Prevalence, Virulence Genes, Antimicrobial Susceptibility, and Genetic Diversity of Bacillus cereus Isolated From Pasteurized Milk in China

Bacillus cereus is a common and important food-borne pathogen that can be found in various food products. Due to low-temperature sterilization for a short period of time, pasteurization is not sufficient for complete elimination of B. cereus in milk, thereby cause severe economic loss and food safety problems. It is therefore of paramount importance to perform risk assessment of B. cereus in pasteurized milk. In this study, we isolated B. cereus from pasteurized milk samples in different regions of China, and evaluated the contamination situation, existence of virulence genes, antibiotic resistance profile and genetic polymorphism of B. cereus isolates. Intriguingly, 70 samples (27%) were found to be contaminated by B. cereus and the average contamination level was 111 MPN/g. The distribution of virulence genes was assessed toward 10 enterotoxigenic genes (hblA, hblC, hblD, nheA, nheB, nheC, cytK, entFM, bceT, and hlyII) and one emetic gene (cesB). Forty five percent strains harbored enterotoxigenic genes hblACD and 93% isolates contained nheABC gene cluster. The positive rate of cytK, entFM, bceT, hlyII, and cesB genes were 73, 96, 75, 54, and 5%, respectively. Antibiotic susceptibility assessment showed that most of the isolates were resistant to β-lactam antibiotics and rifampicin, but susceptible to other antibiotics such as ciprofloxacin, gentamicin and chloramphenicol. Total multidrug-resistant population was about 34%. In addition, B. cereus isolates in pasteurized milk showed a high genetic diversity. In conclusion, our findings provide the first reference on the prevalence, contamination level and characteristics of B. cereus isolated from pasteurized milk in China, suggesting a potential high risk of B. cereus to public health and dairy industry.

Bacillus cereus is a common and important food-borne pathogen that can be found in various food products. Due to low-temperature sterilization for a short period of time, pasteurization is not sufficient for complete elimination of B. cereus in milk, thereby cause severe economic loss and food safety problems. It is therefore of paramount importance to perform risk assessment of B. cereus in pasteurized milk. In this study, we isolated B. cereus from pasteurized milk samples in different regions of China, and evaluated the contamination situation, existence of virulence genes, antibiotic resistance profile and genetic polymorphism of B. cereus isolates. Intriguingly, 70 samples (27%) were found to be contaminated by B. cereus and the average contamination level was 111 MPN/g. The distribution of virulence genes was assessed toward 10 enterotoxigenic genes (hblA, hblC, hblD, nheA, nheB, nheC, cytK, entFM, bceT, and hlyII) and one emetic gene (cesB). Forty five percent strains harbored enterotoxigenic genes hblACD and 93% isolates contained nheABC gene cluster. The positive rate of cytK, entFM, bceT, hlyII, and cesB genes were 73, 96, 75, 54, and 5%, respectively. Antibiotic susceptibility assessment showed that most of the isolates were resistant to β-lactam antibiotics and rifampicin, but susceptible to other antibiotics such as ciprofloxacin, gentamicin and chloramphenicol. Total multidrug-resistant population was about 34%. In addition, B. cereus isolates in pasteurized milk showed a high genetic diversity. In conclusion, our findings provide the first reference on the prevalence, contamination level and characteristics of B. cereus isolated from pasteurized milk in China, suggesting a potential high risk of B. cereus to public health and dairy industry.

