Prevalence and Characterization of Food-Related Methicillin-Resistant Staphylococcus aureus (MRSA) in China

Methicillin-resistant Staphylococcus aureus (MRSA) is an emerging pathogen that is difficult to treat due to the multiresistance of the bacteria upon infection. From 2011 to 2016, 1581 S. aureus strains were isolated from 4300 samples from retail foods covering most provincial capitals in China. To determine the prevalence of food-related MRSA and its genetic background in China, antibiotic resistance, staphylococcal toxin genes, staphylococcal cassette chromosome mec (SCCmec) typing, spa-typing and MLST were carried out in this study. In total, 108 (7.4%) isolates were confirmed for MRSA by phenotyping (cefoxitin) and genotyping (mecA/mecC gene). A total of 52.8% (57/108) of the MRSA isolates belonged to clonal complex 59 (CC59) (ST59, ST338, and ST3355), which was the predominant clone in this study. These CC59 isolates carried SCCmec elements of type IV, V, or III and exhibited spa type t437, t441, t543, t163, t1785, or t3485, and half of them carried major virulence genes, such as the Panton-Valentine leucocidin (PVL) gene. The secondary clones belonged to ST9 (15.7%, 17/108) with a type of t899-SCCmec III and showed a broader range of antimicrobial resistance. The remaining MRSA isolates (31.5%, 34/108) were distributed in 12 different STs and 18 different spa types. All isolates harbored at least one of the enterotoxin genes, whereas only 4 isolates (3.70%) were positive for the toxic shock syndrome toxin tsst alleles. For antibiotic susceptibility testing, all isolates were resistant to more than three antibiotics, and 79.6% of the isolates were resistant to more than 10 antibiotics. Amoxycillin/clavulanic acid, ampicillin, cefoxitin, penicillin, ceftazidime, kanamycin, streptomycin, clindamycin, and telithromycin was the most common antibiotic resistance profile (55.6%, 60/108) in the study. In summary, the results of this study implied that the major food-related MRSA isolate in China was closer to community-associated MRSA, and some of the remaining isolates (ST9-t899-SCCmec III) were supposed to livestock-associated MRSA. In addition, most MRSA isolates showed resistance to multiple drugs and harbored staphylococcal toxin genes. Thus, the pathogenic potential of these isolates cannot be ignored. In addition, further studies are needed to elucidate the transmission routes of MRSA in relation to retail foods and to determine how to prevent the spread of MRSA.


INTRODUCTION
Methicillin-resistant Staphylococcus aureus (MRSA) is a pathogen of increasing importance in hospitals as well as in the community and livestock. It can be resistant to several antibiotics and quickly disseminates worldwide. In recent years, MRSA has been attributed to an estimated 5400 extra deaths and over one million extra days of hospitalization (Kobayashi et al., 2015). S. aureus became MRSA because of the acquisition of the gene mecA or mecC, which encodes the low-affinity penicillin-binding protein 2a (PBP2a), which, unlike other PBPs, remains active and allows for cell wall biosynthesis at otherwise lethal β-lactam concentrations. The mec operon is carried by the staphylococcal cassette chromosome mec (SCCmec) and most likely originated from horizontal transfer from coagulase-negative staphylococcal species (Banerjee et al., 2008;Garcíaálvarez et al., 2011).
