB40NclfAp	Apr, 500-bp chromosomal fragment covering the clfA promoter region of COL into pSB40N	23
pSB40NfnbAp	Apr, 250-bp PCR fragment covering the fnbA promoter region of BB255 into pSB40N	This study

^aAbbreviations: Ap^r, ampicillin resistant; Cm^r, chloramphenicol resistant; Em^r, erythromycin resistant; Mc^r, methicillin resistant; Tc^r, tetracycline resistant; ORF, open reading frame.

Northern blot analyses.

Total RNA was isolated as described by Cheung et al. (7). An 8-μg portion of total RNA from each sample was electrophoresed through a 1.5% agarose-0.66 M formaldehyde gel in morpholinepropanesulfonic acid (MOPS) running buffer (20 mM MOPS, 10 mM sodium acetate, 2 mM EDTA [pH 7]). RNA was blotted onto a positively charged nylon membrane (Roche, Basel, Switzerland) with a vacuum blotter (Pharmacia, Uppsala, Sweden). The intensity of the 23S and 16S rRNA bands stained with ethidium bromide was verified to be equivalent in all the samples before transfer. Labeling and hybridization were done by using the digoxigenin labeling and detection kits as specified by the manufacturer (Roche). The following primers were used to generate the digoxigenin-labeled DNA probes by PCR labeling: clfA for, 5′-CGATTGGCGTGGCTTCAGTGC-3′; clfA rev; 5′-CGTTGTTGAAACATTTTCCGC-3′ (nucleotides 345 to 365 and 806 to 826 of accession no. Z18852); fnbA for, 5′-GGGATGGGACAAGATAAAGAAGC-3′; fnbA rev., 5′-ACGACACGTTGACCAGCATG-3′; (nucleotides 174924 to 174902 and 174366 to 174347 of accession no. AP003137.2).

Construction of plasmid pSB40NfnbAp.

For pSB40NfnbAp, 250 bp of the fnbA promoter region of BB255 was synthesized by PCR using the upstream primer 5′-GCGGATCCGTTTTAAATTAGATAATGATG-3′ (the BamHI linker is underlined) and the downstream primer 5′-CGCCTCGAGCCCTTTAAATGCAAAATTCATT-3′ (the XhoI linker is underlined). The PCR product was digested with BamHI and XhoI and cloned into promoter probe plasmid pSB40N (29) to obtain pSB40NfnbAp. Sequence analysis confirmed the identity of the insert. Plasmid pSB40NfnbAp was subsequently transformed into Escherichia coli XL1Blue containing either plasmid pAC7-sigB or pAC7 (23). The two-plasmid testing was performed as described by Homerova et al. (23).

Functional expression of fibrinogen and fibronectin activity.

Adherence of S. aureus cells to solid-phase fibrinogen or fibronectin was measured by two methods. In the first method, flat-bottom microtiter plates (MaxiSorp; Nunc, Roskilde, Denmark) were coated with serial dilutions (in phosphate-buffered saline PBS) of either human fibrinogen (Sigma, Buchs, Switzerland) or fibronectin (Sigma) and blocked with 2 mg of bovine serum albumin per ml in phosphate-buffered saline (PBS). Then 50 μl of bacterial cells, corresponding to 5 × 10⁸ CFU, was added to each well, and the mixtures were incubated at 37°C for 2 h. To monitor the effects of proteases on S. aureus adherence, cells were precultured and assayed in either the presence or absence of 0.4 U of the global protease inhibitor α₂-macroglobulin (Roche Diagnostics, Rotkreuz, Switzerland) per ml. Adherent cells were fixed with warm air (60°C) for 30 min and stained with 0.5% (wt/vol) crystal violet for 45 min. After addition of 200 μl of citrate buffer (pH 4.3) and incubation for 45 min at room temperature, the absorbances were read in an enzyme-linked immunosorbent assay reader at 570 nm. Each batch of assays included at least one control strain as well as blank wells.

In a second assay the fibrinogen- and fibronectin-binding capacities of the BB255 derivatives were determined by direct binding of washed cells (10⁴ CFU) to solid-phase fibrinogen and fibronectin in six-well plates by using a method previously described by Kupferwasser et al. (31).

