Toxic Shock Syndrome Toxin 1 Evaluation and Antibiotic Impact in a Transgenic Model of Staphylococcal Soft Tissue Infection

Staphylococcal toxic shock syndrome (TSS) is a life-threatening illness causing fever, rash, and shock, attributed to toxins produced by the bacterium Staphylococcus aureus, mainly toxic shock syndrome toxin 1 (TSST-1). TSS was in the past commonly linked with menstruation and high-absorbency tampons; now, TSS is more frequently triggered by other staphylococcal infections, particularly of skin and soft tissue. Investigating the progress and treatment of TSS in patients is challenging, as TSS is rare; animal models do not mimic TSS adequately, as toxins interact best with human immune cells. We developed a new model of staphylococcal soft tissue infection in mice producing human immune cell proteins, rendering them TSST-1 sensitive, to investigate TSS. The significance of our research was that TSST-1 was found in soft tissues and immune organs of mice and that early treatment of mice with the antibiotic clindamycin altered TSST-1 production. Therefore, the early treatment of patients suspected of having TSS with clindamycin may influence their response to treatment.

KEYWORDS antibiotics, dissemination, HLA-DQ8, nonmenstrual toxic shock syndrome, Staphylococcus aureus, TSST-1, transgenic mice S taphylococcal toxic shock syndrome (TSS) is a potentially lethal illness characterized by fever, rash, desquamation, organ dysfunction, and shock. The syndrome is attributed to superantigens produced by Staphylococcus aureus, in particular, toxic shock syndrome toxin 1 (TSST-1). TSST-1 has been associated with almost all menstrual TSS (mTSS) and half of nonmenstrual TSS (nmTSS) cases (1), while staphylococcal enterotoxins A, B, and C (SEA, SEB, and SEC) are implicated in the remaining nmTSS cases (2,3). Superantigens bind simultaneously to the HLA class II molecule on antigen-presenting cells and the T-cell receptor, causing massive T-cell activation, expansion, and cytokine release (2). In the United Kingdom, nmTSS is now more common than mTSS. Skin and soft tissue infections (SSTI) are the most frequent trigger for nmTSS, which, in the United Kingdom, is associated with TSST-1-producing strains in 41% of cases (4). Due to its rarity, TSS pathogenesis is poorly understood, and there is a paucity of clinical data to guide treatment choices. Notwithstanding a lack of clinical trial or in vivo data, combination antimicrobial treatment with ␤-lactams and protein synthesis inhibitors is recommended for staphylococcal TSS (5), based solely upon in vitro studies and extrapolation from observational studies of streptococcal TSS.
Murine models of TSS may provide insight into TSS pathogenesis and antimicrobial efficacy but are hampered by low-affinity interactions between murine major histocompatibility class II (MHC II) and staphylococcal superantigens. Prior sensitization with lipopolysaccharide (6) or D-galactosamine (7) has been used to induce superantigenmediated lethality, though the pathological changes incurred may differ markedly from those induced by superantigen alone. Transgenic expression of human HLA class II can render mice superantigen sensitive and allows investigation of superantigen-associated inflammation without the need for sensitization (8), removing potential experimental confounders. There are few recent studies of staphylococcal TSS infection using contemporary clinical strains and none that evaluate disease progression and toxin release in superantigen-sensitive mice. We developed a humanized transgenic model of superantigen-associated SSTI using a clinical TSST-1-producing CC30 methicillinsensitive S. aureus (MSSA) TSS-associated isolate to investigate the pathogenesis and treatment of nmTSS.
(This work was presented in part at the 55th Interscience Conference on Antimicrobial Agents and Chemotherapy [ICAAC], San Diego, CA, in September 2015.)

RESULTS
HLA-DQ8 transgenic mice are superantigen and TSST-1 sensitive. The proliferation of mouse spleen cells in response to superantigens was assessed to determine the superantigen sensitivity of transgenic mice in comparison to that in wild-type mice to recapitulate human immune responses to superantigens, such as those which occur in TSS. Spleen cells of HLA-DQ8 transgenic mice were markedly more sensitive to purified TSST-1 than spleen cells from wild-type C57BL/6 mice (Fig. 1A). Indeed, HLA-DQ8 spleen cells were more sensitive to all staphylococcal superantigens tested than cells from either HLA-DR4 or wild-type mice (see Fig. S2 in the supplemental material). Because of this, all further experiments were performed using HLA-DQ8 mice.
