Adherence of Staphylococcus aureus to Endothelial Cells: Influence of Capsular Polysaccharide, Global Regulator agr, and Bacterial Growth Phase

Abstract

The adherence of Staphylococcus aureus to human endothelial cells (EC) is probably an important step in the pathogenesis of systemic staphylococcal infections. We examined the influence of type 5 capsular polysaccharide (CP5) production, the global regulator agr, and the bacterial growth phase on S. aureus adherence to EC. Whereas S. aureus Newman showed maximal adherence to EC in the logarithmic phase of growth, an isogenic agr mutant showed maximal adherence in the stationary growth phase. S. aureus adherence to EC and CP5 expression were negatively correlated: a mutation in the agr locus diminished CP5 production and led to increased adherence. Likewise, induction of CP5 expression by addition of NaCl to the growth medium resulted in reduced staphylococcal adherence to EC. S. aureus Newman cells that adhered to EC did not express CP5. A Newman cap5O mutant was acapsular and showed significantly greater adherence to EC than the parental strain did (P < 0.005). Complementation of the cap5O mutation in trans restored CP5 expression and reduced EC adherence to a level similar to that of the parental strain. The enhanced adherence shown by the cap5O mutant was similar in magnitude to that of the agr mutant or the cap5O agr double mutant. Cells of the cap5O mutant and cap5O agr double mutant harvested from stationary-phase cultures adhered significantly better than did cells harvested in the exponential growth phase. These data are consistent with the postexponential and agr-independent expression by S. aureus of at least one putative EC adhesin, whose binding domain may be masked by CP5.

Staphylococcus aureus is responsible for a broad range of human diseases, including foreign body infections, bacteremia, abscesses, and wound infections (41). The ability of S. aureus to adhere to extracellular matrix proteins is thought to be essential for colonization and the establishment of infections (17, 41, 52). The interaction of S. aureus with endothelial cells (EC) may play a major role in the pathogenesis of endovascular infections, including endocarditis and metastatic infections (26, 41).

S. aureus can express several cell wall-associated adhesins specific for recognizing matrix proteins; these adhesins are referred to as MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) (52). MSCRAMMs with specific binding activity toward fibronectin (31, 62), fibrinogen (45, 47), and collagen (53) have been cloned and biochemically characterized. Whether MSCRAMMs are also responsible for the adherence of S. aureus to host cells is still unclear. A role for fibronectin-binding proteins in staphylococcal adherence to bovine epithelial cells (13) and mammary gland cells (36) was recently described; however, other factors were also postulated to mediate epithelial cell adherence (13).

The expression of most cell wall adhesins and extracellular virulence determinants in S. aureus is coordinately regulated during the growth cycle (57). Some of the growth cycle-dependent expression patterns can be attributed to the global regulator agr, which acts as a quorum-sensing system (4, 28, 29). The capsular polysaccharide (CP) is an important surface component that is positively regulated by agr (10). More than 90% of S. aureus strains express 1 of 11 CP types (33, 65), and most strains colonizing and infecting humans produce either CP type 5 (CP5) or CP8 (2, 24). Specific environmental and/or host signals result in variation of CP expression both in vitro (9, 40, 55, 68, 70) and in vivo (23, 35, 40).

