Vaginal community state types (CSTs) alter environmental cues and production of the Staphylococcus aureus toxic shock syndrome toxin-1 (TSST-1)

ABSTRACT

Menstrual toxic shock syndrome (mTSS) is a rare but life-threatening disease associated with the use of high-absorbency tampons. The production of the Staphylococcus aureus toxic shock syndrome toxin-1 (TSST-1) superantigen is involved in nearly all cases of mTSS and is tightly controlled by regulators responding to the environment. In the prototypic mTSS strain S. aureus MN8, the major repressor of TSST-1 is the carbon catabolite protein A (CcpA), which responds to glucose concentrations in the vaginal tract. Healthy vaginal Lactobacillus species also depend on glucose for both growth and acidification of the vaginal environment through lactic acid production. We hypothesized that interactions between the vaginal microbiota [herein referred to as community state types (CSTs)] and S. aureus MN8 depend on environmental cues and that these interactions subsequently affect TSST-1 production. Using S. aureus MN8 ΔccpA growing in various glucose concentrations, we demonstrate that the supernatants from different CSTs grown in vaginally defined medium (VDM) could significantly decrease tst expression. When co-culturing CST species with MN8 ∆ccpA, we show that Lactobacillus jensenii completely inhibits TSST-1 production in conditions mimicking healthy menstruation or mTSS. Finally, we show that growing S. aureus in “unhealthy” or “transitional” CST supernatants results in higher interleukin 2 (IL-2) production from T cells. These findings suggest that dysbiotic CSTs may encourage TSST-1 production in the vaginal tract and further indicate that the CSTs are likely important for the protection from mTSS.

IMPORTANCE

In this study, we investigate the impact of the vaginal microbiota against Staphylococcus aureus in conditions mimicking the vaginal environment at various stages of the menstrual cycle. We demonstrate that Lactobacillus jensenii can inhibit toxic shock syndrome toxin-1 (TSST-1) production, suggesting the potential for probiotic activity in treating and preventing menstrual toxic shock syndrome (mTSS). On the other side of the spectrum, “unhealthy” or “transient” bacteria such as Gardnerella vaginalis and Lactobacillus iners support more TSST-1 production by S. aureus, suggesting that community state types are important in the development of mTSS. This study sets forward a model for examining contact-independent interactions between pathogenic bacteria and the vaginal microbiota. It also demonstrates the necessity of replicating the environment when studying one as dynamic as the vagina.

INTRODUCTION

Staphylococcus aureus is a Gram-positive bacterial colonizer of human mucosal surfaces, including the vagina. This bacterium encodes a wide array of virulence factors, yet the superantigens are a unique subset of exotoxins that trigger an unconventional immune response. Superantigens force the activation of T cells through the cross-linking of major histocompatibility complex class II (MHC-II) molecules to the T cell receptor but irrespective of peptide specificity (1). With up to 20% of the entire T cell repertoire potentially becoming activated (2), superantigens can initiate a highly pro-inflammatory state known as a cytokine storm that may lead to toxic shock syndrome (TSS). Staphylococcal TSS is a serious illness with symptoms including rash, high fever, hypotension, desquamation, and multiple organ dysfunction syndrome, which can become fatal if untreated (3).

The most concerning staphylococcal pathology in the vagina is menstrual TSS (mTSS). The causative agent behind mTSS is the superantigen known as toxic shock syndrome toxin-1 (TSST-1) (4). The production of TSST-1 by S. aureus is tightly controlled by a complex orchestration of regulators and environmental signals (5–7). Recently, it was discovered that the carbon catabolite protein A (CcpA) is the principal repressor of TSST-1 in the vaginal environment as a response to high glucose levels (8). Typically, the vaginal environment contains high glucose levels leading up to menstruation, at which point glucose concentration decreases significantly. The fluctuation in glucose level, therefore, likely contributes to the restriction of mTSS to the time period of menstruation as a result of CcpA-mediated repression.

Other conditions that sustain the production of TSST-1 in this environment are high oxygen and carbon dioxide levels (9, 10), near-neutral pH, and increased protein levels (11). Many of these conditions are met at menstruation, specifically with the use of high-absorbency tampons that can increase the oxygen levels in the typically anaerobic vagina (11, 12). As a result, the environmental cues present in the vaginal environment are critical for the restraint—or in the over-production—of TSST-1.

The vaginal environment is strongly shaped by the microbiota. Classifications of the vaginal microbiota have been grouped into five community state types (CSTs), each fore fronted by a prominent bacterial species (13). Lactobacillus crispatus, Lactobacillus gasseri, Lactobacillus iners, and Lactobacillus jensenii dominate CST-I, CST-II, CST-III, and CST-V, respectively, with CST-I and CST-III being the most common (13). Vaginal lactobacilli are often ascribed a protective function, with the exception of L. iners that is considered a transitional state (14). The final group, CST-IV, is a polymicrobial community composed mostly of strict anaerobes, as well as the prominent member Gardnerella vaginalis (15). CST-IV is strongly associated with dysbiosis and bacterial vaginosis (BV), the most common vaginal disorder among women of childbearing age (16).