The response of HLA-DQ8 splenocytes to superantigens was compared to that of human peripheral blood mononuclear cells (PBMCs). HLA-DQ8 mice were sensitive to TSST-1 at micromolar concentrations and to SEB at nanomolar concentrations. Human PBMCs were sensitive at picomolar concentrations to both superantigens (Fig. S3).
Modeling soft tissue infection in HLA-DQ8 mice. Having demonstrated responsiveness to TSST-1, HLA-DQ8 transgenic mice were infected subcutaneously (s.c.) with TSST-1-producing S. aureus, and groups were euthanized at 24, 48, and 72 h. By 24 h, there was visible abscess formation at the inoculation site. Abscess volume and bacterial load decreased by 72 h. Bacteria disseminated to the spleen at all time points in 17/18 mice, though spleen bacterial loads decreased during infection. Two to three mice in each group had detectable bacteremia at each time point (Fig. 2). Weight loss was maximal 24 h following infection (median, 9.3%; range, 0 to 15.9%).
tst mRNA transcripts in pus obtained from the abscess were detectable in 5/6 mice at 24 h, 1/6 mice at 48 h, and 2/6 mice at 72 h and were maximal at 24 h (Fig. 3A). Due to the use of abscess samples for RNA and other analyses, measurement of TSST-1 protein was undertaken for just one mouse at each time point; TSST-1 protein was, however, detected by Western blotting in the 24-h pus sample (8 g/ml) but not at 48 or 72 h. Human PBMCs were sensitive to TSST-1 at nanomolar concentrations (Fig. 3B). Pus recovered from abscesses demonstrated sustained mitogenic activity toward human PBMCs at 24 and 48 h, but not at 72 h, when diluted 1:100. Strong mitogenic activity toward human PBMCs was also detected in all sera at 24 h despite infrequent bacteremia, consistent with the presence of superantigen in the mouse serum ( Fig. 3C and D).
On histological analysis of abscess sections from single mice, bacteria were detected on each day of infection, accompanied by heavy subcutaneous infiltration by neutrophils, with inflammation ( Fig. S4).
Draining lymph node involvement during S. aureus infection. To determine whether S. aureus infection involved draining inguinal lymph nodes, four HLA-DQ8 mice infected with the CC30 tst-positive S. aureus strain HSS357 were euthanized 24 h following infection and dissected, with careful removal of inguinal lymph nodes. Bacteria were detected in the subcutaneous abscess (median, 1.3 ϫ 10 7 CFU; range, 0 to 2.8 ϫ 10 7 CFU/abscess), the ipsilateral inguinal lymph node (median, 8.5 CFU; range, 0 to 1.4 ϫ 10 4 CFU/lymph node), and spleen (median, 4.0 ϫ 10 Ϫ1 CFU; range, 2.0 ϫ 10 Ϫ1 to 2.6 ϫ 10 3 CFU/mg spleen), but not the contralateral inguinal lymph node or blood. The greatest bacterial burden was in the subcutaneous abscess, but there was also abscessation in the ipsilateral inguinal lymph nodes of all mice. TSST-1 was detected at the highest level in pus from the subcutaneous abscesses of all four mice and was detected in the ipsilateral inguinal lymph nodes from 2/4 mice but not in any contralateral inguinal lymph node (Fig. 4A). Mitogenicity was elicited (in descending order of magnitude) from the subcutaneous abscess pus, ipsilateral inguinal lymph node, and serum from infected mice (Fig. 4B), consistent with the presence of superantigen.