MRSA strains have been reported from various sources. To distinguish epidemiological groups of MRSA, it was divided into hospital-associated MRSA (HA-MRSA), community-associated MRSA (CA-MRSA) and livestock-associated MRSA (LA-MRSA) (Petinaki and Spiliopoulou, 2012). MRSA was first recognized as HA-MRSA in 1961 and then spread into the community and, later on, into healthcare facilities; in 1990, this type became recognized as CA-MRSA. LA-MRSA has always been associated with animals and is linked to a jump from animals to humans (Petinaki and Spiliopoulou, 2012). In general, the genetic backgrounds differ among different types of MRSA. In previous research, CA-MRSA strains frequently harbor SCCmec type IV or V and often produce potent toxins/virulence factors, such as Panton-Valentine leucocidin (PVL), arginine catabolic mobile element (ACME) and phenol-soluble modulins, while HA-MRSA strains typically possess larger-size SCCmec type I-III and are more resistant to other classes of antibiotics (An and Otto, 2008;Chuang and Huang, 2013). For multilocus sequence typing (MLST), sequence type 5 (ST5), ST8, ST22, ST36, and ST45 spread successfully to different regions of the world and caused substantial nosocomial disease (Deleo et al., 2010), whereas CA-MRSA showed five lineages worldwide: ST1-IV (USA400), ST8-IV (USA300), ST30-IV (Pacific/Oceania), ST59-IV and V (USA1000, Taiwan) and ST80-IV (European) (Skov, 2009). In contrast, LA-MRSA strains exhibit co-resistance to many nonβ-lactam antimicrobials (e.g., antibiotics and metals), including those commonly used in animal production, and many of them belong to CC398 or CC9, as determined by MLST (Bens et al., 2006;Neela et al., 2009). The SCCmec elements of LA-MRSA are different from those carried by other MRSA genotypes commonly found in the community and healthcare settings (Li et al., 2011). In addition, the majority of LA-MRSA isolates lack toxins, such as PVL and other enterotoxins (Hallin et al., 2011).
Nowadays, MRSA strains have been reported from various foods sources, such as poultry, pork, beef, milk and vegetables, suggesting that foods may serve as reservoirs (Wang et al., 2014;Wu et al., 2018). As is commonly known, foods have many different origins, and different types of MRSA are present in foods of different origin in different countries. In China, large-scale studies of the prevalence of S. aureus in food are scarce. From July 2011 to June 2016, we collected 4300 retail food samples from supermarkets, fairs and farmers' markets that covered most of the provincial capitals of China (Supplementary Figure 1) and found 1063 (24.7%) S. aureus-positive samples from all sampling sites. To better understand the genetic background among foodrelated MRSA isolates in China, this study aimed to identify the MRSA isolates from our previous study and to characterize these MRSA isolates for their antimicrobial resistance profiles, their virulence genes and their genotypic types (SCCmec-, MLST, and spa types).

Bacterial Isolates
A total of 1581 S. aureus isolates were collected from 4300 retail food samples in 39 Chinese cities (Supplementary Figure 1), comprising 469 isolates from meat and meat products (bacon/sausage, poultry, pork, mutton and beef), 511 isolates from aquatic products (freshwater fish, shrimp and seafood), 368 isolates from quick-frozen products (frozen dumplings/steamed stuffed buns and frozen meat), 148 isolates from ready-to-eat food (cold vegetable/noodle dishes in sauce, fried rice/sushi, roast meat, sausage, and ham), 42 isolates from edible mushrooms, 30 isolates from vegetables and 13 isolates from pasteurized milk. These isolates were obtained between July 2011 and June 2016 according to the GB 4789.10-2010 food microbiological examination of S. aureus (National Food Safety Standards of China) and the most probable number (MPN) method (Gombas et al., 2003). They were identified by Gram stain, catalase and oxidase tests and API STAPH test strips (BioMerieux, Marcy-1'Etoile, France). Each isolate was incubated at 37 • C overnight in brain heart infusion (BHI) broth. Genomic DNA was extracted using a genomic DNA extraction kit (Magen Biotech, Guangzhou, China) according to the manufacturer's instructions. The concentration of genomic DNA was determined at 260 nm using a NanoDrop-ND-1000 UV-Vis spectrophotometer (Thermo Fisher Scientific, MA, United States).

