Prevalence, Antimicrobial Resistance, and Whole Genome Sequencing Analysis of Shiga Toxin-Producing Escherichia coli (STEC) and Enteropathogenic Escherichia coli (EPEC) from Imported Foods in China during 2015–2021

Shiga toxin-producing Escherichia coli (STEC) and enteropathogenic Escherichia coli (EPEC) are foodborne pathogens that cause hemolytic uremic syndrome and fatal infant diarrhea, respectively, but the characterization of these bacteria from imported food in China are unknown. A total of 1577 food samples from various countries during 2015–2021 were screened for STEC and EPEC, and the obtained isolates were tested for antimicrobial resistance and whole genome sequencing analysis was performed. The prevalence of STEC and EPEC was 1.01% (16/1577) and 0.51% (8/1577), respectively. Antimicrobial resistances to tetracycline (8%), chloramphenicol (8%), ampicillin (4%), ceftazidime (4%), cefotaxime (4%), and trimethoprim-sulfamethoxazole (4%) were observed. The antimicrobial resistance phenotypes corresponded with genotypes for most strains, and some resistance genes were related to mobile genetic elements. All 16 STEC isolates were eae negative, two solely contained stx1 (stx1a or stx1c), 12 merely carried stx2 (stx2a, stx2d, or stx2e), and two had both stx1 and stx2 (stx1c + stx2b, stx1a + stx2a + stx2c). Although they were eae negative, several STEC isolates carried other adherence factors, such as iha (5/16), sab (1/16), and lpfA (8/16), and belonged to serotypes (O130:H11, O8:H19, and O100:H30) or STs (ST297, ST360), which have caused human infections. All the eight EPEC isolates were atypical EPEC; six serotypes and seven STs were found, and clinically relevant EPEC serotypes O26:H11, O103:H2, and O145:H28 were identified. Two STEC/ETEC (enterotoxigenic E. coli) hybrids and one EPEC/ETEC hybrid were observed, since they harbored sta1 and/or stb. The results revealed that food can act as a reservoir of STEC/EPEC with pathogenic potential, and had the potential ability to transfer antibiotic resistance and virulence genes.


Introduction
Shiga toxin-producing E. coli (STEC) are major foodborne pathogens, which can cause watery diarrhea, bloody diarrhea, or hemorrhagic colitis, and even life-threatening hemolytic uremic syndrome (HUS). It is estimated that STEC leads to 2,801,000 acute isolated these bacteria from food imported from different continents, and performed the antimicrobial resistance and whole genome sequencing analysis, so as to provide technical support for food safety investigations and monitor the emergence of E. coli in foods.

