Cytolysins Augment Superantigen Penetration of Stratified Mucosa

Abstract

Staphylococcus aureus and Streptococcus pyogenes colonize mucosal surfaces of the human body to cause disease. A group of virulence factors known as superantigens are produced by both of these organisms, which allows them to cause serious diseases from the vaginal (staphylococci) or oral mucosa (streptococci) of the body. Superantigens interact with T cells and antigen presenting cells to cause massive cytokine release to mediate the symptoms collectively known as toxic shock syndrome. Here we demonstrate that another group of virulence factors, cytolysins, aid in the penetration of superantigens across vaginal mucosa as a representative nonkeratinized stratified squamous epithelial surface. The staphylococcal cytolysin α toxin and the streptococcal cytolysin streptolysin O enhanced penetration of toxic shock syndrome toxin-1 and streptococcal pyrogenic exotoxin A, respectively, across porcine vaginal mucosa in an ex vivo model of superantigen penetration. Upon histological examination, both cytolysins caused damage to the uppermost layers of the vaginal tissue. In vitro evidence using immortalized human vaginal epithelial cells demonstrated that although both superantigens were proinflammatory, only the staphylococcal cytolysin α toxin induced a strong immune response from the cells. Streptolysin O damaged and killed the cells quickly, allowing only a small release of interleukin-1β. Two separate models of superantigen penetration are proposed: staphylococcal α toxin induces a strong proinflammatory response from epithelial cells to disrupt the mucosa enough to allow for enhanced penetration of toxic shock syndrome toxin-1, whereas streptolysin O directly damages the mucosa to allow for penetration of streptococcal pyrogenic exotoxin A and possibly viable streptococci.

Introduction

Superantigens produced by the gram positive pathogens Streptococcus pyogenes and Staphylococcus aureus are the major factors in a variety of severe infections, such as toxic shock syndrome (TSS) (1–7), scarlet fever (8), necrotizing fasciitis (9), purpura fulminans (10), and necrotizing pneumonia. Superantigens made by S. pyogenes include the streptococcal pyrogenic exotoxins (SPE) A, C, G, H, I, J, K, L, M, streptococcal superantigen (SSA), and streptococcal mitogenic exotoxin Z (SMEZ). SPE A and SPE C have been implicated in most cases of streptococcal TSS (12–17). Of the superantigens made by Staphylococcus aureus, the staphylococcal enterotoxins (SE) B and C are responsible for half of the cases of non-menstrual TSS, while the other half are caused by toxic shock syndrome toxin-1 (TSST-1) (18–22). TSST-1 is also responsible for the majority of menstrual TSS (mTSS) cases. Superantigens were aptly named by Marrack and Kappler in 1990 due to their unique mechanism of T cell stimulation (23). These exotoxins bind to the variable region of the β chain of the T cell receptor (Vβ-TCR) and major histocompability complex II (MHC II) on antigen presenting cells, such as macrophages (24–28). This interaction leads to the proliferation and activation of a large number of T cells and the release of cytokines from both cell types. The production of tumor necrosis factors-α and -β (TNF-α, TNF-β) results in capillary leakage, interleukin-1β (IL-1β) causes fever, and IL-2 and interferon-gamma (IFN-γ) cause rash (8, 29–33). The effects of the massive cytokine release caused by superantigens will eventually lead to hypotension, shock, and if not treated properly, death.

Much is known about the interaction of superantigens with immune cells systemically, but their role at the mucosal surface is unknown. Both S. pyogenes and Staphylococcus aureus typically colonize nonkeratinized stratified squamous epithelia to initiate infections, with S. pyogenes initiating severe infections often following oral mucosal colonization and mTSS initiating from vaginal mucosa (34, 35); an important difference in streptococcal TSS and staphylococcal mTSS is that streptococci typically invade systemically, whereas Staphylococcus aureus remains on the mucosal surface (3, 4, 35). Subsequent to initial bacterial colonization, superantigens may interact with the mucosa in at least two ways: 1) superantigens may penetrate through the mucosa to reach adaptive immune cells located in the underlying tissues, and 2) superantigens may exert effects directly on epithelial cells which contribute to the course of the infection. In fact, TSST-1 is known to penetrate the vaginal mucosa to cause mTSS, even while Staphylococcus aureus remains localized at the vaginal surface (3, 4). It has been shown that there are differences among the superantigens in their ability to penetrate mucosal surfaces in order to cause TSS and lethality in a rabbit model (36). When administered vaginally to rabbits, TSST-1 is the most lethal, followed by SEC1 and then SPE A, whereas all three superantigens are lethal when administered intravenously (36). High doses of TSST-1 are also lethal when administered orally; thus TSST-1 appears to penetrate mucosal surfaces better than other superantigens (36).

Porcine vagina ex vivo has also been used to analyze the ability of TSST-1 to penetrate the mucosa. Ex vivo porcine tissue is an excellent model of human vaginal tissue; vaginal tissue from both human and pig is a nonkeratinized stratified squamous epithelium with intercellular lipids, including ceramides, glucosyl ceramides, and cholesterol located in the surface layers (34, 37–43). As tight junctions are not present between the cells of the vaginal mucosa, the intercellular lipids constitute a permeability barrier. TSST-1 labeled with ³⁵S-methionine has been shown to cross the ex vivo porcine vaginal mucosa, with significant amounts of toxin remaining within the tissue (44). Studies by Peterson et al. demonstrated that the penetration of TSST-1 is enhanced by the presence of heat-killed and live Staphylococcus aureus (45). Live bacteria were better able to enhance penetration of TSST-1 than heat-killed bacteria, indicating that secreted factors made by the bacteria may be contributing to the disruption of the mucosal barrier. The authors suggested that cytolysins may play a role in this process due to their probable ability to provoke inflammatory responses.