A protective role for vaginal lactobacilli is mainly attributed to their critical role in the acidification of the vaginal environment via the metabolism of glycogen into lactic acid. The appearance of lactobacilli in the vaginal microbiota at puberty, and their maintenance until menopause, is further postulated to be due to the role of estrogen in vaginal epithelial cell differentiation and release of glycogen (17, 18). For instance, the presence of L. crispatus has been directly correlated with estrogen levels (19), and the use of hormonal therapy in menopausal women (20) and the use of estrogen rings among transgender men have restored Lactobacillus levels (21). However, the inability to explicitly quantify a relationship between estrogen and cell-free glycogen is hypothesized to be due to a lag time in the maturation of basal vaginal epithelial cells and the subsequent release of glycogen from superficial cells (22, 23). Nonetheless, free glycogen is positively correlated with lactobacilli levels and low vaginal pH, with pre-menopausal women having higher levels of lactobacilli than post-menopausal women (24). The inverse relationship between glycogen and pH corresponds to the strong association between lactobacilli with both high levels of free glycogen and with low pH (22, 23). At menstruation, the decrease in glucose concentration results in a decrease in the abundance of lactobacilli and an increase in microbial diversity, thereby increasing the vaginal pH (15).

Considering that the production of TSST-1 is favored in a low glucose environment with increased pH, the drop in estrogen and subsequent decrease in lactobacilli at menstruation create conditions allowing for mTSS to occur. In this study, we aimed to investigate how CSTs I–V may alter S. aureus MN8 TSST-1 production in various in vitro conditions that mimic the vaginal environment. Using a tst transcriptional reporter and anti-TSST-1 immunoblot assays, we found that supernatants from representative CST spp. can significantly decrease TSST-1 expression in the absence of CcpA-mediated glucose repression. We further demonstrate that L. jensenii can overcome a variety of environmental stressors to inhibit the production of TSST-1 by S. aureus MN8 ∆ccpA and may be a rational probiotic candidate to pursue in order to prevent re-occurrence of mTSS.

RESULTS

Supernatants from CSTs have repressive activity hidden behind glucose repression in VDM

Vaginally defined medium (VDM) was developed and adapted to mimic the vaginal environment and its secretions, containing high glucose and glycogen concentrations (25, 26). Although VDM represents the conditions found in the vagina, one major environmental driver that is absent in the media is the signaling from the resident microbiota. To investigate the effects of CSTs I–V on the expression of tst, each CST representative was grown in VDM at various glucose concentrations, and the supernatants were collected for the use in a luciferase reporter assay that measures tst transcription. To account for the potential presence of potent inhibitory molecules in the supernatant and depletion of nutrients, the supernatants were diluted with their respective fresh media prior to the assays. The glucose concentrations of 60, 0.7, and 0 mM were used in accordance with the development of VDM and its ability to sustain bacterial growth (25) and with previous in vitro work examining the relief of CcpA repression (8). Blood glucose is affected by the menstrual cycle phase, with a gradual increase during ovulation, a peak during the luteal phase (~6 mM) and a sharp decline at menstruation (27). While it is postulated that the vagina also contains a glucose concentration of greater than 5 mM (28), which indeed represses tst (8), 60 mM was used to ensure that glucose was not depleted in these in vitro experiments containing high bacterial density.

In supernatants containing 60 mM glucose, repression of tst was evident in the wild-type S. aureus MN8 luminescent reporter (Fig. 1A). To relieve glucose repression, the assay was repeated with CST supernatants grown in VDM containing 0.7 or 0 mM glucose. Under these de-repressed conditions compared with control media, most CST supernatants significantly decreased tst expression (Fig. 1B and C). As a second method of relieving glucose repression, the MN8 ccpA mutant strain was used, and similar trends compared with wild-type were observed where all the CST supernatants except L. gasseri decreased tst transcription in high (60 mM) and low (0.7 mM) glucose (Fig. 1D and E); however, without glucose in the CST supernatants, there was no significant decrease except with the L. iners supernatant (Fig. 1F). Overall, these data indicate that different representative CST supernatants have repressive activity for TSST-1 transcription but that this is hidden behind CcpA-mediated glucose repression.

Fig 1.

Supernatants from CSTs I–V have repressive activity hidden behind glucose repression in VDM. CSTs I–V were grown in standard VDM (60mM glucose), low glucose VDM (0.7mM glucose), or VDM lacking glucose (0mM glucose). CST supernatants were filtered (0.2 µm) and diluted with the respective fresh media at a 1/4 dilution. Wild-type MN8 (A, B, and C) and MN8 ∆ccpA (D, E, and F) containing the luminescence reporter plasmid pAmilux::P_tst were then grown in the diluted supernatants. The assay was performed in the Synergy H4 reader at 37°C and continuous shaking on the medium setting. The relative luminescence units were standardized to the growth (OD₆₀₀) (Fig. S1 and S2) to determine relative P_tst expression. Error bars represent mean ± SD. Ordinary one-way analysis of variance was performed with GraphPad Prism 9, no significance not shown (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001).

Supernatants from stable CSTs do not eliminate the production of the TSST-1 protein by MN8

While the luciferase reporter assay highlights repression of TSST-1 at the transcriptional level, the fully functional protein must be produced for mTSS to occur. To investigate whether stable CST supernatants (I, II, and V) have similar activity at the transcriptional and translational levels, wild-type MN8 and MN8 ∆ccpA were grown in the respective diluted supernatants. TSST-1 production was assessed by Western blot 4 hours after inoculation (Fig. 2), as transcriptional data indicated that this is when peak tst expression occurs (Fig. S2), as well as after overnight growth (Fig. 2). At both the 4-hour and overnight time points, TSST-1 production by wild-type MN8 was not detected in 60 mM glucose experiments (Fig. 2A and D), yet TSST-1 production was recovered by decreasing the glucose concentration (Fig. 2B, C, E, and F). Using MN8 ∆ccpA, TSST-1 was detected in all conditions tested, regardless of glucose level (Fig. 2G through L), indicative of the role of CcpA in glucose-mediated repression. However, in L. crispatus and L. jensenii 60 mM glucose supernatants at the 4-hour time point, the TSST-1 bands were noticeably less intense relative to the medium control (Fig. 2G). Examining the raw luminescence reads (Fig. S2D), this reduction is consistent with a 1- to 2-hour delay and reduction in peak tst expression when exposed to L. crispatus and L. jensenii supernatants. Altogether, these results indicate that the restriction of tst by the CST supernatants does not abrogate the production of the protein but rather may cause a delay in some instances, which is overcome as protein accumulates beyond 4 hours.

