Staphylococcus aureus Fibronectin-Binding Proteins Contribute to Colonization of the Female Reproductive Tract

ABSTRACT

Methicillin-resistant Staphylococcus aureus (MRSA) is an opportunistic pathogen and frequent colonizer of human skin and mucosal membranes, including the vagina, with vaginal colonization reaching nearly 25% in some pregnant populations. MRSA vaginal colonization can lead to aerobic vaginitis (AV), and during pregnancy, bacterial ascension into the upper reproductive tract can lead to adverse birth outcomes. USA300, the most prominent MRSA lineage to colonize pregnant individuals, is a robust biofilm former and causative agent of invasive infections; however, little is known about how it colonizes and ascends in the female reproductive tract (FRT). Our previous studies showed that a MRSA mutant of seven fibrinogen-binding adhesins was deficient in FRT epithelial attachment and colonization. Using both monolayer and multilayer air-liquid interface cell culture models, we determine that one class of these adhesins, the fibronectin binding proteins (FnBPA and FnBPB), are critical for association with human vaginal epithelial cells (hVECs) and hVEC invasion through interactions with α₅β₁ integrin. We observe that both FnBPs are important for biofilm formation as single and double fnbAB mutants exhibit reduced biofilm formation on hVECs. Using heterologous expression of fnbA and fnbB in Staphylococcus carnosus, FnBPs are also found to be sufficient for hVEC cellular association, invasion, and biofilm formation. In addition, we found that an ΔfnbAB mutant displays attenuated ascension in our murine vaginal colonization model. Better understanding of MRSA FRT colonization and ascension can ultimately inform treatment strategies to limit MRSA vaginal burden or prevent ascension, especially during pregnancy and in those prone to AV.

INTRODUCTION

Staphylococcus aureus is a commensal of approximately 20% of the healthy adult population primarily found in the anterior nares and an opportunistic bacterial pathogen able to cause a wide variety of infections throughout the body (1). The prevalence of S. aureus has increased due to higher rates of colonization, immunosuppressive conditions, increased use of surgical implants, and dramatic increases in antibiotic resistance (2, 3). S. aureus is a common colonizer of the skin and mucous membranes, including those of the vaginal tract, and has been reported to vaginally colonize up to 25% of pregnant individuals (4–9). Compared to antibiotic-susceptible strains, methicillin-resistant S. aureus (MRSA) infections exhibit elevated mortality rates, require longer hospital stays, and exert a higher financial burden, all highlighting the severity of the growing MRSA problem (10). In the past few decades, MRSA strains have expanded from health care settings to infect otherwise healthy individuals in the community (“community-associated” MRSA [CA-MRSA]) (11). USA300 isolates are one of the most problematic lineages of CA-MRSA that have emerged in the United States and has spread throughout the world, reaching epidemic levels in many hospital settings (12–19). Reports suggest an increasing prevalence of USA300 in pregnant and postpartum individuals, coinciding with increased incidences in the NICU and in newborn nurseries (8, 14, 20–25). To understand how USA300 colonizes the vaginal tract, it is of paramount importance to understand how it physically associates with mucosal epithelial surfaces.

To establish colonization in the human body, MRSA possesses a suite of surface-expressed adhesins called MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) that allow for initial attachment to extracellular matrix components, including fibrinogen, fibronectin, collagen, cytokeratin, and elastin, among others. These adhesins are also known to play a significant role in S. aureus aggregation and biofilm formation (reviewed in references 26 and 27). Biofilms are communities of bacteria encased in a matrix which typically act as a protective mechanism against outside assaults such as antibiotics and immune factors. This protection is conferred by the presence of a physical barrier and reduced metabolic activity of bacteria within the biofilm community, and for these reasons, biofilms have been shown to reduce efficacy of treatments for vaginal conditions such as bacterial vaginosis (28–30). Many MRSA strains, including USA300, are known to be robust biofilm formers (27, 31). Furthermore, S. aureus biofilms have been detected in vaginal wash samples of menstruating women, in dual-species biofilms with Candida albicans from the vaginal tract, and on explanted human gestational membranes (32–34). While MRSA biofilm formation in the vaginal tract has not been extensively studied, vaginal overgrowth of S. aureus is associated with nearly half of all cases of aerobic vaginitis (AV), a condition characterized by depletion of native Lactobacillus spp., inflammation, and thinning of the vaginal epithelium (35, 36). These factors can contribute to MRSA ascension into upper reproductive tissues. Consequences of MRSA ascension include squamous intracellular lesions of the cervix, infertility, pelvic inflammatory disease, spontaneous miscarriage, increased risk of preterm birth, preterm premature rupture of membranes (PPROM), and chorioamnionitis (37–41). Again, while the consequences of S. aureus ascension are known, the mechanism of ascension and bacterial factors that contribute to this movement have not been uncovered. It is also unknown whether MRSA utilizes invasion of FRT epithelia to move through cell layers during ascension or to escape detection by antimicrobials and the immune system.