MRSA Confirmation
Cefoxitin disks (30 µg) were used for detecting methicillinresistant isolates. S. aureus ATCC 25923 was used as a control. The mecA/mecC gene, which has been shown to confer methicillin resistance to S. aureus (MRSA), was also detected by PCR using primers as described previously (Perez-Roth et al., 2001;Stegger et al., 2012).

Detection of Staphylococcal Toxin Genes
All MRSA isolates were tested by PCR for the presence of 20 genes coding for staphylococcal enterotoxins (sea, seb, sec, sed, see, seg, seh, sei, sej, sek, sel, sem, sen, seo, sep, seq, ser, and seu), and the tsst-1 gene encoding the TSST (Varshney et al., 2009) and the lukSF-PV (pvl) genes were determined by PCR according to a previously published method (Jarraud et al., 2002). The primers and PCR conditions are presented in Supplementary  Table 1. The amplicons were stained with Goldview, loaded, and electrophoresed in 1.5% agarose at 120 V for 0.5 h and visualized under a UV transilluminator gel imaging system (GE Healthcare, WI, United States). The images were saved as TIFF files for analysis.

Molecular Typing of MRSA Isolates
All MRSA isolates were subjected to SCCmec typing, spa-typing and MLST. The SCCmec typing method was performed on the isolates by multiplex PCR as previously described (Zhang et al., 2005). The S. aureus protein A (spa) repeat region was amplified according to a published protocol (Shopsin et al., 1999). The MLST scheme used to characterize MRSA isolates is based on the sequence analysis of the following seven housekeeping genes: arcC, aroE, glpF, gmk, pta, tpi, and yqil (Enright et al., 2013). The DNA fragments were purified using a PCR purification kit (Qiagen, Genman) and sequenced in each direction with Big Dye fluorescent terminators on an ABI 3730XL sequencer (Applied Biosystems). The spa types were randomly assigned using the SpaServer website 1 . For each MLST locus, an allele number was given to each distinct sequence variant, and a distinct ST number was attributed to each distinct combination of alleles at the seven genes. STs were determined using the S. aureus MLST database 2 . Clonal complex (CC) analysis was performed in eBURST v.3 as previously described (Feil et al., 2004). The minimum spanning tree (MST) was constructed with Bionumerics 7.6 software (Applied Maths, Sint-Martens-Latem, Belgium).

Antibiotic Resistance Profiles of MRSA Isolates
The antibiotic susceptibility results of 108 MRSA isolates are shown in Table 3. All MRSA isolates were resistant to more than three antibiotics, including 33.3% of the isolates that were resistant to 4-10 antibiotics, 46.3% of the isolates that were resistant to 11-15 antibiotics and 20.4% of the isolates that were resistant to 16-26 antibiotics. The isolates were susceptible to linezolid, vancomycin and daptomycin, and the frequencies of resistance to individual agents were 100% for ampicillin and penicillin G, followed by 97.2% for ceftazidime, 87.0% for amoxicillin/clavulanic acid, 83.3% for erythromycin, 79.6% for clindamycin, 75.9% for kanamycin, 74.1% for telithromycin, 67.6% for streptomycin, 65.7% for cefepime and tetracycline, 38.0% for chloramphenicol, 27.8% for gentamycin, ciprofloxacin and fusidic acid, 25.9% for norfloxacin, 22.2% for amikacin, 13.95% for quinupristin/dalfopristin, 13.0% for trimethoprim/sulphamethoxazole 1:19, 7.4% for rifampicin, 2.8% for nitrofurantoin and 0.9% for teicoplanin. In different types of food products, the resistance to most antibiotics were equally distributed. Norfloxacin, clindamycin, quinupristin/dalfopristin, gentamycin, and trimethoprim/sulphamethoxazole 1:19 resistance was rare in MRSA isolated from non-animal sources (i.e., edible mushrooms and vegetables). It is worth noting that amikacin was the most commonly observed resistance antibiotic, whereas quinupristin/dalfopristin, gentamycin, ciprofloxacin and norfloxacin resistance were observed in low frequencies in aquatic product-related MRSA isolates compared to other types of food isolates (Supplementary Figure 2). However, AMC-AMP-FOX-P-CAZ-K-S-DA-TEL was the most common antibiotic resistance profile (55.6%, 60/108) in this study. All selected antibiotics were grouped into 15 classes of agents. Five MRSA isolates exhibited resistance to only β-lactam antibiotics, whereas 97 isolates exhibited a multidrug resistance phenotype, with resistance to ≥3 classes of antimicrobial agents.