Prevalence of STEC and EPEC
Of the 1577 imported food samples from 2015 to 2021, 34 (2.16%) and 41 (2.60%) were tested positive for stx and eae, respectively. A total of 16 (1.01%) food samples were confirmed to be contaminated with STEC, and 8 (0.51%) food samples contaminated with EPEC ( Table 1). The STEC/EPEC prevalence in different imported food samples (i.e., frozen beef, frozen pork, and frozen mutton) are shown in Table 1.
For the majority of the time, the strains carrying antimicrobial resistance genes showed the respective resistance/intermediate resistance phenotype (Table 2). There was one strain harboring the sulfonamide resistance gene, sul1, but did not show resistance or intermediate resistance to trimethoprim-sulfamethoxazole. The strain 1509-1 carried the ciprofloxacin resistance gene qnrS1, thus the ciprofloxacin MIC reached 0.25, while for other strains it was 0.03 ( Table 2).
The multiple antimicrobial resistant strains also had multiple resistance genes. The location of these resistance genes was evaluated. For 1053l-2, the aadA1, sul1, and qacE were found on the same contig 181; aph(3′)-Ib and aph(6)-Id were on contig 321; aac(3)-IV and aph(4)-Ia were on contig 279; the co-occurrence of aadA1, cmlA1, and sul3 on contig 102; aadA8b, aadA2, and dfrA12 on contig 103 were found for 1095a; and for strain 1509-1, the co-occurrence of qnrS1 and blaTEM-1B were observed on contig 54. Furthermore, the relations of these antimicrobial resistance genes to mobile genetic elements were evaluated. The aac(3)-IV and aph(4)-Ia were associated with insertion sequence ISEc59. The gene blaTEM-1B was related to insertion sequence ISKpn19 while qnrS1 was related to unit transposon Tn2.   For the majority of the time, the strains carrying antimicrobial resistance genes showed the respective resistance/intermediate resistance phenotype (Table 2). There was one strain harboring the sulfonamide resistance gene, sul1, but did not show resistance or intermediate resistance to trimethoprim-sulfamethoxazole. The strain 1509-1 carried the ciprofloxacin resistance gene qnrS1, thus the ciprofloxacin MIC reached 0.25, while for other strains it was 0.03 ( Table 2).
The multiple antimicrobial resistant strains also had multiple resistance genes. The location of these resistance genes was evaluated. For 1053l-2, the aadA1, sul1, and qacE were found on the same contig 181; aph(3 )-Ib and aph(6)-Id were on contig 321; aac(3)-IV and aph(4)-Ia were on contig 279; the co-occurrence of aadA1, cmlA1, and sul3 on contig 102; aadA8b, aadA2, and dfrA12 on contig 103 were found for 1095a; and for strain 1509-1, the co-occurrence of qnrS1 and blaTEM-1B were observed on contig 54. Furthermore, the relations of these antimicrobial resistance genes to mobile genetic elements were evaluated. The aac(3)-IV and aph(4)-Ia were associated with insertion sequence ISEc59. The gene blaTEM-1B was related to insertion sequence ISKpn19 while qnrS1 was related to unit transposon Tn2.
The phylogenetic tree based on cgMLST showed that the strains in this study were diverse. Regardless of the pathotype (STEC/EPEC), serotype, or ST, all the strains were grouped according to their phylogroups, as isolates belonging to different phylogroups (B2, D/E, A, C, and B1) clustered together (Figure 2). Some virulence genes were previously found in phylogroup B1, for instance, lpfA, ehxA, iha, subA, espP, and epeA. Certain stx subtypes were found to be associated with food type or serotype. For instance, stx2e was mainly found in pork, stx2a and stx2d were more commonly found in beef. Genotype stx1c + stx2b was observed in O128:H2 from mutton. There was no apparent correlation between country, food type, serotype, and ST; however, five isolates from beef in country A showed highly similar characteristics, regarding serotype, ST, and virulence genes.

