A total of 541 clinical strains of S. aureus, including 256 methicillin-resistant S. aureus (MRSA) and 285 methicillin-susceptible S. aureus (MSSA), isolated from patients with various diseases between 1981 and 1996 from 10 medical institutions in Japan were included in the analysis of TSST-1 production. We have also analyzed seven tst-positive strains isolated from non-mTSS patients between 1999 and 2017 in Japan (Table 1) and seven tst-positive strains isolated between 2015 and 2018 in Japan from patients who did not develop TSS. The diagnosis of TSS was made clinically based on the CDC criteria (CDC, 1990; Supplementary Table 1). In addition, N315 (MRSA, ST-5) isolated from patients who did not develop TSS was included as a control strain (Kuroda et al., 2001).

S. aureus strains used in this study were cultivated in Todd Hewitt (TH) broth (Becton, Dickinson and Company; BD, NJ, United States), brain heart infusion (BHI) broth (BD), or BHI agar (BD) with aeration at 37°C, unless indicated otherwise. Escherichia coli DH5α and BL21 were aerobically grown in Luria–Bertani (LB) broth (BD) and tryptic soy agar (TSA, BD) at 37°C. In order to maintain the plasmids of S. aureus and E. coli, 10 μg/ml of chloramphenicol (Nacalai Tesque, Inc., Kyoto, Japan) was supplemented in culture medium. The bacterial culture was stocked at −80°C in a final concentration of 50% glycerol (Wako Pure Chemical Industries, Ltd., Tokyo, Japan) until further use.

S. aureus isolates grown overnight in BHI broth were lysed by lysostaphin (Merck KGaA, Darmstadt, Germany) and achromopeptidase (Wako Pure Chemical Industries, Ltd.). PCR was performed by Quick Taq ^® HS DyeMix (Toyobo Co., Ltd., Osaka, Japan) using the cell lysate as a template. Primer set GTSSTR-1 and GTSSTR-2 (Supplementary Table 2) was used to amplify a 326-bp region of tst. The thermal cycling conditions included initial denaturation at 95°C for 3 min followed by 35 cycles of 98°C for 10 s, 50°C for 30 s, and 68°C for 1 min. The amplified products were electrophoresed on 1% agarose gel, stained with ethidium bromide, and visualized using AE-6933FXES Printgraph (Atto Co., Tokyo, Japan).

Phenol–chloroform-based extraction of genomic DNA from S. aureus and purification of DNA using a QIAamp DNA mini kit (Qiagen, Hilden, Germany) was described previously (Watanabe et al., 2018). Genomic sequencing and assembly were performed as previously described with slight modifications (Watanabe et al., 2019), and default parameters were used except otherwise noted. Briefly, sequencing was performed using a combination of Oxford Nanopore Technologies (ONT) platform [MinION Mk-1B devise (ONT, Oxford, United Kingdom) integrated with a FLO-MIN106 (R9.4.1) flow cell (ONT)] and Illumina MiSeq platform [Nextera XT library preparation kit (Illumina) and the MiSeq reagent kit version 3 (Illumina)]. Raw data obtained from the ONT platform were base called and demultiplexed by the MinKNOW software (ONT) and porechop (version 0.2.41¹), respectively. After read-trimming by seqkit (v0.10.1) (Shen et al., 2016), the reads were de novo assembled into single contigs by flye (v2.8.1-b1676) (Kolmogorov et al., 2019) and racon (v1.4.12) (Vaser et al., 2017). The assembled sequences were polished with short-read sequencing data of MiSeq by pilon (v1.22) (Walker et al., 2014). The collected whole-genome sequences were finally annotated by prokka (v1.14.6) (Seemann, 2014). Assembled genome sequences of 998 tst-positive S. aureus strains were downloaded from the National Center for Biotechnology Information (NCBI).² Phylogenetic analysis was performed by kSNP3.0 (Gardner et al., 2015).