Antibiotic impact on lesion size and superantigen toxin production. S. aureusinfected HLA-DQ8 mice were treated with a single dose of flucloxacillin (FCX), clindamycin (CLD), FCX-CLD, or sterile PBS at 24 h postinfection and were euthanized at 30 h, i.e., 6 h after antibiotic or PBS administration, to assess the impact of antibiotics on TSST-1 production and the host immune response. Within 6 h of antibiotic administration, mice treated with CLD-containing regimens had smaller abscesses and reduced local TSST-1 production compared to those of mice treated with PBS or FCX alone ( Fig. 5A and B). Accordingly, there was a clear reduction in pus and serum sample mitogenicity ( Fig. 5C and D). The pus sample was not subjected to quantitative real-time PCR (qRT-PCR) of tst transcripts due to previously low levels of transcript detection.
Cytokine differences between control and antibiotic-treated groups were negligible; however, the level of IL-2 was reduced in mice treated with CLD-containing regimens compared to those in mice treated with PBS or FCX, consistent with reduced T-cell stimulation (Fig. 6). The single antibiotic dose did not impact the numbers of CFU detected in the abscess, lymphoid organs, or blood compared with those detected in mice given PBS (Fig. 7). In particular, CLD did not confer any additional benefit with regard to bacteriological clearance during this short treatment time.

DISCUSSION
We describe an S. aureus subcutaneous-abscess model in HLA-DQ8 transgenic mice. These mice demonstrated sensitivity to TSST-1, which expanded the murine T-cell receptor V␤ subsets TCR V␤3 and TCR V␤13. S. aureus disseminated to the draining inguinal lymph nodes and spleen, while TSST-1 production was detected in not only the subcutaneous abscess but also the draining inguinal lymph node, signaled by mitogenic activity in the abscess pus, the ipsilateral inguinal lymph node, and serum. Clindamycin-containing antibiotic regimens reduced abscess volume, TSST-1 production, and the overall mitogenic activity of the lesion within just 6 h of a single treatment.
Experimental comparison of infections with different strains of transgenic mice was not possible in this study. It was notable that HLA-DQ8 mice appeared to be more sensitive to TSST-1 than HLA-DR4 mice in vitro, despite previous work indicating that TSST-1 binds to HLA-DR molecules with levels of affinity greater than (9) or equal to (10) its levels of affinity to HLA-DQ molecules. Polymorphisms in HLA-DR are also known to affect transgenic-mouse responses to TSST-1 (11). It is likely that differential expression of the HLA transgenes and endogenous H2 or responding T-cell subset repertoires may explain the enhanced responses in HLA-DQ8 mice, as we noted that responsiveness to concanavalin A (ConA) and medium alone was also greater in HLA-DQ8 mice. Notwithstanding the in vitro findings, infection with a CC30 S. aureus isolate that produces a high level of TSST-1 led to marked suppuration, abscess formation, and cytokine production.  We believe that the suppuration observed was in part related to TSST-1 responsiveness. It was not possible to evaluate an isogenic, tst-deficient strain to determine which of the observed effects were due solely to TSST-1; the CC30 lineage of S. aureus is challenging to transform, although recent tools may now allow for its manipulation (12). HSS357 carried (aside from tst) sei and seg, which are found in the enterotoxin gene cluster (egc). The egc gene cluster is widespread in S. aureus strains (13) and unlikely to have any specific association with TSS.
