A Study on Prevalence and Characterization of Bacillus cereus in Ready-to-Eat Foods in China

Bacillus cereus is widely distributed in different food products and can cause a variety of symptoms associated with food poisoning. Since ready-to-eat (RTE) foods are not commonly sterilized by heat treatment before consumption, B. cereus contamination may cause severe food safety problems. In this study, we investigated the prevalence of B. cereus in RTE food samples from different regions of China and evaluated the levels of bacterial contamination, antibiotic resistance, virulence gene distribution, and genetic polymorphisms of these isolates. Of the tested retail RTE foods, 35% were positive for B. cereus, with 39 and 83% of the isolated strains harboring the enterotoxin-encoding hblACD and nheABC gene clusters, respectively. The entFM gene was detected in all B. cereus strains. The cytK gene was present in 68% of isolates, but only 7% harbored the emetic toxin-encoding gene cesB. Antimicrobial susceptibility testing revealed that the majority of the isolates were resistant not only to most β-lactam antibiotics, but also to rifamycin. Multilocus sequence typing (MLST) revealed that the 368 isolates belonged to 192 different sequence types (STs) including 93 new STs, the most prevalent of which was ST26. Collectively, our study indicates the prevalence, bacterial contamination levels, and biological characteristics of B. cereus isolated from RTE foods in China and demonstrates the potential hazards of B. cereus in RTE foods.


INTRODUCTION
Ready-to-eat (RTE) foods, such as cooked meats and poultry, cold vegetable dishes in sauce, cold noodles, and fried rice, are very popular as they are intended for direct consumption. Although they are very convenient for consumers, RTE foods have been shown to be frequently contaminated with pathogenic bacteria such as Bacillus cereus, Listeria monocytogenes, and Staphylococcus aureus (Batchoun et al., 2011;Hwang and Park, 2015;Wu et al., 2016;Yang et al., 2016).
Antibiotic treatment is still the main method for treating bacterial infections, including those caused by B. cereus; however, the extensive use of antimicrobials has led to the emergence of antibiotic-resistant strains, including those resistant to multiple antibiotics, which can cause routine treatments to fail (Friedman et al., 2016). Thus, determining the antibiotic resistance profile of B. cereus is important for informing drug selection for treatment regimens.
The contamination of RTE foods by pathogenic bacteria such as B. cereus is a major food safety concern; thus, it is necessary to monitor and characterize B. cereus contamination in RTE foods. This study investigated the potential pathogenicity, contamination levels, molecular characteristics, and antibiotic resistance profiles of B. cereus isolated from RTE foods in China, providing important information about the prevalence of B. cereus in RTE foods.

Sample Collection
A total of 860 RTE food samples were collected from retail markets and supermarkets in 39 major Chinese cities (Supplementary Figure S1) between 2011 and 2016 according to the general guidelines of the National Food Safety Standard in Sample Collection (The Hygiene Ministry of China, 2010). The samples included cooked meat (656 samples), cold vegetable dishes in sauce (85 samples), and rice/noodles (119 samples). All samples were placed in separate sterile bags, transferred to the laboratory on ice within 2 days, and kept below 4 • C.
Qualitative and Quantitative Detection of B. cereus Bacillus cereus was qualitatively and quantitatively detected according to the bacteriological analytical manuals of the U.S. Food and Drug Administration and the National Food Safety Standard of China (The Hygiene Ministry of China, 2003;Tallent et al., 2012). In brief, 25 g samples were randomly collected from each RTE food sample and put into sterile blender jar with 225 mL Trypticase Soy Broth (TSB) with polymyxin (Huankai, Guangzhou, China), then blended for 2 min at high speed (10,000 to 12,000 rpm). Homogenates were incubated 48 ± 2 h at 30 ± 2 • C. Afterward, a loop of the resulting cultures was streaked onto mannitol egg yolk polymyxin agar plates (MYP) (Huankai), which were incubated 24 h at 30 • C. Single colonies were then streaked onto chromogenic B. cereus agar plates (Huankai). Different presumptive colonies from the chromogenic B. cereus agar plates were picked for further biochemical characterization using a B. cereus biochemistry assessor (Huankai) to identify authentic colonies. The most probable number (MPN) method was used for the quantitative detection of B. cereus. A three-tube MPN series was inoculated into TSB with polymyxin, using 1 mL inoculums of 10 −1 , 10 −2 , and 10 −3 dilutions of each sample, with three tubes at each dilution. The tubes were incubated for 48 ± 2 h at 30 ± 2 • C and observed for turbid growth typical of B. cereus. The cultures from turbid, positive tubes were streaked onto MYP agar plates and incubated for 24 h at 30 • C. One or more pink, lecithin-positive colonies were selected from each MYP agar plate and further confirmed on chromogenic B. cereus agar plates. The number of tubes confirmed as positive for B. cereus was used to calculate the MPN of B. cereus per g (mL) sample, expressed as MPN/g (mL) using the MPN table.