Discussion
The characterization of STEC and EPEC from imported food in China has not been reported previously. Therefore, in this study, we isolated these bacteria from various kinds of food, mainly animal source frozen meat, imported from different countries, and studied their antimicrobial resistance, genetic diversity, and virulence profile.
The overall PCR screening rates for STEC (2.16%) and EPEC (2.60%) and the isolation rates for STEC (1.01%) and EPEC (0.51%) in this study were both lower than in the previous reports [4,[19][20][21]. For instance, the PCR screening rates for STEC were 19.5% for locally produced retail raw meats in China [4], 49.3% for ground beef in Chile [19], 8.5% for ground beef and 13.4% for ground pork in the United States [20], and 8.4% for fresh beef in Italy [21]; the STEC isolation rates were 6.8% for retail raw meats in China [4], 10% for ground beef in Chile [19], 5.2% for both ground beef and ground pork in the United States [20], 3.7% for fresh beef in Italy [21], and 2% for cow's milk in Spain [18]. EPEC was detected in 8.5% of ready-to-eat samples in China [13] and in 6% of cow's milk in Spain [18]. The food samples in this study were mainly frozen meat, which is not suitable for STEC/EPEC survival or multiplication. This may be one reason for the low STEC/EPEC prevalence in this study. In addition, good hygiene control measures may be taken by the overseas food manufactures. Furthermore, different enrichment and isolation methods may be used by different studies. Nevertheless, similar to previous studies, the STEC isolation rates were lower than PCR screening rates. The interfering high levels of background microflora, the presence of other bacteria carrying stx, the low levels of STEC in the samples, or the presence of free Stx phages can lead to the failure of STEC isolation [22]. Currently, there is no suitable selective isolation agar for STEC, therefore the isolation of the suspected STEC colonies from selective/differential agar media is challenging. Most STEC isolates in this study were obtained from MacConkey and TBX, not from Chromagar TM STEC media, which has inhibitory effects on uncommon non-O157 serotypes [23,24]. Therefore, we recommend using Chromagar™ STEC together with E. coli differentiating agar, e.g., TBX or MacConkey, to facilitate the isolation.
Drug resistance in E. coli has become a worldwide issue. Overall, the antimicrobial resistance is not severe in the studied isolates. Resistance to β-lactam, tetracycline, chloramphenicol, and sulfonamide are most frequently detected in this study, quite different from the situation in China, for which streptomycin (46.94%) and ciprofloxacin (20.41%) resistances were most common for the STEC from retail meats [5]. It is worth noting that some multi-drug resistant isolates carrying multiple antimicrobial resistance genes on the same contigs were found in this study. The co-occurrence of multiple antibiotic resistance genes could show the extensive administration of antimicrobials over many years and it may have led to the development of multiple resistances by mobile genetic elements, resulting in co-selection. The mobile genetic element-associated resistance genes were also found in this study, and this poses a great challenge to the combat against bacterial antimicrobial resistance, since they can easily be horizontally transferred among bacteria. The antimicrobial resistance phenotypes are in accordance with the genotypes for most antimicrobials; however, there is a mismatch between them in a few isolates, a finding that has also been reported in other species [25,26]. The observed phenotypic resistance strains lacking resistance genes might be due to non-resistance genetic factors, while resistance genes detected in susceptible isolates might be considered as "silent" or unexpressed genes [26].
The virulence potential of STEC should be assessed based on various factors, including serotype, stx subtype, virulence gene, phylogroup, and ST. We detected several STEC serotypes that have been related to human infections, such as O8:H19, O128:H2, O100:H30, and O130:H11, based on previous reports [27] and the EnteroBase database. Certain subtypes of the stx2 subtypes (stx2a, stx2c, and stx2d) were reported to be linked with serious human diseases [6]. Nine isolates in this study possessed the above subtypes, thus posing a health threat. Stx2e-producing STEC strains have also been isolated from patients with acute diarrhea and HUS, thus the clinical significance of the four isolates carrying stx2e in this study should not be neglected. Specifically, the two stx2e-positive O8:H19phylogroup C isolates belonged to ST360, which is principally found in human diseases based on the EnteroBase database, and possessed adhesion and colonization factor lpfA; thus, they should also be attached importance. Strains belonging to E. coli phylogroups B1, C, and E2(O157) are often pathogenic and of interest to medical research [28]. We found that some genes were highly associated with phylogroup B1, such as lpfA, ehxA, iha, subA, espP, and epeA. Strains positive for these virulence genes may have a pathogenic potential. Iha is an adherence conferring protein and a siderophore receptor distributed among STEC, and it may be involved as an alternative mechanism of adhesin in eae-negative STEC strains [29]. Subtilase cytotoxin, encoded by the subA and subB genes, is harbored by the O113:H21 outbreak strain and other eae-negative strains associated with human diseases [30]. Strains harboring stx2a and eae/aggR were assessed to be on the highest level for their estimated potential to cause diarrhea, bloody diarrhea, and HUS [9]. We did not detect the virulence combination in the studied isolates. However, the two O130:H11 isolates within phylogroup B1 possessing stx2a + iha + subA + ehxA + lpfA may have a high pathogenic potential.
None of the isolates obtained in this study were typical EPEC, which is in line with the previous results on various kinds of food [13,31]. The aEPEC strains are considered as emerging entero-pathogens detected worldwide [18]. Recent studies indicated that typical EPEC cases of diarrhea have been replaced with atypical EPEC in both developing and industrialized countries [32]. The stx genes are carried by lambdoid phages, which are highly mobile genetic elements, and the horizontal transfer and the dissemination, as well as the loss of the stx genes, are facilitated [33]. The aEPEC can also include EHEC and EPEC that have lost the stx genes and bfp genes during passage through a host or the environment or after culture in the laboratory; thus, these bacteria should not be underestimated. Clinically relevant serotypes O26:H11 and O103:H2 were found for two EPEC isolates in phylogroup B1, thus posing a great health threat. Although sharing the same serotype, O26:H11, the virulome was quite different for those belonging to B1 (12360) and A (ST48), indicating that their virulence potential may be distinct.
Notably, we found two STEC and one EPEC carrying sta1 and/or stb, which were the virulence markers of enterotoxigenic Escherichia coli (ETEC), and thus were STEC/ETEC and EPEC/ETEC hybrids, respectively. The E. coli genome is dramatically plastic and this accelerates the adaptation of this species into various environments, which provides numerous opportunities for new variants to emerge via the gains and losses of genes. Recently, hybrid E. coli pathotypes are representing emerging public health threats with enhanced virulence from different pathotypes. The most notorious hybrid was the STEC/EAEC strain O104:H4, which caused a large outbreak with numerous HUS cases and deaths in Germany in 2011 [34]. STEC/ETEC hybrids have been reported to be associated with diarrheal disease and hemolytic uremic syndrome (HUS) in humans [35,36], and EPEC/ETEC hybrid types have also been reported from patients [37]. The horizontal transmission of stx and/or sta/stb genes by the independent acquisition of the Stx-phages and/or plasmids carrying these genes lead to the emergence of STEC/ETEC hybrids. The STEC/ETEC hybrid (826a) identified in this study had the same virulence characteristics as the STEC/ETEC strain causing diarrhea in an 82-year-old patient in Sweden. For instance, both strains belonged to O100:H30, ST993, and phylogroup A and carried stx2e + sta1; therefore, it also had a high potential to cause human diseases [38]. It is also worth noting that the STEC/ETEC hybrid 908e2 possessing two st variants, sta1 and stb, had a distinctly mucoid and thread-drawing morphology on the nutrient agar. The genetic mechanism should be further explored.