INTRODUCTION
The opportunistic pathogen Bacillus cereus is known to cause food-borne outbreaks in humans. B. cereus generally causes two types of gastrointestinal illness including emesis and diarrhea after consumption of a contaminated food, which contains more than 10 4 -10 5 spores or vegetative cells of B. cereus per gram (Jensen et al., 2003;Bamnia and Kaul, 2015). The investigation during 1960 to 1992 had shown that food-borne outbreaks associated with B. cereus ranged from 1 to 22% in Europe, Japan, and North America (Griffiths and Schraft, 2002). B. cereus has emerged as the second food-borne pathogen in France after Staphylococcus aureus (Santé Publique France, 2015; Glasset et al., 2016), and has ranked third in China (Mao et al., 2010).
Antibiotic therapy is still the primary treatment method for the infections of B. cereus. However, emergence of antibiotic resistant B. cereus strains, mainly due to antibiotic misusage (Barbosa and Levy, 2000) or acquisition of resistance genes through horizontal gene transfer (Bogdanova et al., 1998;Agerso et al., 2002;Brown et al., 2003), results in the failure of antibiotic treatment. Thus, obtaining the B. cereus antibiotics resistance profile is highly relevant to public health.
Bacillus cereus has been found in milk with a high contamination rate (Rather et al., 2011;Gundogan and Avci, 2014;Chaves et al., 2017;Owusu-Kwarteng et al., 2017;Saleh-Lakha et al., 2017;Lan et al., 2018), especially pasteurized milk using low-temperature sterilizing process that cannot fully eliminate B. cereus spores (Zwietering et al., 1996). Consequently, it could lead to a variety of milk defects and foodborne diseases (Stenfors Arnesen et al., 2008). Besides, B. cereus in milk is probably associated with a higher risk of hazard. For example, thermophilic strains of B. cereus from milk had higher toxicity (Miller et al., 2016) and milk-originated B. cereus sensu lato strains harbored Bacillus anthracis-like plasmids, suggesting that B. cereus could gain more pathogenic genes through the interflow among Bacillus groups in milk (Bartoszewicz and Marjańska, 2017).
B. cereus contamination in dairy products have been reported (Gundogan and Avci, 2014;Yobouet et al., 2014;Drean et al., 2015;Chaves et al., 2017;Owusu-Kwarteng et al., 2017;Saleh-Lakha et al., 2017;Lan et al., 2018); however, the overall scale, especially in pasteurized milk, is quite small. Since pasteurization has a low inactivation rate of B. cereus spores and detection of B. cereus is not required for the dairy microorganism test of Chinese food security standard (The Hygiene Ministry of China, 2010b), it may increase the risk of B. cereus in dairy products. Thus, it is necessary to assess the prevalence and microbiological traits of B. cereus in pasteurized milk products. Here we used enterobacterial repetitive intergenic consensus sequences polymerase chain reaction (ERIC-PCR) to analyze genotypic diversity of B. cereus isolated from pasteurized milk products in China and combined with the pathogenic potential, antimicrobial resistance characters, aiming to provide an overview of the risk assessment for B. cereus isolated from pasteurized milk in China.

Sample Collection
From July 2011 to May 2016, two hundred and seventy six pasteurized milk samples were collected from major cities in China (Supplementary Figure S1). The investigation process was divided into three stages. The first-stage investigation mainly focused on cites in southern china, and the second stage included six cities in northern China. The third one was a national wide survey that included other 15 cities. Details of sample distribution were shown in Table 1.

Qualitative and Quantitative Detection of B. cereus
The qualitative and quantitative detection of B. cereus were performed according to National Food Safety Standard (The Hygiene Ministry of China, 2010a) with minor modification. In brief, 25 ml of sample was mixed and homogenized with 225 ml Trypticase-soy-polymyxin (TSB) broth (Huankai, China) at 30 • C for 48 h. Then cultures were streaked on the Mannitolegg yolk-polymyxin (MYP) agar plate (Selective media; Huankai, China) and Chromogenic plate (Huankai, China) and incubated at 30 • C for 24 h. Colonies with pink sparkle in blue or bluegreen precipitation on Chromogenic plate were picked for further biochemical identification using the B. cereus biochemistry assessor (Huankai, China). The quantitative detection assay was conducted by B. cereus most probable number (MPN) counting method in Food Safety Standards (The Hygiene Ministry of China, 2010a).

Detection of Virulence Genes
The genomic DNA of various strains isolated from pasteurized milk samples was extracted using the HiPure Bacterial DNA Kit (Magene, United States) in accordance with the manufacturer's instruction.

Antimicrobial Susceptibility Testing
Antimicrobial susceptibility of all B. cereus isolates was evaluated by the Kirbye-Bauer disk diffusion method according to performance standards for antimicrobial susceptibility testing  Supplementary Table S1.