ERIC-PCR
The ERIC-PCR were conducted by using the primers listed in Supplementary Table S1 with previously described thermal profiles (Forghani et al., 2014;Gao et al., 2018). The PCR mixture (25 µL) contained of 50 ng of genomic DNA, 2 µM of each primer, and 12.5 µL of PCR Premix TaqTM (Takara, China). The amplification and agarose gel electrophoresis analysis were performed as described previously (Gao et al., 2018). DNA fingerprints were analyzed by Bionumerics software version 7.6 (Applied Maths, Belgium). The result of ERIC-PCR fingerprinting was used to characterize isolates from the same sample in order to exclude clonal isolates.

Antimicrobial Susceptibility Testing
The sensitivity of B. cereus to 20 antimicrobial agents was tested using the standard Kirby-Bauer disk diffusion method (Park et al., 2009; The Clinical and Laboratory Standards Institute [CLSI], 2010; Kim et al., 2015;Gao et al., 2018;Osman et al., 2018;Yu et al., 2019). The B. cereus isolates were streaked on the nutrition agar plate and grown 16-18 h at 37 • C. The colony was then picked and suspended using 0.85% physiological saline to 0.5 McFarland standard and spread on the surface of a Mueller-Hinton agar plate. After the inoculum was dried, the antimicrobial disks were put to the surface of the plates. The Mueller-Hinton agar plates were incubated 16-18 h at 35 ± 2 • C, and the inhibition zone was measured. The isolates were classed as susceptible (S), intermediate (I), or resistant (R) according to CLSI guidelines and the inhibition zones were measured and interpreted according to the zone diameter interpretation criteria for S. aureus in Supplementary Table S2.

Housekeeping Gene Amplification, Sequencing, and Sequence Type Determination
The genetic diversity of the B. cereus isolates was characterized using multilocus sequence typing (MLST) with primers specific for seven housekeeping genes (glp, gmk, ilv, pta, pur, pyc, and tpi; Supplementary Table S1). The seven housekeeping genes were sequenced by BGI (Shenzhen, China) and submitted to the PubMLST database for allele number identification. Each isolate was assigned a sequence type (ST) according to their combination of the seven housekeeping gene alleles. New STs were validated by the MLST database curator. Clonal complexes (CCs) were defined as single locus variants (SLV) of two or more independent isolates that shared identical alleles at six or seven loci (Drewnowska and Swiecicka, 2013) and were identified using the geoBURST tool (Francisco et al., 2009) with bootstrap resampling (1,000). Evolutionary relationships between the isolates were evaluated using a minimum spanning tree constructed by PHYLOViZ 2.0 software (Instituto de Microbiologia, Portugal; Ribeiro-Goncalves et al., 2016).

Prevalence of B. cereus in RTE Foods
In this study, 302 of the 860 collected samples (35%) were positive for B. cereus, including 224 of the 656 cooked meat samples (34%), 59 of the 119 rice/noodle samples (50%), and 19 of the 85 cold vegetable dishes in sauce samples (22%). 368 different strains of B. cereus were identified based on the results of biochemical analysis and ERIC-PCR fingerprinting (Supplementary Figure S2). Based on quantitative analysis, 68% (206/302) of the positive samples were contaminated at levels ranging between 3 and 1100 MPN/g; however, 10% (29/302) of the samples exceeded 1100 MPN/g, including 18 cooked meat samples, 10 rice/noodles samples and one cold vegetable dish in sauce sample ( Table 1).

Distribution of Virulence Genes Among B. cereus Isolates
The hblACD gene cluster was found in 39% of the B. cereus isolates (Figure 1 and Supplementary Table S3), with hblA, hblC, and hblD present in 46, 49, and 50% of the isolates, respectively. The NHE genes nheA, nheB, and nheC were found in 89, 99, and 94% of the isolates, respectively. Moreover, 68% of the strains possessed the cytK gene, whereas the cesB gene was the least frequently observed toxin gene, present in just 7% of the strains. However, all of B. cereus isolates possessed the entFM gene, a higher rate than that of other enterotoxin genes.

