Introduction

Even though many countries have food safety guidelines and foodborne disease regulations, large-scale foodborne disease outbreaks continue to increase worldwide (Lee & Yoon, 2021). Foodborne diseases can be divided into two categories, infection or poisoning, which can be further categorized according to etiological agents such as viruses, bacteria, parasites, and chemicals (Bari & Yeasmin, 2018). Among these agents, pathogenic Escherichia coli is one of the most common cause of foodborne disease outbreaks reported in the United States, Japan, and South Korea (Dewey-Mattia et al., 2018; Lee & Yoon, 2021). The Ministry of Food and Drug Safety of South Korea estimates that 5,582 outbreaks of foodborne-related illness occurred in South Korea from 2002 to 2022, resulting in 139,630 patients suffering from foodborne diseases. Norovirus (29.4%) and pathogenic E. coli (22.2%) are two of the most frequently reported etiological agents associated with foodborne disease outbreaks in South Korea (https://www.foodsafetykorea.go.kr/). The most serious outbreak of E. coli-caused foodborne illness was reported in Germany in 2011, which subsequently spread to many other countries, resulting in 3,816 identified cases of enterohemorrhagic E. coli infection and 54 associated deaths worldwide (Buchholz et al., 2011; Frank et al., 2011).

Pathogenic E. coli are categorized into the five pathotypes based on virulence factors, toxin production, host cell attachment, and invasiveness; enteropathogenic E. coli (EPEC), enterohemorrhagic E. coli (EHEC), enteroinvasive E. coli (EIEC), enterotoxigenic E. coli (ETEC), and enteroaggregative E. coli (EAEC) (Yang et al., 2017). Previous studies in South Korea, Qatar, and Iran have reported that most pathogenic E. coli strains from foodborne disease outbreaks are resistant to third-generation cephalosporins (Aminshahidi et al., 2017; Eltai et al., 2020; Kim et al., 2014). Although antibiotics remain the most commonly used treatment against pathogenic E. coli in clinical cases, excessive use of these antimicrobials has increased the emergence of multidrug-resistant strains (Pasberg-Gauhl, 2014). According to the Centers for Disease Control and Prevention, over 2.8 million individuals become infected with antibiotic-resistant pathogens each year in the United States, and over 35,000 die as a consequence of these infections (CDC, 2019). The emergence and rapid spread of E. coli carrying genes encoding extended-spectrum β-lactamases (ESBLs) or carbapenemases are considered urgent health problems worldwide (Zhang et al., 2018). ESBLs are a group of plasmid-encoded enzymes that hydrolyze β-lactams, including penicillin, cephalosporins, and monobactam, but not carbapenem (Park et al. 2019; Rawat & Nair 2010). Derivatives of ESBL genes including blaTEM (named after patient Temoniera), blaSHV (sulfhydryl variable active site), blaCTX (cefotaximase)-M, and blaCMY (cephalomycinase) are known and their distribution is geographically different (Pitout & Laupland, 2008; Hawkey & Jones, 2009; Lalak et al., 2016; Ghenea et al., 2022). CTX-M-type ESBLs have spread rapidly among E. coli strains, with the blaCTX-M-15 gene being identified as the predominant type (Carvalho et al., 2021).

In this study, to investigate the distribution of ESBLs and their genomic diversity in ESBL-producing pathogenic E. coli strains isolated from patients with food poisoning, we examined the resistance of 80 ESBL-producing pathogenic E. coli strains to various antibiotics and analyzed their genomic diversities by whole genome sequencing (WGS). This information will help us to better understand the molecular epidemiology of antimicrobial-resistant pathogenic E. coli isolated from foodborne disease outbreaks.

Materials and methods

Isolation and identification of ESBL-producing pathogenic E. coli

A total of 495 pathogenic E. coli strains, including 80 ESBL-producing pathogenic E. coli, were isolated from the diarrhea patients (one sample per person) by standard rectal swab sample collection method in Gyeonggi-do, South Korea, by the Research Institute of Health & Environment from 2014 to 2018. General information on 80 ESBL-producing pathogenic E. coli strains was listed in Table S1. For isolation of pathogenic E. coli, each swab sample was incubated at 37 °C for 16 h in tryptone soy broth (TSB; Oxoid, Basingstoke, UK). An aliquot of each TSB enrichment was streaked on MacConkey Agar (Oxoid, Basingstoke,UK) and incubated at 37 °C for 24 h. The pink single colonies were sub-cultures on tryptic soy agar (TSA; Oxoid, Basingstoke,UK) and confirmed using the VITEK 2 system with gram-negative (GN) identification card (bioMérieux, Marcy, France). For screening of ESBL-producing pathogenic E. coli strains, identified E. coli strains were cultured on CHROMagar ESBL (CHROMagar, Paris, France) at 37 °C for 24 h. The dark pink to reddish single colonies were sub-cultured on TSA (Oxoid). ESBL production was confirmed using VITEK 2 system with AST-N169 card (bioMérieux).