A previous study from our group found that a multimutant of S. aureus adhesins was deficient in its ability to associate with vaginal and cervical epithelial cells, as well as colonization of the murine female reproductive tract (42). Because each of the mutated adhesins (ClfA, ClfB, SdrCDE, FnBPA, and FnBPB) has fibrinogen-binding capabilities, this mutant was termed the “fibrinogen-binding (ΔFg-binding) mutant.” Here, we show that one class of adhesins within the ΔFg-binding mutant, the fibronectin binding proteins (FnBPA and FnBPB), were critical for cellular association with hVECs. To build on this finding, we used in vitro and in vivo models to further investigate how MRSA FnBPs contribute to FRT colonization through biofilm formation, cellular invasion, and ascension. Fluorescent microscopy of MRSA grown on hVEC monolayers showed that both FnBPs are needed for optimal biofilm formation on vaginal epithelia. In vitro invasion assays and antibody blocking of human α₅β₁ integrin were used to determine that MRSA efficiently invades hVEC cells through a previously characterized pathway involving fibronectin (Fn) bridging with α₅β₁ integrin. We also determined that FnBPs remain important for cellular association and invasion of hVECs in a multilayer air-liquid interface cell culture system. In addition, heterologous expression of MRSA FnBPs in Staphylococcus carnosus is sufficient for fibronectin binding, hVEC association and invasion, and biofilm formation on human plasma. Finally, FnBPs are important for ascension into upper reproductive tissues in a murine model of vaginal colonization. This study is an important step in understanding factors required for MRSA FRT colonization and ascension and provides insights into mechanistic targets for potential therapeutics aimed at mitigating MRSA vaginal colonization in at-risk individuals.

RESULTS

FnBPs are important for association with hVECs.

To characterize the ability of MRSA to associate with cells of the lower female reproductive tract, we performed in vitro quantitative cellular association assays with human vaginal epithelial cell (hVEC) monolayers. This assay quantifies bacteria that are both adhered to the outer surfaces of human cells and bacteria that have invaded the intracellular space. A previous study from our group found that a USA300 MRSA mutant of multiple adhesins with known fibrinogen (Fg)-binding roles (ClfA, ClfB, SdrCDE, and FnBPs), referred to as the ΔFg-binding mutant, is deficient in its ability to associate with hVECs and human cervical cell lines and has a colonization defect in the murine FRT (42). To determine the separate contribution of these factors, mutations in genetic loci for clfA, clfB, sdrCDE, and fnbAB were compared to MRSA wild-type (WT) and the ΔFg-binding mutant strains in their ability to associate with hVECs. Of the mutants tested, only the ΔfnbAB mutant phenocopied the ΔFg-binding mutant and displayed significantly lower hVEC association compared to the WT (Fig. 1A). To determine the contribution of the individual fibronectin-binding proteins, FnBPA and FnBPB, single transposon mutant strains (Fig. 1B) as well as ΔfnbAB mutant strains complemented with either fnbA or fnbB (pFnbA4 and pFnbB4 [43]) (Fig. 1C) were assessed by the same quantitative cellular association assay. Both transposon mutants and complemented strains displayed similar levels of hVEC cellular association to the WT and significantly greater cellular association than the ΔfnbAB mutant (Fig. 1B to C).

FIG 1.

Fibronectin-binding proteins are important for MRSA association with vaginal epithelia. (A) Association of MRSA WT and adhesin mutants with hVECs. (B) Association of MRSA WT, ΔfnbAB and single FnBP mutants with hVECs. (C) Association of MRSA WT, ΔfnbAB, and plasmid complemented strains with hVECs. The data are expressed as the percent cell-associated MRSA relative to the initial inoculum. Experiments were performed in triplicate, with three technical replicates averaged per experiment. Error bars represent the standard errors of the mean (SEM). Statistical analyses were performed by ordinary one-way analysis of variance (ANOVA) with multiple comparisons, and the results for all analyses with P < 0.05 compared to WT (A) or P < 0.05 between all comparisons (B and C) are shown.

FnBPs contribute to MRSA biofilm formation on hVECs.

FnBPs are known to be important for MRSA biofilm formation on abiotic surfaces and in device-associated infections, and S. aureus biofilm investigations are usually focused on foreign bodies (44–47). Although S. aureus biofilms have been found on tampons and tissue from vaginal washes and coisolated in biofilms with C. albicans on vaginal mucosa (33, 34), S. aureus biofilm formation is rarely studied directly on mucosal epithelial surfaces. Therefore, we sought to investigate whether FnBPs contribute to MRSA biofilm formation on hVECs. To assess biofilm formation, MRSA WT, ΔfnbAB, fnbA::Tn, and fnbB::Tn strains were constructed to express DsRed fluorescent protein on a plasmid. Growth and fluorescence of each of these strains were monitored for 24h in tryptic soy broth (TSB) to confirm that all strains grow and fluoresce similarly (see Fig. S1). hVEC monolayers were inoculated with the DsRed-expressing MRSA strains mentioned above and incubated statically for 18 h to allow for robust biofilm formation beyond initial attachment. The mean fluorescent intensity of DsRed was imaged, measured, and compared between strains. Since there is no standard marker for determining biofilm formation in S. aureus, biofilm formation in all images was determined by visible clustering of bacteria surrounded by extracellular matrix that did not stain with DAPI (4′,6′-diamidino-2-phenylindole) or DsRed but obscured the hVEC cellular body in a bright-field overlay. The ΔfnbAB mutant was significantly attenuated in its ability to form biofilm on hVECs compared to WT, as would be expected (Fig. 2A to C). While single mutants of fnbA and fnbB associated with hVECs at the same level after a 30-min incubation (Fig. 1B), these mutants unexpectedly were deficient in their ability to form biofilm on hVECs compared to the WT (Fig. 2A, B, D, and E). Thus, both FnBPs contribute to overall biofilm structure and degree of clustering of the bacteria on human cell monolayers (Fig. 2B to E).

FIG 2.