The main objective of the present study was to determine the role of two cytolysins, streptolysin O (SLO) and α toxin, in the penetration of superantigens across vaginal mucosa as a model of nonkeratinized stratified squamous epithelium. SLO is a thiol-activated toxin made by S. pyogenes that belongs to a group of cytolysins known to bind cholesterol and form pores in the membranes of eukaryotic cells (46). Seventy to eighty monomers of SLO oligomerize on cell membranes to form large pores, up to 30 nm in diameter, which can be lethal for most cell types (47, 48). α toxin, made by Staphylococcus aureus, is a heptamer pore-forming toxin that creates small (2.6 nm diameter) pores in eukaryotic membranes (49). Many species-specific cell types can be killed by α toxin, including erythrocytes, platelets, mononuclear immune cells, endothelial cells, and epithelial cells. In addition to cellular damage, both cytolysins can elicit cytokine responses from epithelial cells, which may contribute to disruption of the mucosa. Dragneva et al. showed that low amounts of α toxin induce secretion of IL-8 from epithelial cells and monocytes (50). Non-lethal doses of SLO stimulate the release of IL-6 and IL-8 from HaCaT human keratinocytes, while a SLO-deficient S. pyogenes mutant induces lower levels of IL-1β, IL-6, and IL-8 from these cells, compared to wild type bacteria (51, 52).

Superantigens may also interact with cells of the mucosa to induce inflammation, which may independently disrupt the mucosal barrier. A few studies have shown that superantigens can induce cytokine responses from other cell types. The staphylococcal superantigen TSST-1 has been shown to bind both endothelial and epithelial cells, and in some cases is internalized by the cells (53–56). More recently, it was demonstrated that TSST-1 and SEB induce cytokine responses from epithelial cells. TSST-1 incubated with vaginal epithelial cells induces TNF-α, MIP-3α, and IL-8, and it induces TNF-α and IL-8 production from bronchial epithelial cells (45, 57). SEB has been shown to induce an IL-8 response from nasal epithelial cells and to alter permeability of rabbit maxillary sinus epithelium, indicating that the enterotoxin can specifically interact with epithelial cells (58, 59). Rajagopalan et al. and Herz et al. have shown that SEB also induces systemic inflammatory responses when administered from mucosal surfaces, including nasal, conjuctival, and vaginal mucosae (60–64). Rajagopalan et al. also demonstrated a similar effect when the streptococcal superantigen SPE A was administered nasally to HLA-transgenic mice, however not much is known about the ability of streptococcal superantigens to stimulate a cytokine response from the epithelium itself (64). Peterson et al. demonstrated that, like TSST-1, SPE A can induce proinflammatory cytokine production from human vaginal epithelial cells (HVECs) (45). The second objective of the present study was to evaluate the ability of superantigens (TSST-1 and SPE A) and cytolysins (α toxin and SLO) to induce cytokine responses from HVECs.

In the present study we show that α toxin and SLO act to enhance penetration of TSST-1 and SPE A, respectively, across ex vivo porcine vaginal mucosa in a model of superantigen penetration. Both cytolysins cause localized damage and inflammation in the ex vivo porcine tissue; however purified cytolysins added to HVECs induced cell damage and death to different extents. Superantigens incubated with HVECs induce proinflammatory cytokine and chemokine responses from the cells, but these responses are altered when cytolysin is present. Based on these data, we propose two separate models for streptococcal and staphylococcal superantigen penetration across vaginal mucosa.

Materials and Methods

Penetration studies

An ex vivo porcine vaginal permeability model for superantigens has been previously described (34, 44, 45). Briefly, porcine vaginal mucosa was isolated from pigs at slaughter and used within 3 hours of harvest. Tissue discs (8–10 mm in diameter) were mounted between two halves of continuous-flow perfusion chambers, exposing approximately 0.2 cm² of the epithelial surface to the donor compartment. Internally labeled ³⁵S-TSST-1 (10–50 μg/ml) in the absence and presence of α toxin (5 μg/ml or 50 μg/ml) or ³⁵S-SPE A (40 μg/ml) in the absence and presence of SLO (5 μg/ml or 50 μg/ml) was added to the upper compartment in PBS. Dithiothreitol (DTT, Roche Diagnostics Corporation, Indianapolis, IN) at a final concentration of 10 mM was added to all specimens receiving SLO to maintain a reducing environment for the oxygen-labile cytolysin. Seven replicates were used for each condition. PBS was continuously pumped through the lower compartment as a collection fluid for up to 8 hourly samples and the dpm (disintegrations per minute) of the samples counted in a scintillation counter. An aliquot of each radiolabeled solution was also counted to determine the dpm per ng toxin applied. The total amount of toxin to traverse the tissue was determined by converting the dpm of the samples to ng toxin using the dpm/ng conversion factor determined for the radiolabeled toxin solutions (44). Tissue discs from each treatment group were removed from the chambers at the conclusion of the experiment, fixed in formalin, wax embedded, cross-sectioned, and stained with hematoxylin and eosin for histological examination. One specimen from each group was snap-frozen in liquid nitrogen, cut in cross-section at 14 μm, and sections placed on x-ray film and developed to visualize radioactive toxin remaining in the tissue.

Dutch-belted rabbits are highly sensitive to TSST-1 when administered continuously for 7 days in subcutaneous miniosmotic pumps (16). However, rabbit susceptibility to TSST-1 is most easily measured by the capacity of TSST-1 to synergize with lipopolysaccharide up to 10⁶-fold, through acceleration of cytokine release; the animals succumb within 48 hours (65) whether TSST-1 is administered intravenously or intra-vaginally (36). Thus, we performed an experiment in female rabbits to assess the ability of α toxin to facilitate TSST-1 vaginal penetration following intra-vaginal administration of TSST-1 + α toxin, followed 4 hours later by intravenous lipopolysaccharide. We have experimentally determined that intravenous administration of 0.01 μg/kg of TSST-1 to rabbits followed at 4 hours with 50 μg/kg lipopolysaccharide is 100% lethal, whereas administration of these agents is not lethal when TSST-1 is administered intra-vaginally (0.01 μg/kg) followed intravenously with lipopolysaccharide (50 μg/kg). Incidentally, the lethal dose of TSST-1 alone by either route is >3.5 mg/kg and for lipopolysaccharide intravenously is >500 μg/kg. α toxin (0.05 μg/kg) is not lethal to rabbits whether given intravenously or intra-vaginally. Thus, we compared the ability of 0.01 μg/kg of TSST-1 ± α toxin (both intra-vaginal) to synergize with lipopolysaccharide (50 μg/kg) intravenously 4 hour later. Intra-vaginal administrations of TSST-1 and α toxin in PBS were made through catheters threaded into the vaginas of rabbits after anesthesia with ketamine and xylazine (36); lipopolysaccharide (Salmonella enterica serovar Typhimurium) was injected intravenously through the marginal ear veins. Animals were monitored for development of TSS over 48 hours. In agreement with the University of Minnesota IACUC, rabbits that failed to exhibit escape behavior and could not right themselves were considered to have lethal TSS and were euthanized.

