Bacillus cereus Isolated From Vegetables in China: Incidence, Genetic Diversity, Virulence Genes, and Antimicrobial Resistance.

Bacillus cereus is a food-borne opportunistic pathogen that can induce diarrheal and emetic symptoms. It is widely distributed in different environments and can be found in various foods, including fresh vegetables. As their popularity grows worldwide, the risk of bacterial contamination in fresh vegetables should be fully evaluated, particularly in vegetables that are consumed raw or processed minimally, which are not commonly sterilized by enough heat treatment. Thereby, it is necessary to perform potential risk evaluation of B. cereus in vegetables. In this study, 294 B. cereus strains were isolated from vegetables in different cities in China to analyze incidence, genetic polymorphism, presence of virulence genes, and antimicrobial resistance. B. cereus was detected in 50% of all the samples, and 21/211 (9.95%) of all the samples had contamination levels of more than 1,100 MPN/g. Virulence gene detection revealed that 95 and 82% of the isolates harbored nheABC and hblACD gene clusters, respectively. Additionally, 87% of the isolates harbored cytK gene, and 3% of the isolates possessed cesB. Most strains were resistant to rifampicin and β-lactam antimicrobials but were sensitive to imipenem, gentamicin, ciprofloxacin, kanamycin, telithromycin, ciprofloxacin, and chloramphenicol. In addition, more than 95.6% of the isolates displayed resistance to three kinds of antibiotics. Based on multilocus sequence typing, all strains were classified into 210 different sequence types (STs), of which 145 isolates were assigned to 137 new STs. The most prevalent ST was ST770, but it included only eight isolates. Taken together, our research provides the first reference for the incidence and characteristics of B. cereus in vegetables collected throughout China, indicating a potential hazard of B. cereus when consuming vegetables without proper handling.