Fibronectin-binding proteins contribute to MRSA biofilm formation on hVECs. (A) TRITC channel mean fluorescent intensity measurements of DsRed-expressing MRSA strains incubated on hVEC monolayers for 18 h. Ten images were taken from three separate slide chambers (n = 30) for each strain. Statistical analyses were performed by ordinary one-way ANOVA with multiple-comparison tests. (B to E) Median representative images of DsRed-expressing MRSA on DAPI-stained hVECs shown for WT (B), ΔfnbAB (C), fnbA::Tn (D), and fnbB::Tn (E) strains. All images were obtained at ×20 magnification with a 1/50-s exposure for DAPI and a 1/25-s exposure for TRITC.

MRSA invades hVECs through interactions with human α₅β₁ integrin.

FnBPs are also known to contribute to intracellular invasion of several endothelial and epithelial cell types (48–51), but it is not known how FnBPs contribute to MRSA cellular invasion in the vaginal tract. To investigate the contribution of FnBPs to invasion of vaginal epithelial cells, we performed in vitro quantitative cellular invasion assays with hVEC monolayers. While WT MRSA was able to invade hVECs at 25 to 65% efficiency, we recovered <0.5% of the ΔfnbAB mutant from the intracellular compartment (Fig. 3A). The fnbA::Tn mutant also displayed attenuated invasion compared to both the fnbB::Tn mutant and the WT, while the fnbB::Tn mutant displayed more similar levels of invasion to the WT (Fig. 3A).

FIG 3.

MRSA invasion of hVECs is mediated by FnBPs and human α₅β₁ integrin. (A) Invasion of hVECs by MRSA. (B and C) Total cellular association with hVECs (B) and invasion of hVECs (C) by WT MRSA after 30-min treatment with anti-α₅β₁ integrin antibody, nonspecific IgG antibody, or no antibody. The data are expressed as the percent cell-associated (B) and cell-invaded (A and C) MRSA relative to the initial inoculum. Experiments were performed in quadruplicate, and the means of three technical replicates are shown for each experiment; error bars represent the SEM. Statistical analyses were performed by ordinary one-way ANOVA with Tukey’s multiple-comparison tests, and all comparisons with P < 0.05 are shown.

It is known that the main pathway for S. aureus FnBP-mediated invasion of epithelial and endothelial cells involves bridging between FnBPs and human α₅β₁ integrin by fibronectin (52–55). We used flow cytometry with an anti-α₅β₁ integrin antibody to confirm the presence of surface-expressed α₅β₁ integrin on our hVEC cell line (see Fig. S2). To assess whether MRSA also uses the α₅β₁ integrin pathway to invade hVECs, we pretreated confluent hVEC monolayers with either murine anti-human α₅β₁ antibody, a nonspecific murine IgG control, or no antibody for 1 h before performing concurrent total cell association and invasion assays with WT MRSA. While blocking α₅β₁ integrin on the surface of hVECs had no effect on the overall ability of MRSA to associate with hVECs (Fig. 3B), the percentage of invading bacteria was reduced by 20- to 30-fold (Fig. 3C). This result indicates that bridging with α₅β₁ integrin is likely the main pathway used by MRSA to invade hVECs without affecting MRSA extracellular adherence.

Results can be recapitulated in a multilayer air-liquid interface hVEC culture model.

While standard monolayer vaginal epithelial cell culture is an invaluable tool for the characterization of interactions between human cell surfaces and bacteria, these models are unable to assess bacterial ability to penetrate stratified epithelium, the environment in which colonization is normally established in vaginal tissue (56, 57). Recently, models have been established using Transwell insert systems to grow polarized and differentiated multilayer cultures of vaginal epithelial cells (58–61). To supplement knowledge gained from studies on hVEC monolayers, we used an air-liquid interface Transwell cell culture system to induce vaginal cell differentiation and maturation (61). This method of cell culture has been shown to induce polarized cell layer differentiation, form apical anucleate cells, produce both glycogen and mucus that gathers to the apical surface, and show similar expression of Toll-like receptors and other surface proteins that respond to pathogens (56, 61–63). To confirm hVEC cell layer growth and differentiation in our air-liquid interface culture system, several Transwell inserts were excised for sectioning and staining at two different time points (Fig. 4A and B). Compared to the flat squamous cells seen on the polycarbonate membrane on day 1 (Fig. 4A), the cells at day 8 can be seen to form a multilayer culture that ranges from four to eight cells deep (up to 12 in some areas [not shown]) with a layer of thin desquamated cells at the air interface (Fig. 4B). Total cell association and invasion assays were repeated with WT, ΔfnbAB, fnbA::Tn, and fnbB::Tn strains on these differentiated hVEC cultures. Roughly 60 to 70% of the MRSA WT was able to associate with the differentiated cultures after 30 min, while 12 to 25% was able to invade after 3 h. The single FnBP mutants were able to associate with the differentiated hVEC cultures similarly to the MRSA WT (Fig. 4C). Invasion also showed a similar trend, as the fnbA::Tn mutant was less able to invade than the WT and fnbB::Tn mutant (Fig. 4D). As in the monolayer-based assays, the ΔfnbAB mutant had a more pronounced defect in associating with hVECs than all other strains tested and was not able to invade the differentiated culture (Fig. 4C and D).

FIG 4.

MRSA FnBPs are important for cell association and invasion in a three-dimensional cell culture model of differentiated hVECs. H&E-stained images at ×40 magnification of a cross section of day 1 (A) and day 8 (B) vaginal cultures grown in transwells. (C and D) MRSA total cell association with hVEC cultures (C) and invasion of day 8 to 9 differentiated hVEC cultures (D). The data are expressed as the percent cell-associated (C) and cell-invaded (D) bacteria relative to the initial inoculum. Experiments were performed in triplicate, and means for each experiment are shown; error bars represent the SEM. Statistical analyses were performed by ordinary one-way ANOVA with Tukey’s multiple-comparison tests.