Conclusions
The contamination frequency of STEC and EPEC from imported foods during the period study was relatively low, and the antimicrobial resistance was not severe. However, multidrug-resistant isolates harboring the respective multiple antimicrobial resistance genes, which are related to mobile genetic elements, were identified, and thus have the potential to transfer antibiotic resistance. Some isolates carried the virulence factors described in pathogenic strains and thus have a high pathogenic potential. STEC/ETEC and EPEC/ETEC hybrid strains were identified. Since the virulence genes of E. coli are usually located on plasmids or prophages, the emerging E. coli hybrids in foods may have enhanced virulence and thus should be attached with great importance.

Sample Collection
A total of 1577 samples from different kinds of foods imported from different continents were collected from the containers or airplanes at Shanghai port from 2015 to 2021, including frozen beef (n = 1066), frozen pork (n = 172), fresh and frozen aquatic products (n = 198), frozen mutton (n = 102), milk products (n = 27) and frozen chicken (n = 12). The samples were collected according to the National Sampling Plan by authorities and sent directly to the laboratory for testing.

Strain Isolation and Identification
The enrichment method was based on the ISO/TS 13136:2012 [39]. Briefly, a 25 g portion of each sample was transferred into a sterile sample filter bag containing 225 mL of sterile modified tryptone-soya broth (mTSB) (Land Bridge, Beijing, China), then incubated at 37 • C for 18-24 h. After enrichment, 1 mL of culture was centrifuged and performed for DNA extraction by using the bacterial genomic DNA extraction kit (TIANGEN biotech co., Beijing, China), following the manufacturer's instructions. DNA was screened for the presence of stx and eae according to the method of USDA FSIS MLG-5 [40]. The 25 µL reaction mixture contained 12.5 µL of real-time PCR premix (Takara, Dalian, China), 1.26 µM of stx (stx1 and stx2) primers, 1 µM of eae primers, 0.16 µM of 16 s primers, 0.25 µM of stx1 probe, 0.25 µM of stx2 probe, 0.2 µM of eae probe, 0.1 µM of 16 s probe, and 5 µL of DNA template. The PCR reaction conditions were as follows: 95 • C pre-denaturation for 2 min, followed by 40 cycles of 95 • C denaturation (5 s) and 60 • C (34 s) annealing. The EHEC O157: H7 strain EDL933 and nucleotide-free water were included as the positive and blank controls, respectively. For the stx positive samples, the enrichment was streaked onto three solid media: CHROMagar™ STEC agar (CHROMagar, Paris, France), MacConkey agar (Land Bridge, Beijing, China), and TBX agar (Oxoid, UK); for the eae positive samples, the enrichment was streaked onto two solid media: MacConkey agar (Land Bridge, Beijing, China) and TBX agar (Oxoid, UK). Approximately 40 colonies with E. coli morphology were picked from the above agars for further stx or/and eae detection. Each stx/eae-positive isolate was confirmed to be E. coli by the API 20 E system (bioM'erieux, Lyon, France).