In an additional experiment, S. pyogenes strain MNBU (M-type 3 clinical isolate, 8 × 10¹⁰ colony-forming units [CFU]) or Staphylococcus aureus strain MN8 (vaginal TSS isolate, 1 × 10⁹ CFUs) were added to the upper chamber in the absence or presence of SLO and α toxin (5 μg/ml each), respectively. DTT was added to all conditions receiving SLO at a final concentration of 10mM. Four to five replicates were used for each condition. Perfusate was collected every 2 hours for 8 hours. Samples were concentrated to obtain a bacterial pellet (14,000 rpm, 5 min), then resuspended in 400 μl PBS, and plated out on Todd-Hewitt (TH) agar to determine the amount of bacteria that penetrated the epithelium. The total CFUs were determined for each sample and averaged among the replicates for each time point. The total CFUs at the conclusion of the experiment was determined by adding the average total CFU for each time point (2, 4, 6, and 8 hours).

Toxin preparation

Superantigens TSST-1 and SPE A were purified as previously described (22, 66). Briefly, TSST-1 was isolated from Staphylococcus aureus strain RN4220 (pCE107) and SPE A was isolated from Bacillus subtilis strain IS75 (pJS103 MiniKC) grown in beef heart medium (67). The cultures were precipitated with ethanol at 4°C, the precipitate resolubilized in water, and toxin purified by isoelectric focusing. Isoelectric focusing was carried out in two phases; the first phase utilized a pH gradient of 3.5 to 10, followed by another using a pH gradient based on the isoelectric point of the toxin. The isoelectric point of TSST-1 is 7.2, so the second IEF used a gradient of 6 to 8, while the isoelectric point of SPE A is 5.2 and the second IEF used a gradient of 4 to 6 (22, 68). Each superantigen was identified in a double immunodiffusion assay based on its specific reactivity with a polyclonal antibody generated against the exotoxin (69). Purity was confirmed by SDS-PAGE, which demonstrated a single protein band at a molecular weight of 22,000 (TSST-1) or 26,000 (SPE A). Purified toxins were quantified using the BioRad protein assay (BioRad Co., Hercules, CA) with the superantigen staphylococcal enterotoxin B used for standard curve generation.

Both TSST-1 and SPE A were internally labeled with ³⁵S-methionine for penetration studies. In previous labeling studies with TSST-1, approximately 10⁷ dpm/ug protein has been achieved. SPE A has three naturally-occurring Met residues, all of which can be labeled. The purification is the same as that described above, but bacterial strains were cultured in 50 ml media containing 10 mCi ³⁵S-methionine.

α toxin was purified from Staphylococcus aureus strain MNJA as described for the superantigens. The second isoelectric focusing step was done using a gradient from 7 to 9, since the isoelectric point of α toxin is 8.5 (70). SLO was purchased from Sigma-Aldrich (St. Louis, MO). In both cases, purity was confirmed by SDS-PAGE and silver staining to identify homogenous protein samples.

Culture of immortalized cells

The use of immortalized HVECs have been previously described (45, 71, 72). HVECs were maintained in Keratinocyte Serum Free Medium (KSFM; Gibco, Invitrogen, Carlsbad, CA) supplemented with bovine pituitary extract and epidermal growth factor as provided by the manufacturer and a 1% final volume of penicillin-streptomycin (Sigma-Aldrich) and amphotericin B (Fungizone; Gibco, Invitrogen). The cells were grown at 37°C in the presence of 7% CO₂. On days of experimentation, antimicrobials were not used since it has been observed that amphotericin B reduces cytokine production by these cells.

Cytokine assays

Purified exotoxins (SPE A and/or SLO, TSST-1 and/or α toxin) were added to the cell culture medium and incubated with HVECs at 37°C in 7% CO₂ for 2–6 hours. DTT (Roche) at a final concentration of 10 mM was added to all specimens receiving SLO to maintain a reduced environment for the oxygen-labile cytolysin. At the conclusion of each experiment, the media were collected and analyzed by ELISA (45) using human Quantikine® kits (R and D Systems, Minneapolis, MN). The following cytokines and chemokines were measured: IL-1β, IL-6, IL-8, MIP-3α, and TNF-α, with the minimum detectable dose ranging from 0.70 pg/ml to 3.5 pg/ml.

Uric acid assay

Uric acid release from injured cells was measured using the QuantiChrom^™ kit available from BioAssay Systems (Hayward, CA). This assay utilizes the compound 2,4,6-tripyridyl-s-triazine, which forms a blue colored complex with iron only in the presence of uric acid. The intensity of the color change is proportional to the amount of uric acid present in the sample.

Trypan blue cell staining

After incubation with exotoxins, HVECs were rinsed and subjected to trypsin (Gibco, Invitrogen) treatment and centrifugation (200 × g, 5 min) to obtain a cell pellet. Cells were then resuspended in 0.5 ml KSFM and 0.1 ml 0.4% trypan blue staining solution (Sigma-Aldrich) for 15 minutes. HVECs were counted using a hemacytometer and survival percentages were calculated.

Statistics

In all cases where statistical analysis was necessary, mean values and standard errors of the mean were determined. Statistical difference between means was determined using the Student’s unpaired t test with normally distributed data. Fishers exact test was used to assess differences in TSS survival rates between experimental and control rabbit groups.