It is likely that specific adhesins mediating adherence to EC are differentially expressed and/or masked by differentially expressed extracellular surface structures such as CP. In this study, we report the influence of agr, CP production, and bacterial growth conditions on the adherence of S. aureus to EC. Our findings indicate that mutations and growth conditions that reduce or eliminate CP expression enhance the adherence of S. aureus to EC. In the absence of CP expression, S. aureus adherence to EC is enhanced in the postexponential growth phase but does not appear to be influenced by the agr regulatory locus. The importance of differential expression and exposure of an EC adhesin(s) to the pathogenesis of staphylococcal infections is discussed.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The strains and plasmids used in this study are listed in Table 1. S. aureus type 5 strain Newman was grown in medium composed of 1% Casamino Acids, 1% yeast extract, 0.5% NaCl, 0.5% glucose, and 60 mM glycerophosphate (CYGP) at 37°C with shaking. Chloramphenicol (10 μg/ml) or tetracycline (5 μg/ml) was added to the culture medium when indicated. Bacteria were inoculated to an initial optical density at 600 nm of 0.05 in CYGP and grown to the exponential phase (optical density at 600 nm = 0.6; 2.5 h) or the stationary phase (14 h) of the growth cycle. Before each experiment, the bacteria were washed twice and resuspended in sterile phosphate-buffered saline (PBS).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant genotype or phenotype	Source or reference
S. aureus strains
Newman	Wild type, CP5 positive	NCTC 8178 (12)
ALC355	Newman Δagr::tetM, reduced CP5 expression	76
KKO22	Newman Δcap5O, CP5 negative	This study
CW22	Newman Δagr::tetM, Δcap5O, CP5 negative	This study
Plasmids
pLI50	Shuttle vector (Apr, Cmr)	37
pKBK22	3.4-kb BamHI-EcoRI fragment (cap5MNP Δcap5O) in pTS1	Portoles et al., submitted
pKBK24	pLI50 containing a 2.4-kb PCR amplicon carrying cap5O and flanking DNA	Portoles et al., submitted

Construction of the S. aureus cap5O mutant.

pKBK22 was constructed as described previously (M. Portoles, K. B. Kisier, N. Bhasin, K. H. N. Chan, and J. C. Lee, submitted for publication). In brief, a 4.1-kb DNA fragment comprising cap5MNOP was subcloned from the cap5 gene region of S. aureus Reynolds. A 727-bp HpaI deletion, encompassing nucleotides 9 to 734 of the 1,260-bp cap5O gene (61), resulted in a +1 frameshift. Thus, pKBK22 contains a 3.4-kb insert comprising the cap5O deletion and flanking cap5 sequences (cap5MN and cap5P) in the temperature-sensitive shuttle vector pTS1.

pKBK22 was electrotransformed into S. aureus RN4220 and then transduced with phage 80 alpha into S. aureus Newman, in both instances selecting for Cm^r colonies at 30°C. The mutation was introduced into the chromosome by allelic exchange. In brief, plasmid integrants in the chromosome were selected by plating of cells on tryptic soy agar plus chloramphenicol (5 μg/ml) at 42°C. Single colonies were passaged at 30°C (no antibiotic selection) and then grown at 42°C to eliminate the excised plasmids. Cm^s colonies were screened for CP5 production by colony immunoblotting with CP5-specific polyclonal rabbit antiserum. CP5-negative colonies were examined by PCR and Southern blotting to confirm the presence of the deletion in cap5O. The CP5-negative cap5O mutant of strain Newman was designated KKO22. Lack of CP5 production by KKO22 was confirmed by immunodiffusion of capsular extracts with CP5-specific rabbit antiserum. The mutant strain did not differ in other phenotypic markers, e.g., hemolytic activity, from the parental strain.

To complement the mutation in KKO22, the cap5O gene and flanking cap5 sequences were amplified by PCR with primers KK14 (5′-ACAAGGATCCAAAAAGTTCGCTGAACAAGCATTACAAG-3′) and KK15 (5′-ACAAGGATCCAGTAATAAAGATACGCTCTTTGTCTTTG-3′) and the ELONGASE enzyme mix (Life Technologies, Inc., Gaithersburg, Md.) for 25 cycles of 94°C for 30 s, 60°C for 1 min, and 68°C for 3 min. (The BamHI sites in primers KK14 and KK15 are underlined.) The 2.4-kb amplicon was digested with BamHI and ligated into pLI50 to create plasmid pKBK24. pLI50 and pKBK24 were each transformed into RN4220 and then transduced into Newman or KKO22, as described above.

Construction of the S. aureus cap5O agr double mutant.