The measurement of TSST-1 production in vivo provides valuable contextual information to inform future superantigen research. In some experiments, we detected higher levels of TSST-1 in subcutaneous abscess pus than could be detected during broth culture of the same strain (4) and higher levels than previously reported in the abscesses and kidney extracts of mice infected with TSST-1-producing S. aureus (14,15), consistent with upregulation of superantigen production during suppurative infection. Microbial spread and suppuration were noted in the ipsilateral inguinal lymph node, a tissue rich in superantigen-sensitive T cells, and TSST-1 protein was detected here too 24 h after infection. The detection of bacteria and TSST-1 in ipsilateral, but not contralateral, inguinal nodes is consistent with transit in afferent lymphatic vessels rather than blood. During infection, abscessation was present in all the ipsilateral inguinal lymph nodes, including those without detectable viable bacteria, as S. aureus may have been killed by the neutrophils creating the pus. We cannot determine whether TSST-1 was produced within the lymph node by S. aureus or transferred from the site of infection via lymphatics. The proximity of TSST-1-producing S. aureus to such lymphoid tissues may be pivotal to events occurring during TSS; the extent to which this occurs in clinical disease is unknown. It is widely believed that TSS results from the systemic dissemination of superantigens into the bloodstream and consequent interaction with leukocytes in the lympho-reticular system; in mice, the spleen is known to act as a major source of cytokines following systemic superantigen administration (16). Our results raise the possibility that superantigen exposure within secondary lymphoid organs, such as the lymph nodes, may contribute to the cytokine storm underlying TSS pathogenesis during infection. Previous studies using HLA transgenic mice have reported IL-6, IFN-␥, and IL-2 serum responses in HLA-DQ8 mice exposed to aerosolized SEB (17) and IL-6 and IFN-␥ production by HLA-DR3 spleen cells exposed to SEB (18). We detected cytokine responses to a bolus of toxin and to infection; IL-6, IFN-␥, KC, and MCP-1 were the most consistent markers of inflammation.
Lethal shock was not observed or expected in this model; we note that sensitization agents were not used and that bacterial clearance occurred over the 72-h experimental period. Thus, this model does not replicate TSS as observed in humans but served to elucidate potential pathways of TSST-1 production, dissemination, and abscess progression in a model that reproduces some of the immunological responses to staphylococcal infection seen in humans. Previous models of TSS in HLA-DR1 mice required D-galactosamine pretreatment to elicit liver failure, an event that is entirely related to tumor necrosis factor (TNF)-induced hepatocyte apoptosis in the D-galactosamine setting, while SEB alone results in cytokine release only (16). Changes in serum cytokine levels were observed in both HLA-DQ6 and HLA-DQ8 mice at 4 h following SEB and streptococcal pyrogenic exotoxin A (SPEA) challenge (19). In our study, weight loss, tst transcripts, TSST-1 protein, and cytokines were maximal at 24 h following infection, consistent with a marked systemic inflammatory response that might be like that observed in nmTSS.
TSST-1 expanded TCR V␤ subsets 3 and 13 in HLA-DQ8 mouse splenocytes. TSST-1-induced TCR V␤15 and -17 subset expansion, as was previously reported in earlier murine studies, could not be evaluated in the current study, as the assay used did not detect them (20,21). Further work to determine whether TSST-1 results in specific T-cell V␤ subset expansion and cytokine release within lymphoid organs in the context of S. aureus infection would provide novel insight into nmTSS pathogenesis.
Notwithstanding findings in local lymph nodes, the mitogenicity assays strongly pointed to the presence of superantigen in the sera of infected mice. While we could not directly quantify TSST-1 in serum, parallel standard TSST-1 bioactivity assays yielded data suggesting that 1 to 10 pg/ml of TSST-1 was present in the blood. TSST-1 at 0.2 pg/ml is reported to cause half-maximum proliferation of human T cells (2). We cannot rule out the possibility that low levels of S. aureus were present in the blood (limit of detection, 200 CFU/ml). However, the absence of a detectable bacteremia supports the assertion that TSST-1 may disseminate from the initial infection site to systemic circulation either by transcytosis across cellular barriers to reach the blood (22) or via the lymphatic system, enabling activation of T cells distant to the site of infection.