A phylogenetic tree based on the 7 concatenated MLST sequences (Figure 2) shows the relatedness between the MRSA strains. Two different clusters were observed in this study (designated as A and B). Cluster A included ST1, ST3304, ST188, ST9, ST15, ST6, ST5, ST7, ST943, ST630, and ST88, and cluster B included ST398, ST10, ST338, ST3355, and ST59, which showed distant genetic relationships. STs correlated well with spa types, such as ST1-t127, ST188-t189, ST9-t899, ST7-t091, ST59-t437, and ST338-t437. Overall, the genetic diversity among MRSA isolates was higher based on different cities and different food sources. For different food products, more than three subtypes were found in each type of food, except edible mushrooms and vegetables. CC59-t437 (45.4%, 49/108) was the predominant clone in this study, but ST9-t899-SCCmec IVb was the predominant clone in quick-frozen meat and was found only in animal-derived food (raw meat, quick-frozen meat and readyto-eat meat). In addition, 80% (4/5) of SCCmec III-ST338-t437 isolates were found in aquatic products from the city of Kunming. However, some strains showed correlations among geographical locations, such as Sta223-2 (isolated from Shenzhen), Sta251 (isolated from Guangzhou), Sta403 (isolated from Shantou), Sta487 (isolated from Heyuan), and Sta1753 (isolated from Zhanjiang), which were isolated from neighboring cities in this study and clustered into one subtype (ST9-t899-SCCmec IVb).

DISCUSSION
MRSA is a significant public health concern in humans and animals. The rate of mortality due to MRSA infections has remained high in recent years. In hospitals, the prevalence rates of MRSA in some Asian countries, such as Taiwan, China, Japan, and South Korea, can reach 70-80% (Chuang and Huang, 2013). For CA-MRSA, the prevalence varies substantially worldwide and ranges from less than 1% to more than 50% in different countries (Tristan et al., 2007;Deurenberg and Stobberingh, 2008). MRSA     has also been isolated from animals, as reported in many previous studies (Leonard and Markey, 2008). It is important to identify the origin of food-related MRSA and to evaluate the potential pathogenicity of these MRSA isolates. From July 2011 to June 2016, 4300 retail food samples were collected from supermarkets, fairs and farmers' markets, covering most of the provincial capitals of China. This wide-scale and systematic investigation of S. aureus from retail food in China supplements nationwide qualitative and quantitative data on the prevalence and levels of S. aureus. In this study, we determined the MRSA isolates from these food-related S. aureus isolates and found 108 MRSA isolates (6.83%) in various types of food products (raw meat, aquatic products, quick-frozen products, ready-to-eat food, edible mushrooms and vegetables) from most of the sampling cities (29/39, 74.4%) in China, which suggested that retail food in China could be contaminated with MRSA. Many studies have evaluated the presence of MRSA in retail food. In China, among studies that sampled retail food, 6.07% of MRSA isolates were found in quick-frozen dumpling samples of Shaanxi province (Hao et al., 2015) and in 1.7% of chicken samples , whereas MRSA was present in 29.5% of grain products, meat products and dairy products in southwest China. Overall, these rates varied from our results in this study. This incidence may be attributed to a number of factors, such as the sample size, sampling site, types of samples or isolation methods. In 2015,  collected 1150 S. aureus isolates from retail markets from 203 cities in 24 provinces in China and found 91 isolates (7.9%) that were identified as MRSA by PCR. The MRSA isolates were distributed in raw meat, rice and flour products, vegetable salads, sandwiches, meat products, and eggs. Compared with other countries' studies, the prevalence of MRSA from retail foods in China was not low. For instance, Ge et al. (2017) conducted a 1-year survey in 2010-2011 from 3520 retail meats in eight U.S. states and found that 1.9% of samples were positive for MRSA. An Italian survey found that 6 of 160 (3.75%) foods of animal origin harbored MRSA (Normanno et al., 2007). MRSA was present in 3.6% (15/421) of retail meat in Korea (Lee, 2003), 1.6% (5/318) of food animals in Spain (Lozano et al., 2009), 0.75% (20/2662) in Switzerland (Huber et al., 2010), 3% (11/367) in Greece (Papadopoulos et al., 2018), and 1.9% (2/103) in the United Kingdom (Hadjirin et al., 2015). Therefore, Chinese food safety regulators should improve hygiene and supervision efforts.