The agr mutation was transduced from agr mutant ALC355 into strain Newman(pLI50) and cap5O mutant KKO22(pLI50) with φ11. Transductants were selected on medium containing tetracycline and chloramphenicol. Successful construction of the mutants was confirmed by Southern blotting with agr-specific probes. The agr deletion resulted in decreased hemolytic activity compared to those of the parental strains.

EC culture.

EC were obtained from human umbilical cord veins by the method of Jaffe et al. (27) with modified protease digestion. In brief, cells were isolated with 0.025% trypsin–0.01% EDTA (Cell Systems, St. Katharinen, Germany), washed, and grown in endothelial growth medium (Cell Systems) in 0.2% gelatin-coated tissue culture flasks (Becton Dickinson, Heidelberg, Germany). The cells were maintained at 37°C in an atmosphere containing 5% CO₂. They were seeded in 24-well plates (Becton Dickinson) or culture slides (Becton Dickinson), and only confluent monolayers of the third to sixth passages were used for adherence assays.

Adherence assays.

Confluent monolayers of EC (∼10⁵ EC per well) were washed three times with M199 (Sigma Chemical Co., Deisenhofen, Germany) to remove antibiotics and incubated for 20 min at 37°C with 3% bovine serum albumin (Sigma) in M199 to minimize background adherence. The EC were then incubated with bacterial inocula ranging from ∼10⁶ to ∼10⁹ CFU of S. aureus at 37°C in an atmosphere containing 5% CO₂. To exclude contamination, negative controls with uninfected EC were included in each experiment. The supernatants were removed after 2 h, and nonadherent bacteria were analyzed for CP production by immunofluorescence. EC were subsequently washed three times to remove remaining nonadherent bacteria. The plates were treated with trypsin-EDTA and 0.25% Triton X-100 (Sigma) to lyse EC and release adherent bacteria. Adherent bacteria released by EC lysis were evaluated for CP5 production by immunofluorescence, and serial dilutions were plated onto sheep blood agar for enumeration of CFU. The results of representative experiments are shown. Each assay was repeated at least once, and within one experiment all samples were processed at least in quadruplicate.

For microscopic evaluation, adherence assays were performed on culture slides containing confluent monolayers. Washed monolayers with adherent bacteria were fixed with 50% methanol and stained with crystal violet. Bacterium-to-EC ratios were determined by counting the number of adherent bacteria in three microscopic fields with 50 EC per field.

Evaluation of CP5 production.

CP5 production was detected by a modified indirect immunofluorescence technique (23). In brief, bacteria were fixed by heat, incubated for 30 min with human immunoglobulin G (IgG) (Sigma) (0.2 mg/ml in PBS–0.05% Tween 20), and washed. The bacteria were incubated for 1 h at ambient temperature with mouse IgM monoclonal antibodies to CP5 (25) diluted 1:50 in PBS–0.05% Tween 20. Bacteria were washed three times before being incubated with fluorescein isothiocyanate-conjugated rabbit F(ab)₂ fragments to mouse IgM (Dako, Hamburg, Germany). In a subsequent step, the bacteria were stained with 4′,6-diamidino-2-phenylindole (DAPI; 2 μg/ml) (Sigma) for 5 min at room temperature. Three microscopic fields with 50 bacteria per field were evaluated, and the percentage of CP5-positive bacteria was determined by comparing the number of fluorescent bacteria with the total number of DAPI-stained bacteria. To control for nonspecific reactions of antibodies with staphylococcal protein A, immunofluorescence studies were also performed with the cap5O mutant and with the parental strain and the second antibody only.

Statistical analysis.

The data were analyzed by Student's t test for unpaired data to determine statistically significant differences.

RESULTS

Effect of agr and growth phase on adherence of S. aureus to EC.

In preliminary studies of staphylococcal adherence to EC, we observed that S. aureus Newman adhered to EC but that in contrast to the invasive strain 8325-4, <1% of the inoculum invaded the EC. Similarly, the agr mutant ALC355 was noninvasive. After incubation for 2 h with the staphylococci, no obvious changes in EC morphology were evident microscopically.