Results

Cytolysins augment penetration of superantigens across vaginal mucosa

An ex vivo porcine model of superantigen penetration of vaginal mucosa has been used previously to demonstrate that staphylococcal TSST-1 penetrates the mucosa in small amounts, with most remaining on the tissue (45). In the presence of Staphylococcus aureus, however, the amount of superantigen that penetrates is greatly increased. When comparing the amount of TSST-1 that penetrates in the presence of live or heat-killed Staphylococcus aureus more superantigen can penetrate vaginal mucosa when live bacteria are present, suggesting that the bacteria may be making an exoprotein that augments superantigen penetration. It has been previously suggested by our laboratory that cytolysins, such as α toxin, may act to disrupt the vaginal mucosa to allow for better penetration of TSST-1. It is also possible that this may be the case for streptococcal superantigens; therefore we chose to examine the ability of the streptococcal cytolysin SLO to augment penetration of SPE A.

Table I shows the penetration through porcine vaginal tissue of radiolabeled ³⁵S-TSST-1 and ³⁵S-SPE A in the absence and presence of cytolysins α toxin and SLO, respectively. In both cases, the presence of cytolysins increased the total amount of superantigen that penetrated the vaginal mucosa. In the case of TSST-1 and α toxin, both concentrations of cytolysin significantly enhanced the amount of TSST-1 that penetrated the epithelium (p < 0.05). In the case of SPE A and SLO, however, only the lower dose of SLO (5 μg/ml) was able to significantly augment the penetration of SPE A (p <0.05). A larger starting concentration of SPE A was used (40 μg/ml compared to 10 μg/ml TSST-1) because previous studies using rabbit vaginal mucosa showed that SPE A was unable to penetrate as well as TSST-1 (36), however, we also measured the amount of SPE A at 10 μg/ml that penetrated the porcine tissue and found that SPE A was able to penetrate as well as TSST-1 in this model.

Table I. Cytolysins augment penetration of superantigens.

Superantigens (TSST-1 or SPE A) were internally labeled with ³⁵S-methionine and applied to the epithelial surface of fresh porcine vaginal mucosa mounted in perfusion chambers in the absence or presence of cytolysins α toxin or SLO (5 or 50 μg/ml). In both cases, cytolysins acted to disrupt the mucosa and augment superantigen penetration. TSST-1 was added at 10 μg/ml, whereas SPE A was added at a higher concentration of 40 μg/ml based on previous studies done on rabbit vaginal mucosa which indicated that SPE A was not able to penetrate the mucosa as well as TSST-1 (36). An additional set of chambers was used to determine the penetration of only 10 μg/ml SPE A; this set demonstrated that SPE A penetrates the porcine vaginal mucosa just as well as TSST-1. SEM = standard error of the mean for 3–7 replicates. Statistical difference calculated using Student’s unpaired t test with normally distributed data (compared to amount of superantigen able to penetrate on its own). NS = not significant.

Condition	Concentration of Toxin (total ng ± SEM) After 8 Hours	Statistical Difference When Cytolysin is Present
TSST-1 (10 μg/ml)	25 ± 2.33
TSST-1 (10 μg/ml) + α toxin (5 μg/ml)	39 ± 6.78	p <0.05
TSST-1 (10 μg/ml) + α toxin (50 μg/ml)	55 ± 11.59	p <0.05
SPE A (40 μg/ml)	112 ± 21.05
SPE A (40 μg/ml) + SLO (5 μg/ml)	176 ± 12.67	p <0.05
SPE A (40 μg/ml) + SLO (50 μg/ml)	150 ± 7.33	NS
SPE A (10 μg/ml)	28 ± 6.23

At the conclusion of each experiment, tissue specimens were sectioned and stained with hematoxylin and eosin to visualize damage to the mucosa. α toxin at both the low (5 μg/ml) and high (50 μg/ml) doses damaged the surface layers of the vaginal mucosa, with obvious sloughing of the epithelium seen at the higher dose (Figure 1). SLO also damaged the mucosa at both concentrations (5 and 50 μg/ml) with significant intra-epithelial separation again seen at the higher dose (Figure 2). SPE A alone, on the other hand, did not damage the epithelium, compared to a PBS only control.

Figure 1. α toxin disrupts vaginal mucosa.

TSST-1 (10 μg/ml) and α toxin A) 5 μg/ml or B) 50 μg/ml were applied to ex vivo porcine vagina and incubated for 8 hours. Tissue was sectioned and stained with hematoxylin and eosin for histological examination. Both concentrations of α toxin appear to disrupt the surface layers of the vaginal epithelium, with the higher concentration causing sloughing of the uppermost epithelial layers.

Figure 2. SLO damages vaginal tissue.

SPE A (40 μg/ml) in the absence or presence of SLO (5 or 50 μg/ml) was applied to ex vivo porcine vagina for 8 hours. Tissues were sectioned and stained with hematoxylin and eosin for histological examination. A) SPE A alone does not damage the mucosa. B) A lower dose of SLO (5 μg/ml) shows damage to the epithelial surface, C) but more distinct damage, including intra-epithelial separation, can be seen at the higher dose (50 μg/ml). D) Control tissue that has been incubated in PBS alone for 8 hours.

In order to determine where in the tissue the non-penetrating superantigen was located, tissue sections were placed on x-ray film and developed. This allowed us to visualize radioactivity remaining on or within the tissue, and it indicated that the majority of the radiolabeled superantigen was “trapped” in the uppermost, epithelial layers of mucosa (Figure 3). It is interesting to note that even though cytolysins augment superantigen penetration, the majority of superantigen still remained in the surface layers of the tissue and did not penetrate throughout. It is important to remember that only microgram amounts of superantigen are required to cause TSS in humans (73), therefore although the differences between the amount of superantigen that penetrates in the absence and presence of cytolysin may appear small, the differences may explain the development of TSS or simple clearance of superantigen without TSS.

Figure 3. Most SPE A remains in the tissue.

