The experiment was strictly conducted according to the Guide for Care and Use from the Research Ethics Committee of Soochow University (20210220). All procedures involving human participants were performed to the ethical standards. Patients were given informed consent to participate in the study.

Information on the food poisoning incidents was gathered from the Center for Disease Control and Prevention of Suzhou, China. A total of 896 stool samples (n = 850) and vomit samples (n = 46) from diarrheal patients were collected from 22 hospitals and community health centers between 2011 and 2021 and forwarded for testing using the Chinese National Foodborne Disease Surveillance Manual (Li et al., 2018). These 22 facilities were dispersed among ten regions, between 30°93’N to 35°55’N and 120°21′E to 121°64′E in Eastern China. The collected samples were tested for pathogenic bacteria including Salmonella, Shigella, Staphylococcus aureus, Bacillus cereus, Vibrio parahaemolyticus, diarrhea-causing Escherichia coli, and Listeria monocytogenes. Food poisoning incidents with Staphylococcus aureus as the dominant pathogen were enrolled for further analysis. Detail descriptions of CC, ST types, isolation of year and geographic data of S. aureus were shown in Supplementary Table S1.

The present research was conducted during the period from 2011 to 2021. In the initial step, 10 g of stool samples and vomit samples were collected and transferred to the laboratory on ice. Then samples were homogenized in 0.1% peptone saline in a filter bag (Bkmam, Changde, China). After that, 100 ul were cultured onto Baird Parker agar (HopeBio 4,115, Beijing, China) and CHROMagar™ MRSA agar (Becton Dickinson, Franklin Lakes, NJ). The plates were incubated in the carbon dioxide incubator overnight at 37°C. Then, a loop full of bacterial culture from incubated tubes was streaked separately into the Baird Parker agar, and the plates were examined and studied carefully for the presence of characteristic colonies of S. aureus.

The S. aureus strains were identified by 16 s rRNA sequencing and MALDI-TOF MS (Bio-Merieux, Craponne, France). Generally, a 1.4-kb fragment of the 16S rRNA gene was amplified by PCR with universal primers (27F, 5’-AGAGTTTGATCMTGGCTCAG-3′, 1492R, 5’-TACGGYTACCTTGTTACGACTT-3′). The PCR product was purified and sequenced in both directions by use of conserved-region primers on the platform of Honsunbio company (Shanghai, China). Purified sequencing results were processed and edited by ABI 3100 (Applied Biosystems) and Sequencher (Gene Codes, Ann Arbor, Mich.), respectively. The altered sequences were identified by comparing them to GenBank via Blast. In the case of bacteria with a low identity, MALDI-TOF MS was enrolled for detection. Briefly, each sample was inoculated as triplicates on the target plate and covered by the freshly prepared matrix (Bruker Daltonik GmbH, Bremen, Germany). Mass spectra were compared with spectra obtained from the associated database (Hansen and Lee, 2017; Holzknecht et al., 2018).

Susceptibility testing of confirmed S. aureus strains was performed according to the Clinical and Laboratory Standards Institute disk diffusion method (CLSI 2022). Staphylococcus aureus ATCC 25923 was used as a control strain. Disks from Oxoid were used. The following antimicrobial disks were tested: ampicillin (10 μg), penicillin (10 units), cefoxitin (30 μg), ceftazidime (30 μg), chloramphenicol (30 μg), clindamycin (2 μg), erythromycin (15 μg), gentamicin (10 μg), linezolid (30 μg), and tetracycline (30 μg). Cefoxitin was tested as a surrogate marker for the detection of methicillin resistance.