FACS analysis.

The abundance of ClfA on the cell surface was measured by fluorescent-activated cell sorter (FACS) analysis using purified F(ab')₂ fragments from anti-ClfA rabbit polyclonal immunoglobulin G (kindly provided by T. Foster, Dublin, Ireland) as described previously (49). Briefly, S. aureus cells were washed, suspended in PBS containing 10% fetal calf serum-antibody solution (10 μg of antibody/ml), and incubated for 2 h at 4°C. The cells were then washed twice with PBS-10% fetal calf serum, suspended in PBS-10% fetal calf serum-antibody solution containing F(ab')₂ fragments of a fluorescein isothiocyanate-labeled swine anti-rabbit antibody (DakoCytomation, Glostrup, Denmark), incubated for 1 h at 4°C, washed, fixed with 4% formaldehyde solution for 30 min at 4°C, and resuspended in PBS-10% fetal calf serum before being subjected FACS analysis. S. aureus DU5880 (35), a clfA-negative strain, was used as control.

Attachment to human platelet-fibrin clots in vitro.

Platelet-fibrin clots were produced by pouring 0.5 ml of plasma into 30-mm-diameter cell culture plates (Corning Coster, Corning, N.Y.) in the presence of 100 μl of bovine thrombin solution (5,000 National Institute of Health units per ml) (Diagnotec AG, Liestal, Switzerland) and 100 μl of a 0.2 mM CaCl₂ solution. The clots were dehydrated overnight at 30°C and kept at 4°C until used. Bacterial adherence was determined by adding 2 ml of saline containing 10⁴ CFU per ml to the wells and agitating the plates for 3 min at 120 rpm on a rotating incubator. After gentle decantation of the fluid, the clots were washed twice for 5 min with 2 ml of PBS to remove nonadherent bacteria. They were then overlaid with 3 ml of Columbia agar, which was allowed to solidify prior to incubation at 37°C. The number of adherent bacteria that gave rise to colonies was determined after 24 h of incubation and expressed as a multiple of adherent organisms relative to the original inoculum (adherence ratio = [number of adherent bacteria/inoculum size] × 10⁴). The statistical significance of the different attachment of the test organisms to platelet-fibrin clots was evaluated by the one-way analysis of variance with Bonferroni's correction for multiple group comparisons.

Experimental endocarditis.

Catheter-induced vegetations of the aortic valve were produced in rats as previously described (22, 40). Groups of animals were inoculated 24 h after catheterization by intravenous injection of 0.5 ml of saline containing increasing numbers of organisms from the late exponential growth phase. The animals were sacrificed 16 h after the bacterial challenge, and quantitative vegetation and spleen cultures were performed. Vegetation and spleen tissues were serially diluted and plated onto both antibiotic-free tryptic soy agar plates and tryptic soy agar plates containing the appropriate antibiotic, allowing characterization of the stability of all mutants after animal passage. In a second set of experiments, animals were sacrificed 72 h after the bacterial challenge and vegetation and spleen cultures were processed as described above. Statistical differences comparing the rates of aortic valve and spleen infections were evaluated by the χ² test with the Yates correction. Differences between median bacterial densities in infected tissues were analyzed by the Mann-Whitney test. Data were considered significant when P values were <0.05 by the use of two-tailed significance levels.

RESULTS AND DISCUSSION

Effect of σ^B on clfA and fnbA expression.

Microarray-based analysis of the transcriptional profiles of three genetically distinct S. aureus strains indicated that clfA and fnbA were positively influenced by σ^B activity (4). In those studies, clfA transcription appeared to be strongly up-regulated by σ^B during later growth stages whereas no significant differences in clfA expression were observable during early growth. We confirmed these findings by Northern blot analyses. As shown in Fig. 1, the clfA transcript was produced in a σ^B-dependent manner. A 2.9-knt transcript was strongly produced in the rsbU-complemented strain, GP268, but was very weak in the isogenic ΔrsbUVWsigB mutant, IK181 (Fig. 1A). In addition to the σ^B-dependent transcript, a larger clfA-encoding transcript (4.5 knt) was observed during early growth in both GP268 and IK181, suggesting that basal transcription of clfA is governed in S. aureus by a σ^B-independent process during early growth stages. The finding that clfA expression increased with growth is striking and contradicts the common paradigm that MSCRAMM-encoding genes are expressed predominantly during the early growth stages.