DNA Extraction
Genomic DNA from each strain was extracted from overnight cultures using the bacterial genomic DNA extraction kit (TIANGEN biotech co., Beijing, China), following the manufacturer's instructions. The DNA concentration was determined using the Qubit™ dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). The DNA integrity was determined by 1% gel electrophoresis. The qualified DNA was stored in −20 • C until use.

Whole Genome Sequencing and Contig Assembly
The library was constructed using NEB Next ® Ultra™ DNA Library Prep Kit for Illumina (NEB, England). The genomes of the strains sequenced used an Illumina HiSeq sequencer (Illumina, San Diego, CA, USA), with the 2 × 150 bp pair-end chemistry according to the manufacturer's instructions, at approximately 350X average coverage. The sequencing reads were quality-control processed and quality evaluated with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/, accessed on 21 October 2021). The processed reads were assembled de novo with SPAdes (version: 3.12.0) in "careful mode".

Molecular Characterization of the Strains Based on WGS
The serotype of each strain was determined using the genes deposited in the Center for Genomic Epidemiology (http://www.genomicepidemiology.org, accessed on 21 October 2021) for E. coli as part of their web-based serotyping tool (SerotypeFinder 2.0-https://cge.cbs.dtu.dk/services/SerotypeFinder/, accessed on 21 October 2021), with a similarity of 85% and minimum length of 60%; the ST of each strain was in silico analyzed using the MLST E. coli#1 approach (dnaE, gyrB, recA, dtdS, pntA, pyrC, and tnaA) provided by the Center for Genomic Epidemiology (https://cge.cbs.dtu.dk/services/MLST/, accessed on 21 October 2021); and the phylogroup was determined by the using the E.coli phylotype analysis available in the EnteroBase website (https://enterobase.warwick. ac.uk/species/ecoli/search_strains, accessed on 21 October 2021). The virulence genes present in each strain were determined using the genes deposited in the Center for Genomic Epidemiology for E. coli as part of their web-based VirulenceFinder 2.0 tool (https://cge.cbs.dtu.dk/services/VirulenceFinder/, accessed on 21 October 2021), with the similarity of 90% and minimum length of 60%; the resistance genes present in each strain were identified using the genes deposited in the Center for Genomic Epidemiology for E. coli as part of their ResFinder 4.1 tool (https://cge.cbs.dtu.dk/services/ResFinder/, accessed on 21 October 2021), with a similarity of 90% and minimum length of 60%. The relations of the antimicrobial resistance genes to the mobile genetic elements were evaluated using the mobile element finder v1.0.3 tool provided by the Center for Genomic Epidemiology (https://cge.cbs.dtu.dk/services/MobileElementFinder/, accessed on 30 November 2021).

Phylogenetic Analysis
The phylogenetic relationship of the strains was assessed by a core genome multilocus sequence typing (cgMLST) analysis. The core genome of all the E. coli strains was described using REALPHY 1.12 [42]. Specifically, one of the E. coli strains was randomly selected as reference, and the genome sequences of all the other strains were then mapped to the reference genome to identify their core genome. A core genome phylogenetic tree was constructed using RAxML Version 8.2.4 with GTRGAMMA option and E. albertii was used as the outgroup [43]. The phylogenetic tree was visualized using the Interactive Tree of Life (iTOL) [44,45].