After growing the isolates for 24 h at 37°C in Tryptone Soya Broth (AOBOX 02–049, Beijing, China), genomic DNA was extracted and purified using a HiPure Bacterial DNA Kit (D3146, Meiji Biotechnology Co., Ltd., Guangzhou, China). Library construction was performed with Vazyme TruePrep DNA Library Prep Kit TD501 (Vazyme, Nanjing, China). A Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies, Wilmington, DE, United States) and a 1.0% (w/v) agarose gel were used to assess the quality of the extracted DNA. Purified DNA was whole-genome sequenced using Illumina HiSeq 4,000 platform (150 bp paired-end reads with ~200-fold average coverage) on the Honsunbio platform (Shanghai, China). The raw reads were trimmed, and genome was assembled by EToKi v1.0 (Zhou et al., 2020). The QUAST v2.3 was utilized to assess the quality of the genome assembly (Gurevich et al., 2013). The raw sequencing reads were uploaded to the China National GenBank with the accession number of CRA010922. The sequences can be accessed at https://bigd.big.ac.cn/gsa/browse/CRA010922.

To classify STs into clonal complexes (CCs), the Illumina read files were subjected to multilocus sequence typing (MLST.20) and eBURST v3 analysis (Urwin and Maiden, 2003; Feil et al., 2004). A maximum likelihood core-genome phylogenetic tree was constructed using 117 strains based on 547 core genes (~96,496 SNPs). The S. aureus ATCC 25923 was employed as the reference genome for SNP analysis. The select minimum depth at SNP positions was set to 10x, while select minimum relative depth at SNP positions was set to 10%. The select minimum distance between SNPs (prune) was set to 10 bp and select minimum SNP quality was set to 30. The select minimum read mapping quality was set to 25 and the select minimum Z-score was set to 1.96 in CSI Phylogeny v1.4 (Kaas et al., 2014).

Antimicrobial-resistant genes and virulent factors were determined using the ResFinder, Mobile ElementFinder, and VFDB databases, with a minimum of 60% nucleotide identity retained in all algorithms (Liu et al., 2019; Bortolaia et al., 2020; Johansson et al., 2021). The detection of arg, icaA, icaB, icaC, icaD, icaR, luxS, ΦSa3 and ΦAVβ (SAAV_2008, SAAV_2009) genes was blast in MyDbFinder 2.0 with the select threshold of 98% and select minimum length of 60% on the platform of Centre for Genomic Epidemiology.¹

The line chart and donut chart were made using GraphPad Prism 7, and statistical significance was assessed using One-way ANOVA with p < 0.05. The genotypic data were visualized in Grapetree and iTOL (Urwin and Maiden, 2003; Letunic and Bork, 2019).

In this investigation, the food poisoning incidents collected for this study were distributed between 30°93’N and 35°55’N and 120°21′E to 121°64′E (Figure 1). A total of 117 strains of S. aureus were isolated from 896 stool and vomit samples collected from diarrheal patients who registered to the 22 hospitals and community health centers between 2011 and 2021 in Eastern China. Among the 117 isolates, 20 isolates were identified as methicillin-resistant S. aureus (MRSA), with the majority of the strains collected from Region IV and Region II. Among them, six strains were isolated from vomit samples and the remaining 111 strains were detected from diarrheal stool. Further epidemiological analysis revealed that the food poisoning incidents in this study were divided sporadically into ten regions, with Suzhou (Region IV) and Kunshan (Region II) being the dominant regions of food poisoning incidents (Figure 1).

The MLST typing of the 117 foodborne S. aureus genomes identified 19 unique CC/ST types, namely CC1, CC5, CC6, CC7, CC8, CC12, CC15, CC22, CC25, CC59, CC72, CC88, CC188, CC398, CC1281, ST672, ST944, ST1920, and ST2315 (Figure 2, Table 1). MRSA was detected in 20 isolates including CC6, CC22, CC59, CC88, and CC398, while MSSA isolates (n = 97) were widely distributed in the majority of the CC/ST types, indicating the diversity of S. aureus contamination in the current food poisoning incidents. Details of the housekeeping genes in MLST typing were described in Supplementary Table S2.

An intriguing finding from this study was the considerable shift in CC types from CC1 to CC398 between 2011 and 2021 (Figure 3B). Before 2018, food poisoning incidents were primarily linked to community-associated CC1. However, the subsequent five years witnessed an increase in the CC398 strains associated with livestock settings, indicating the intricate nature of foodborne S. aureus outbreaks.