MLST and Clonal Complex Analysis
The genetic diversity of the 368 B. cereus isolates was analyzed by MLST with the internal fragment sequences of seven housekeeping genes. In total, 192 different STs were identified, 93 of which were novel and designated ST2150-ST2361. ST26 was the most abundant ST (28 isolates), followed by ST205 (14 isolates

Prevalence and Genetic Diversity of B. cereus Isolates
Bacillus cereus is a foodborne pathogen that causes various symptoms and is found on multiple types of food. In this study, 302 of the 860 RTE samples (35%) were positive for B. cereus, indicating that retailed RTE foods are a potential risk to consumers. The rate of contamination observed in this study was higher than has been reported in some countries, for example Korea (Cho et al., 2011) and Morocco (Merzougui et al., 2014), but lower than in others such as India (Sudershan et al., 2012). Around 90% of the samples had <1100 MPN/g of B. cereus, the generally accepted threshold for B. cereus contamination; however, the remaining samples all exceeded 1100 MPN/g, indicating that some RTE foods may cause food poisoning due to high levels of contamination. Cooked meats are commonly prepared and sold for direct consumption in different nations. In this study, B. cereus was present in 34% (220/656) of the cooked meat samples, consistent with the frequency reported by Tewari et al. (2015). Research has shown that open-air stalls increases the opportunity for environmental pollution (Ng et al., 2013); for instance, cooked meat sold in open-air stalls can be exposed to dust-containing spores, increasing the chance of B. cereus contamination. This could explain the high prevalence of B. cereus in our cooked meat samples.
Vomiting was first associated with B. cereus in the United Kingdom in 1971 which was caused by fried rice (Mortimer and McCann, 1974). Vomiting due to B. cereus infection is typically associated with starch-containing foods (Fricker et al., 2007;Delbrassinne et al., 2015) which are believed to promote the growth of B. cereus and its production of emetic toxins (Griffiths and Schraft, 2017). In this study, 50% (59/119) of the rice/noodle samples tested positive for B. cereus, which is higher than that reported in other studies of starchcontaining foods (Chang et al., 2011;Delbrassinne et al., 2012;Merzougui et al., 2014).
Cold vegetable dishes in sauce, also known as Chinese salad, are a favorite RTE food in China. B. cereus was detected in 22% (19/85) of the cold vegetable dishes in sauce samples, a higher prevalence than observed in other reports (Valero et al., 2007), but lower than that observed in vegetable salads in Korea (Chon et al., 2015). Although cold vegetable dishes in sauce RTEs undergo processing steps, they are usually not subjected to heat processing; thus, microorganisms cannot be completely eliminated on the fresh vegetables. Some studies have indicated that the growth, harvest, processing, and packaging of vegetables are likely to increase the possibility of microbial contamination (Sagoo et al., 2003). Since B. cereus is widespread in nature, particularly in soil, vegetables are easily contaminated by this bacterium; previously, we showed that B. cereus was present in up to 50% of vegetables (Yu et al., 2019). Taken together, these results suggest that the consumption of cold vegetable dishes in sauce made from vegetables contaminated with B. cereus could increase the potential risk to the public.  Some studies have indicated that RTE foods can become contaminated by different bacterial pathogens via food preparation surfaces (Christison et al., 2008;Cho et al., 2011;de Oliveira et al., 2011;Shiningeni et al., 2019). Many pathogenic bacteria such as B. cereus are capable of forming viscous, highly heat-and drought-resistant spores. These properties increase spore retention and make their removal from preparation surfaces on production lines difficult. Following its preparation, the load of B. cereus in an RTE can increase rapidly and reach dangerous levels (above 10 3 CFU/g or mL); thus, consumers face a higher risk of foodborne illness from RTE foods, which are usually eaten without additional heat treatment. Consequently, regulations have been imposed on the number of B. cereus cells in RTE foods in different countries and regions based on standard guidelines. Hong Kong classifies RTE foods as either satisfactory (<10 3 CFU/g), acceptable (10 3 -10 5 CFU/g), or unsatisfactory (>10 5 CFU/g), and the sale of "unsatisfactory" RTE foods must be halted immediately (Food and Environmental Hygiene Department, 2014). The Food Standards of Australia and New Zealand classes RTEs into four levels: satisfactory, acceptable, unsatisfactory, and potentially harmful (>10 4 CFU/g of B. cereus) (New South Wales Food Authority, 2009), whereas in England "unsatisfactory" RTE foods are those with >10 5 CFU/g of B. cereus (Health Protection Agency, 2009). In this study, 29 positive samples (29/302, 10%) had more than 1100 MPN/g of B. cereus, which may exceed some of these guideline limits.