Heterologous expression of MRSA FnBPs in nonadherent S. carnosus is sufficient for Fn binding, hVEC cellular association and invasion, and biofilm formation.

It is known that the nonpathogenic S. carnosus strain TM300 does not possess any fibronectin-binding surface proteins and that heterologous expression of S. aureus FnBPs in S. carnosus confers invasiveness of primary human kidney embryonic cells (64). Therefore, this strategy was used to test whether FnBPs are sufficient for hVEC cellular association, invasion, and biofilm formation. To reduce any cellular association or biofilm phenotypes resulting from other S. carnosus adhesins, we used a TM300-derived strain lacking the clfA gene and a putative adhesin, SCA_2092 (AH5687; TM300 ΔclfA ΔSCA_2092::gfp), to express either the pFnbA4 or pFnbB4 plasmids. AH5687 strains expressing the MRSA FnBPs (abbreviated as “S.c. pFnbA” and “S.c. pFnbB” in Fig. 5) were confirmed to bind fibronectin significantly more than both the parental strain and MRSA (Fig. 5A). Expression of each FnBP in S. carnosus is sufficient to associate with and invade hVECs significantly better than the nonadherent S. carnosus WT strain (Fig. 5B and C). However, only expression of FnBPA, not FnBPB, confers hVEC invasiveness to the levels of the MRSA WT (Fig. 5C). We were unfortunately unable to measure biofilm formation on hVECs using these S. carnosus strains. Because AH5687 expresses GFP chromosomally at lower levels than the S. aureus plasmid-expressing strains (Fig. 2), the background fluorescence from the epithelial cells obscured our ability to measure biofilm formation on hVECs. Therefore, chamber slides prepared with 20% human plasma in phosphate-buffered saline (PBS) were inoculated with one of the three strains, incubated, and imaged. Quantification of the green fluorescent protein (GFP) channel indicated that each FnBP is sufficient for robust biofilm formation on plasma-coated plates compared to the S. carnosus adhesin-deficient empty vector strain (Fig. 5D to G).

FIG 5.

Heterologous expression of FnBPs in S. carnosus is sufficient for Fn binding, hVEC adherence and invasion, and biofilm formation. (A) Binding to immobilized human Fn measured by determining the optical density of crystal violet-stained bacteria attached to Fn-coated plate after 1 h. (B and C) Total cell association with hVECs (B) and invasion of hVECs (C) by WT MRSA and S. carnosus. (D) GFP channel mean fluorescent intensity measurements of GFP-expressing S. carnosus strains incubated for 18 h on human plasma-coated chamber slides. Ten images were taken from two separate slide chambers (n = 20) for each strain. (E to G) Median representative images are shown for S. carnosus empty vector, S. carnosus pFnbA, and S. carnosus pFnbB (S.c. empty vector [E], S.c. pFnbA [F], and S.c. pFnbB [G]). The data are expressed as the percent cell-associated (B) and cell-invaded (C) bacteria relative to the initial inoculum. Means of three experiments (A to C) or means of 20 images (D) are shown; error bars represent the SEM. Statistical analyses were performed by ordinary one-way ANOVA with Tukey’s multiple comparisons. **, P < 0.01; ***, P < 0.001.

FnBPs contribute to ascension into upper reproductive tract tissues in a murine model.

Because the MRSA ΔFg-binding mutant was less able to colonize all FRT tissues in vivo in a previous study (42), we hypothesized that FnBPs contribute to FRT colonization and ascension in our murine model of bacterial vaginal colonization. Mice were first treated with β-estradiol to synchronize their estrous cycles. The next day, mice were vaginally inoculated with 10⁷ CFU of MRSA in PBS, and the following day, the mice were sacrificed. Homogenized vaginal, cervical, and uterine tissues were then plated on selective media to assess MRSA bacterial burden. A high burden of MRSA was observed in reproductive tract tissues in nearly all mice after 24 h, which is consistent with previous results seen in this model. Compared to the WT strain, we recovered significantly less of the ΔfnbAB strain from vaginal, cervical, and uterine tissues (Fig. 6).

FIG 6.

Fibronectin-binding proteins contribute to MRSA colonization of murine FRT tissues. CFU counts of MRSA recovered from murine vaginal (A), cervical (B), and uterine tissues (C) at 24 h postinoculation with either WT or ΔfnbAB strains are depicted. The data are expressed as the log CFU per gram of excised tissue. Experiments were performed in triplicate with 8 to 10 mice per inoculation group. Statistical analyses were performed by unpaired t tests, and P values are shown for each comparison.

DISCUSSION

Because individuals colonized with S. aureus are most likely to become infected with strains that they carry (2, 4, 23), it is important to study bacterial colonization at mucosal reservoirs, such as the nares and vaginal tract. Though the consequences of MRSA vaginal colonization and ascension have been explored (9, 11, 14, 65–68), few studies have examined the mechanisms responsible for these processes. Adherence to host tissues is a prerequisite to colonization and establishment of bacterial infection (69), and in MRSA, this process is mediated by sortase-anchored surface adhesins that bind to extracellular matrix molecules. We used in vitro and in vivo models of vaginal colonization to explore the role of FnBPs in establishing MRSA colonization in the vaginal tract and ascending into upper reproductive organs. The results of this study show that MRSA FnBPs significantly contribute to association with vaginal epithelia, biofilm formation at the epithelial surface, invasion of hVECs, and colonization of FRT tissues in vivo.