Region IV and Region II were identified as the most prevalent regions among the current food poisoning incidents, with CC398 (n = 25), CC5 (n = 14), CC6 (n = 10), and CC15 (n = 10) being the dominant types (Figure 2). Historically, S. aureus CC8, CC15, and CC45 had been the predominant strains causing foodborne diarrhea in European countries (Baumgartner et al., 2014). Analysis of 1850 food products from China revealed that S. aureus CC1 (10.7%) was the most common type, followed by CC7 (10.6%) and CC5 (4.8%), suggesting a different major variation compared to the food poisoning incidents we found in this study (Wu et al., 2018). This disparity could potentially be attributed to differences in Chinese and Western food cultures, as well as variations in cooking practices.

CC398 was identified as the most predominant clonal complex in this study (Figure 2, Table 1). Previous retrospective analysis revealed that human and animal infections with S. aureus CC398 occurred in nations throughout West Europe and Eastern Asia, with the initial cases harboring the strain being identified in the Netherlands (Stegger et al., 2010; Laumay et al., 2021). Since then, the annual growth in the detection rate of CC398 had been increasing, reaching 20.0% in 2006 in Europe (Laumay et al., 2021). In China, the positive rate for CC398 strains in livestock-associated food products varied between 4.6 to 33.3% depending on the region (Wu et al., 2018; Li W. et al., 2021). Similarly, CC398 was detected in meat products with a prevalence of 11 to 31%, indicating that CC398 had entered the food chain and performed as the primary epidemic strain circulating over the world (Verkade and Kluytmans, 2014).

S. aureus CC5 was a common community-associated MRSA lineage circulating in poultry (Aires-de-Sousa, 2017). Studies identified that the majority of S. aureus isolated from poultry belonged to the avian-associated spectrum of CC5, which emerged from a human-to-poultry host jump and was characterized by numerous signatures of adaptation to the avian host including carriage of the ΦAVβ prophage genes (Matuszewska et al., 2020). In the present study, a local blast was performed to detect the ΦAVβ genes (SAAV_2008, SAAV_2009) in S. aureus strains (Lowder et al., 2009; Tang et al., 2017). All CC5 isolates carried ΦAVβ genes, which was consistent with the identification of these isolates in poultry products from current food poisoning incidents (Figure 2, Table 1).

According to previous studies, CC6 was identified as a prevalent pathogen causing food poisoning in adults. Previous study indicated that a total of 868 S. aureus isolates were collected from meat products in China, of which 47 strains were CC6, accounting for about 5.4% (Wu et al., 2018). There was evidence that S. aureus CC6 variations were isolated from diarrhea, and virulence varied between enterotoxin-encoding mobile genetic elements in Eastern Asia (Suzuki et al., 2014). In this study, CC6 isolates carried antimicrobial resistance genes such as aadD1, blaZ, mecA, ermB, and tetM, indicating a multidrug-resistant (MDR) spectrum in China (Figure 4, Table 1). As a result, it is imperative to develop tools for accurate screening for MDR S. aureus in foods.

The final predominant strain that caused diarrhea in food poisoning incidents was S. aureus CC15. Previous studies reported that a total of 8 CC types and 12 ST types of S. aureus were isolated from food poisoning incidents, with CC15 accounting for 5% of the total and carrying numerous enterotoxin genes such as sec, sed, and see (Lv et al., 2021). However, CC15 isolated in this study was not found to carry any enterotoxin genes, nor as lukF-PV and lukS-PV, suggesting a low pathogenicity but a high prevalence of hypotonic S. aureus contamination in food poisoning incidents in Eastern China (Figure 4, Table 1).