SPE A was radiolabeled with ³⁵S-methionine prior to application to the porcine vagina ex vivo. Whole pieces of tissue were sectioned and exposed to x-ray film to determine where the superantigen remained in the tissue. In all conditions, SPE A primarily remained in the uppermost layers of the epithelium, which can be seen on the x-ray film (top). Stained tissue sections are also shown for reference (bottom).

Superantigens and cytolysins induce different cytokine responses from HVECs

Although cytolysins were shown to damage the vaginal mucosa, it was possible that both superantigens and cytolysins induced proinflammatory responses that disrupt the barrier and facilitate greater penetration of molecules across the vaginal surface. In fact, our laboratory has previously shown that TSST-1 (100 μg/ml) can induce the production of proinflammatory cytokines and chemokines (IL-8, MIP-3α, and TNF-α) from an immortalized human vaginal epithelial cell (HVEC) line (45). However, here we show that a lower dose of TSST-1 (10 μg/ml versus 100 μg/ml) induced only a small amount of IL-1β and IL-6 from HVECs (Figure 4). When α toxin was incubated with HVECs at two concentrations (50 μg/ml and 5 μg/ml) it induced all cytokines and chemokines tested (IL-1β, TNF-α, IL-6, IL-8) in a dose-dependent manner, with the exception of MIP-3α, which was not detected. When TSST-1 (10 μg/ml) was co-incubated with α toxin (5 or 50 μg/ml) on the HVECs, very little IL-1β or TNF-α was detected, but IL-6 and IL-8 were induced to a greater extent, indicating synergy between TSST-1 and α toxin. In general, when both TSST-1 and α toxin were incubated with the cells, the extent of the immune response was dependent on the concentration of α toxin present. An experiment was also conducted using a higher dose of TSST-1 (100 μg/ml) with both concentrations of α toxin (5 or 50 μg/ml) that showed the same trend, however the overall response was reduced compared to that seen with a lower dose of TSST-1 (data not shown).

Figure 4. α toxin induces a proinflammatory cytokine and chemokine response from HVECs.

Cells were incubated with TSST-1, α toxin, or combinations of both for 6 hours; cell culture supernates were collected and assayed by ELISA for IL-1β, TNF-α, IL-6, IL-8, and MIP-3α. Peterson et al. (45) demonstrated that a higher concentration of TSST-1 (100 μg/ml) can induce IL-8 and MIP-3α from HVECs, as well as low levels of TNF-α. Here we show that a lower concentration (10 μg/ml) of TSST-1 induced only low levels of IL-1β and IL-6, whereas α toxin at two doses (50 μg/ml and 5 μg/ml) induced a wider range of proinflammatory cytokines and chemokines. When administered simultaneously (TSST-1 10 μg/ml and α toxin 50 μg/ml or 5 μg/ml) IL-1β and TNF-α were no longer detected, but more IL-6 and IL-8 were being produced. All concentrations are given as a difference from a media only control. Error bars represent the standard error of the mean.

In contrast to the proinflammatory response to α toxin from the HVECs, SLO alone (10 μg/ml) induced only a low level of IL-1β from the cells (Figure 5). Lower doses of SLO (1 and 0.1 μg/ml) were shown to elicit similar, but dose-dependent, responses from the HVECs (data not shown for 0.1 μg/ml dose). SPE A, like TSST-1 (at 100 μg/ml), induced strong IL-8 and MIP-3α responses from the cells, and this response was found to increase over time (data not shown). SPE A also induced IL-6, which was only detected in response to the lower TSST-1 concentration. A lower concentration of SPE A (10 μg/ml) demonstrated a similar trend to that seen with the higher dose. When SPE A (100 or 10 μg/ml) and SLO (10 μg/ml or 1 μg/ml) were incubated with the HVECs, only an IL-1β response was seen. This response was stronger than that to SLO alone; however it is important to note that any cytokines and chemokines induced by SPE A alone were no longer detected.

Figure 5. SPE A elicits a proinflammatory response from HVECs, but only IL-1β can be measured when SLO is present.

SPE A and/or SLO was added to HVECs for 6 hours; cell culture supernates were collected and analyzed by ELISA for cytokine and chemokine production by the cells. SLO alone (10 μg/ml) induced only a small IL-1β response from the cells, whereas SPE A alone (100 μg/ml) induced large amounts of IL-6, IL-8, and MIP-3α. When administered simultaneously to the HVECs at three different combinations of concentrations [high SLO (10 μg/ml) and high SPE A (100 μg/ml), high SLO (10 μg/ml) and low SPE A (10 μg/ml), and low SLO (1 μg/ml) and high SPE A (100 μg/ml)] only IL-1β was detected and all cytokines and chemokines induced by SPE A were no longer detectable. All concentrations are given as a difference from a media only or DTT only control. Error bars represent the standard error of the mean.

Cytolysins cause different amounts of damage to the HVECs

The cytolysins may have been directly damaging or killing the epithelial cells, which may have triggered the inflammatory response seen. Three methods were used to assess cellular damage: trypan blue staining of dead cells, measurement of uric acid release from damaged cells, and comparison of total cell numbers in the absence and presence of cytolysin after 6 hours. The release of uric acid from cells and tissues has been shown to correlate with cellular damage, and extracellular uric acid acts as a danger signal to the immune system by triggering dendritic cells that phagocytose the molecules to become activated (74). The high dose of α toxin (50 μg/ml) killed the HVECs only after 6 hours, as seen by only a 10% survival rate when the total cell number was compared to that of a media only control (yellow hatched column), but no uric acid release was detected under any condition (Figure 6). Of those cells remaining adherent after 6 hours in the presence of the high dose of α toxin only 33% survived (yellow column). The lower dose of α toxin (5 μg/ml) did not induce uric acid release or kill the cells at any time point. SLO, on the other hand, induced uric acid release from the HVECs after 2 hours, but lower levels of uric acid were detected at 4 and 6 hours, indicating that uric acid may not remain stable in the cellular medium. SLO was shown to kill the cells after 4 hours (green columns), with 53% survival at higher doses (10 and 50 μg/ml, data not shown). After 6 hours with SLO, cells still adherent to the flask were alive, as indicated by trypan blue staining (blue columns), however based on total cell counts only 38% of the cells remained compared to a media only control (blue hatched column). The lower dose of SLO (1 μg/ml) did not affect survival, despite inducing a strong uric acid release from the cells. In contrast, neither superantigen (100 μg/ml) induced uric acid release from the HVECs, nor did they directly kill the cells as indicated by trypan blue staining and total cell counts (data not shown).