Fig 2.

Supernatants from CSTs I and V can delay production of TSST-1. CSTs I, II, and V were grown in standard VDM (60mM glucose), low glucose VDM (0.7mM glucose), or VDM without glucose (0mM glucose). CST supernatants were filtered (0.2 µm) and diluted with the respective fresh media at a 1/4 dilution. Wild-type MN8 and MN8 ∆ccpA were then grown in the diluted supernatants for either 4 hours (A–C and G–I) or overnight (D–F and J–L). Supernatants were TCA precipitated (standardized to 12 OD₆₀₀ units), followed by SDS-PAGE and anti-TSST-1 Western blotting.

Supernatants from transitional and dysbiotic CSTs may increase the production of TSST-1

To examine how transitional or dysbiotic CST supernatants (III and IV) affect TSST-1 production, supernatant experiments were performed using L. iners and G. vaginalis, respectively, at the overnight time point. In standard VDM (60 mM glucose), TSST-1 was not produced by wild-type MN8, although production was recovered as glucose levels decreased (Fig. 3A through C). Using the MN8 ∆ccpA mutant, TSST-1 was produced across all glucose conditions in both fresh VDM and CST supernatants (Fig. 3D through F). In 0.7 mM glucose VDM conditions, the TSST-1 bands from MN8 and MN8 ∆ccpA grown in L. iners and G. vaginalis supernatants appeared more intense relative to the VDM control (Fig. 3B and E), overall suggesting that TSST-1 production may be augmented in the presence of L. iners and G. vaginalis supernatants.

Fig 3.

Supernatants from CSTs III and IV may augment production of the TSST-1 protein. CSTs III and IV were grown in standard VDM (60mM glucose), low glucose VDM (0.7mM glucose), or VDM lacking glucose (0mM glucose). CST supernatants were filtered (0.2 µm) and diluted with the respective fresh media at a 1/4 dilution. Wild-type MN8 (A, B, and C) and MN8 ∆ccpA (D, E, and F) were then grown in the diluted supernatants overnight. Supernatants were TCA precipitated (standardized to 12 OD₆₀₀ units), followed by SDS-PAGE and anti-TSST-1 Western blotting.

Transitional and dysbiotic CST supernatants are capable of increasing T cell activation by S. aureus

Although CST supernatants downregulated tst at the transcriptional level during the exponential growth phase (Fig. 1), evaluating TSST-1 protein at the stationary phase demonstrated that this exotoxin continued to be produced (Fig. 2 and 3). This suggests that although the microbiota may have restricting effects on tst expression, once the environmental conditions for TSST-1 production are met, the toxin will still be expressed. This led us to inquire whether the CST supernatants could alter the T cell activation capacity of S. aureus. We exposed peripheral blood mononuclear cells (PBMCs) to the various titrated S. aureus supernatants to evaluate T cell responses. To ensure toxin production, the assay was performed using samples from experiments at 0.7 mM glucose VDM with aeration to mimic environmental conditions typical of mTSS. With wild-type MN8 and the ccpA mutant, all supernatants from S. aureus grown in “stable” CST supernatants were capable of inducing dose-dependent interleukin 2 (IL-2) production similar to S. aureus grown in the VDM control (Fig. 4A and B). However, S. aureus grown in L. iners and G. vaginalis supernatants were able to induce IL-2 concentrations at ~10-fold higher dilution, indicating stronger potency (Fig. 4A and B). Examining the 10⁻⁶ dilution, S. aureus exposed to L. iners and G. vaginalis supernatants had significantly higher IL-2 production for both wild-type S. aureus (Fig. 4C) and the ΔccpA mutant (Fig. 4D). Among the three blood donors, all displayed the same trends in T cell activation; however, one donor was more sensitive to S. aureus exposed to L. iners and G. vaginalis supernatants. These results suggest an interesting trend that supernatants from transitional or dysbiotic CSTs may promote stimulation of T cells by S. aureus.

Fig 4.

Transitional and dysbiotic CST supernatants enhance activation of T cells by S. aureus MN8. Representative CST species were grown in VDM 0.7mM glucose. CST supernatants were filtered (0.2 µm) and diluted with the fresh medium at a 1/4 dilution. Wild-type MN8 (A) and MN8 ∆ccpA (B) were grown in the diluted supernatants with aeration, to mimic mTSS. The S. aureus supernatant was filtered and added to human PBMCs for 18 hours. Cell supernatants from PBMCs were assayed for IL-2 by ELISA as a measure of Tcell activation. Data represent the mean values at the 10⁻⁶ dilution from three healthy donors expressed as percentage relative to the control S. aureus MN8 supernatant for both wild-type S. aureus MN8 (C) and the S. aureus MN8 ΔccpA mutant (D). Each point represents one healthy donor, and the bars represent mean ± SEM. Ordinary one-way analysis of variance was performed with GraphPad Prism 9, no significance not shown (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001).