A previous study identified that a multimutant of several families of fibrinogen-binding proteins attenuated the ability for USA300 to associate with vaginal and cervical epithelia and colonize the murine female reproductive tract (42). The present study identified FnBPs as the critical family of adhesins for vaginal cell association and found that expression of either FnBPA or FnBPB is sufficient for USA300 association with vaginal epithelia. FnBPs are known to be critical surface proteins for S. aureus cellular association and biofilm formation (48, 70, 71), but their role in vaginal colonization was not previously characterized. While FnBP expression is typically limited to exponential growth phase in methicillin-sensitive S. aureus, they have been shown to be robustly expressed well into stationary phase in MRSA (72). FnBPs contribute to cellular association by binding to various extracellular matrix proteins. FnBPs bind Fn by their 10 (FnBPB) or 11 (FnBPA) Fn-binding repeats (FnBRs), which each have various binding affinities for Fn. Both FnBPA and FnBPB are also known to bind both fibrinogen (Fg) and elastin with their A domain. Since Fg binding is known to be important for bacterial reproductive tract colonization (42, 73), it is also possible that the Fg-binding capabilities of S. aureus FnBPs could work alongside Fn binding to contribute to FRT colonization. While Fn binding appears to be the primary function of FnBPA, FnBPB is known to have high binding affinity for loricrin, which can promote corneocyte attachment (74), as well as affinity for plasminogen and histones, which can protect S. aureus from neutrophil extracellular traps (NETs) (75, 76). While ligand binding can be an important defense mechanism in the reproductive tract, binding of bacteria to one another in the form of a biofilm is also beneficial for survival in many niches (26, 77, 78).

It is important to understand the determinants of MRSA biofilm formation in the vaginal tract as biofilm formation has been implicated in treatment failure and recurrence of vaginal infections (30). Multiple studies have shown that FnBPs are important for S. aureus biofilm formation on abiotic surfaces and in device-associated infections, that FnBPs are expressed at higher levels in biofilms than in broth culture, and that biofilm formation is facilitated by low-affinity homophilic bonds between the A domains of FnBPA or FnBPB proteins (44, 46, 47, 72, 79). While many studies have investigated Staphylococcal biofilms on abiotic or serum/plasma-coated surfaces, few have investigated the possibility of biofilm formation at mucosal epithelial barriers. Though S. aureus biofilms have been detected on epithelial tissue from vaginal washes and grown on human gestational membranes (32, 33), no previous studies have investigated mechanisms of S. aureus vaginal biofilm formation to our knowledge. Our study is the first to show that MRSA can form robust, FnBP-dependent biofilms attached to human vaginal epithelial cells when grown statically in vitro. A previous study of USA300 LAC FnBP-mediated biofilms on an abiotic surface showed that complementation with either FnBPA or FnBPB in an fnbAB mutant fully restored biofilm formation under flow chamber conditions (46), while our study showed that on vaginal epithelia, optimal biofilm formation is not observed in strains expressing either FnBPA or FnBPB. These differences could be due to lower availability of FnBPs for homophilic interactions after binding to hVECs or differences between biofilm capabilities of single mutant fnbA and fnbB strains in this study compared to the multicopy complementation plasmids used in the previous study. Better understanding of MRSA biofilm formation in the vaginal tract can inform anti-biofilm treatment strategies to decrease their presence in colonized individuals at-risk for health complications.

FnBPs are also known to be involved in promoting MRSA invasion of a variety of nonphagocytic cells (52, 80, 81). In this work, we show that while FnBPB can facilitate low levels of invasion, FnBPA is the predominant adhesin to facilitate invasion of MRSA into human vaginal epithelial cells. In addition, we confirm that MRSA is internalized by the well-studied mechanism of FnBP Fn-bridging with human α₅β₁ integrin (50, 82). It is still unclear why MRSA invades nonphagocytic cells and what function it would serve in colonization of the reproductive tract. It has been proposed that this internalization could provide protection from antimicrobials, the acidic environment, or from immune factors that are abundant in vaginal epithelia (57, 70, 83). In addition, intracellular reservoirs of S. aureus have been associated with diseases such as recurrent osteomyelitis and rhinosinusitis (84, 85). Thus, it is possible that internalization into vaginal epithelia could serve as a reservoir for persistent colonization in the FRT.

Multilayer air-liquid interface cell culture models have been used to model Staphylococcal colonization of airway epithelia, vaginal colonization of common vaginal commensals, and FRT colonization of opportunistic pathogens such as Mycoplasma genitalium, Chlamydia trachomatis, and Neisseria gonorrhoeae (59–61, 86). In our study, we observed similar cell layer differentiation to that of another hVEC multilayer air-liquid interface model that was able to support the growth of vaginal Lactobacilli and common opportunistic pathogens (60). We also noted similar trends in cellular association and invasion phenotypes between the WT and fnbAB mutant tested in monoculture and multilayer systems; however, association and invasion percentages were lower overall for strains tested in multilayer culture compared to monolayer. This difference in colonization is likely a consequence of the layer of anucleate, desquamated cells at the air interface providing some protection of the more susceptible, actively replicating cells near the multilayer base (57). Though this barrier seems to confer some protection, it is worthy of note that all MRSA mutants in this study are still able to associate with the vaginal multilayers and that all strains except the fnbAB double mutant can invade hVECs. We believe that this multilayer air-liquid interface model and its ability to recapitulate the activity of stratified epithelium more accurately than monolayer culture will be important for future studies on bacterial FRT colonization.