Currently, MRSA isolated from food-producing animals has been recognized as LA-MRSA. The worldwide emergence of LA-MRSA since 2005 has prompted many of the surveys to focus on retail meat as a potential vehicle for this new MRSA clone (Ge et al., 2017). In general, ST398 was recognized as the most typical type in LA-MRSA. ST398 in swine, cattle and other animal species has been analyzed in several publications (Walther et al., 2009;Fessler et al., 2010;Graveland et al., 2011). It reported that 24-100% of pig farmers, 37% of poultry farmers, 30-38% of cattle farmers and up to 45% of veterinarians are colonized with MRSA CC398 in the nares (Köck et al., 2014). In this study, most MRSA isolates (88.9%, 96/108) were isolated from animal-derived food (raw meat, aquatic products, quickfrozen products and ready-to-eat food), but only 7 MRSA isolates belonged to this ST type. STs of MRSA isolated from retail food focused on CC59 (ST59, ST338, and ST3355). This showed significant genetic uniformity with the predominant Asian CA-MRSA lineage, which can reach 35.8-76.7% with CA-MRSA in China . In general, the CC59 clone always carried SCCmec IV/V and was concentrated in spa t437 and t441. In 2013, Li et al. (2013) analyzed the 110 CC59 isolates from invasive and noninvasive diseases in China by MLST, SCCmec typing and spa-typing and found that 65.5% of the clones were ST59-t437-IVa. From February 2016 to January 2017 Yang et al. (2017), collected S. aureus strains in Beijing Children's hospital from the respiratory tract, skin and soft tissue, and sterile sites in 104 child cases. Of these, 54.8% were categorized as community-associated SA (CA-SA) infections, and ST59-SCCmec IV-t437 (61.7%) was the most prevalent MRSA genotype. In addition to China, ST59 has also been reported in Vietnam, Japan, Australia and other countries with CA-MRSA infection (Tang et al., 2007;Coombs et al., 2010;Higuchi et al., 2010). In contrast, ST239 and ST5 were found in a nationwide study to be two major MRSA clones with unique geographic distributions in Chinese hospitals (Liu et al., 2009). Accordingly, these results implied that the major food-related MRSA in China was closer to CA-MRSA, a finding that should be brought to public attention.