Genetic Biodiversity Assay
The ERIC-PCR method was performed for typing and comparing the pasteurized milk isolates. ERIC-PCR was carried out using the primers of ERIC-F and ERIC-R (Table 1) as described by Versalovic et al. (1991). The PCR reaction mixture (25 µl) contained 12.5 µl of ExTaq Mix (Takara, China), 1.0 µM of each primer, 50-100 ng genomic DNA. Amplification was performed as follows: an initial denaturation at 94 • C for 3 min; 35 cycles each consists of 30 s at 94 • C, 40 s at 45 • C and 3 min at 72 • C; and a final extension at 72 • C for 10 min. Thereafter the amplicons (10 µl) were electrophoresed on 1.5% agarose gel. The DNA fingerprint was analyzed by BioNumerics 7.1 software (Applied Maths, Belgium). ERIC-PCR cluster analysis was assessed by Simpson's diversity index (Hunter and Gaston, 1988).

Prevalence Analysis of Bacillus cereus in Pasteurized Milk
Bacillus cereus contamination was found to occur in pasteurized milk samples from 32 out of 39 cities in China (Supplementary Figure S1).  Figure S1).

Distribution of Virulence Genes Among B. cereus Isolates
The distribution of virulence genes was evaluated and summarized in Figure 1 and Supplementary Table S2.
According to the pathogenic characteristics of B. cereus, the virulence genes are divided into two categories, namely enterotoxin genes (nheABC, hblACD, cytK, bceT, hly II, entFM) and cereulide synthetase genes (cesB). In enterotoxin genes, the gene cluster nheABC encoding the non-hemolytic enterotoxin (Nhe) complexes present in most of the strains (93%) with only a small portion of strains missing nheC or nheA. Forty five percent strains harbored enterotoxigenic genes hblACD. But the detection rate of hblA (46%) was significantly lower than hblC (66%) and hblD (67%). CytK, entFM and bceT were detected in more than 70% of the strains. In contrast, the detection rate of cereulide synthetase gene cesB was found to be only 5%, demonstrating that diarrheal strains in pasteurized milk samples are more common than emetic strains. Based on the distribution of virulence genes, all isolates were divided into 32 virulence genes profiles. As shown in Figure 2, only two isolates (2833-2A-Bc in G2 and 3732-Bc in G5) contained all 11 virulence genes, and one isolate harbored the least virulence gene profile (2583-Bc in G3, nheA-nheB-entFM). The main gene profile was hblA-hblC-hblD-nheA-nheB-nheC-cytK-hlyII-entFM-bceT (28%), revealing that a large number of potentially diarrheal strains exist in the collected samples.

Clinical, and Laboratory Standards Institute [CLSI] (2010/2015)
for Kirbye-Bauer disk diffusion method. There were 34 antimicrobial resistance profiles for all isolates. As shown in Figure 2, the most susceptible strain resisted to only two antibiotics, while the most resistant strain resisted to ten antibiotics. AMP-AMC-P-KF-FOX-RD was the most common antimicrobial resistance profile. Multidrug resistance (MDR) profiles, defined as simultaneously resistant to more than three types of antibiotics (Magiorakos et al., 2012), were also evaluated. Thirty four percent of all isolates displayed resistance to three or more antibiotics, and 4% of the isolates displayed resistance to over five types of antibiotics (Table 3 and Figure 3).