Figure 6. SLO induces uric acid release from dead and damaged cells, but only a high dose of α toxin kills the cells without inducing the release of uric acid.

To assess the ability of the cytolysins to damage or kill HVECs, uric acid release (◆) from the cells was measured as an indicator of cell damage (72) and trypan blue staining (columns) was done to determine the percentage of cells alive after each time point. SLO (10 and 1 μg/ml) and the DTT only control induced a strong uric acid release from the cells after only 2 hours (red columns), indicating cellular damage. At 4 hours (green columns), not as much uric acid was detected, but only 53% of the cells were still alive as indicated by trypan blue staining. After 6 hours, the cells still adherent to the flask were alive (blue columns), and very little uric acid could be detected, indicating that the uric acid is not stable in the cellular medium. The percent of cells remaining after 6 hours is shown by a blue hatched column, as determined by a difference in total cell number compared to a media only control. α toxin did not induce uric acid release from the cells at any time point, but the high dose (50 μg/ml) killed the majority of the cells after 6 hours (yellow columns). The percent of total cells remaining after 6 hours is shown by a yellow hatched column. Cell death due to α toxin was not seen at 2 (orange columns) or 4 (violet columns) hours. The low dose of α toxin (5 μg/ml) did not induce uric acid release or kill the cells. A media only control is shown (white column) to demonstrate that cells typically do not release uric acid. Error bars on uric acid data represent the standard error of the mean.

SLO augments penetration of S. pyogenes, while α toxin does not enhance penetration of Staphylococcus aureus

We hypothesized that the damage caused by SLO in the absence of a strong proinflammatory response may also allow S. pyogenes to penetrate the epithelium. This may help to explain why S. pyogenes is often found in the bloodstream of streptococcal TSS patients, whereas Staphylococcus aureus remains localized on the mucosal surface during TSS. In order to test this, we monitored the penetration of S. pyogenes and Staphylococcus aureus across ex vivo porcine vagina in the absence and presence of SLO and α toxin, respectively. Figure 7 shows the amount of bacteria that penetrated the epithelium after 8 hours. SLO was clearly able to enhance the ability of S. pyogenes to penetrate the vaginal epithelium; the amount of bacteria that penetrated was significantly higher than all other conditions (p<0.05). S. pyogenes alone was better able to penetrate than Staphylococcus aureus alone, but was only significantly higher in concentration after 2 hours (p<0.001). No bacteria were detected to penetrate when α toxin was present with Staphylococcus aureus; however there was no significant difference between the amounts of Staphylococcus aureus that penetrated in the absence and presence of the cytolysin.

Figure 7. SLO enhances ability of S. pyogenes to penetrate vaginal epithelium, while α toxin does not.

S. pyogenes (8 × 10¹⁰ cells) and Staphylococcus aureus (1 × 10⁹ cells) were added to ex vivo porcine vagina and incubated for 8 hours in the absence and presence of SLO and α toxin (5 μg/ml each), respectively. Perfusate was collected every two hours, concentrated, and plated on TH agar plates to determine the amount of viable bacteria that penetrated the epithelium. Total bacterial concentration was determined by adding total CFU for each two-hour sample, up to 8 hours. While S. pyogenes was able to penetrate the epithelium on its own, the presence of SLO allowed for a significantly higher amount of bacteria to penetrate (p value <0.05). There was no significant difference between the amounts of Staphylococcus aureus that was able to penetrate in the absence and presence of α toxin.

α toxin augments penetration of TSST-1 in a rabbit vaginal model of TSS

Unlike the relatively thick human and porcine vaginal mucosa, the rabbit vaginal mucosal is only 3–5 epithelial cell layers in thickness. The ability of α toxin to facilitate TSST-1 penetration was assessed following intra-vaginal TSST-1 ± intra-vaginal α toxin (Table II). Development of lethal TSS was assessed by synergy of cytokine production due to subsequent intravenous lipopolysaccharide given 4 hours after intra-vaginal TSST-1 ± intra-vaginal α toxin. Rabbits that received intra-vaginal TSST-1 plus intra-vaginal α toxin succumbed to TSS (4/4) when challenged with lipopolysaccharide. In contrast, rabbits that received intra-vaginal TSST-1 alone did not succumb to TSS (0/4) when challenged with lipopolysaccharide (p<0.03), suggesting the cytolysin facilitated TSST-1 penetration of the vaginal mucosa.

Table II. α toxin enhances the lethality of TSST-1 in rabbits.

A rabbit model of TSS which monitors the ability of TSST-1 to synergize with lipopolysaccharide to cause shock through the acceleration of cytokine release was used to assess the ability of α toxin to enhance the penetration of TSST-1 across rabbit vaginal epithelium. TSST-1 (0.01 μg/kg) in the absence or presence of α toxin (0.05 μg/kg) was administered intra-vaginally to young adult female Dutch-belted rabbits; four hours later lipopolysaccharide (50 μg/kg) was administered IV and rabbits were monitored for 48 hours for signs of TSS. Fishers exact test was used to assess differences in TSS survival rates between experimental and control rabbit groups.