Lactobacilli in co-culture with S. aureus MN8 lacking CcpA repression can inhibit the production of TSST-1 in conditions mimicking healthy menstruation and mTSS

The luciferase transcriptional reporter assays and supernatant experiments examined the regulation of TSST-1 by growing both actors independently of each other; however, the vagina harbors a multitude of bacterial species simultaneously. These bacterial species may have interactions dependent on signaling that only occur when both species are present, as opposed to solely the bacterial products found in spent media. In order to investigate the production of TSST-1 in a context that better resembles interactions that take place within the vagina, S. aureus MN8 was co-cultured with each of the stable CST representatives.

Co-cultures were performed with S. aureus on one side of a 0.65 µm membrane, and one of the lactobacilli on the other, to allow for contact-independent signaling (Fig. S3). Each side of the co-culture contained VDM with various glucose concentrations, and experiments were performed aerobically to mimic mTSS and microaerophilically to mimic the otherwise low oxygen state of the vagina. As we aimed to identify the restriction of TSST-1 production, transitional and dysbiotic CSTs were excluded from co-culture experiments given the evidence suggesting that they may instead promote TSST-1 production (Fig. 4). Co-cultures were concentrated and normalized to 12 OD₆₀₀ units in order to detect protein, as without concentration, TSST-1 could not be detected by immunoblot (data not shown).

In co-cultures performed with aeration, wild-type MN8 did not produce TSST-1 at 60 mM glucose (Fig. 5A), yet this repression was again relieved at 0.7 mM glucose (Fig. 5E) and 0 mM glucose (Fig. 5I). Under microaerophilic conditions, TSST-1 was not produced at 60 (Fig. 5B) or 0.7 mM glucose (Fig. 5F), but without glucose TSST-1 was faintly detectable in co-cultures with L. crispatus and L. gasseri (Fig. 5J).

Fig 5.

Environmental conditions affect regulation of TSST-1 by healthy lactobacilli co-cultured with S. aureus MN8. A co-culture apparatus separated two bacterial species with a 0.65 µm filter, with each side containing 20 mL of media. Lactobacilli representing CSTs I, II, and V were co-cultured with either wild-type MN8 or MN8 ∆ccpA in VDM at the indicated glucose concentrations. Control co-cultures were performed with the same S. aureus strain on either side of the membrane. Aerobic co-cultures to mimic mTSS environmental conditions were performed at 250 rpm and 37°C (A, E, I, C, G, and K), while microaerophilic co-cultures to mimic healthy menstruation environmental conditions were performed without shaking at 37°C (B, F, J, D, H, and L). TCA precipitations were performed the following day, standardized to 12 OD₆₀₀ units, followed by SDS-PAGE and anti-TSST-1 Western blotting.

Using the MN8 ∆ccpA mutant in aerobic conditions, TSST-1 was produced in all co-cultures, except with L. jensenii at 60 mM glucose (Fig. 5C). When glucose was reduced, this repressive activity by L. jensenii was lost (Fig. 5G and K). In microaerophilic co-cultures, L. jensenii maintained its repressive activity at both 60 (Fig. 5D) and 0.7 mM glucose (Fig. 5H), while L. crispatus gained this activity (Fig. 5D and H) relative to the aerobic co-cultures (Fig. 5C and G). At 0 mM glucose, a switch between repressive activities was seen among the lactobacilli, as L. crispatus and L. jensenii co-cultures supported toxin production, while the L. gasseri co-culture did not (Fig. 5L).

Given that L. jensenii strongly restricted the production of TSST-1 in co-culture conditions that would mimic mTSS, we performed quantitative anti-TSST-1 enzyme-linked immunosorbent assays (ELISAs) to confirm the qualitative Western blot analyses. In conditions with high aeration, co-cultures containing MN8 ∆ccpA on either side of the membrane produced relatively high amounts of TSST-1, as average TSST-1 production at both 0.7 and 60 mM glucose was ~1,500 ng/mL (Fig. 6A and C). However, TSST-1 was below the detection limit of ~300 ng/mL of this assay from aerobic co-cultures containing L. jensenii (Fig. 6A and C). In contrast, microaerophilic co-cultures had limited TSST-1 production in all conditions, indicative of the strong repression that occurs when oxygen is limited (Fig. 6B and D), likely due to the regulator SrrAB (9). Overall, these results reveal that L. jensenii is able to significantly inhibit TSST-1 production in conditions mimicking mTSS when glucose repression is relieved.

Fig 6.

L. jensenii in co-culture with S. aureus MN8 ∆ccpA inhibits production of TSST-1. A co-culture apparatus separated MN8∆ccpA from L. jensenii with a 0.65 µm filter, with each side containing 20 mL of VDM. Control co-cultures were also performed with MN8 ∆ccpA on either side of the membrane. Aerobic co-cultures that mimic mTSS were performed at 250 rpm and 37°C (A and C), while microaerophilic co-cultures that mimic healthy menstruation were performed without shaking at 37°C (B and D). S. aureus supernatants were collected the following day, filter-sterilized, and stored at −20°C until used in a quantitative TSST-1 ELISA. The dotted line represents the limit of detection. Unpaired t-tests were performed using GraphPad Prism 9; no significance was shown (***P ≤ 0.001).