Frozen stock of S. aureus strain was revived in 4 ml of BHI broth for one night. To measure TSST-1 production without induction, 2 μl of the overnight culture was added into 1 ml of BHI broth. For serum induction assay, 2 μl of the overnight culture was added into 200 μl of BHI broth supplemented with 800 μl of horse serum (Thermo Fisher Scientific, MA, United States) or pooled human serum (Kohjin Bio Co., Ltd., Saitama, Japan). The culture was incubated with shaking at 37°C in 5% CO₂ for 16 h.

TSST-1 production was determined by reversed passive latex agglutination kit (TST-RPLATM, Denka Seiken Co., Ltd., Tokyo, Japan) or sandwich enzyme-linked immunosorbent assay (ELISA). Latex agglutination assay has a detection limit of 1–2 ng/ml. In latex agglutination assay, 25 μl of twofold serially diluted culture supernatant was mixed with an equal volume of latex particles conjugated with antibodies against TSST-1 in V-bottomed microtiter plates. After 20 h of incubation at room temperature, the agglutination reaction was observed. An S. aureus strain was considered positive for TSST-1 production if aggregation was observed in wells with titers greater than 2. For ELISA assay, rabbit anti-TSST-1 antibody was used as a capture antibody, and chicken anti-TSST-1 immunoglobulin Y was used as a detection antibody. Using the avidin–biotin system, o-phenylenediamine’s absorbance was detected at a wavelength of 490 nm with VersaMax microplate reader (Molecular Devices, San Jose, CA, United States).

To assess the activity of tst promoter, the tst promoter regions of representative strains (N315, JMUB2909, JMUB3011, and JMUB3038) were amplified by PCR using the primer set pKAT-N315-tst-prom-F2 and egfp-N315-tst-prom-R (Supplementary Table 2). In addition, egfp and pKAT vector (Kato et al., 2017) were amplified with primer sets egfp-F/pKAT-egfp-R and pKAT-F/pKAT-R, respectively (Supplementary Table 2). The three PCR fragments were assembled by NEBuilder (New England Biolabs, MA, United States) to construct four plasmids: pKAT-9T-tst-promoter-gfp, pKAT-8T-tst-promoter-gfp, pKAT-7T-tst-promoter-gfp, and pKAT-6T-tst-promoter-gfp. These plasmids were transformed into E. coli DH5α. After extraction, the plasmids were electroporated into competent cells of N315 strain, as described previously (Sato’o et al., 2018). One microliter of the precultured N315 transformant was transferred into each 100 μl of BHI supplemented with 400 μl of human serum and 500 μl of BHI. A final concentration of 10 μg/ml of chloramphenicol was added into each medium to maintain plasmids. After 16 h of incubation, each culture was centrifuged at 10,000 rpm for 10 min, and the pellet was dissolved in 50 μl of PBS. The fluorescence was measured at an emission wavelength of 538 nm after excitation with a wavelength of 485 nm by microplate fluorometer Fluoroskan Ascent FL (Thermo Fisher Scientific). The fluorescence intensity was normalized by the turbidity of the bacterial culture.

The tst promoter regions of JMUB3007 and N315 strains were exchanged using a temperature-sensitive plasmid vector pIMAY (Monk et al., 2012). The tst promoter regions of both strains were amplified by PCR using primers pIMAY-tst-UP and pIMAY-tst-DN2 (Supplementary Table 2). On the other hand, pIMAY was amplified with primer set pIMAY-DW-F and pIMAY-DW-R (Supplementary Table 2). The DNA fragments were assembled by In-Fusion HD cloning kit (Takara Bio Inc., Shiga, Japan) and transformed, first, into E. coli DH5α and followed by E. coli BL21. E. coli B derivative BL21 strain was used as an intermediate to increase S. aureus transformation efficiency as our previous research has shown that the plasmids extracted from BL21 have an increased transformation efficiency of > 2 log10 folds (Sato’o et al., 2018). After extraction, the plasmids were electroporated into the respective S. aureus strains. The plasmid was then removed by culturing the transformants in a medium supplemented with anhydrotetracycline. The sequences of the replaced tst promoter regions were confirmed by Sanger-sequencer 3730xl DNA Analyzer (Thermo Fisher Scientific).