We compared the adherence of S. aureus Newman cells and isogenic agr mutant ALC355 cells harvested from different phases of the growth cycle. As shown in Fig. 1, strain Newman cells harvested during the exponential growth phase showed significantly (P < 0.05) greater adherence than those harvested in the stationary growth phase. This effect was apparent at multiplicities of infection ranging from 10:1 to 10,000:1 (ratio of CFU to EC). In contrast, the agr mutant grown to stationary phase showed greater dose-dependent adherence than did the same strain grown to exponential phase. The agr mutant ALC355 was significantly (P < 0.001) more adherent than the parental strain Newman in the stationary growth phase, but the two strains showed similar levels of adherence in the exponential growth phase (Fig. 1).

FIG. 1.

Effect of S. aureus growth phase and agr on adherence. Approximately 10⁵ EC were incubated for 2 h with S. aureus cells at multiplicities of infection ranging from 10 to 10,000 CFU per cell. Strain Newman and the agr mutant ALC355 were grown to the exponential or stationary growth phase. The percent adherence was expressed as 100 × (CFU of adherent bacteria/CFU of inoculated bacteria). Each point represents the mean of four determinations, with the error bars representing the standard deviations.

For microscopic evaluation, adherence assays were performed on culture slides with S. aureus strains harvested in the stationary phase (Fig. 2A and B). Quantification of adherent bacteria per EC yielded results that correlated with those obtained by the plating technique (Fig. 1); e.g., at a multiplicity of infection of 100, 11 ± 2.5 strain Newman bacteria per cell and 49 ± 3.2 agr mutant bacteria per cell were counted. For both S. aureus strains, some EC had many adherent bacteria whereas other cells were totally free of bacteria.

FIG. 2.

Microscopic evaluation of staphylococcal adherence and capsule expression. Culture slides were stained with crystal violet to show the distribution of bacteria adherent to EC (A and B). Strain Newman (A, C, and D) and the agr mutant ALC355 (B, E, and F) were grown to stationary phase. DNA from S. aureus cells of the inoculum were stained with DAPI (C and E), and CP was detected by indirect immunofluorescence (D and F). Magnifications, ×200 (bar, 50 μm) in panels A and B and ×1,000 (bar, 10 μm) in panels C to F.

Inverse correlation between CP expression and adherence.

The decreased adherence of stationary-phase bacteria might be explained by an inhibitory factor produced late in the growth cycle. S. aureus CP, which is expressed postexponentially (11), is a candidate for such an inhibitory factor; moreover, CP5 expression is positively regulated by agr (10). We confirmed the agr-mediated activation of CP5 expression by indirect immunofluorescence experiments with monoclonal antibodies to CP5. If the bacterial cells from the stationary growth phase were examined, CP5 was detected on 39% of strain Newman cells but on only 6% of agr mutant cells (Fig. 3). In contrast to the heterogeneous distribution of CP5 on S. aureus Newman cells, expression of protein A (evaluated without blocking with human serum and with the use of fluorescein isothiocyanate-labeled rabbit IgG) was uniform within the bacterial population (data not shown).

FIG. 3.

Inverse correlation between CP expression and adherence. Strain Newman and the agr mutant ALC355 were grown to stationary phase in culture medium containing 0.5 or 5% NaCl. Adherence assays were performed by incubation of ∼10⁵ EC with 10⁷ CFU of S. aureus for 2 h. The percent adherence was expressed as 100 × (CFU of adherent bacteria/CFU of inoculated bacteria). CP5 expression was determined by indirect immunofluorescence and given as a percentage of all bacteria being DAPI positive. The error bars represent the standard deviations of the means of four (adherence) or three (CP5 expression) determinations.