One of the significant causes of diarrhea and other gastrointestinal problems was S. aureus enterotoxin genes (SEs). Genes for enterotoxins were found in 55 isolates in this investigation, with a prevalence of 27.8% (27/97) for MSSA and 3.5% (7/20) for MRSA (Table 2). CC1, CC5, CC25, and CC59 commonly had part of the sea, seb, sec, seh, selk, sell, and selq genes, however, neither clade A nor clade D contained any enterotoxin genes, except for five sea genes present in CC6 (Figure 4, Table 2).

The carrier rate of foodborne S. aureus enterotoxins varied greatly. A study examining the rate of classical staphylococcus enterotoxins revealed that 39.3% of the foodborne isolates were enterotoxin-positive, with sec and sea as the dominant genes (Zhang et al., 2022). Moreover, a correlation was observed between the prevalence of specific enterotoxin genes, e.g., sea and seh, with the severity of the associated illness (Argudín et al., 2010). This finding underscores the importance of monitoring the prevalence of enterotoxins in foodborne S. aureus and its relevance to outbreak tracing.

The pathogenicity of these S. aureus isolates differed from that of other virulent factors. A total of 14 virulent factors, including the functional components Panton-Valentine leucocidin, toxic shock syndrome toxin, staphylokinase, exfoliative toxin, and enterotoxin, were examined among 117 foodborne S. aureus strains. Of the 117 isolates, only two exfoliative toxin-positive genes (eta) and four tsst-1-positive genes were detected, showing that S. aureus isolates from the foodborne epidemic was hypotonic and enterotoxins were the predominant pathogenicity in the context of food safety (Figure 4, Table 2).

The pvl gene was detected only in four isolates (4/117, 3.4%) in this study, which was surpassed by prior research that found Panton-Valentine leucocidin to be detectable in MRSA at a rate of 24.1% (Wu et al., 2019). Such a high PVL carriage rate among MRSA isolates was observed for food in China, suggesting considerable diversity in the frequency of PVL across foodborne S. aureus strains.

Furthermore, a total of 69 isolates were confirmed to contain scn genes, including 18 isolates from the MRSA group (Figure 4, Table 2). As a recognized marker of the immune evasion cluster (IEC), the scn gene was found in high frequency in human hosts, indicating that the gene may be utilized to differentiate strains that are transferred to humans from environments and animals (de Jong et al., 2018). The scn gene appeared to be more prevalent in the MSSA group. An earlier study conducted in China found an 81.3% (52/64) prevalence of scn in MSSA isolates from the intestinal tracts of adult patients (Li et al., 2022). In our study, the scn gene was detected in more than 59.0% of the isolates, demonstrating that the causes of infections in food poisoning incidents were widespread, and included both human-to-human transmission and environmental and livestock-associated contaminations.

Genes associated with the biofilm formation were predicted in the S. aureus strains. Interestingly, the icaC, icaD, icaR, and capsule genes (capA to capP) were found in majority of the 117 isolates, whereas icaB was found in 107 strains (Supplementary Table S4, Supplementary Figure S1). Furthermore, arg and luxS genes were found in all isolates, which were linked to pathogenicity, such as Quorum Sensing, demonstrating the potential ability to form biofilm in connection to pathogenicity in S. aureus (Reading and Sperandio, 2006) (see Table 3).

Finally, prophage ΦSa3 and ΦAVβ were identified in this study (Supplementary Table S4, Supplementary Figure S1). The majority of human-associated S. aureus harbored β-hemolysin negative-converting bacteriophages, which are classified as ΦSa3 (Goerke et al., 2009). In this study, prophage ΦSa3 was found in 69 isolates, which was consistent with the findings of the scn genes and indicated that S. aureus has jumped its host from animals to humans. Additionally, ΦAVβ prophage genes were the indicators of the human-to-poultry host jump (Lowder et al., 2009). A total of 14 isolates were found to harbor the ΦAVβ genes (SAAV_2008, SAAV_2009) in CC5 strains, which showed that chicken items may have been a source of contamination for the current food poisoning incidents. However, further research is required to elucidate the bacteriophages of pathogenic S. aureus in food poisoning incidents.