Condition	Rabbit lethality after 48 hours
TSST-1 (0.01 μg/kg)	0/4
TSST-1 (0.01 μg/kg) + α toxin (0.05 μg/kg)	4/4 (p<0.03)

Discussion

Both Staphylococcus aureus and Streptococcus pyogenes can initiate TSS from nonkeratinized stratified squamous mucosal surfaces; Staphylococcus aureus causes mTSS while remaining localized on vaginal epithelia, whereas S. pyogenes induces TSS often associated with initial oral mucosal colonization and subsequent penetration of the organism and superantigens systemically. In the case of mTSS, Staphylococcus aureus locally produces the superantigen TSST-1 which must then penetrate the mucosal barrier in order to interact with adaptive immune cells to induce the cascade of events that leads to shock. Previous research done in our laboratory demonstrated that live bacteria could enhance superantigen penetration better than heat-killed bacteria, indicating that live Staphylococcus aureus may be actively secreting factors that aid in disrupting the barrier (45). The main purpose of this study was to determine the role of cytolysins in the penetration of superantigens across vaginal mucosa. Although S. pyogenes can occasionally be found vaginally and induce TSS from this site (75), it is more commonly found in the oral mucosa. This study used only vaginal mucosa as a representative nonkeratinized stratified squamous epithelium to examine both staphylococcal and streptococcal superantigen penetration, however previous studies have indicated a high similarity between vaginal and oral epithelium in both structure and function, including permeability (41, 43).

In this study, we demonstrated that both the staphylococcal cytolysin α toxin and the streptococcal cytolysin SLO are capable of augmenting the penetration of their respective superantigens. In the case of α toxin, there was a dose-dependent increase in the amount of TSST-1 that was able to penetrate ex vivo porcine vaginal tissue. A greater concentration of SLO, on the other hand, did not seem to further enhance the ability of SPE A to penetrate. This may be due to the difference in the amount of monomers required to form a pore for each cytolysin. SLO requires between 70 and 80 monomers, whereas α toxin only requires 7, therefore SLO may cause site-specific damage to the mucosa whereas α toxin can cause a broader range of damage with an increase in concentration. Histological examination showed disruption of the mucosa due to the presence of cytolysin in both cases. Although more superantigen was able to penetrate through the mucosa when cytolysin was present, we demonstrated for SPE A that the majority of the radiolabeled toxin remains in the uppermost layers of the tissue. This entrapment of superantigen in what appears to be the mucosal epithelium may serve as a reservoir for additional toxin during the infection process. We have previously demonstrated a reservoir effect in porcine oral mucosa with transforming growth factor-β3 (TGF-β3), a molecule similar in size to the superantigens (76).

Although damage to the mucosa was seen in histological staining of tissues exposed to cytolysins, it was unclear whether that was due to direct damage by the cytolysins or was the result of an inflammatory process caused by the cytolysins. To clarify this, we examined the ability of α toxin and SLO to induce proinflammatory responses from immortalized epithelial cells. Although α toxin was shown to induce a strong proinflammatory response from the cells, SLO only induced a low level of IL-1β. In order to address the possibility that cytolysins and superantigens act to synergistically induce an inflammatory response from epithelial cells, we measured the amount of cytokines produced in response to both α toxin and TSST-1 or SLO and SPE A. Although α toxin alone was proinflammatory, there was a differential immune response to α toxin and TSST-1 when administered simultaneously. In the presence of α toxin and TSST-1, the cytokines IL-1β and TNF-α were no longer detected, but higher amounts of IL-6 and IL-8 were seen, indicating some synergy between the two toxins. In contrast, there was a stronger IL-1β response to both SLO and SPE A; however the cytokines and chemokines induced by SPE A alone were not detected when SLO was present.

In order to examine the direct damage to HVECs caused by α toxin and SLO, we measured uric acid release (an indicator of cell damage and a known “danger” signal for the immune system (74)) and stained the cells with trypan blue to identify dead cells, as well as compared total cell numbers in the presence and absence of cytolysin after 6 hours. Interestingly, α toxin did not cause uric acid release from the cells at 2, 4 or 6 hours, but the high concentration of α toxin (50 μg/ml) killed approximately 90% of the cells after 6 hours. The strong proinflammatory response we saw from the cells incubated with α toxin indicates that the cells are able to make and secrete large amounts of cytokines and chemokines before the majority of them succumb to the cytotoxic effects of the toxin. SLO killed almost half of the cells by 4 hours, but cell damage was evident after only 2 hours, as indicated by a strong release of uric acid from the cells. The cells remaining adherent after 6 hours incubation with SLO were still alive, however most of the cells were no longer adherent (only 38% of the cells remained). We believe that the low IL-1β response that SLO induces from HVECs is due to the release of preformed IL-1β from the cells and not due to de novo synthesis of new cytokines as we saw an IL-1β response as early as 2 hours after SLO was added to the cells (data not shown).

The ability of SLO to damage epithelial cells without eliciting a strong proinflammatory response led us to hypothesize that SLO may also act to augment penetration of S. pyogenes across the epithelium. Using ex vivo porcine vaginal epithelium, we demonstrated that SLO does in fact enhance the ability of S. pyogenes to penetrate the epithelium. In the case of Staphylococcus aureus, however, the presence of α toxin does not increase the ability of the bacteria to penetrate the vaginal tissue. This is interesting because Staphylococcus aureus typically remains on the vaginal surface in cases of menstrual TSS, whereas S. pyogenes is often found in the bloodstream in cases of streptococcal TSS. Therefore, it is possible that SLO causes enough damage to the mucosa to allow for penetration of both SPE A and S. pyogenes, whereas α toxin only acts to enhance penetration of TSST-1 across the epithelium. This is consistent with in vivo experiments conducted in rabbits that demonstrated the ability of α toxin to enhance penetration of TSST-1 across the vaginal epithelium to cause lethal TSS.