L. jensenii alters S. aureus virulence profile within 4 hours of exposure

The finding that L. jensenii could dramatically decrease the production of TSST-1 in the most substantial number of environmental conditions led us to perform transcriptomic analysis on these notable co-cultures. For these experiments, we compared the transcriptional profile of the control co-culture to that of S. aureus MN8 ∆ccpA grown with L. jensenii at 4 hours after inoculation, close to the time point for peak tst expression under the different conditions (Fig. S2). We found that co-cultures mimicking mTSS—containing 0.7 mM glucose and high aeration—had 404 downregulated genes and 429 upregulated genes for a total of 833 genes being differentially expressed. Of these genes, not only was there a downregulation of tst expression, but there was also an overall downregulation of multiple secreted virulence factors including PSMα genes, γ-hemolysin, and leukocidins (Fig. 7). The major transcriptional activator of TSST-1, SaeR, was also downregulated (Fig. 7). Notable genes that were upregulated included a variety of nitrate metabolism genes, sugar phosphotransferase systems, and the superantigen-like protein 1 (SSL1). Overall, this analysis suggests that the transcriptional profile of S. aureus MN8 ∆ccpA can be dramatically altered to a “less virulent” state within as little as 4 hours of exposure to L. jensenii in an environment that would otherwise support mTSS.

Fig 7.

L. jensenii in co-culture with S. aureus MN8 ∆ccpA downregulates virulence factor expression in conditions mimicking mTSS. A co-culture apparatus separated MN8 ∆ccpA from L. jensenii with a 0.65 µm filter, with each side containing 20 mL of VDM 0.7 mM glucose. Control co-cultures were also performed with MN8 ∆ccpA on either side of the membrane. Co-cultures were performed aerobically at 250 rpm and 37°C to mimic mTSS and grown to the 4-hour time point. Each dot in the volcano plot represents a single transcript, and those above the horizontal threshold indicate P < 0.001, while vertical thresholds indicate log fold changes greater than 2 in the presence of L. jensenii.

DISCUSSION

The role of environmental cues in the regulation of the superantigen TSST-1 has been well established in vitro (8, 9, 29, 30). Through a wide range of repressive signals, the expression of TSST-1 appears to be highly controlled by the causative bacterium S. aureus such that mTSS can occur only in very specific conditions. In this study, we have further established that the contributions of CSTs I–V are likely to also play a role in the prevention, or exacerbation, of mTSS.

While we previously established the importance of glucose concentration as a key environmental determinant in TSST-1 expression (8), we aimed to revisit the role of glucose in the context of the vaginal microbiota. At puberty, the increase in estrogen is believed to coincide with the accumulation of glycogen in vaginal epithelial cells and the subsequent appearance of lactobacilli in the vaginal environment (17). However, glucose varies throughout the menstrual cycle with a large decrease at menstruation due to exfoliation of the vaginal epithelium (8, 31). Through the use of menstrual products, an increase in oxygen levels at this point will alleviate repression through SrrAB (9), allowing for the perfect environment for TSST-1 production. As a result, we were interested in understanding the protective role of the vaginal microbiota as a last resort.

In examining these interactions, it was critical to acknowledge the paradox: glucose represses tst yet it is a required substrate for the optimal growth of lactobacilli, creating a challenge to study this phenomenon in vitro. Consequently, the use of a ccpA-null mutant was necessary to account for this incongruity, allowing us to examine the activity of the CST representatives without being masked by glucose repression. Through the use of various glucose levels and both wild-type and MN8 ∆ccpA, we modeled the “in-betweens” where the environment transitions from one supporting Lactobacillus growth and tst repression, to one supporting tst expression and decreased Lactobacillus growth. Moreover, the presence or development of ccpA mutants in the circulating S. aureus populations is uncharacterized; however, strains with mutations in other tst regulatory systems such as srrAB have been isolated, which produce TSST-1 even in the absence of required signals (32). Therefore, it is not unlikely that S. aureus strains may circulate with mutations in regulatory systems allowing for excess production of TSST-1.

Through the use of VDM at various glucose concentrations, we confirmed the importance of CcpA in repressing TSST-1 (7, 8), as the expression from the tst promoter and production of the protein increased in the absence of glucose (Fig. 1C, 2C, and 5I) or in the MN8 ∆ccpA mutant (Fig. 1D, 2G, and 5C). Given that L. crispatus and L. jensenii restricted the production of TSST-1 in most conditions, including both the presence and absence of glucose in media (Fig. 5C, D, and H), this would strongly suggest that glycogen in the media is being catabolized in sufficient quantities to support lactobacilli growth. Although glycogen is critical for growth of lactobacilli and acidification of the vaginal environment, most Lactobacillus species do not encode a known enzyme capable of degrading glycogen (33–36). Instead, these lactobacilli must rely on amylases of human and other microbial origins to catabolize the polymer into smaller sugars such as glucose and maltose, which can then be used to produce lactic acid (37, 38). While L. crispatus isolates have been identified to encode a pullulanase that sustains the growth of the bacteria on glycogen (35, 36, 39), many strains contain non-functional copies. The L. crispatus strain used in this study does not contain the associated deletions; however, it demonstrated reduced growth in the absence of glucose (Fig. S4A) that was nearly identical to L. jensenii that does not encode a pullulanase (Fig. S4C). This suggests that the L. crispatus pullulanase is non-functional in the conditions tested. However, an α-amylase has been previously characterized and expressed from S. aureus ATCC12600 (40), and if expressed by S. aureus MN8, this may indicate that the small sugars released would be able to cross the co-culture membrane and support the growth of the lactobacilli (Fig. 5).