Bacillus cereus is a food-borne opportunistic pathogen that can induce diarrheal and emetic symptoms. It is widely distributed in different environments and can be found in various foods, including fresh vegetables. As their popularity grows worldwide, the risk of bacterial contamination in fresh vegetables should be fully evaluated, particularly in vegetables that are consumed raw or processed minimally, which are not commonly sterilized by enough heat treatment. Thereby, it is necessary to perform potential risk evaluation of B. cereus in vegetables. In this study, 294 B. cereus strains were isolated from vegetables in different cities in China to analyze incidence, genetic polymorphism, presence of virulence genes, and antimicrobial resistance. B. cereus was detected in 50% of all the samples, and 21/211 (9.95%) of all the samples had contamination levels of more than 1,100 MPN/g. Virulence gene detection revealed that 95 and 82% of the isolates harbored nheABC and hblACD gene clusters, respectively. Additionally, 87% of the isolates harbored cytK gene, and 3% of the isolates possessed cesB. Most strains were resistant to rifampicin and β-lactam antimicrobials but were sensitive to imipenem, gentamicin, ciprofloxacin, kanamycin, telithromycin, ciprofloxacin, and chloramphenicol. In addition, more than 95.6% of the isolates displayed resistance to three kinds of antibiotics. Based on multilocus sequence typing, all strains were classified into 210 different sequence types (STs), of which 145 isolates were assigned to 137 new STs. The most prevalent ST was ST770, but it included only eight isolates. Taken together, our research provides the first reference for the incidence and characteristics of B. cereus in vegetables collected throughout China, indicating a potential hazard of B. cereus when consuming vegetables without proper handling.
Vegetables are an indispensable part for human food and nutrition. The World Health Organization recommends taking 400 g of fresh vegetables and fruits daily to promote human health (World Health Organization, 2003). Vegetables are often consumed directly or only with minimal processing that does not eliminate pathogenic bacteria, such as B. cereus (Goodburn and Wallace, 2013;Mogren et al., 2018). Since consumption of fresh vegetable has increased dramatically over the last few decades (Olaimat and Holley, 2012;Hackl et al., 2013), more foodborne outbreaks resulting from contaminated vegetables have simultaneously emerged (Berger et al., 2010;Castro-Ibáñez et al., 2016). For example, food poisoning outbreaks associated with vegetables contaminated by foodborne pathogens in Korea increased from 119 in 1998 to 271 in 2010 (Park et al., 2018). Therefore, it is necessary to monitor the contamination level of B. cereus in vegetables.
Antimicrobial treatment is the main method to eliminate foodborne pathogens, including B. cereus, in patients with food poisoning. However, antibiotic resistance in B. cereus has already emerged due to the abuse of antibiotics. The therapeutic effect of some antibiotics against antimicrobial-resistant isolates decreases or even disappears, leading to the failure of clinical treatment (Brown et al., 2003;Friedman, 2015;Torkar and Bedenić, 2018). As the consumption of vegetables contaminated with antimicrobial-resistant isolates may lead to more severe infection (Berthold-Pluta et al., 2017), it is important to test the antibiotic resistance of B. cereus in vegetables for food safety and human health.
B. cereus is widely distributed in nature and can contaminate foods primarily through soil and air (European Food Safety Authority [EFSA], 2005;Arnesen et al., 2008). Vegetables are generally planted in fields, where they are exposed to soil, and they are exposed to air during transportation and sale; they can therefore be easily contaminated by this pathogenic bacterium. B. cereus contamination in vegetables has been reported. The contamination rate in different vegetables ranged from 29.0 to 70.0% in South Korea (Chon et al., 2012(Chon et al., , 2015Kim Y. J. et al., 2016). In Mexico City, B. cereus was identified in 57% of the 100 analyzed samples (Flores-Urban et al., 2014). The contamination rate in different vegetables in southeast of Spain varied greatly (Valero et al., 2002).
As an essential daily nutrient, to date, no study has evaluated the occurrence rate of B. cereus in vegetables accounting for the whole of China. Therefore, in this study, we analyzed the contamination, genotypic diversity, pathogenic potential, and antimicrobial resistance of B. cereus isolated from vegetables in China to obtain an overview on the potential risk.  Figure S1) from 2011 to 2016, according to the general sample collection guidelines of the National Food Safety Standard (The Hygiene Ministry of China, 2010b). The samples were placed in sealed bags, transferred to the laboratory in a low-temperature (below 4 • C) sampling box, and immediately subjected to microbiological analysis after sending back to the laboratory.

Isolation and Identification of B. cereus
Twenty-five grams of vegetable samples was cut into pieces, transferred into a sterile homogenizer containing 225 ml phosphate buffered saline (PBS, 0.01 mol/L), and then homogenized at 8,000 rpm for 2 min using a rotary blade homogenizer. The homogenized solution and its 1/10 and 1/100 dilutions were used to detect B. cereus qualitatively and quantitatively according to the B. cereus test rules given by the National Food Safety Standard (The Hygiene Ministry of China, 2010a), as described previously (Gao et al., 2018). Mannitol yolk polymyxin (MYP) agar plate test, parasporal crystal observation (to distinguish between B. cereus and Bacillus thuringiensis), root growth observation, hemolysis test, catalase test, motility test, nitrate reduction test, casein decomposition test, lysozyme tolerance test, glucose utilization test, and acetyl methyl alcohol test were conducted for species detection. The most probable number (MPN) method was adopted for quantitative detection of species. Briefly, 1 ml of the homogenized solution and its 1/10 and 1/100 dilutions were inoculated into three tubes each

Antimicrobial Resistance Testing
The

Distribution of Virulence Genes Among B. cereus Isolates
The presence of toxin genes is summarized in Figure 1. The cereulide synthetase gene cesB was detected in only 3% of isolates. In contrast, the rate of enterotoxin gene detection was very high. hbl genes encoding the Hbl toxin complex were detected in 81% of the samples. Additionally, 99, 100, and 96% of all isolates harbored nheA, nheB, and nheC, respectively. However, only 95% of the isolates harbored the integrated Nhe-encoding gene cluster nheABC. cytK was detected in 87% of the strains.
In contrast, the most sensitive strain, 3763, showed resistance to only two antibiotics (FOX-TE). AMP-KF-FOX-P-AMC-RD was the most common antimicrobial resistant profile (91 of 294 strains). We also evaluated the multidrug resistance (MDR; Magiorakos et al., 2012) profiles and found that 95.6, 76.2, and 35.4% of isolates displayed simultaneous resistance to more than three, four, and five types of antimicrobials, respectively (Figure 2).