FIG. 1.

Northern blot analyses of clfA and fnbA. Total RNAs of strains GP268 (BB255 rsbU⁺) and IK181 (BB255 ΔrsbUVWsigB) grown in LB medium and harvested at the time points indicated were probed with clfA (A) and fnbA (B), respectively. Relevant transcripts and their sizes are indicated.

In contrast to clfA, microarray analysis (4) indicated that the presence of a functional sigB operon promotes fnbA transcription mainly during the early growth stages but not in later stages. Northern blot analysis identified two distinct fnbA-encoding transcripts (Fig. 1B), with the shorter transcript (3.1 knt) being σ^B dependent and the larger transcript (3.7 knt) being independent of σ^B activity. The shorter transcript was observed only in GP268 during early growth, while the larger transcript was detectable in both BB255 derivatives analyzed until mid-log phase, corroborating the microarray results (4).

Two-plasmid testing of the clfA and fnbA promoter regions.

We used a two-plasmid system, which has proved useful in identifying promoters recognized by RNA polymerase containing S. aureus σ^B (23), to test whether the fnbA promoter region that gives rise to the 3.1-knt transcript is recognized by σ^B. The clfA promoter, which was already identified to be σ^B dependent by this method, was used as positive control (23). E. coli strains harboring plasmids pAC7sigB and pSB40NclfAp yielded dark blue colonies upon selection on LBACX-ARA plates, confirming that there is a physical interaction between the S. aureus σ^B-E. coli RNA polymerase holoenzyme hybrid and the clfA promoter region. However, no such interaction was observable for the E. coli transformant harboring plasmids pAC7sigB and pSB40NfnbAp, suggesting that σ^B does not directly mediate fnbA transcription. Interestingly, E. coli cells harboring pSB40NfnbAp and the control plasmid, pAC7, yielded light blue colonies on LBACX- ARA plates, indicating that the fnbA promoter was recognized by the σ^A-containing E. coli RNA polymerase holoenzyme. These findings, together with the transcriptional profile of fnbA during in vitro growth (4) and the lack of a significant σ^B consensus promoter sequence (23) in the upstream region of the fnbA open reading frame, strongly suggest that the positive effect of σ^B on fnbA transcription is most probably indirect and occurs by an as yet unidentified mechanism. It is known that fnbA transcription is enhanced by SarA (12,54, 55) and down-regulated by agr (47, 55). It has been suggested that σ^B activity has a positive effect on the expression of the sarA locus (3, 4, 18), as well as a negative effect on the expression of the agr locus (3, 4, 24). Thus, the two global regulators may, at least in part, account for the up-regulatory effect of σ^B activity on fnbA transcription. However, the mechanisms by which σ^B modulates SarA production are not completely understood (3, 4, 9, 10, 18, 24).

σ^B dependence of surface-bound ClfA.

To determine whether the σ^B-dependent increase in clfA expression correlates with an increase in the amount of surface-bound ClfA, we used FACS analysis to determine the ClfA abundance on the surfaces of a series of BB255 derivatives with differing abilities to produce σ^B activity. As indicated in Fig. 2, FACS shifts were clearly observed for the rsbU-complemented strain, GP268 (Fig. 2C), and the BB255 derivative, MB138 (Fig. 2D), with high σ^B activity due to an amber mutation within the gene encoding the anti-σ^B factor RsbW (2). In contrast, BB255 (Fig. 2A) and its ΔrsbUVWsigB mutant, IK181 (Fig. 2B) demonstrated very slight FACS shifts. A clfA-negative S. aureus strain, DU5880, was negative by FACS analysis (data not shown), indicating that the shifts observed for the BB255 derivatives were ClfA specific.

FIG. 2.