A protective role of lactobacilli in the vaginal environment has been attributed to the decrease in vaginal pH through the production of l- and d-lactic acid via glycogen and glucose metabolism (22, 41–43). Given the significant reductions in tst expression demonstrated in the luciferase assays (Fig. 1), on the surface, this would suggest that a decrease in pH may be the mechanism of action through which lactobacilli both restrict S. aureus growth and also limit tst expression in the vaginal niche. However, these protective lactobacilli are all capable of decreasing the pH of VDM to similar values (Table S1), yet throughout the study, we show inter-species differences in the restriction of TSST-1 production. Recently, Schlievert et al. (44) found that the probiotics Lactobacillus acidophilus LA-14 and Lacticaseibacillus rhamnosus HN001 can inhibit TSST-1 production, with the latter doing so in a pH-dependent manner via the SrrAB two-component system (44). This activity appears to be distinct from L. jensenii as srrAB expression was not affected in our transcriptome analysis. It may be possible that different d-/l-lactic acid ratios play a role in the interspecies differential regulation of tst, as L. jensenii is the sole lactobacilli in this study that only produces the d isomer, which has been implicated in protective effects relative to the l isomer (45, 46). Nonetheless, luciferase assays demonstrate decreased tst expression even in conditions of low or no glucose (Fig. 1B, C, E, and F), which are environments that do not support sufficient acidification via lactic acid production (Table S1). Ultimately, this suggests that pH is not likely the principal method of repression and some, if not all, lactobacilli used in our study have additional mechanisms through which TSST-1 production is restrained. For example, it was found that the vaginal isolate Lactobacillus reuteri RC-14 produces cyclic dipeptides that act in Agr-dependent and Agr-independent manners to repress tst (47); however, it is unclear whether the lactobacilli in this study—or any other species and strains—produce these dipeptides.

One of the most striking results from this study is the ability of L. jensenii, as well as L. crispatus, to nearly entirely eliminate the production of the toxin in co-cultures with MN8 already lacking repression from CcpA and glucose (Fig. 5C, D, and H and 6). This indicates that the TSST-1 repressive mechanisms employed by lactobacilli can be hidden behind glucose repression and may function as a “backup” defense mechanism. Conditions where this repression by lactobacilli occurred were mostly microaerophilic, which is consistent with knowledge that lactobacilli thrive in low oxygen environments. However, given that L. jensenii completely inhibited TSST-1 production in a highly aerobic environment, this may indicate that it is the most protective lactobacilli as it can overcome environmental stressors (Fig. 5C and 6A and C). Of note is the ability of L. jensenii to drastically reduce the production of TSST-1 in a contact-independent manner, as opposed to L. jensenii RC-28 that requires physical contact with S. aureus MN8 (48). This indicates there must be key genomic differences between L. jensenii RC-28 and L. jensenii ATCC 25258, which appear to confer different protective functions in the context of mTSS.

While TSST-1 expression in co-cultures was measured using both Western blotting and ELISAs, some discrepancies were found. For example, TSST-1 was detected via Western blotting in microaerophilic control co-cultures (Fig. 5D and H) but not in the ELISAs (Fig. 6B and D). The detection of TSST-1 following trichloroacetic acid (TCA) precipitation is likely attributed to the concentration of the supernatants to 12 OD₆₀₀ units, as without concentrating the samples, TSST-1 was not detected by immunoblotting. Ultimately, the TCA precipitations increase the qualitative detection capabilities of Western blotting, yet the ELISAs provide a more quantitative measure without requiring concentration. Overall, the co-cultures demonstrate that L. jensenii offers protection in the most environmental conditions compared to the other CST lactobacilli and merits further investigation as a potential probiotic candidate, specifically for those who have experienced mTSS in order to prevent re-occurrence.

The transcriptomic analysis revealed that most known tst regulators were not differently expressed in the presence of L. jensenii (Fig. 7), with the exception of the major activator SaeR. Given that activation from SaeRS is required for TSST-1 expression (5), this provides us with an indication of the pathway that is targeted by L. jensenii in order to limit TSST-1 production. The decrease in saeR expression is also consistent with a massive alteration of the S. aureus virulence profile, as seen by the downregulation of numerous other virulence factors such as α-hemolysin, CHIPS, γ-hemolysin, and leukocidins, which are known to be regulated by SaeRS [as reviewed in Liu et al. (49)]. While most SSLs were downregulated in the presence of L. jensenii, SSL1 was upregulated. SSL1 has been shown to have enhanced expression in nutrient-deficient media (50), which is consistent with the low glucose and nutrient competition occurring in our co-cultures. Evidence of nutrient competition is further substantiated by the upregulation of glucose and sugar phosphotransferase systems. Furthermore, nitrate metabolism was upregulated in these co-cultures, highlighted by increased expression of the oxygen-sensing system NreABC, and the nitrate reductase genes activated by this system. NreABC is typically used in anaerobic gene expression (51); considering the upregulation of nitrogen metabolism genes occurred in an aerobic setting, this would indicate that the presence L. jensenii alters environmental cues.

Why L. jensenii has apparently evolved a tst restrictive mechanism remains undetermined. L. gasseri and L. jensenii are found as the dominating CSTs in 6.3% and 5.3% of women, respectively, while L. crispatus is the dominating CST in 26.2% of women (13). These frequencies indicate that L. crispatus may have adapted to the vaginal niche better than other species, which is supported by the presence of the glycogen-degrading pullulanase in its genome, but not in those of L. gasseri or L. jensenii (35, 36). However, given the trends on S. aureus at the transcriptional and translational levels between L. crispatus and L. jensenii, this may suggest that the mechanism these two species use to limit the production of TSST-1 is conserved but distinct from L. gasseri. A study examining the microbiota compositions of tampons from women with or without tst⁺ S. aureus did not find an association to specific CSTs, and none of the women sampled had L. gasseri- or L. jensenii-dominated CSTs (52). Therefore, it is unknown what the selective pressure to maintain a tst restrictive mechanism is for L. jensenii. It would be curious to examine which CST predominates following an episode of mTSS, as it would provide a strong indication for whether the TSST-1 limiting mechanisms used by the lactobacilli offer a competitive advantage in the vaginal niche.