Multilocus Sequence Typing and Cluster Analysis
Genetic diversity was analyzed by the MLST method. Among all 294 strains, 210 STs were assigned, and 145 strains were assigned to 137 new STs. Additionally, 175 of all the 210 (83%) STs included a single strain, 35 STs included two to eight isolates, and only ST-770 included eight isolates, followed by ST-1605, which included seven strains. Five isolates belonged to ST-26, which is associated with clinical isolates. All 210 STs were grouped into six clonal complexes (CCs) and 189 singletons. The ST-142 complex was most frequent, including 41 isolates, while the ST-18, ST-23, FIGURE 1 | Detection rate of virulence genes in B. cereus from vegetables. The number at the top of the bars represents the positive rate of corresponding toxin genes. hblACD and nheABC mean that the strains are positive for hblA, hblC, and hblD or for nheA, nheB, and nheC at the same time, respectively. "All eight genes" presents the strains with all the detected toxin genes.

Prevalence of B. cereus Isolates in Vegetables
Few studies have evaluated pathogenic B. cereus in vegetables worldwide and no report has focused on the whole of China. In our study here, we found that 50% of all vegetable samples collected from 39 major cities in China contained B. cereus.
The contamination level was more or less the same as those of previous surveys in other countries, i.e., 20-48% in Korea (Chon et al., 2015;Park et al., 2018), 57% in Mexico City (Flores-Urban et al., 2014), and 52% in the southeast of Spain (Valero et al., 2002). These reports, together with ours, indicate that B. cereus contamination in vegetables is common in several countries and suggest that consumption of vegetables contaminated with B. cereus is a potential health hazard (Valero et al., 2002;Kim Y. J. et al., 2016). The high level of contamination by B. cereus may be partly attributed to contact with soil or air during field planting (European Food Safety Authority [EFSA], 2005;  Arnesen et al., 2008) and exposure to air during transportation and sale. Upon contamination, B. cereus may form a biofilm on the surface of vegetables, resulting in its persistent difficulty to be eliminated (Majed et al., 2016). B. cereus-positive samples may also contaminate other vegetables by contact transmission. According to the Microbiological Guidelines for Food of Hong Kong, China (Food and Environmental Hygiene Department, 2014), even though the amounts of B. cereus between 10 3 and 10 5 CFU/g in ready-to-eat foods are considered to be "acceptable, " they pose potential risks, and the raw materials, processing period, and environment should be examined to investigate the reason why these foods are contaminated; if the amounts of B. cereus in ready-to-eat foods are more than 10 5 CFU/g, their quality is considered to be "unsatisfactory" and their sale should be stopped. The standards of microbiological limits for ready-to-eat foods in Australia and New Zealand (New South Wales Food Authority, 2009), however, stipulate that the "acceptable" level of B. cereus is 10 2 -10 3 CFU/g, and an "unsatisfactory" level is 10 3 -10 4 CFU/g. The United Kingdom microbiological testing standards (Health Protection Agency, 2009) for ready-to-eat foods stipulate that the "acceptable" level of B. cereus in ready-to-eat foods is 10 3 -10 5 CFU/g, and a level of more than 10 5 CFU/g of B. cereus is considered to be "unsatisfactory." Of all the B. cereus-positive samples in this study, 9.95% (21/211), mainly of C. sativum L and var. ramosa Hort., had contamination levels of more than 1,100 MPN/g. The contamination levels of these samples were at least at the "acceptable" level according to Hong Kong and United Kingdom standards and at the "unsatisfactory" level according to Australia and New Zealand standards. This suggests that B. cereus-contaminated vegetables pose a potential risk of causing foodborne disease, and care needs to be taken when consuming them directly or with minimal processing.