Histograms of ClfA expression on the surface of S. aureus cells. FACS analysis was performed with both irrelevant antibodies (gray areas) and purified anti-ClfA F(ab′)₂ fragments (white areas). (A) BB255 (rsbU); (B) IK181 (BB255 ΔrsbUVWsigB); (C) GP268 (BB255 rsbU⁺); (D) MB138 [BB255 rsbW7(Am)].

Functional expression of fibrinogen and fibronectin activity.

The findings that clfA and fnbA expression are positively influenced by σ^B raises the question whether and how σ^B affects S. aureus adherence to the host cell matrix proteins fibrinogen and fibronectin. The BB255 derivatives were therefore analyzed for their abilities to attach to fibrinogen- or fibronectin-coated surfaces in vitro (Fig. 3). GP268 and MB138 exhibited at least five times greater attachment to either fibrinogen-coated (Fig. 3A) or fibronectin-coated (Fig. 3B) surfaces than BB255 did. Moreover, the binding capacities correlated with σ^B activity; the greatest binding was observed with the σ^B overproducer, MB138, and the lowest binding was observed with the ΔrsbUVWsigB mutant IK181. No obvious differences in adherence were detectable between BB255 and IK181, corroborating previous findings suggesting that BB255, although producing a significant amount of σ^B, is unable to activate this alternative transcription factor due to its inability to generate the positive regulator of σ^B activity, RsbU (3, 19).

FIG. 3.

Bacterial adherence of BB255 and its derivatives to increasing concentrations of immobilized fibrinogen (A) and immobilized fibronectin (B). Strains used were BB255 (squares), IK181 (BB255 ΔrsbUVWsigB, circles), GP268 (BB255 rsbU⁺, triangles), and MB138 [BB255 rsbW7(Am), diamonds].

Similar results were observed when bacterial binding to solid fibrinogen and fibronectin was determined by a method described by Kupferwasser et al. (31). The numbers of CFU bound to solid fibrinogen and fibronectin were again significantly increased (P < 0.05) in both GP268 and MB138 compared to those in BB255 and IK181 (Fig. 4). Both experimental outcomes are consistent with previous findings showing a positive effect of σ^B on the attachment of S. aureus to either soluble fibrinogen (41) or soluble fibronectin (3).

FIG. 4.

Bacterial adherence of BB255 and its derivatives to immobilized fibrinogen (50 μg/ml) (A) or immobilized fibronectin (10 μg/ml) (B). Adherence was expressed as the percentage of bacteria bound to the immobilized host cell matrix proteins from an initial inocolum of 10⁴ CFU.

Effect of the universal protease inhibitor α₂-macroglobulin on the adherence phenotype of the BB255 derivatives.

Karlsson et al. (27) reported that in sarA mutant cells, which produced large amounts of the four major extracellular proteases of S. aureus, the levels of cell-bound fibronectin-binding proteins A and B were low in comparison to those in wild-type cells. This was found to occur independently of fnbA or fnbB transcription. More recently, the same authors showed that extracellular protease production among clinical isolates of S. aureus varied mainly due to differences in sarA expression. They further indicated that the level of σ^B-dependent expression of sarA determined the level of protease production; there was an inverse correlation between σ^B activity and protease activity (28). Because synthesis of extracellular proteases is activated by agr (5, 25, 33) and because agr expression is negatively regulated by σ^B activity (3, 4, 24), it is reasonable that σ^B activity down-regulates extracellular protease production in an indirect manner. This σ^B-mediated decrease in extracellular protease levels is likely to increase the amount of wall-anchored fibronectin-binding protein and thus may be, at least for fibronectin, the primary reason for the observed increase in the adherence of sigB⁺ strains. To determine whether σ^B-SarA-mediated differences in protease activity might account for the adherence differences observed between sigB⁺ and sigB mutant strains, we repeated the adherence experiments in the presence of the universal protease inhibitor α₂-macroglobulin. The fibrinogen and fibronectin adherence profiles obtained in presence of α₂-macroglobulin were almost identical to those shown in Fig. 3. This finding does not agree with the recent findings by Xiong et al. (55), who observed a 30% increase in fibronectin-binding activity for strains grown in the presence of α₂-macroglobulin compared to strains grown in its absence. However, Xiong et al. noted that the predominant mechanism dictating phenotypic fibronectin binding seems to occur at the level of fnbA transcription. Our findings support this hypothesis and indicate that the differences in adherence observed between sigB⁺ and sigB mutant strains were not likely to be due to σ^B-SarA-mediated alterations in the protease activity of the respective strains but were caused by the increased σ^B-dependent expression of clfA and fnbA. Furthermore, fibrinogen- and fibronectin-binding protein homologues, such as ClfB and FnbB, whose expression is not considered to be affected by σ^B activity, were apparently not able to mask the strong decrease in adherence to fibrinogen and fibronectin observed for the σ^B-activity-defective strains, illustrating the importance of ClfA and FnbA for the capacity of S. aureus to adhere to these host cell matrix proteins.