As shown in Fig. 3, we could modulate CP5 expression by varying the salt concentration in the medium. The addition of 5% NaCl to CYGP culture medium resulted in a significant increase in the numbers of CP5-positive Newman cells (P < 0.005) and agr mutant ALC355 cells (P < 0.0001) over the numbers when the same strains were grown in CYGP containing 0.5% NaCl. Most strain Newman cells cultivated in the presence of added NaCl were encapsulated but poorly adherent (Fig. 3). In contrast, the agr mutant cultivated under low-salt conditions showed minimal CP5 expression and maximal adherence. These results suggest that CP5 inhibits the adherence of S. aureus Newman to EC.

Detection of CP5 on adherent and nonadherent S. aureus.

To determine whether there was a difference in CP5 expression by adherent and nonadherent S. aureus cells, nonadherent bacteria were separated from cell-associated bacteria after the adherence assay. As shown in Table 2, all adherent bacteria were negative for CP5 by indirect immunofluorescence whereas nonadherent bacteria showed higher CP5 expression than that of the input inoculum; this result reflects a high affinity of the unencapsulated bacteria for EC. The inhibitory effect of CP5 on adherence may have led to the separation of a heterogeneous bacterial population such that encapsulated bacteria were nonadherent and unencapsulated cells were adherent.

TABLE 2.

CP5 expression by S. aureus strain Newman and agr mutant ALC355 before (inoculum) and after the adherence assay

Strain	% of bacteria staining positive for CP5a
	Inoculum	Adherent	Nonadherent
Newman	39 (1.4)	0 (0.0)	70 (1.7)
ALC355	6 (2.1)	0 (0.0)	41 (1.9)
Newman (NaCl)b	84 (1.6)	0 (0.0)	95 (2.4)
ALC355 (NaCl)b	69 (1.2)	0 (0.0)	88 (1.6)

^aMeans of three determinations given as the percentage of bacteria found to be DAPI positive. Standard deviations are given in parentheses. 

^bGrowth medium supplemented with 5% NaCl. 

Adherence of the cap5O mutant to EC.

We hypothesized that CP production was responsible for the reduced adherence of the agr-positive strain compared with that of the agr mutant (Fig. 1). Therefore, we investigated the influence of a cap5O mutation on adherence in both agr-positive (KKO22) and agr-negative (CW22) genetic backgrounds. The cap5O gene encodes a UDP-N-acetylmannosamine dehydrogenase, an essential enzyme in the CP5 biosynthetic pathway (Portoles et al., submitted). A cap5O mutation was created by deletion of 727 bp in the 5′ end of the cap5O gene and was introduced by allelic exchange into the chromosome of S. aureus Newman. The mutant KKO22 was determined to be CP5 negative by immunodiffusion of capsular extracts, immunofluorescence, and colony immunoblotting. Adherence assays were performed with staphylococci harvested from the stationary growth phase in CYGP medium supplemented with either 0.5 or 5% NaCl (Fig. 4). Like the agr mutant ALC355, the cap5O mutant KKO22 showed significantly (P < 0.005) greater adherence than the parental strain Newman did. Complementation of the cap5O mutation with pKBK24 (carrying an intact cap5O gene) restored CP5 expression to strain KKO22 and resulted in the reduction of EC adherence to a level similar to that of the wild-type strain (Fig. 4). The cap5O agr double mutant showed no further increase in adherence. In contrast to that of CP-positive strains, the adherence of the CP-negative strains KKO22 and CW22 was unaffected by the addition of NaCl to the growth medium.

FIG. 4.

Influence of CP5 expression on S. aureus adherence to EC. Strains were grown to stationary phase in culture medium containing 0.5 or 5% NaCl to the culture medium. The adherence assay was performed by incubation of ∼10⁵ EC with 10⁷ CFU of S. aureus for 2 h. Percent adherence was expressed as 100 × (CFU of adherent bacteria/CFU of inoculated bacteria). The error bars represent the standard deviations of the means of four determinations.

Adherence of S. aureus from exponential and stationary growth phases in a CP-negative background.