Multilocus Sequence Typing and Genetic Diversity
MLST is a crucial epidemiological typing method based on the sequences of seven different housekeeping gene loci; it is used in studies of evolution and population diversity of B. cereus isolates (Erlendur et al., 2004;Cardazzo et al., 2008;Liu et al., 2017;Yang et al., 2017). In this study, we employed MLST to analyze genetic polymorphism in isolates from vegetables. Most of the isolates were assigned to singleton (Figure 3). The six CCs were distributed in separate samples, except ST-23 complex and ST-97 complex. ST-18 complex, ST-23 complex, ST-142 complex, and ST-205 complex even crossed with some singletons, indicating high genetic diversity of the isolates. However, we could not find any unique STs that existed in only one particular vegetable variety. Five strains, two of which were isolated from var. ramosa Hort, one from C. sativum L, and the remaining two from C. sativus L., were assigned to ST26, the same molecular type of clinical isolates NC7401 and F4810/72 (Agata et al., 2002;Fricker et al., 2007). Three out of these five isolates were identified as potential emetic strains. As the preformed cereulide in foods is persistent and may also lead to food poisoning (Agata et al., 1995), there is a potential risk when consuming these vegetables directly or with minimal processing. Interestingly, we found that all isolates that belonged to the ST-18 complex, ST-97 complex, and ST-142 complex harbored the same virulence gene profile (hblA-hblC-hblD-nheA-nheB-nheC-cytK), while other CCs showed no such phenomenon. Additionally, 30 of 32 isolates belonging to ST-18 and ST-97 complexes showed resistance to cefotetan (CTT, 30 µg), whereas only 10 of 73 isolates belonging to the other four CCs showed similar resistance, which may be explained by the properties of founder clones of different CCs.

Virulence Gene Detection and Potential Toxicity
Diarrhea caused by B. cereus is attributed to different enterotoxins produced by these strains in the small intestine (Jeßberger et al., 2015), including Hbl, Nhe (Lund and Granum, 1997), and CytK (Lund et al., 2000). In this study, we evaluated seven enterotoxin genes in B. cereus, and the positive rates of nheABC, hblACD, and cytK were 95, 81, and 87%, respectively (Figure 2 and Supplementary Table S1). The positive rates of nheABC, hblACD, and cytK were higher than those found in Korea (69.5, 41.7, and 74.2%;Park et al., 2018), but the hblA gene was detected in only 82% of isolates, which is lower than that reported in Mexico City (Flores-Urban et al., 2014). When considering different kinds of food, the nheABC frequency in our vegetables was lower than in Sunsik from Korea (Chon et al., 2012) or in pasteurized milk from China (Gao et al., 2018), but the cytK detection rate was much higher than those in rice and cereals from Korea (55%; Park et al., 2009), in Sunsik from Korea (77%; Chon et al., 2012), and in pasteurized milk from China (73%; Gao et al., 2018). Owing to the properties of foods consumed raw or processed minimally, the wider distribution of diarrheal B. cereus in these vegetables and their potential hazard cannot be neglected.
Emetic symptoms are caused by the emetic toxin cereulide. The positive rate of cesB was 3%, which is higher than those reported in Korea and Mexico City (Flores-Urban et al., 2014;Chon et al., 2015;Park et al., 2018), almost the same as that (2.9%) in Sunsik of Korea (Chon et al., 2012), but slightly lower than the 5% in pasteurized milk of China (Gao et al., 2018). Although the positive rate of cesB was quite low when compared with the rates of enterotoxins, cereulide is very persistent and heat-stable. Even emetic toxin remaining in sterilized food can cause emetic symptoms accordingly (Agata et al., 1995), the emetic isolates in these raw consuming vegetables are still potential risks.
Antimicrobial Resistance of B. cereus Isolates B. cereus infection may lead to diarrhea, vomiting, and even death (Hilliard et al., 2003;Evreux et al., 2007;Jean-Winoc et al., 2013;Ramarao et al., 2014). Antibiotic resistance test may provide a theoretical reference for the clinical treatment of B. cereus food poisoning and infection. Detailed antimicrobial resistance information of the isolates from vegetables is shown in Supplementary Table S2. More than 72.5% of all isolates were susceptible to six classes of antimicrobial agents, including aminoglycosides (CN, K), ketolide (TEL), glycopeptides (TEC), quinolones (CIP), phenylpropanol (C), tetracyclines (TE), and folate pathway inhibitors (SXT). A total of 60.2 and 99.7% of the strains were also susceptible to third-generation cephalosporin (CTT) and penems (IPM), respectively. More than 83.0% of the isolates showed resistance to other β-lactam antibiotics, including penicillins (AMP, P), β-Lactam/β-lactamase inhibitor combinations (AMC), cephems (KF, FOX), and ansamycins (RD). The high resistance of B. cereus to β-lactam antimicrobials has been widely reported and may be due to the synthesis of β-lactamase (Philippon et al., 1998;Chen et al., 2004;Park et al., 2018). Notably, 83.0% of the isolates showed resistance to rifampin (RD), which is much higher than the resistance rates of vegetables in Korea (only 48.7%; Park et al., 2018) and even higher than those of different kinds of foods, including rice and cereal (62%; Park et al., 2009), ready-to-eat foods (0%; Agwa et al., 2012), and traditional dairy products (0%; Owusu-Kwarteng et al., 2017). This may be ascribed to the usage of antibiotics in different countries or the evolution of strains toward rifampicin resistance. These results emphasize the need for caution when using β-lactams and ansamycins (such as RD) for the clinical treatment of B. cereus. On the other hand, 95.6, 76.2, and 35.4% of isolates showed resistance to more than three, four, and five classes of antibiotics simultaneously, respectively, suggesting the need to monitor multiple drug resistance in B. cereus.