Adherence of S. aureus to platelet-fibrin clots.

Since cardiac vegetations contain numerous proteins and platelet factors, which were not present in the in vitro ligand assay described above, the adherence experiments were repeated with platelet-fibrin clots, which more closely mimic cardiac vegetations of recovering damaged valves (39). Adherence of the σ^B-activity-producing strain GP268 to platelet-fibrin clots was higher than that of BB255 and its ΔrsbUVWsigB mutant, IK181, and was further increased in the σ^B overproducer MB138 (P < 0.05), indicating a positive correlation of σ^B activity with the capacity of S. aureus to adhere to platelet-fibrin clots (Fig. 5).

FIG. 5.

In vitro adherence of S. aureus organisms to platelet-fibrin clots. The adherence ratio was expressed as a function of the original inoculum (see Materials and Methods). The results represent the means and standard deviations for 12 independent determinations. Asterisk, P < 0.05 versus all the other groups.

Experimental endocarditis.

Binding of S. aureus to platelets has been considered crucial for the development of infective endocarditis by affecting vegetation formation and septic embolization (50, 51). Additionally, several in vivo studies have implicated ClfA, ClfB, and FnbA as pathogenic factors in experimental S. aureus infection (14, 40, 44, 52). Thus, the σ^B-activity-associated increase in binding of S. aureus to immobilized fibrinogen, fibronectin, and platelet-fibrin clots suggests that σ^B activity is likely to facilitate the development of infective endocarditis as well. This hypothesis was tested in a rat catheter-induced aortic vegetation model of infection. As in previous experiments (14, 40), the rate of infection 16 h postinoculation was inoculum dependent (Fig. 6). The BB255 derivatives caused endocarditis in 20 to 40% and 90 to 100% of rats at low and high inocula, respectively. No statistically significant difference in infectivity could be observed among any of the strains analyzed (P >0.05), even though the σ^B-activity-producing strains tended to cause slightly more endocarditis than did BB255 or its ΔrsbUVWsigB mutant (Fig. 6).

FIG. 6.

Infectivity titer determination for the various test organisms in the rat model of experimental endocarditis 16 h postinoculation. The figure shows a compilation of results from six separate experiments. (A) An inoculum size of 10³ CFU was used to obtain results in four experiments. Pooled results are shown. In two experiments, the rats were challenged with bacterial inocula of 2 × 10³ to 3 × 10³ CFU/ml. In the two other experiments, the rats were challenged with 4 × 10³ 6 × to 10³ CFU/ml. The results were similar with both inocula. (B) An inoculum size of 10⁴ CFU was used to obtain results in two experiments. The columns indicate the percentage of positive vegetations and spleen cultures. The number of animals per group (n) and the S. aureus strain are indicated at the bottom of the columns. Statistical differences were evaluated by the χ² test with the Yates correction. *, P < 0.05 compared with BB255 and IK181.

The bacterially challenged animals were also analyzed for positive spleen cultures (Fig. 6). In general, spleen cultures were positive more often than aortic cultures were. This demonstrated that the animals had been appropriately inoculated and suggested that spleen colonization was not necessarily dependent on valve infection. Importantly, in contrast to aortic vegetations, the frequency of positive spleen cultures correlated with increased σ^B activity (Fig. 6) and was most pronounced in rats challenged with the σ^B overproducer MB138 (P < 0.05). All spleen cultures became positive as the inoculum size was increased 10-fold (Fig. 6B). Although spleen colonization at 16 h after inoculation is primarily indicative of inoculation and does not necessarily indicate a spleen infection, it might become relevant in terms of the in vivo survival capacity of the bacteria, in combination with the respective bacterial density that is reached in this organ (see below).