Current TSS treatment recommendations advise a combination of ␤-lactam and lincosamide antibiotics, until culture results are known (5). This is based on in vitro studies, extrapolation from observational studies of streptococcal TSS, and in vivo evaluations of the effects of protein synthesis inhibitors in rabbit models of pneumonia using Panton-Valentine leukocidin-producing S. aureus (23). There is a lack of published in vivo data on the effect of clindamycin on TSST-1 production in any infection model. We elected to evaluate this effect and whether using clindamycin at the outset of TSS management might impact disease progression. We chose to treat a time point when tst transcripts and TSST-1 protein were maximal and cytokines detectable. Unsurprisingly, a single dose of antibiotic did not reduce abscess bacterial burden, consistent with previous findings (24). Notably, however, mice treated with clindamycincontaining regimens had smaller abscesses, reduced TSST-1 production, and diminished mitogenicity of pus and serum compared to those of mice treated with other regimens. This is the first work to demonstrate the superantigen-inhibitory effects of clindamycin in vivo. The findings suggest that clindamycin may have an indirect effect on disease and abscess progression, potentially by reducing TSST-1 synthesis, despite little measurable effect on bacterial counts by this model. Previous reports do suggest that abscessation may be enhanced by superantigens; hepatic abscesses are known to develop in HLA transgenic mice exposed to SEA-producing S. aureus (25). Although clindamycin can exert an inhibitory effect on superantigen-induced host cytokine production in vitro (26), serum cytokine levels in our study did not demonstrate a clear antibiotic effect, perhaps due to the timing of analysis following one antibiotic dose.
Our findings support the adjunctive use of clindamycin to modify disease progression in the treatment of suspected staphylococcal TSS, to reduce superantigen toxin production more rapidly, and to potentially reduce abscessation. Further studies are required to increase our understanding of TSS pathogenesis and the role of lymph node superantigen expression and to explore the efficacy of treatment with other immune modulators, such as intravenous immunoglobulin, to limit the lethal potential of this syndrome.
Bacterial culture. HSS357, a clinical tst-positive CC30 MSSA strain that caused TSS, was selected based on highest in vitro TSST-1 production (187 ng/ml following overnight culture in 5 ml of brain heart infusion [BHI] broth) among clinical tst-positive CC30 MSSA isolates causing TSS (4). Overnight culture of strain HSS357 in 50 ml BHI yielded 400 ng/ml of TSST-1, although transcription of tst peaked at 8 h and diminished thereafter (see Fig. S1 in the supplemental material). HSS357 was sensitive to all antibiotics, including clindamycin and flucloxacillin, with the exception of penicillin. Antibiotic MICs were determined by British Society for Antimicrobial Chemotherapy methods (http://www.bsac.org.uk) and interpreted in accordance with European Committee on Antimicrobial Susceptibility Testing guidelines (http://www.eucast.org). HSS357 carried the superantigen genes seg and sei in addition to tst, determined by toxin gene profiling (sea to see, seg to sej, tst, and pvl) by multiplex PCR (29,30).
For in vivo administration, HSS357 was cultured overnight in BHI broth at 37°C with agitation at 200 rpm and then centrifuged, washed, and resuspended in sterile phosphate-buffered saline (PBS). Inocula, pus, and tissue samples for culture were plated onto Luria broth (LB) agar and incubated overnight at 37°C prior to quantification of CFU per milliliter.
To determine superantigen sensitivity in vivo, 80 g of TSST-1 in 100 l of PBS was administered intraperitoneally to HLA-DQ8 and C57BL/6 mice. Blood was taken by tail bleed at 2 h and by cardiac puncture at 6 h. Serum was separated and stored at -20°C for cytokine analysis.
Bacterial infection. HLA-DQ8 mice were infected subcutaneously on a shaved area of the right flank with 1 ϫ 10 9 CFU of tst-positive S. aureus in 100 l sterile PBS or with PBS alone as a control. To investigate antibiotic impact, mice were given intraperitoneal flucloxacillin (FCX; 12.5 mg/kg of body weight), clindamycin (CLD; 10 mg/kg), or flucloxacillin with clindamycin (FCX-CLD, 12.5 mg/kg and 10 mg/kg, respectively) in 100 l of PBS or 100 l PBS as a control 24 h postinfection.

ACKNOWLEDGMENTS
We acknowledge D. Kioussis and R. A. Flavell for transgenic mice and D. M. Altmann for access, and we thank Nur S. C. Ahmad for technical contributions.
We acknowledge support from the NIHR Biomedical Research Centre (BRC) to the Imperial College Healthcare Trust and the NIHR BRC-funded Tissue Bank. This work was also supported by the UK Clinical Research Collaboration through a research training fellowship (G0800777/1 to H.S.).
We have no conflicts of interest to declare.