Understanding whether FnBPs are sufficient for processes involved in vaginal tract colonization can allow us to predict whether other strains which have acquired FnBPs may be able to colonize this environment. Heterologous expression of surface proteins in S. carnosus is a valuable tool to study the acquisition of virulence traits in a nonpathogenic staphylococcal strain (87, 88). A previous study has shown that expression of either FnBPA or FnBPB in S. carnosus TM300 conferred invasiveness and cellular association when incubated on primary human embryonic kidney cells that was equal to the ability of the S. aureus strain Cowan 1 (64). In our study, cellular association with hVECs was only partially complemented by heterologous expression of FnBPA or FnBPB, and the FnBPB-expressing strain was significantly less invasive than both WT MRSA and the FnBPA-expressing strain (Fig. 5B and C). This could partly be due to the use of Cowan 1, a methicillin-sensitive strain, in the previous study compared to our parent strain USA300 LAC, an invasive MRSA isolate with high FnBP expression, and the difference in human cell types used for association and invasion studies. One condition where these studies did overlap, however, is the ability for FnBPA-complemented S. carnosus to invade equally to the parent S. aureus strain. This S. carnosus pFnbA invasion phenotype, combined with the observation that the MRSA fnbA mutant was more severely attenuated in its ability to invade hVECs than the fnbB mutant (Fig. 3), reinforces the idea that FnBPA-mediated invasion through Fn bridging interactions with human integrin is the main process mediating S. aureus invasion of hVECs. It is also important to note that heterologous expression of FnBPs was able to confer increased biofilm formation capabilities to S. carnosus on plasma-coated plates, as similar biofilm phenotypes have been observed in another study of USA300, mentioned above, expressing the same plasmid-expressed FnBPs on abiotic surfaces (46).

In this study, we report the use of a mouse model for investigating the contribution of FnBPs to colonization of the murine FRT. While similar to previous models used for studying Group B Streptococcus (GBS), Enterococcus faecalis, and S. aureus vaginal colonization (42, 89, 90), this model focuses on early colonization and initial ascension into upper reproductive tissues. Using this model, we found that the ΔfnbAB mutant is less able to colonize FRT tissues compared to WT MRSA. Based on the findings from our in vitro studies, the mutant’s decreased ability to associate with vaginal epithelia, form biofilms, or invade epithelial cells could make them more susceptible to microbial competition or host defenses within the FRT, leading to their reduced ability to colonize. It is noteworthy, however, that while the loss of FnBPs causes a steep change in MRSA’s ability to associate with hVECs in vitro, the difference between the WT and ΔfnbAB mutant in vivo is less drastic. This phenomenon is likely due to the more complex environment in vivo that includes endogenous microbiota and mucosal immunity. In addition, while our study identified that FnBPs contribute to ascension in the murine model, the mechanism of MRSA FRT ascension remains unknown. Another FRT-colonizing bacterium, GBS, is known to ascend by inducing vaginal epithelial exfoliation through interactions with human β₁ integrin (91). It remains to be investigated whether MRSA induces a similar process though interactions between human and MRSA surface proteins to ascend.

In summary, this study identifies several mechanisms by which FnBPs contribute to MRSA colonization of the FRT. Understanding the role of FnBPs for cellular association, biofilm formation, invasion, and ascension in this niche underscores the importance of S. aureus adhesins in colonization of mucosal surfaces. Further studies to identify bacterial factors contributing to MRSA colonization and ascension as well as characterizing interactions with the immune system and endogenous microbes will be critical for defining how MRSA persists and progresses to infection in the FRT.

MATERIALS AND METHODS

Bacterial strains.

The bacterial strains and plasmids used in this study are described in Table 1. S. aureus strain USA300 LAC (12) was used for all MRSA experiments. S. aureus was grown in tryptic soy broth (TSB) at 37°C, and growth was monitored by measuring optical density at 600 nm (OD₆₀₀). For selection of S. aureus mutants, tryptic soy agar (TSA) was supplemented with 10 μg/mL chloramphenicol (Cm), 10 μg/mL erythromycin (Erm), or 2 μg/mL tetracycline (Tet).

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Genotype or description	Source or reference
Strains
S. aureus
AH1263	USA300 LAC WT	12
AH4416	ΔclfA clfB::Tn ΔsdrCDE::tet ΔfnbAB	42
AH4037	ΔclfA	97
AH2905	clfB::Tn	This study
AH3157	ΔsdrCDE::tet	This study
AH4392	ΔfnbAB	42
AH5902	fnbA::Tn	This study
AH5901	fnbB::Tn	This study
S. carnosus
AH5687	TM300 ΔclfA ΔSCA_2092::gfp	98
Plasmids
pFnbA4	Multicopy plasmid expressing fnbA; Ampr Cmr	43
pFnbB4	Multicopy plasmid expressing fnbB; Ampr Cmr	43
pHC48	S. aureus constitutive DsRed expression vector; Cmr	95

Phage transduction with phage 11 was used to move mutations and plasmids between S. aureus strains (92). To generate the fnbA::Tn, fnbB::Tn, and clfB::Tn mutants, mariner-based transposon bursa aurealis mutations from the Nebraska Transposon Mutant Library (93) in strain JE2 (NE186, fnbA::Tn; NE728, fnbB::Tn; NE391, clfB::Tn) were transduced into LAC and selected with Erm. The ΔsdrCDE::tet mutation (AH3157) was also made by phage transduction with phage 11 from Newman DU5973 (94) to LAC and selected with Tet. Similarly, the multicopy plasmids pFnbA4 and pFnbB4 (43) were moved from strain DU5883 into LAC and selected with Cm. A phage 11 stock containing the DsRed expression plasmid pHC48 (95) was transduced into the LAC WT, ΔfnbAB, fnbA::Tn, and fnbB::Tn to make DsRed-expressing strains AH5925 to AH5928, respectively. Similar growth and expression of DsRed was assessed in TSB over 24 h and confirmed (see Fig. S1).