The sarA gene of the N315 strain was eliminated using temperature-sensitive plasmid pIMAY. About 1-kbp upstream and downstream flanking DNA fragments of sarA were amplified by PCR using primer sets pIMAY-sarAKO-DN1/sarAKO-fPCR-UP1 and pIMAY-sarAKO-UP1/sarAKO-fPCR-DN1 (Supplementary Table 2). A DNA fragment comprising a pIMAY vector backbone was amplified with primers pIMAY-DW-F and pIMAY-DW-R2 (Supplementary Table 2). The three DNA fragments were assembled by In-Fusion HD cloning kit and transformed into E. coli BL21. After extraction of knockout plasmid pIMAY sarAKO, the plasmid was electroporated into N315. The sarA gene of N315 was then eliminated by allelic exchange as described previously (Monk et al., 2012). The sequence of N315 ΔsarA was confirmed by Sanger-sequencer 3730xl DNA Analyzer. Finally, pKAT-9T-tst-promoter-gfp and pKAT-8T-tst-promoter-gfp were electroporated into N315 ΔsarA for use in reporter assay.

Reporter assay was performed in 50 μl of total reaction volume. Briefly, 0.2 μl of overnight culture was inoculated into 10 μl of BHI supplemented with 40 μl of human serum or 50 μl of BHI; both culture medium contained 10 μg/ml of chloramphenicol for plasmid maintenance. After 16 h of incubation, each culture was centrifuged at 15,000 rpm for 5 min, and the pellet was resuspended in 50 μl of PBS. The fluorescence was measured at an emission wavelength of 538 nm after excitation with a wavelength of 485 nm by a microplate fluorometer Fluoroskan Ascent FL. The fluorescence intensity was normalized against the turbidity of the bacterial culture.

Student’s t-test was used for the analysis of normally distributed continuous variables.

The genome sequence data were registered in the DNA Data Bank of Japan (DDBJ).³ The accession numbers of CC-5 S. aureus Sak-1, JMUB3007, JMUB3024, JMUB4687, and JMUB4716 are DRR303938, DRR307742, DRR307743, DRR307744, and DRR307745, respectively. All accession numbers for S. aureus used in this study are shown in Supplementary Table 3.

The study protocols were approved by the institutional review board of Shizuoka General Hospital (SGHIRB #2016091). Consent to participate was not required, following the ethical guidelines for medical and health research involving human subjects by the Ministry of Health, Labor and Welfare, Japan, since this study analyzed bacteria, which were isolated as clinical specimens, and patients’ personal health information could not be accessed.⁴ Bacterial isolates were taken from hospitals as part of the standard patient care and used anonymously.

S. aureus has a complex regulatory system controlling tst, which varies with the strain background (Novick, 2003; Andrey et al., 2015). In order to identify the specific bacterial populations, which can be triggered to produce TSST-1 during infection, we analyzed 541 clinical S. aureus strains isolated from patients with various diseases between 1981 and 1996 from 10 medical institutions in Japan. First, we assessed the ability of the 541 strains to produce TSST-1 under in vitro conditions by passive latex agglutination kit. A total of 157 out of the 541 strains (29%) produced TSST-1 after overnight culture in TH broth. The prevalence of TSST-1-producing strains was similar to previous data, for example, 22.7% of strains isolated in China were found to be able to produce TSST-1 (Zhao et al., 2019). The measured TSST-1 titer showed unimodal distribution from 4 to 4,096 with a median of 512 (Figure 1A). Next, we evaluated if the production of TSST-1 by TSST-1-producing strains was inducible by serum. Supplementation of horse serum into culture medium altered TSST-1 production, and the titers of TSST-1 produced by the 157 strains were distributed from 4 to 32,768 with bimodal peaks at 16 and 16,384 (Figure 1B). Thirty-nine strains (24.8%) produced TSST-1 at titers of equal or more than 8, 192, and 88 strains (56.1%) produced TSST-1 at titers less than 32. These results indicated that the TSST-1-producing strains could be classified into three groups: (1) strains showing increased production of TSST-1 following supplementation of horse serum, (2) strains producing a similar amount of TSST-1 with or without supplementation of horse serum, and (3) strains showing decreased production of TSST-1 following supplementation of horse serum. Interestingly, six MRSA strains isolated from fatal TSS cases, including Sak-1 strain (see below), belonged to the first group, suggesting that there might be an association between TSST-1 induction by serum and development of TSS.