ERIC-PCR Typing and Cluster Analysis
Molecular typing is a good way for tracing the sources and understanding the epidemiology of food-borne pathogens, such as random amplification of polymorphic DNA (RAPD, Nilsson et al., 1998), multi-locus sequence typing (MLST, Helgason et al., 2004), amplified fragment length polymorphism (AFLP, Hill et al., 2004), pulse-field gel electrophoresis (PFGE, Merzougui et al., 2014) and so on. Among these methods, ERIC-PCR (Shangkuan et al., 2000) is a simple and efficient approach to study the genetic diversity among strains that can explain the association of phenotypic and genotypic characters quite well. For our samples, we explored ERIC-PCR for genetic diversity analysis. The fingerprints gained by ERIC-PCR consisted of 2-11 distinct bands ranging in FIGURE 2 | Dendrogram of Bacillus cereus isolated from pasteurized milk in China. Similarity (%) between fingerprints generated by ERIC-PCR was calculated by using the Dice index. The data were sorted by the UPGMA method. Different colors represent different groups, and the virulence genes profile and antimicrobial resistance characteristics of all isolates are listed in this dendrogram.
Frontiers in Microbiology | www.frontiersin.org size from 250 base pairs to 5000 base pairs. Together with the distribution of virulence genes and antimicrobial susceptibility, the dendrogram was generated by using the software BioNumerics 7.0 (Figure 2). All isolates showed 83 ERIC-PCR patterns. As shown in the Figure 2, when the relative similarity coefficient is 65%, 103 isolates could be divided into 8 clusters (G1 to G8). The G1 strains were dominant in our national isolates of which the fingerprints consisted of 6 to 8 bands and shared the common bands in sizes of about 1500 base pairs, 3000 base pairs and 4000 base pairs. As a discriminatory index, Simpson's diversity is used to evaluate different typing methods (de la Puente-Redondo et al., 2000;Almeida et al., 2016;Wasfi et al., 2016), which produces values in the ranges of 0.0-1.0. The value 1.0 indicates that a typing method is able to distinguish each member of a population. Conversely, 0.0 indicates that all members of a population are of an identical type (Hunter and Gaston, 1988). If Simpson's index (DI value) of a typing method is greater than 0.90, it is suggested that the typing method has generated a good result to distinguish all isolates. In our study, the DI value was 0.996, indicating the ERIC-PCR typing method could discriminate all the isolates well. Moreover, G1 (25.2%), G4 (16.5%), and G5 (16.5%) were the dominant groups. The fingerprints of G4 and G5 were quite similar, as well as their drug resistance spectra and virulence gene profiles. These two groups accounted for 41% of all strains, with hblA/C/D-nheA/B/C-cytK-hlyII-entFM-bceT as the main virulence gene type and AMP-AMC-P-KF-FOX-RD as the main drug resistance spectrum, indicating that the genetic diversity of these isolates were somehow conservative.

The Prevalence and Genetic Diversity of B. cereus Isolates
Till now, the prevalence studies of pathogenic B. cereus in pasteurized milk in China are scant. In this study, B. cereus was detected in 27% of pasteurized milk samples collected from major cities in China, suggesting a potential risk of B. cereus. Compared to previous surveys in other countries, it is a medium contamination level compared with 27% in Abidjan (Yobouet et al., 2014), 27.37% in Brazil (Reis et al., 2013), 14% in Slovakia (Acai et al., 2014), 47% in Ghana (Owusu-Kwarteng et al., 2017) and 37% in India (Rather et al., 2011). Traced to the source of B. cereus contamination in pasteurized milk, heat stable B. cereus spores in raw milk and the post-pasteurization contamination along the milk processing lines were major sources (Lin et al., 1998;Saleh-Lakha et al., 2017). Besides, the environments for milk production, handling and processing could introduce B. cereus into dairy products (Cui et al., 2016a;Kumari and Sarkar, 2016). Some studies reported the storage temperature of dairy products also affected the number and toxicity of B. cereus (Fermanian et al., 1997;Notermans et al., 1997) as toxic strains could produce toxin even at 8 • C. Together, these analyses imply the high prevalence of B. cereus and existence of potential hazards in consuming of the contaminated pasteurized milk.
The epidemiological typing is considered a crucial tool for studying the prevalence of food-borne bacteria. ERIC-PCR is based on the targeting of repeated DNA sequences with oligonucleotide primers, which has been broadly employed to perform the epidemiological typing of micro-organisms and widely applied into the risk survey of pathogenic bacteria such as Salmonella Typhimurium (Almeida et al., 2016), Staphylococcus aureus (Miao et al., 2016), Klebsiella pneumonia (Wasfi et al., 2016) and Arcobacter spp. (Vicente-Martins et al., 2018) as well as B. cereus strains (Hsueh et al., 1999;Shangkuan et al., 2000;Lopez and Alippi, 2007;Freitas et al., 2008;Avsar et al., 2017). In this study, we used ERIC-PCR to analyze the genetic and biological diversity of B. cereus isolates from pasteurized milk in China. The results showed that ERIC typing was suitable for studying the relationship between genetic and biological characteristics (Figure 2). The dominant genotype defined by ERIC fingerprints had the main virulence gene profile hblA/C/D-nheA/B/C-cytK-hlyII-entFM-bceT and drug resistance spectrum AMP-AMC-P-KF-FOX-RD, indicating the correlation between these different traits. Previous studies showed that fingerprinting patterns were used to distinguish different species such as AFLP typing in grouping B. anthracis, B. cereus, and B. thuringiensis isolates (Hill et al., 2004) and RAPD typing for discrimination of the B. cereus groups (Kuwana et al., 2012). However, there are very few studies available on fingerprint typing to correlate the genetic characteristics with biological traits of different strains. Here, in this study, we found that ERIC-PCR had great advantages in distinguishing and classifying strains with different biological characteristics.