To reinvestigate the impact of bacterial growth phase on the adherence of S. aureus to EC, we compared the adherence of strain Newman with that of isogenic strains with mutations in agr and/or cap5O. As shown in Fig. 5, all bacterial strains harvested from the exponential growth phase showed similar levels of adherence. This finding is consistent with the observation that little CP5 is expressed by cells in the exponential phase. However, mutants defective in agr and/or cap5O adhered significantly (P < 0.05) better to EC if they were derived from stationary-phase rather than exponential-phase cultures. This observation suggests postexponential activation of at least one adhesin that is masked by CP in the parental strain.

FIG. 5.

Effect of staphylococcal growth phase on adherence to EC. S. aureus strains were grown to exponential or stationary growth phase. The adherence assay was performed by incubation of ∼10⁵ EC with 10⁷ CFU of S. aureus for 2 h. The percent adherence was expressed as 100 × (CFU of adherent bacteria/CFU of inoculated bacteria). The error bars represent the standard deviations of the means of four determinations.

DISCUSSION

In this study we analyzed the role of CP, growth phase, and agr on the adherence of S. aureus Newman to human EC. The evidence suggests that CP5 inhibits the adherence of bacteria to EC by masking the major cell wall adhesin(s). This possibility is supported by our findings that (i) adherence was negatively correlated with CP expression, (ii) only unencapsulated bacterial cells were adherent, and (iii) the CP5-negative isogenic cap5O mutant showed significantly greater adherence than the parental strain did. Our results are in direct contrast to those of Soell et al. (64), who reported that S. aureus CP5 and CP8 can bind to EC. We believe that the putative adhesin may have been a contaminant of their CP preparation.

Previous studies have shown that the expression of S. aureus extracellular polysaccharide(s) interferes with the adherence process by masking adhesins (8, 18, 20, 42–44, 50). However, masking was seen only for heavily encapsulated strains belonging to capsular type 1 or 2 or was due to an uncharacterized capsule-like material detected after growth of strains in milk whey (43). This periodate-sensitive material might be the recently described poly-N-succinyl-β1→6-glucosamine of S. aureus, which was preferentially expressed during infection (46); alternatively, it might be CP5 or CP8, each of which is released from whole bacterial cells by periodate treatment (J. Lee, unpublished observations). Evidence to support the masking of MSCRAMMs by microcapsules such as CP5 is scant (18, 48, 63). Whereas CP did not influence staphylococcal binding to collagen (18, 63), capsular mutants derived from strain Reynolds showed a tendency to bind more fibronectin and bone sialoprotein than did the parental strain (48). Since the major adhesin responsible for adherence to EC is not known, we can only speculate that the respective binding domain of this putative adhesin is covered by the capsular material.

Masking effects of extracellular polysaccharides leading to decreased adherence to host cells have been described for other pathogenic bacteria, such as Streptococcus pneumoniae (69), group A streptococci (21), Klebsiella pneumoniae (16), Neisseria meningitidis (66), Escherichia coli (59), and Haemophilus influenzae type b (67). Because bacterial adherence is thought to be a prerequisite for infection, the CP-mediated inhibition of adherence suggests that encapsulated strains are less virulent. However, CP exhibits antiphagocytic activity (32, 48, 70) that protects the bacterium during infection. The relative virulence of the clinically prominent CP types 5 and 8 and their acapsular mutants depends on the animal model of infection chosen for study. Early studies indicated that microencapsulated S. aureus strains were no more virulent for rodents than were unencapsulated mutants (1, 3, 38). More recently, a protective effect of specific antibodies to CP5 has been documented in different animal models of staphylococcal infection (15, 39). In addition, mortality was higher (48, 70) and bacteremia (70) and arthritis (48) were more frequent among mice inoculated with serotype 5 strain Reynolds than among mice inoculated with unencapsulated mutants. The results of virulence studies were influenced by the method used for preparation of the challenge inoculum, i.e., growth on solid medium or in liquid growth medium that affects CP expression (70). CP expression in vitro is highly sensitive to various environmental signals (9, 23, 40, 55, 56, 68) and is probably influenced by the environment in vivo as well. Recently, we showed that CP expression is down-regulated during chronic lung infections in patients with cystic fibrosis (23) but is up-regulated in a rat model of endocarditis (40) and in a mouse model of nasal colonization (35). Furthermore, we and others (23, 56) found S. aureus cultured in broth to be heterogeneous for CP expression. Although the molecular basis for the heterogeneity of CP expression by S. aureus is not known, phase variation in the expression of extracellular polysaccharides has been described for other bacteria (22, 34, 58, 77) and is thought to be important in bacterial immune evasion by antigenic variation. Encapsulated and unencapsulated S. aureus cells may occupy different niches in the host by virtue of their different adherence capacities. Furthermore, the ability to adhere may be more important for the onset of infections than later, when the bacterial population is organized in a biofilm community (57). Whether this correlates with the exponential and stationary growth phases in vitro remains to be proven. Therefore, studies evaluating the conditions of cell adherence as well as CP expression during an actual staphylococcal infection are needed. Kiser et al. demonstrated that S. aureus bacteria colonizing the nares of mice were CP5 positive (35). However, whether the CP5-positive bacteria in the nose were cell associated was not addressed.