The antimicrobial susceptibility pattern observed in S. aureus isolates from various outbreaks revealed that only ten isolates displayed resistance to at least one antimicrobial, including penicillin, tetracycline, and macrolides (Figure 4, Table 2, Supplementary Table S5). Antimicrobial resistance indicators were substantially more prevalent in CC398 isolates than in other CC types. Macrolide, Lincosamide, and Streptogramin B genes were found in 36.8% of MSSA isolates and 25.0% of MRSA isolates, respectively. Similar to previous research, we found that 93.0% of the S. aureus isolates in our investigation were penicillin-resistant (Table 2) (Li et al., 2022).

The frequency of antimicrobial resistance genes from S. aureus in diarrhea patients had been extensively investigated. For instance, a total of 187 S. aureus clinical isolates were collected in South China from 2010 to 2016 (Liang et al., 2019). Among them, 103 isolates were identified as MRSA with resistance to erythromycin (64.1%), clindamycin (48.5%), gentamicin (36.9%) and ciprofloxacin (34.0%). These findings demonstrated that the therapeutic management of hypervirulent MRSA infection may be complicated by MDR isolates colonizing livestock.

Intriguingly, tetracycline resistance is a hallmark of the CC398 clade, and tetracycline accumulation in CC398 has been reported to be associated with rapid radiation from humans to livestock (Price et al., 2012). However, all the CC398 strains were tetracycline susceptible in this study. This may be explained by the fact that the current strains were all isolated from human infection instead of livestock origination. In a prior study, we conducted a sampling of pork in Beijing and found that the livestock-associated CC398 harbored a high number of tetracycline resistance genes including tetK, and tetM (Li H. et al., 2021). Altogether, our findings support the differential carriage of tetracycline resistance in S. aureus between humans and livestock reported by Price et al.

Furthermore, CC59 was found to harbor a bunch of antimicrobial resistance and virulence genes including eta, etb, scn, sea,seb,selk, selq, ant(6)-la, blaZ, erm(B), tet(K) and others in this study. CC59 was a predominant clonal lineage of community-acquired S. aureus circulating in Asia (Pang et al., 2020). Previous research described two distinct clones of the ST59 sequence type, PVL-negative/SAK-positive and PVL-positive/SAK-negative (Hung et al., 2016). In this study, a similar mode of distinct clones was observed that seven CC59 strains were PVL-negative/SAK-positive and the rest one was PVL-positive/SAK-negative. Moreover, CC59 isolated from food chain were reported to be resistant to ampicillin, penicillin, erythromycin, tetracycline and others (Pang et al., 2020), which was consistent with the results in this study.

In addition, vancomycin-resistant S. aureus strains were reportedly isolated from food products in Egypt, with a detection rate of 64.7% for vanA and 29.4% for vanB (Saber et al., 2022). However, none of the 117 foodborne S. aureus isolates in this study tested positive for vanA (Supplementary Table S3). The S. aureus virulence factors linked to staphylococcal food poisoning are of importance in terms of food safety. In this study, the enterotoxin A, B, and C genes (sea/seb/s) were shown to be the most frequent enterotoxin genes in the current foodborne S. aureus isolates. These enterotoxins have been linked to staphylococcal food poisoning as well as toxic shock syndrome in humans (Benkerroum, 2018). Although our results suggested that these diarrhea episodes were hypotonic and merely transient low-MDR infections, however, further research for continued surveillance given the detection of virulence and antimicrobial resistance determinants is required to elucidate the genomic characteristics of pathogenic S. aureus in food poisoning incidents in the context of public health.

This study has some limitations. This investigation primarily focused on the events produced by S. aureus due to the range of pathogenic bacteria leading to food poisoning incidents. Nevertheless, diarrhea caused by Salmonella and Shigella was more frequent in most cases (Kotloff, 2022). Additionally, the sample size and strain counts were limited and only covered the period from 2011 to 2021 in terms of food poisoning incidents in Eastern China. To enhance the utilization of genetics in food safety research and manufacturing, it is crucial to gather additional data on foodborne S. aureus in future studies.