In the present study, 15.7% (17/108) of isolates belonged to ST9, the secondary clones of which were found only in animal-derived food (raw meat, quick-frozen meat and readyto-eat meat). According to previous studies, ST9 was the predominant S. aureus and MRSA genotype in pigs and related workers in Asia (Chuang and Huang, 2015). In 2008, ST9 MRSA was first found in Chinese pigs, and farm workers carried ST9-t899-SCCmec III-PVL-negative . Studies in Taiwan, Hong Kong, Malaysia and Thailand have since reported the prevalence of this type of LA-MRSA (Neela et al., 2009;Graveland et al., 2011;Larsen et al., 2012;Lo et al., 2012). ST9 is the most prevalent LA-MRSA in most Asian countries and differs from the European pig-associated clone (ST398) with regard to clonal type, SCCmec content and resistance profile (Ye et al., 2016). In China, ST9 strains always showed SCCmec III with spa t899. These characteristics were in accordance with our results, which showed that the ST9 MRSA isolated from our study showed ST9-t899-SCCmec III. Therefore, this portion of the food-related MRSA isolates was supposed to be LA-MRSA. However, it is worth noting that in this study, all ST9 MRSA isolates were resistant to more than 15 antibiotics and showed a broader range of antimicrobial resistance than ST59 MRSA (Figure 2). More than 80% of ST9 MRSA was resistant to erythromycin, ciprofloxacin, gentamicin, tetracycline and clindamycin. Currently, more evidence has implicated animals as reservoirs of antimicrobialresistant bacteria and has indicated that animals can potentially transmit resistance genes to humans (Liu et al., 2018). Thus, more attention should be focused on this type of strain among food chains.
Except for CC59 and ST9, the remaining strains belonged to ST1 (8.4%, 9/107), ST398 (6.5%, 7/107), ST7 (4.7%, 5/107), ST630 (2.8%, 3/107), and ST188 (1.9%, 2/107). Of these strains, most STs correlated well with spa types, but seven ST398 isolates distributed in four different spa types. In addition, SCCmec types were not detected in most (Figure 2). This finding is consistent with the results of a previous study suggesting that divergent SCCmec elements were inserted into the (clonal) ST398 MSSA (van Duijkeren et al., 2008;Smith and Pearson, 2011). Therefore, further studies evaluating MSSA ST398 in retail food are needed to determine the reason for the correlation. In this study, ST1-t127 and ST7-t091 were also detected. These STs showed high genetic diversity among MRSA isolates based on different cities and different food sources, which is a common finding in isolates of human and animal origin (Franco et al., 2011;Hummerjohann et al., 2014). Thus, these types of S. aureus isolates have been relevant to a variety of clinical infections and have theoretical pathogenic potential.
As is usually observed with CA-MRSA strains, CA-MRSA ST59 isolates had significantly more pronounced virulence in various animal infection models than the geographically matched HA-MRSA clones ST5 and ST239 . For the strains of ST59, the evolutionary acquisition of PVL, the higher expression of α-toxin and, possibly, the loss of a large portion of the β-haemolysin-converting prophage probably contribute to its higher pathogenic potential (Chen et al., 2013;Chen and Huang, 2014). In this study, we also investigated the PVL gene and found 24.07% PVL-positive MRSA isolates. All PVL-positive MRSA isolates belonged to CC59, including 22 ST59 isolates and four ST338 isolates. For LA-MRSA ST9 isolates, certain important virulence factors, such as PVL, are absent. Despite a lack of virulence factors, ST9 strains have been found to cause disease in humans (van Loo et al., 2007;Liu et al., 2009;Chuang and Huang, 2015). Therefore, the hazards of these strains for consumers cannot be ignored.
As an important foodborne pathogen, S. aureus is involved in most staphylococcal food poisoning (SFP) events due to staphylococcal enterotoxins, the virulence factors that are heat stable and proteolytic or demonstrate emetic activity (Grumann et al., 2014). In this study, we also investigated most of the enterotoxin genes of S. aureus. All food-related MRSA isolates harbored at least one of the SE genes. The percentages of MRSA isolates containing sea, seb, and sec all exceeded 50%, whereas sed and see were detected in only fourteen MRSA isolates (12.96%). Except for the egc cluster, which accounted for 29.4% (245/833) of the detected genes in the present study, other SE genes showed 48.4% (403/833) of the detected genes. Generally, SEA, followed by SED, is the enterotoxin most frequently associated with SFP, although outbreaks caused by SEB, SEC, and SEE have also been reported (Argudín et al., 2012). In contrast to classical SEs, the relationship between the novel SEs and SFP is not fully understood, but most of them (SEG, SEH and SEI, SER, SES, and SET) have been shown to be emetic after oral administration in a primate model (Argudín et al., 2010). In this study, 94 of 108 MRSA isolates (87.04%) harbored egc cluster genes (seg, sei, sem, sen, seo, and seu), and 94.4% (102/108) of isolates harbored one or more genes for other novel SEs or tsst-1 genes. It is suggested that attention should not only be paid to classical enterotoxins but also to novel ones since an increasing number of foodborne outbreaks have been associated with novel enterotoxins.