S. aureus adherence properties are influenced by additional factors other than capsule production. The global regulator agr is thought to be an inhibitor of adhesin expression (57), given the up-regulation of cell wall-associated protein A (51) and fibronectin-binding proteins in agr mutants (60). However, we found no evidence that agr regulates the adhesin(s) responsible for the adherence of strain Newman to EC, since no difference in adherence was apparent between the cap5O mutant and the cap5O agr double mutant. The increased adherence capacity of the agr mutant can be fully explained by decreased CP5 expression, which would serve to unmask an adhesin that might be buried beneath the capsular layer. Other investigators have characterized the adherence of S. aureus 8325-4 and its agr mutant to mesothelial cells (54) and to mammary gland epithelial cells (36) and have described the internalization of strain 8325-4 by EC (75). However, strains Newman and 8325-4 differ in numerous properties. Although S. aureus 8325-4 is negative for CP expression (74), it shows greater adherence to EC than does the capsule-negative mutant of strain Newman (unpublished observations). Whether different adhesins or different quantities of the same adhesin account for these strain-specific adherence properties is under investigation.

The agr operon acts as a quorum-sensing system that regulates growth phase-dependent expression of virulence factors and adhesins. However, the growth phase of the bacterial inoculum is not described in most reports of S. aureus adherence to EC (5, 30, 49, 72), while in other studies the assays were performed only with bacteria harvested in the stationary growth phase (6, 7, 14). Tompkins et al. (71) reported that S. aureus cells from the exponential growth phase were more adherent to human EC than were bacteria harvested in the stationary growth phase. Similarly, we found that a greater percentage of strain Newman cells from the exponential growth phase than from a later point in the growth cycle (when capsule production was maximal) adhered to EC. In contrast, the cap5O mutant and the CP5-deficient agr mutant were more adherent to EC when harvested in the stationary growth phase. Thus, the major EC adhesin(s) in strain Newman appears to be expressed postexponentially. Accordingly, other investigators have shown that some S. aureus strains exhibit an increased adherence to cultured cells when derived from the stationary growth phase (73). Whether the different, strain-dependent patterns of adhesion to cell monolayers are due to differential expression of various adhesins or CP remains to be determined.

We conclude that strain Newman produces at least one adhesin that is masked by CP, is expressed postexponentially, and is independent of agr. Although we believe that this adhesin is probably none of the known MSCRAMMs, our observations do not exclude the possibility that other adhesins may mediate EC adherence early in the growth cycle. In addition to the exponential growth phase-dependent expression of fibronectin-binding proteins (60), collagen-binding protein (19), and clumping factor B (47), other adhesins—such as clumping factor A (a fibrinogen-binding protein)—are preferentially expressed in the stationary phase (76). The relevance of these in vitro findings to the pathogenesis of staphylococcal infections is uncertain and requires further investigation.