S. aureus strains having the ability to produce a high level of TSST-1 in the presence of serum include those isolated from fatal TSS cases, such as Sak-1, which we reported in 1999 (Cui et al., 1999). The Sak-1 strain was isolated from a postoperative TSS patient who had passed away due to severe clinical symptoms. This strain was shown to produce 32 times more TSST-1 when induced by human blood. Therefore, we had suggested that the induction of TSST-1 by human blood might contribute to the development of non-mTSS (Cui et al., 1999). This hypothesis can be supported by the above TSST-1 induction experiments conducted with 157 TSST-1-producing strains. To further clarify this postulation, we carried out a follow-up study with seven S. aureus strains (inclusive of Sak-1) isolated from the non-mTSS patients between 1999 and 2017 in Japan (Supplementary Table 1). Genome analysis revealed that the seven non-mTSS strains belonged to three different sequence types (STs): ST-5 (three strains), ST-8 (three strains), and ST-30 (one strain). All TSS strains carried the TSST-1 gene (tst) as well as other enterotoxin genes (Table 1). This study also included a well-characterized tst-positive strain N315 (SCCmec type II, ST-5) and seven other tst-positive strains isolated between 2015 and 2018 in Japan from patients who did not develop TSS as controls. These tst-positive strains isolated from patients without TSS belonged to four different ST types: ST-5 (one strain), ST-2809 (two strains), ST-8 (three strains), and ST-30 (two strains).

Phylogenetic analysis using whole-genome sequences showed that all the above strains could be roughly classified into three different CCs: CC-5, CC-8, and CC-30 (Figure 2A). Moreover, the phylogenetic analysis indicated that TSS strains and the strains isolated from patients without TSS were closely related to each other when they had the same ST. The ST-5 and ST-2809 strains belonged to CC-5, and they were classified into the same clade of the phylogenetic tree (Figure 2A).

We assessed the ability of non-mTSS strains to produce increased TSST-1 after supplementation of human serum to growth media. The amount of TSST-1 produced by 15 representative strains in the presence and absence of human serum was measured and compared. As shown in Figure 2B, the induction of TSST-1 production by human serum varied among strains of different genotypes. The CC-5 strain Sak-1 had significantly higher TSST-1 production when human serum was added into the media. This result was in agreement with the previous study, which used whole blood as an inducing agent (Cui et al., 1999). Like Sak-1, other CC-5 strains isolated from TSS patients (JMUB3007 and JMUB3024) also produced a significantly increased amount of TSST-1 in the presence of human serum. In contrast, the CC-5 strains from patients who did not develop TSS had decreased TSST-1 production when human serum was added. We previously described that Mu50 (SCCmec type II, ST-5), isolated from patients who did not develop TSS, showed the same characteristics as our studied strains (Cui et al., 1999). On the other hand, there was no remarkable difference in TSST-1 production between the two different culture conditions for CC-8 and CC-30 strains regardless of whether the strains were isolated from TSS patients or not. These results indicated that the association between induced TSST-1 production and development of TSS would vary depending on the bacteria’s genetic background.