Analysis of the bacterial densities obtained after 16 h revealed that all animals had comparable bacterial densities in their aortic vegetations at low inocula (P > 0.05). However, at high inocula, the rats inoculated with the σ^B overproducer MB138 had significantly higher vegetation bacterial densities (P < 0.05) than did those in all other groups (Table 2). Additionally, the bacterial densities of MB138 within colonized spleens were significantly higher (P < 0.05) than those of BB255 and IK181 when the animals were challenged with a low inoculum (Table 2). Since all strains analyzed displayed similar abilities to grow under in vitro conditions, the higher vegetation bacterial densities observed for MB138 after 16 h was unlikely to be due to an increase in growth rate and thus may reflect an abated clearance of the σ^B-overproducing strain by host defense mechanisms at this site. Spleens are known to display a massive recruitment of polymorphonuclear neutrophils (PMNs), which are considered to fulfill major functions in the host defense system (13). Several laboratories have shown that σ^B activity significantly reduces the sensitivity of S. aureus to oxygen radicals and hydrogen peroxide (19, 24, 30), which are released by PMNs to challenge the pathogenic bacteria. Interestingly, SarA influences the capacity of S. aureus to be internalized by PMNs and to survive inside of PMNs (20). The findings that challenging rats with MB138 resulted in statistically significantly larger numbers of positive spleen cultures and higher bacterial densities 16 h postinoculation than did challenging them with BB255 or IK181, combined with the observation that σ^B-activity-possessing strains produce larger quantities of sarA-containing transcripts, suggest that σ^B activity may indeed contribute to the survival capacity of S. aureus under in vivo conditions during the early stages of infection, especially at sites with a high PMN content.

TABLE 2.

Bacterial densities in vegetations and spleens of rats that developed experimental endocarditis 16 h after bacterial challenge

Bacterial strain	Median bacterial densitya (CFU/g) (range) with inoculum of:
	103 CFU		104 CFU
	Vegetations	Spleens	Vegetations	Spleens
BB255	5.63 (4.51-6.22)	1.74 (1.57-2.47)	5.80 (3.67-8.57)	2.99 (2.35-4.29)
IK181	5.61 (3.52-7.88)	1.80 (1.05-2.36)	5.78 (3.01-7.36)	2.73 (2.33-3.53)
GP268	6.09 (2.40-7.20)	2.05 (1.53-2.86)	5.53 (3.98-8.87)	2.98 (2.50-4.59)
MB138	6.18 (4.28-7.93)	2.10 (1.60-3.30)†	7.85 (5.93-8.41)‡	3.28 (3.05-3.51)*

^a†, P < 0.05 versus BB255 and IK181; ‡, P < 0.05 versus all the other groups; *, P < 0.05 versus IK181.

To monitor the evolution of the infection with time, rats were challenged with high inocula and were sacrificed 72 h after bacterial inoculation. By 72 h postinoculation (Fig. 7), the rates of aortic infection slightly decreased relative to the rates at 16 h but remained similar between groups (P > 0.05). The number of CFU recovered in infected vegetations of rats inoculated with either BB255, IK181, or GP268 strains increased relative to the number recovered at 16 h and were larger (although not statistically significantly larger) than the number of CFU recovered from vegetations of rats inoculated with MB138, which decreased relative to the number observed by 16 h.

FIG. 7.

Infectivity titer determination in the rat model of experimental endocarditis 72 h after inoculation with an inoculum size of 10⁴ CFU. The figure shows a compilation of results of separate experiments. Details are as in Fig. 6.