L. iners has been implicated in the activation of innate immunity (53), as well as transitions to a state of BV (54), but it remains unclear whether L. iners is a beneficial or harmful member in the vaginal microbiota. Holm et al. (55) proposed that different strains of L. iners may contribute to different phenotypes in the vaginal environment, which can either support stability or promote dysbiosis, and may be influenced by whether the L. iners strains were isolated from a healthy or dysbiotic environment (55). The strain used in this study, L. iners AB-1, is a clinical isolate from a healthy woman (56); however, the origin of this strain does not appear to confer any protective benefit in the context of TSST-1 production.

G. vaginalis has been highly associated with dysbiosis and BV and was found in the microbial compositions of 97.4% cases of BV, making it quite likely to be the etiological agent (57). A previous study that used G. vaginalis supernatant in a challenge assay for TSST-1 production found a 15% reduction in tst expression (58), which is consistent with the results of our luciferase assays (Fig. 1). In the current study, G. vaginalis enhanced production of TSST-1 from S. aureus at low concentrations of glucose (Fig. 3), which translated into increased Tcell activation (Fig. 4) suggesting that dysbiotic BV may promote the development of mTSS. However, previous work also found increases in tst expression by L. crispatus supernatant (58), which contradicts our results (Fig. 1). The discrepancies here are likely due to the use of a non-specific rich medium (brain heart infusion [BHI]) in combination with high glucose VDM, which is not fully representative of the vaginal environment, and did not consider the knowledge of CcpA-mediated repression that has since emerged (8).

Taken together, this work further reveals that the microbiota may play a critical role in the restraint of the superantigen TSST-1, yet this restriction is intimately linked to environmental cues. We have identified CSTs that may be more at risk than others if tst⁺ S. aureus is present in the microbiota and have identified a potential probiotic candidate, L. jensenii, aimed at limiting mTSS. This study sets forward a model for examining contact-independent interactions of the vaginal microbiota and demonstrates the importance of using environment-mimicking media when studying gene regulation in dynamic environments.

MATERIALS AND METHODS

Bacterial growth conditions

A list and description of bacterial strains used in this study can be found in Table 1. L. crispatus, L. gasseri, and L. jensenii were routinely grown on de Man, Rogosa, and Sharpe (MRS) plates supplemented with 1.5% agar at 37°C in anaerobic conditions (BD GasPak EZ Anaerobe Container System) overnight and cultured microaerophilically in MRS broth at 37°C, unless mentioned otherwise. L. iners and G. vaginalis were grown anaerobically on NYC Medium III (NYC III) plates supplemented with 1.5% agar at 37°C for 2 days and cultured microaerophilically in NYC III broth at 37°C for 1–2 days, unless mentioned otherwise. S. aureus MN8 strains were grown on tryptic soy 1.5% agar (TSA) plates with or without 10-µg/mL chloramphenicol overnight at 37°C and routinely cultured in TSB with or without 10-µg/mL chloramphenicol overnight at 37°C (250 rpm) unless mentioned otherwise. VDM was prepared as previously described with 0.5% proteose peptone (25, 26). Glucose concentrations were adjusted to either 60, 0.7, or 0 mM.

TABLE 1.

Descriptions of bacterial strains used in this study

Strain	Description	Source
S. aureus MN8	Prototypic mTSS strain	Blomster-Hautamaa et al. (59)
S. aureus MN8 ∆ccpA	MN8 with a deletion of ccpA	Dufresne et al. (8)
S. aureus MN8 pAmilux::Ptst	MN8 containing pAmilux::Ptst and Cm10 resistance	Li et al. (47)
S. aureus MN8 ∆ccpA pAmilux::Ptst	MN8 ∆ccpA containing pAmilux::Ptst and Cm10 resistance	Dufresne et al. (8)
L. crispatus ATCC 33820	Representative strain of CST I	ATCC
L. gasseri ATCC 33323	Representative strain of CST II	ATCC
L. jensenii ATCC 25258	Representative strain of CST V	ATCC
L. iners AB-1	Representative strain of CST III	Macklaim et al. (56)
G. vaginalis ATCC 14018	Representative strain of CST IV	ATCC

Supernatant collection and processing

L. crispatus, L. gasseri, and L. jensenii were subcultured microaerophilically at 1% in VDM for overnight growth at 37°C, while L. iners and G. vaginalis were subcultured in the same conditions followed by 48 hours of growth at 37°C. This was repeated for all modified formulations of VDM. Unless otherwise mentioned, supernatants were collected by centrifuging cultures of the five representative CST strains or S. aureus strains at 3,750 × g for 7 minutes (4°C). CST supernatants were filter sterilized with a 0.2 µm syringe filter (Basix) and stored at 4°C, while S. aureus supernatants were filter sterilized and stored at −20°C until ready for processing.

Luciferase reporter assay

Representative CST supernatants were diluted 1/4 with their respective fresh VDM. The dilutions were used to account for changes in pH that would not sustain S. aureus growth, the presence of potential growth inhibitory molecules, and the depletion of nutrients from the media. Previously constructed S. aureus MN8 containing pAmilux::P_tst and MN8 ∆ccpA containing pAmilux::P_tst strains (Table 1) were subcultured at 1% in various formulations of VDM and each of the diluted CST supernatants and incubated for 3 hours. The cultures were brought to an OD₆₀₀ of 0.01 and inoculated in a 96-well plate in triplicate for 18 hours at 37°C in a Biotek Synergy H4 multimode plate reader, with continuous shaking on the medium setting. Measures of bacterial growth (OD₆₀₀) and activity of the tst promoter luminescence production (relative luminescence unit) were recorded once every hour. The relative expression was calculated as the area under the curve (AUC) of relative luminescence unit over the AUC of OD₆₀₀.