The tst gene is located on the S. aureus pathogenicity island (SaPI), which is highly diverse in structure and sequence (Lindsay et al., 1998). The whole-genome sequence analysis of the 15 strains revealed that the tst gene was located either on the SaPI2 in CC-5 and CC-30 strains or on the SaPI3 in CC-8 strains (Figure 3A). However, no remarkable difference was found in the SaPI structure between strains isolated from TSS patients and patients with clinical manifestations other than TSS (Figure 3A). The sequences of the coding region of tst were also found to be identical in all 15 strains. Nonetheless, sequence alignment of the tst promoter region showed that the number of thymidine (T) in the T repeat region at 107–115 bp upstream of the tst transcription start site (Andrey et al., 2010) varied from six to nine among the tested strains (Figure 3B). This region was supposed to be a putative second binding site of global regulator SarA. Among the six CC-5 strains, two of three CC-5-TSS strains [Sak-1 and JMUB3007 (SCCmec type II, ST-5)] carried eight consecutive T (8-T) repeats, while another CC5-TSS strain [JMUB3024 (MSSA, ST-5)] and the rest of the three strains from patients with clinical manifestations other than TSS carried nine consecutive T (9-T) repeats. The three CC-30 strains and six CC-8 strains carried 6- and 7-T repeats, respectively, regardless of whether the strains were isolated from TSS patients or not. The sequences of the tst promoter region, other than the T repeat region located 107–115 bp upstream of the tst transcription start site, were identical in all 15 strains. However, the induction of TSST-1 production by serum addition was different in each CC. Therefore, we hypothesized that the number of T’s in the region is involved in the regulation of TSST-1 and, thus, conducted an intensive study on the correlation of the number of T repeat with serum induction of TSST-1 production.

To confirm whether the number of T repeats in the tst promoter region affects serum induction of TSST-1 production, a reporter assay was performed using green fluorescent protein (GFP) expression vector pKAT (Kato et al., 2017). A set of tst promoter regions containing different number of T repeats (from six to nine) were individually cloned into the upstream of the egfp gene of pKAT to generate a series of eGFP expression plasmids, where the eGFP expression was regulated by the inserts. The constructed plasmids were then transformed into N315 (SCCmec type II, ST-5) strain, and the emitted fluorescence was measured with or without serum supplementation in the culture medium. As shown in Figure 4A, among all the transformants tested, N315 carrying a plasmid with 8-T repeats showed significantly increased eGFP expression compared with the others, a finding which was consistent with the results of serum induction of TSST-1 production observed in the clinical strains (Figure 2B). Since the CC-5 strains can be divided into two groups according to their TSST-1 inducibility, we also evaluated the influence of the number of T repeat on serum induction of TSST-1 using two CC-5 clinical strains, N315 (SCCmec type II, ST-5, carrying 9-T repeat promoter) and JMUB3007 (SCCmec type II, ST-5, carrying 8-T repeat promoter), by exchanging their tst promoters and subsequently culturing them with/without human serum. As expected, we found that serum induction significantly increased TSST-1 production when the 9-T repeat promoter was replaced with 8-T repeat promoter in N315, and vice versa, decreased when the 8-T repeat promoter was replaced with 9-T repeat promoter as seen in JMUB3007 (Figure 4B). These findings confirmed that the number of T repeats in the tst promoter region was at least one of the direct factors involved in regulating the serum induction of TSST-1 production.

Since the T repeat region locates on the putative second SarA-binding region (Figure 3B), we investigated whether SarA positively or negatively regulates the tst expression of CC-5 strains using GFP reporter assay. A sarA-knockout N315 strain was constructed and transformed with plasmids carrying 9-T or 8-T repeat promoters upstream of gfp gene. Our results showed that sarA knockout strains have significantly decreased tst promoter activity even when they harbored 8-T repeat promoters, which was demonstrated earlier in this study to have a higher strength (Figure 4C). This result, hence, indicates that SarA is a positive regulator of tst gene in CC-5 strains.