Based on these results we propose two separate models of penetration for staphylococcal and streptococcal superantigens (Figure 8). In the case of Staphylococcus aureus infections, we have demonstrated that both α toxin and TSST-1 are proinflammatory when incubated with HVECs. Therefore, we propose that it is an inflammatory process induced by α toxin which acts to disrupt the epithelial barrier sufficiently to allow for enhanced penetration of TSST-1. The main cytokines induced by α toxin are IL-6, IL-1β, and TNF-α, which are known to induce the acute phase proteins, activate the vascular endothelium, and to cause local tissue destruction (77). Additional research done in our laboratory has indicated that superantigens are binding a receptor on epithelial cells in order to induce cytokine production from the cells (45). Typically, the outermost epithelial layers of the vaginal mucosa are senescent compared to those cells found adjacent to the mucosal connective tissue, therefore we believe that superantigens must be binding the viable epithelial cells in the lower layers of the mucosal epithelium. Thus, α toxin may act on the outermost layers of the mucosa to cause inflammation, but TSST-1 induces another set of cytokines and chemokines once it reaches the lower layers of epithelium. The chemokines IL-8 (CCL8) and MIP-3α (CCL20) in particular may be responsible for recruiting neutrophils and T cells to the site of infection (78, 79). We believe that recruitment of adaptive immune cells to the sub-epithelial mucosa is necessary for superantigens to encounter an adequate amount of T cells and antigen presenting cells to elicit the massive cytokine production that leads to TSS; we have shown that such recruitment can occur in our ex vivo porcine vaginal model (34).

Figure 8. Two different proposed models for penetration of staphylococcal and streptococcal superantigen penetration of vaginal mucosa.

A) In the case of Staphylococcus aureus infections, both α toxin and TSST-1 are secreted at the vaginal surface. α toxin induces a strong proinflammatory response from HVECs, without necessarily damaging the epithelium (unless at high concentrations). The localized inflammation caused by α toxin acts to disrupt the barrier, allowing TSST-1 to penetrate the epithelium and interact with active epithelial cells in the underlying layers to induce a proinflammatory response of its own. Chemokines elicited by TSST-1 recruit adaptive immune cells to the area, which the superantigen can then interact with to induce T cell proliferation and massive cytokine production, which leads to toxic shock syndrome. B) In the case of Streptococcus pyogenes infections, it follows that both SPE A and SLO are made on mucosal surfaces. SLO directly damages the epithelium without inducing a strong cytokine response from the cells. The damage caused by SLO allows SPE A to better penetrate the epithelium to gain access to lower level epithelial cells, which are more active and therefore can be stimulated to produce proinflammatory cytokines by the superantigen. At this point, the model becomes similar to that proposed for Staphylococcus aureus, in which adaptive immune cells are recruited to the underlying layers of the mucosa and the superantigen starts the cytokine cascade that eventually leads to toxic shock syndrome.

S. pyogenes, although not commonly found in the vaginal tract, is more often found on the oral mucosa. As we have already argued, we believe that the effects shown here can be translated to those occurring at another mucosal surface. SLO is relatively non-inflammatory, but it does induce a large amount of damage to the epithelium through direct interactions with epithelial cells. SPE A, like TSST-1, is proinflammatory, and we believe that the interaction of SPE A with epithelial cells is actually occurring in the lower layers of the epithelium, similar to that suggested for TSST-1. Thus, we propose that SLO acts to directly damage epithelial cells that form the outermost layers of the mucosa without inducing a strong inflammatory response from the cells. This allows for SPE A to better penetrate the superficial permeability barrier, and to bind a receptor on viable epithelial cells that are located in the lower epithelial layers. SPE A induces the chemokines IL-8 (CCL8) and MIP-3α (CCL20) like TSST-1, and it is possible that those chemokines are directly responsible for recruiting adaptive immune cells to the sub-epithelial mucosa. Here, the models are similar in that the superantigen is better able to interact with T cells and macrophages when a strong immune response is actively attracting a large number of cells to the site of infection. Because of the direct cell damage caused by SLO in the absence of a strong inflammatory response, we propose that SLO may also act to enhance penetration of the bacteria through the mucosa. This is important because in streptococcal TSS, S. pyogenes is often found in the bloodstream, compared to mTSS in which Staphylococcus aureus remains localized on the vaginal surface. The lack of inflammatory response to SLO may provide a way for the S. pyogenes to rapidly traverse the superficial mucosal barrier before an appropriate immune response can be mounted against the bacteria.

We showed that the majority of radiolabeled SPE A remains in the uppermost layers of the mucosa, which may act as a reservoir for additional superantigen during infection. Other work done in our laboratory has implicated that superantigens may be “sequestered” in the mucosa through interactions with a receptor on epithelial cells (Brosnahan, A.J., M.M. Schaefers, W.H. Amundson, M.L. Peterson, and P.M. Schlievert, “Toxic Shock Syndrome Toxin-1 (TSST-1) Amino Acid Residues Required for Interaction with Human Vaginal Epithelial Cells (HVECs),” Published abstract for the American Society of Microbiology General Meeting, 2008) (45). Significant interaction between the superantigen and epithelial cells may be required to induce a sufficient immune response to initiate the cascade of events that leads to TSS. Therefore it is possible that cytolysins are only secreted at the mucosal surface to create an initial disruption of the barrier which subsequently allows superantigens to gain access to the more active epithelial cells in the lower layers of the mucosa. There the interaction between superantigens and epithelium leads to production of chemokines required to recruit an ample number of adaptive immune cells to the site of infection. At this point, the superantigens are able to cross-bridge the TCR of T cells and the MHC II of antigen presenting cells so as to induce massive cytokine production, which eventually will lead to TSS.

In sum, these studies have demonstrated the ability of staphylococcal and streptococcal cytolysins to augment penetration of superantigens through porcine vaginal tissue in an ex vivo model of superantigen penetration. The cytolysins were shown to cause damage to the mucosa, particularly the outermost layers of tissue. In vitro human cell studies demonstrated that while both superantigens (TSST-1 and SPE A) were capable of inducing cytokine and chemokine production from HVECs, only the cytolysin α toxin was proinflammatory when incubated with the cells. SLO acted quickly to damage the cells, and we believe that only preformed IL-1β was released. A high concentration of α toxin killed the cells; however this process was slow enough to allow for production and secretion of various cytokines and chemokines before cell death. Finally, we proposed two distinct models of superantigen penetration for Staphylococcus aureus and S. pyogenes that involved disruption of the mucosal tissue through an inflammatory reaction to α toxin and through direct cellular damage by SLO, both of which lead to enhanced superantigen penetration.