By 72 h postinoculation, the rates of spleen infection of rats inoculated with either BB255, IK181, or GP268 slightly decreased relative to the rates observed at 16 h (Fig. 6). In contrast, the frequency of positive spleen cultures remained unchanged (100%) in animals that were challenged with MB138. However, the difference in spleen infection frequency between MB138 and the other derivatives tested was not statistically significant. Surprisingly, while the number of CFU obtained from spleens of rats inoculated with either BB255, IK181, or GP268 increased relative to the number obtained 16 h, the same was not observable for MB138 (Table 3). The number of CFU obtained from the spleens of MB138-inoculated animals after 72 h remained more or less unchanged relative to the number obtained after 16 h and was significantly smaller than the number of CFU recovered from the spleens of BB255, IK181, or GP268-infected animals (P < 0.05). The observed decrease in bacterial densities in both vegetations and spleens by 72 h suggests that high σ^B activity negatively influences the ability of S. aureus to develop high cell densities at the infection site during later stages of infection.

TABLE 3.

Bacterial densities in infected vegetations and spleens of rats 72 h after bacterial challenge with an inoculum of 10⁴ CFU

Bacterial strain	Median bacterial densitya (CFU/g) (range) in:
	Vegetations	Spleens
BB255	8.75 (6.30-9.84)	5.44 (4.49-5.60)†
IK181	8.32 (3.37-9.35)	5.17 (1.78-5.60)†
GP268	9.19 (5.75-9.57)	5.15 (3.05-5.64)†
MB138	6.03 (3.11-9.48)	3.43 (1.96-5.56)

^a†, P < 0.05 versus MB138.

The in vivo data presented here demonstrate that σ^B activity has little to no effect on the development of infective endocarditis, irrespective of its ability to positively influence the expression of clfA and fnbA, somehow contradicting previous publications showing that inactivation of clfA or fnbA had small but significant impacts on the infectivity of S. aureus (40, 52). It is worth noticing that Xiong at al. (55), who investigated the impacts of sar and agr mutations on fnbA expression and on the fibronectin adherence capacity under in vitro and in vivo conditions, found that mutations of sarA and/or agr resulted in a marked effect on fnbA expression but had no statistically significant impact on infectivity in an experimental rabbit endocarditis model. It becomes clear from the transcription and adhesion data shown here that S. aureus is still able to produce significant amounts of fibrinogen- and fibronectin-binding proteins even in the absence of σ^B activity. The amounts of clf- and fnb-encoding transcripts produced by the sigB mutant strains seem to be sufficient to allow the pathogen to bind to the host cell matrix proteins in vivo, giving a simple explanation of why mutations of clfA or fnbA decrease the infectivity of S. aureus but sigB does not. Moreover, our in vivo findings are in agreement with previous findings demonstrating that mutations in sigB had no significant effect on the virulence of S. aureus in four different animal models (6, 24, 41).

Considering that σ^B-dependent promoters are present in the upstream regulatory regions of many virulence-associated loci, it is nevertheless surprising that any attempt to demonstrate an influence of σ^B on the virulence of S. aureus in an animal model has failed to date. Based on the findings that the expression of various virulence-associated factors is influenced by σ^B activity (3, 4, 18, 19, 24, 30, 42) but that no influence on infectivity has been demonstrated so far, we have to conclude that σ^B is more likely to be a modulator of virulence gene expression and not a virulence determinant by itself. Thus, one possible function of σ^B on S. aureus-host interactions might be the global regulation of factors associated with virulence in a way (i) promoting adhesion of the pathogen to host cells, i.e., by upregulation of MSCRAMMS, and (ii) allowing the organism to persist at the attachment site without harming the host tissue, i.e., by down-regulation of exoprotein production, thereby reducing the inflammatory response caused by the host defense system. Such a function of σ^B would fit with our observations that σ^B had a positive impact on the ability of S. aureus to reach high cell densities during the early stages of infection while having a negative influence on the capacity of the pathogen to develop high cell densities during later stages of infection. Further support is given by the recent findings by Bischoff et al. (4), who showed that σ^B had a negative influence on the expression of various excreted virulence factors, such as proteases, lipases, and hemolysins, which all are supposed to be important for the ability of S. aureus to evade the site of infection, once the infection has been established. While promoting adhesion during early stages, σ^B activity may decrease the ability of the pathogen to evade the primary site of infection during later growth stages, thus decreasing its capacity to damage the host cell tissue, and to establish new sites of infection.