Virulence Genes of B. cereus Isolates
Unlike emesis, diarrhea is associated with a series of enterotoxins produced by B. cereus in the small intestine, and the predominant protein responsible for this has not been yet determined other than enterotoxins (Jessberger et al., 2015). Nevertheless, pore-forming cytotoxins haemolysin BL (Hbl), non-haemolytic enterotoxin (Nhe) and cytotoxin K (CytK) have been identified as etiological agents of the diarrheal disease (Lund and Granum, 1996;Lund and Granum, 1999;Lund et al., 2000). In this study, 10 enterotoxigenic genes were detected and the proportion of nheABC, hblACD and cytK in B. cereus were found to be 93, 45, and 73%, respectively, which is quite similar to previous studies in France (nhe 96%, hbl 40%, cytk 42%, Glasset et al., 2016) or in China (nhe 100%, hbl 78.3%, Cui et al., 2016b), but were higher than an investigation of dairy products in Turkey (hbl 13%, nhe 60%, cytk 75%, Yibar et al., 2017), indicating that diarrheal strains have a wider distribution and a higher risk of B. cereus infections exists in consuming pasteurized milk in China.

Antimicrobial Susceptibility of B. cereus
Bacillus cereus may cause severe diseases and infections that even lead to death (Lund et al., 2000;Dierick et al., 2005). Effective antibiotic therapy is considered as a predominant treatment to eliminate B. cereus infections, which has necessitated the investigation of antimicrobial susceptibility of B. cereus. In our study, B. cereus isolates were resistant to β-lactam antibiotics and rifampicin, but were susceptible to quinolones, aminoglycosides and macrolides. Consistent with previous studies, B. cereus were resistant to β-lactam antibiotics (Luna et al., 2007;Fernandes et al., 2014;Kim et al., 2015;Yibar et al., 2017) owing to the β-lactamase production (Bottone, 2010). It is worth mentioning that B. cereus was resistant to cephalosporin (cefoxitin, cephalothin) except for the third generation cephalosporin (cefotetan). Since broadspectrum cephalosporin is the main antibiotic used in the treatment of gastrointestinal diseases caused by bacterial infections, it should be avoided as clinical treatment to gastroenteritis caused by B. cereus (Savic et al., 2016). Nowadays, vancomycin is considered as one of the most proper choice for B. cereus infections (Tatara et al., 2013;Torkar et al., 2016). However, a portion of our isolates (about 13%) detected were not sensitive to this antibiotic, suggesting the existence of potential risk for B. cereus infections. For multiple drug resistance, B. cereus isolates displayed resistance to three or more antibiotics were 34% which should be paid more attention to.

CONCLUSION
Despite the potential health risks associated with B. cereus, the prevalence and molecular analysis of genotypes of them have not been fully explored in pasteurized milk in China. In this study, we showed the high prevalence of B. cereus and its antibiotic resistance characteristics in pasteurized milk all over China. ERIC-PCR analysis demonstrated high genetic diversity of B. cereus in pasteurized milk samples. To the best of our knowledge, this is the first comprehensive investigation about the prevalence of B. cereus virulence factors, antibiotic resistance phenotypes and genotypes by ERIC typing in pasteurized milk from diverse regions of China that revealed a potential high risk of B. cereus to public health and dairy industry.

ACKNOWLEDGMENTS
YD is awarded the 1000-Youth Elite Program (The Recruitment Program of Global Experts in China). We thank Dr. Srinivasan Balamurugan for his comments on the manuscript.