CONCLUSION
Vegetables such as L. esculentum Mill., C. sativus L., var. ramosa Hort., and so on are usually consumed directly or with minimal processing, so potential hazards associated with B. cereuscontaminated vegetables should not be ignored. The results in this study revealed a high incidence of B. cereus in vegetable samples collected from across China, for the first time as we know. Of all the samples, 21/211 (9.95%), mainly of C. sativum L and var. ramosa Hort., had contamination levels of more than 1,100 MPN/g. According to the Microbiological Guidelines for Food of Hong Kong, China (Food and Environmental Hygiene Department, 2014), the United Kingdom microbiological testing standards for ready-to-eat foods (Health Protection Agency, 2009), and the standards of microbiological limits for readyto-eat foods in Australia and New Zealand (New South Wales Food Authority, 2009), the contamination levels of these samples were at the "acceptable" level according to Hong Kong and United Kingdom standards and at the "unsatisfactory" level according to the Australia and New Zealand standards. These contamination levels indicate a potential risk caused by the consumption of B. cereus-contaminated vegetables either directly or with minimal processing and should be kept in mind while assessing the quality of vegetables. If necessary, the reason for the contamination should also be traced. The pathogenic ST ST26 was detected in vegetable isolates. Seven enterotoxin genes associated with diarrheal symptoms were widespread among isolates, and the emetic toxin gene was also detected. In addition, most isolates were resistant to β-lactam antimicrobials, such as amoxicillin-clavulanic acid, penicillin, ampicillin, cephalothin, cefoxitin, and rifampin. Our results indicate a potential risk of consuming vegetables without sufficient processing and the increasing difficulties in eliminating B. cereus with antibiotics. To ensure the health and safety for the public, it is therefore necessary to develop new methods to prevent contamination and consequent potential foodborne outbreak induced by B. cereus in raw consuming vegetables.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2019.00948/full#supplementary-material FIGURE S1 | Sampling cities where the vegetables were collected.