Supernatant experiments

S. aureus MN8 and MN8 ∆ccpA were subcultured at 1% in various forms of VDM including the media with diluted CST supernatants. Following 3 hours, the cells were adjusted to an OD₆₀₀ of 0.01 in the same media and grown either for 4 hours or overnight. Following the designated incubation time, OD₆₀₀ readings were recorded, and supernatant was precipitated by a concentration of 6% TCA. The precipitates were washed with ice-cold acetone followed by centrifugation 15,000 × g at 4°C. The pellet was resuspended in 8M urea and stored at −20°C until usage for SDS-PAGE.

Western blots were performed as previously described (8). Briefly, 12% acrylamide gels were run at 80–85 V for 30 minutes and then 150 V for 1 hour. The gel was transferred onto a polyvinylidene difluoride membrane and blocked overnight at 4°C with 5% skimmed milk, 10% horse serum (Gibco), and 10% fetal calf serum (Wisent) in phosphate-buffered saline. The blot was washed and incubated for 1 hour with 1:1,000 of rabbit polyclonal anti-TSST-1 antibody (60). IRDye 800-conjugated donkey anti-rabbit IgG antibody (Rockland) was added in a 1:20,000 dilution and applied to the membrane, with incubation for 1 hour in the dark. The membrane was imaged using an Odyssey imager (LI-COR Biosciences).

Co-cultures

To investigate the contact-independent regulation occurring between S. aureus and each of the protective CST species, co-cultures were performed in various formulations of VDM. L. crispatus, L. gasseri, and L. jensenii were grown overnight at 37°C in MRS broth. S. aureus MN8 and MN8 ∆ccpA were grown overnight at 37°C in tryptic soy broth (TSB). The co-culture apparatus was prepared with a sterile 0.65 µm nylon membrane (GVS North America). Each side contained 20 mL of VDM and a 1/100 subculture of one CST species or one of the S. aureus strains. The co-cultures were grown overnight at 37°C in a shaking incubator (250 rpm) to mimic the aerobic conditions of mTSS and in a stand-still incubator to mimic low oxygen conditions otherwise seen in the vagina. In total, co-cultures were performed at 60, 0.7, and 0 mM glucose, as well as aerobically and microaerophilically. A diagram of the co-culture set-up can be found in Fig. S3.

The following day, TCA precipitations were performed as previously described, and untreated supernatants were collected for ELISA. TCA precipitations were stored at −20°C until use in Western blotting as previously described.

PBMC assay

IL-2 production from PBMCs was assessed by exposing the cells to S. aureus supernatants collected from VDM 0.7 mM glucose CST supernatant experiments. The isolation of PBMCs was performed as previously described (8). PBMCs were seeded for a final concentration of 1 × 10⁶ cells/mL in a 96-well plate. The S. aureus supernatants were then serially diluted and added to PBMCs in duplicate. Following an 18-hour incubation at 37°C and 5% CO₂, the plates were centrifuged, and the cell supernatants were collected for an IL-2 ELISA, using the manufacturer’s instructions (Invitrogen). The plates were read at 450 and 570 nm using a Biotek Synergy H4 multimode plate reader.

TSST-1 ELISA

To quantify TSST-1, co-cultures of interest were reproduced three times. TSST-1 production from co-cultures of interest was determined through a sandwich ELISA as previously described (5). Briefly, supernatants from the S. aureus side of co-cultures were collected, filter sterilized, and stored at −20°C prior to the assay. Plates were coated with anti-TSST-1 polyclonal rabbit IgG solution, followed by addition of the samples. Detection was performed using an anti-TSST-1 IgG horseradish peroxidase conjugate. The plates were read at 450 and 570 nm using a Biotek Synergy H4 multimode plate reader.

RNA sequencing

Co-cultures of interest were incubated for 4 hours each (OD₆₀₀ approximately 1) prior to RNA extraction. The S. aureus sides of the co-cultures were centrifuged and pelleted, followed by re-suspension in RNAprotect (QIAGEN), and pelleted again. Pellets were lysed using 100µg/mL lysostaphin, and RNA was extracted with the RNeasy Plus Mini Kit (QIAGEN). Extracted RNA was treated with the Turbo DNA-free Kit (Ambion), and the RNA quality was determined using an Agilent Bioanalyzer (Robarts Institute, University of Western Ontario), and only samples with RNA integrity numbers above 9 were sequenced. RNA-sequencing and comparative analyses were performed by SeqCenter (Pittsburgh, USA). Twelve million paired-end Illumina sequencing was performed, followed by analysis as previously described (8). Sequences were compared against the publicly available MN8 genome (GenBank assembly accession no. GCA_024296845.1).

Statistical analysis

All statistical analysis was performed using GraphPad Prism 9. Ordinary one-way analysis of variance was used without correction for multiple comparisons for luciferase assays and IL-2 ELISAs. Unpaired t-tests were used for TSST-1 ELISAs.

ETHICS APPROVAL

Healthy, adult human blood donors were recruited through the Department of Microbiology and Immunology at The University of Western Ontario. Donor blood was anonymized, and informed consent was obtained from all donors. This protocol (110859) was approved by the London Health Sciences Centre Research Ethics Board (University of Western Ontario, London, ON, Canada).

DATA AVAILABILITY

Raw RNA read data were deposited at NCBI under BioProject accession no. PRJNA991630.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jb.00447-23.

Supplemental material. jb.00447-23-s0001.docx.

Figures S1 to S4 and Table S1.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.