To create the S. carnosus strains expressing FnBPA and FnBPB, either plasmid pFnbA4 or pFnbB4 (43) was transformed into electrocompetent AH5687 cells by a previously described electroporation protocol (96). Transformants were selected on TSA plates with 10 μg/mL Cm grown at 30°C for 48 h.

Epithelial cell culture conditions.

Immortalized VK2/E6E7 human vaginal epithelial cells (99) were obtained from the American Type Culture Collection (ATCC; CRL-2616) and were maintained in keratinocyte serum-free medium (KSFM; Gibco) with 0.1 ng/mL human recombinant epidermal growth factor (Gibco) and 0.05 mg/mL bovine pituitary extract (Gibco) at 37°C with 5% CO₂.

In vitro MRSA cell association assays.

VK2 cell monolayers were seeded and grown in KSFM in Costar 48-well plates (Corning) for the completion of in vitro assays. Since this medium is serum-free and no additional serum is added, the below in vitro assays are free from outside sources of serum components, including fibronectin and fibrinogen. Assays to determine total cell-associated and invaded MRSA were performed as described previously (42, 100).

(i) Bacterial total cell association assay. Overnight bacterial cultures were subcultured in 5 mL of TSB (Sigma-Aldrich) and grown to mid-log phase (OD₆₀₀ ~0.4 to 0.5), spun down for 5 min at 3,100 relative centrifugal force (rcf), resuspended in PBS (Fisher), and normalized to an OD₆₀₀ of 0.4. Confluent hVEC monolayers were infected with bacteria at a multiplicity of infection (MOI) of 1, centrifuged at 200 rcf for 5 min, and incubated at 37°C and 5% CO₂ for 30 min. After a 30-min incubation, wells were washed five times with 0.5 mL of sterile PBS/well. Then, 0.05 mL of 0.25% trypsin-EDTA solution was added to each well and incubated for 5 min at 37°C and 5% CO₂ to detach the cells from the Corning plate. Cells were then lysed with 0.2 mL of 0.025% Triton X-100 in PBS by vigorous pipetting. The lysates were then serially diluted and plated on TSA to enumerate bacterial CFU. Experiments were performed at least three times under each condition in triplicate.

(ii) Bacterial cellular invasion assay. Experiments were performed as in the above total cell association assay until the initial incubation following bacterial inoculation. After inoculation, the plate was incubated at 37°C and 5% CO₂ for 2 h. After this incubation, wells were washed three times with 0.5 mL of sterile PBS/well. A solution of 100 μg/mL gentamicin and 5 μg/mL lysostaphin in KSFM was added to each well (0.5 mL/well), followed by incubation at 37°C and 5% CO₂ for 1 h to kill all extracellularly attached MRSA. Then, steps were followed as in total cell association assay, including PBS washing, trypsin treatment, Triton X-100 treatment, and plating for CFU.

(iii) Antibody blocking assays. Antibodies used in these experiments are anti-integrin α₅β₁ clone JBS5 (MAB1969; Sigma-Aldrich) and Ms IgG isotype control (Invitrogen). Fresh stocks of 10 μg/μL antibody in KSFM were made for each experiment. For the antibody blocking bacterial total cell association and invasion assays described above, VK2 monolayers were washed once with PBS, and 100 μL of a 10-μg/μL concentration of antibody in KSFM was added to each well. hVECs were incubated at 37°C and 5% CO₂ for 1 h. Cells were then washed once with PBS before adding 200 μL of fresh KSFM and then inoculated with 50 μL of bacteria in PBS at an MOI of 1. After inoculation, each assay was performed as described above.

Bacterial cellular association microscopy.

hVEC cell monolayers grown in glass chamber slides (Nunc Lab-Tek II Chamber Slide System; Thermo Fisher) were inoculated with fluorescent bacterial strains at an MOI of 10 and incubated statically at 37°C and 5% CO₂ for 18 h to allow MRSA to associate with the hVECs and form biofilms. After incubation, chamber slides were gently rinsed with PBS three times and stained with DAPI (Fluoroshield with DAPI; Sigma-Aldrich) before affixing glass coverslips. Sealed chamber slides were imaged on a Keyence BZ-X Microscope at ×20 magnification. Images were only taken in fields of view with full coverage of VK2 cells indicated by presence of DAPI-stained nuclei visible in DAPI channel. Ten images/chamber were taken for each strain, and a total of three individual chambers were used for fluorescence quantification (n = 30). The mean fluorescence intensity of each TRITC (tetramethyl rhodamine isothiocyanate) channel (red channel, images dsRed fluorescing bacteria) was measured using ImageJ software (W. S. Rasband, ImageJ, U.S. National Institutes of Health, Bethesda, MD [https://imagej.nih.gov/ij/], 1997-2018.). Biofilm formation was measured as a function of mean fluorescence intensity of the TRITC channel at ×20 magnification.