Nowadays, multiple drug resistance is the most important characteristic of MRSA isolates. For the 26 clinically relevant antibiotics investigated, all MRSA isolates were resistant to more than three antibiotics. It seems higher than many previous studies in food-related MRSA isolates (Hanson et al., 2011;Shahraz et al., 2012;Ge et al., 2017;Tang et al., 2017). Interestingly, norfloxacin, clindamycin, trimethoprim/sulphamethoxazole 1:19, gentamycin and quinupristin/dalfopristin resistance were rare in MRSA isolated from non-animal sources (edible mushrooms and vegetables) (Supplementary Figure 2). The reason for this finding may be attributed to the food source. In general, most animal-derived food-related S. aureus came from animal farms that used these antibiotics as food supplements in animal feed (Wang et al., 2014), whereas S. aureus isolated from edible mushrooms or vegetables most likely originated in the environment. Furthermore, amikacin was the most commonly observed resistance antibiotic in aquatic product-related MRSA isolates, whereas gentamycin, quinupristin/dalfopristin, ciprofloxacin, and norfloxacin resistances were observed less frequently than other types of food isolates (Supplementary Figure 2). As we know, amikacin, gentamycin, kanamycin and streptomycin all belong to the aminoglycosides, which exert their bactericidal effects by irreversibly binding to the 30S ribosomal subunits of susceptible bacteria, inhibiting protein synthesis (Hammerberg et al., 1986). The most common clinical resistance mechanism to aminoglycosides is the structural modification of aminoglycosides by aminoglycosidemodifying enzymes (AMEs). Amikacin is a broad-spectrum semi-synthetic derivative of kanamycin and a poor substrate of many AMEs. Furthermore, even if amikacin is modified by AMEs, the modified amikacin can still bind to the 30S ribosomal subunit (Yuan et al., 2013). Therefore, amikacin is one of the most potent classes of antibiotics in S. aureus infection. In fact, the mechanism of MRSA resistance to amikacin is poorly understood. Thus, why amikacin resistance was higher in MRSA isolates from aquatic products should be further studied.

CONCLUSION
Essential for human survival, food is the one of most basic necessities of life. In this study, we investigated food-related MRSA and determined its genetic background in China. MRSA isolates were found in most investigated cities and were observed in different types of food samples. ST59 was the predominant clone in food-related MRSA in China, which indicated that the major food-related MRSA isolates in China were closer to CA-MRSA. Moreover, as the major LA-MRSA in Asian populations, ST9 MRSA was the secondary clone and showed a broader range of antimicrobial resistance. Determination of staphylococcal toxin genes presented their virulence potential, and antimicrobial susceptibility testing further confirmed the severe situation of MRSA isolated in retail food in China. In addition, some antibiotics were also found to be higher in some types of food. However, further studies are required to determine the reason for this correlation and to elucidate the transmission routes of MRSA in relation to retail foods in order to provide the tools for preventing the spread of MRSA.

AUTHOR CONTRIBUTIONS
QW, JZ, SW, and TL conceived and designed the experiments. JH, FZ, and JD performed the experiments. SW, RP, and HZ analyzed the data. XW, LX, MC, and XY contributed reagents, materials, and analysis tools. SW and JW contributed to the writing of the manuscript.