Virulence Gene Profiles of B. cereus Isolates
The most common enterotoxin-encoding genes in the B. cereus strains we identified were the nhe and entFM, consistent with previous reports (Kim et al., 2010;Chon et al., 2015;Hwang and Park, 2015;Glasset et al., 2016;Gao et al., 2018;Yu et al., 2019). CytK is a cytotoxin isolated from a B. cereus strain that caused a severe food poisoning outbreak leading to three deaths in France (Lund et al., 2000). The cytK gene was present in 68% of the B. cereus isolates identified in this study, consistent with earlier studies that found the gene in 40-73% of B. cereus strains isolated from foods (Hwang and Park, 2015;Lee et al., 2017;Gao et al., 2018). Hence, the widespread distribution of diarrheacausing B. cereus in RTE foods and its potential hazards cannot be ignored.
Approximately 7% (25/368) of the strains isolated in this study were emetic strains, similar to the findings of a study performed in Jordan (Batchoun et al., 2011), lower than the levels reported in Holland and Ghana (Biesta-Peters et al., 2016;Owusu-Kwarteng et al., 2017), but higher than found by studies in Korea (Chon et al., 2015;Hwang and Park, 2015). According to previous studies, starch-containing foods promote toxin production by B. cereus (Griffiths and Schraft, 2017). Interestingly, in this study the 18 B. cereus strains isolated from cooked meat contained the emetic toxin gene cesB. Since vomiting intoxication cases have been known to cause death (Mahler et al., 1997;Dierick et al., 2005;Lopez et al., 2015), even this comparatively small number of cesB positive strains cannot be ignored.

Antimicrobial Susceptibility of B. cereus
Beyond food poisoning, B. cereus is also associated with nongastrointestinal infections. Antibiotic susceptibility testing can provide a reference for the clinical treatment of food poisoning. In this study, we exhaustively tested the antimicrobial resistance of all B. cereus isolates. More than 82.34% of the isolates were resistant to most of the β-lactam antibiotics, whilst only 13.59% (50/368) and 0.27% (1/368) were resistant to CTT and imipenem (IPM), respectively. This result is unsurprising, since B. cereus can produce a β-lactamase (Chen et al., 2003;Bottone, 2010). According to the results of our antimicrobial sensitivity testing, suspected B. cereus infections should not be treated clinically with broad-spectrum cephalosporins and penicillin, but with the CTT or IPM. We also found that many B. cereus isolates displayed multiple-drug resistance profiles, suggesting that B. cereus infections in RTE foods pose a significant potential risk.

Genetic Diversity of B. cereus in RTE Foods
Epidemiological typing methods, such as MLST, are considered crucial for studying the prevalence of foodborne bacteria. MLST is a nucleotide sequence-based approach for the unambiguous characterization of isolates that has been broadly used for epidemiological typing and risk analyses of many pathogenic bacteria, such as S. aureus , L. monocytogenes , Cronobacter spp. (Xu et al., 2015), Salmonella (Matheson et al., 2010), and B. cereus (Cardazzo et al., 2008;Yang et al., 2017;Zhang et al., 2017;Yu et al., 2019). In this study, we used MLST to analyze the genetic polymorphisms of B. cereus strains isolated from RTE foods (Supplementary Table S4). Apart from six CCs, most isolates were assigned as singletons; ST-205 and ST-142 were the predominant CCs, consistent with previous reports . Moreover, 28 strains were assigned to ST26, including 18 strains that were found to contain the cesB gene. ST26 is a recognized ST associated with food poisoning that causes vomiting, and includes two clinically isolated strains (NC7401 and F4810/72) (Agata et al., 2002;Priest et al., 2004;Fricker et al., 2007), suggesting that potentially harmful strains may be present in RTE foods in China; however, further studies investigating cereulide production by these strains must be carried out to understand their threat.

CONCLUSION
In this study, we evaluated the prevalence of B. cereus in RTE foods from different regions in China. Our results indicated that RTE foods are highly contaminated with B. cereus and that the isolated strains contained a variety of pathogenic genes which may increase the potential risk of foodborne diseases. In addition, most of the strains exhibited MDR profiles with important implications for clinical treatment. Together with the diverse genotypic polymorphisms we observed, our findings reveal the potential high risk of B. cereus in RTE foods.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
YD, QW, JW, JZ, and SY conceived the project and designed the experiments. SY, PY, JW, CL, HG, CCL, LK, LY, SW, TL, MC, HZ, RP, YZ, and XW performed the experiments. QW and YD supervised the project. SY and YD analyzed the data and wrote the manuscript. QW, JW, and YD complemented the writing.