S. carnosus biofilm microscopy.

Since the GFP-expressing S. carnosus strains were not able to be visualized on hVECs due to background fluorescence, plasma-coated chamber slides were used as a proxy to visualize whether heterologous expression of FnBPs is sufficient for biofilm formation without the presence of vaginal epithelium. Glass chamber slides were coated with 250 μL of 20% human plasma in PBS (pooled human plasma; Innovative Research) and left overnight at 4°C. After the plasma-coated slide was rinsed with PBS, the chambers were inoculated with 10⁷ CFU of S. carnosus and incubated statically at 37°C for 18 h. After incubation, chamber slides were gently rinsed with PBS three times before affixing glass coverslips with mounting media (Cytoseal 60; Richard-Allan Scientific). Ten images/chamber were taken for each strain, and a total of two individual chambers were used for fluorescence quantification (n = 20). Mean fluorescence intensity of the GFP channel was measured using ImageJ software, as described above. Biofilm formation was measured as a function of mean fluorescence intensity of the GFP channel at ×20 magnification.

Fibronectin binding assay.

Wells of a Costar 96-well plate (Corning) were coated with 50 μL of 20-μg/mL fibronectin (fibronectin, human plasma; EMD Millipore) in PBS, and the plate was left overnight at 4°C. The next day, an overnight culture of bacteria was subcultured 1:100 in 10 mL of TSB and grown at 37°C until reaching an OD₆₀₀ of ~1.5. Roughly 30 min after starting bacterial cultures, the fibronectin-coated plate was washed three times with 0.1 mL of PBS per well, and then 0.15 mL of 1% bovine serum albumin (Fisher) in PBS was added to each well. The fibronectin plate was then incubated for 2 h at 37°C. After the bacterial cultures reached an OD₆₀₀ of ~1.5 (~2.5 h), 5 mL of culture was spun down, and the TSB was removed to normalize the concentration in 2 mL of PBS to an OD₆₀₀ of 1.0. The fibronectin plate was then rinsed three times with 0.2 mL of PBS before adding 0.1 mL per well of bacterial solution at an OD₆₀₀ of 1.0, making sure to leave uninfected fibronectin-coated control wells. The plate was then incubated for 1 h statically at 37°C. After incubation, each well is washed once with 0.15 mL of PBS and incubated for 1 h at 60°C to dry all liquid from the wells. Then, 0.08 mL of 0.1% crystal violet in dH₂O was then added to each fibronectin-coated well. The plate was incubated at room temperature for 3 to 5 min before vigorous washing under tap water and full drying by inverted tapping onto paper towels. Next, 0.08 mL of 33% acetic acid was used to dissolve the dried crystal violet stain by gentle pipetting, and then the OD₅₇₀ was measured on a Tecan plate reader. Final OD₅₇₀ values were calculated by averaging technical replicates and subtracting the average value of the fibronectin only blanks.

Differentiated vaginal epithelial cell model.

Costar Transwell permeable support plates (6.5-mm insert, 24 wells, 0.4-μm-pore-size polycarbonate membrane; Corning) were used for the differentiated vaginal epithelial cell model. We added 0.7 mL of supplemented KSFM (see “Epithelial Cell Culture Conditions” above) to the bottom chamber of the Transwell plates before adding 5 × 10⁴ VK2 cells in 250 μL of supplemented KSFM to the Transwell insert. After 48 h, the KSFM in the bottom chamber was changed, and all liquid media were gently aspirated from the insert to create an air-liquid interface. Media in the bottom chamber were then changed, and any medium that leaked into the insert was removed every 2 days for 8 to 10 days. After 6 to 7 days, no more media from the bottom chamber should leak into the insert, indicating a mature multilayer culture made up of several cell layers. The culture is allowed to incubate for 8 to 10 days after initial seeding before being used for in vitro assays. Bacterial total cell association and bacterial invasion assays were performed as described previously (see “In Vitro MRSA Cell Association Assays”). To obtain hematoxylin and eosin (H&E)-stained cross sections of multilayer cultures, day 1 and 8 cultures were removed from the Transwell system, and polycarbonate membranes were removed with a 6-mm biopsy punch (Fisher), placed into histology cassettes, and immersed in 4% formaldehyde (LabChem) for 24 h. Cassettes were then moved into 70% ethanol and submitted to the Gates Center for Regenerative Medicine for sectioning and H&E staining.

Murine model of bacterial vaginal colonization.

A detailed description of this protocol as originally described for Streptococcus agalactiae can be found in the Journal of Visualized Experiments (89). The protocol was later adapted for colonization with MRSA (42). Female 8-week-old CD1 mice (Charles River) were intraperitoneally injected with β-estradiol (0.5 mg in 0.1 mL of sesame oil per mouse) to synchronize the mice in the proestrus stage of their estrous cycles. At this time, mice were also vaginally lavaged with sterile PBS, and lavage fluid was plated on MRSA CHROMagar chromogenic media and incubated at 37°C overnight. Any mice with MRSA colonies in preinoculation vaginal lavage fluid were excluded from the study. At 24 h after synchronization, mice were vaginally inoculated with 10⁷ CFU of mid-log growth phase MRSA USA300 LAC in 0.01 mL of PBS. At 24 h after inoculation, mice were humanely euthanized to collect FRT tissues. The vaginas, cervices, and uteri were dissected, separated, and placed into separate screw-cap homogenization tubes with 1.0-mm zirconia beads. Tubes were weighed before adding tissues and again after 1 min of homogenization at maximum speed in a tissue homogenizer. To quantify bacterial load in each of the tissues, tissue homogenate was serially diluted, and 25-μL portions of the 10⁰ to 10⁻⁴ dilutions were plated on MRSA CHROMagar, followed by incubation at 37°C overnight before enumerating the MRSA CFU.

Statistical analysis.

All statistical analysis was performed using GraphPad Prism software version 9.0, and statistical significance was accepted at P values of <0.05. Specific tests are indicated in the figure legends.