Pathotype determination of pathogenic E. coli

The presence of virulence genes associated with pathogenic properties of E. coli were determined by PCR amplification. The isolates were cultured in TSB at 37 °C for 24 h, and their genomic DNA was extracted from overnight cultures of pathogenic E. coli isolates using the Nextractor NX-48 system and NX-48 bacterial DNA kits (Genolution Inc., Seoul, Korea). PCR was performed using PowerChek™ Diarrheal E. coli 8-plex Detection Kit (Kogenbiotech, Seoul, Korea) according to the manufacturer’s instructions. The amplified virulence genes of pathogenic E. coli strains determined ETEC (st and lt encoding heat-stable and heat-labile enterotoxins), EHEC (VT1 and VT2 encoding verocytotoxin 1 and 2), EPEC (eaeA encoding intimin), EAEC (aggR encoding transcription regulator for aggregative adherence fimbria I), and EIEC (ipaH encoding invasion plasmid antigen H).

Antimicrobial susceptibility tests

The antimicrobial susceptibility tests were performed using the VITEK 2 system with AST-N169 card (bioMérieux) according to the manufacturer’s instructions. Escherichia coli ATCC 25922 was used as a control strain. The AST-N169 card were used for 17 antibiotics tests; ampicillin (AMP), amoxicillin/clavulanic acid (AMC), ampicillin/sulbactam (SAM), cefalotin (CEF), cefazolin (CFZ), cefotetan (CTT), cefoxitin (FOX), cefotaxime (CTX), ceftriaxone (AXO), imipenem (IMI), amikacin (AMK), gentamicin (GEN), nalidixic acid (NAL), ciprofloxacin (CIP), tetracycline (TET), chloramphenicol (CHL), trimethoprim/sulfamethocxazole (SXT). The interpretation of resistant (R), intermediate (I) and sensitive (S) was based on the criteria issued by the Clinical and Laboratory Standards Institute (CLSI, 2021). The minimum inhibitory concentrations (MIC) of ciprofloxacin, tetracycline, chloramphenicol and trimethoprim/sulfamethocxazole were determined using the E-test method (bioMerieux Inc.). Fluoroquinolone, tetracyclines, phenicol, and folate pathway inhibitor MIC values were confirmed according to Clinical and Laboratory Standards Institute guidelines (CLSI, 2021) and European Committee on Antimicrobial Susceptibility Testing guidelines (EUCAST, 2022).

Whole genome sequencing

All ESBL-producing pathogenic E. coli strains (n = 80) were subjected to whole genome sequencing. Genomic DNA of the isolates was extracted a NucleoSpin Microbial DNA kit (Macherey–Nagel, USA) and TissueLyser II (Qiagen, Germany) by following the manufacturer’s instructions. The quality of genomic DNA was determined by NanoDrop spectrophotometer (Thermo-Fisher Scientific, USA), standard agarose gel electrophoresis, and the Qubit 3.0 fluorometer (Thermo-Fisher Scientific). Intact genomic DNA was sheared by a Covaris S220 ultra sonicator (Covaris, USA), and the sequencing library was constructed using the Illumina TruSeq Nano DNA LT library prep kit (Illumina, USA) according to the TruSeq Nano DNA Library Preparation protocol. The quality of the libraries was assessed on a 2100 Bioanalyzer System with DNA 1000 Chip (Agilent Technologies, USA). Sequencing was performed using the NextSeq 500 Sequencing System (Illumina, USA) to generate 2 × 150 bp read length. The contigs of genomic sequences were de novo assembled using CLC Genomics Workbench v20 (Qiagen, USA) with default parameters.

Bioinformatics analysis

To analyze draft genome of each strain, annotation of the assembled genomes was performed using Prokka Web-based tool v1.12 (https://github.com/tseemann/prokka). STs were determined using MLST 2.0 (https://cge.food.dtu.dk/services/MLST/) with 7 housekeeping E. coli genes (adk, fumC, gyrB, icd, mdh, purA, and recA) (Larsen et al., 2012). Antibiotics Resistance Genes (ARGs) were identified using the ResFinder 4.1 database (https://cge.food.dtu.dk/services/ResFinder/) (Bortolaia et al., 2020). Phylogroup was assigned with ClermonTyping (http://clermontyping.iame-research.center/). The E. coli strains were divided into seven main phylogroups termed A, B1, B2, C, D, E and F (Beghain et al., 2018). To specify the degree of overall relatedness among genomes, we estimated the genome-wide ANI using FastANI v1.33. ANI analysis estimates the average nucleotide identity of all orthologous genes shared between any two genomes. Organisms belonging to the same species typically exhibit ≥ 95% ANI (Goris et al., 2007; Jain et al., 2018). Pairwise ANI values were visualized using a heat map generated by ComplexHeatmap v2.2.0 and gplots v3.3.5 in R, dividing the strains into four phylogenetic clusters.

Results and discussion

Isolation of ESBL-producing pathogenic E. coli

A total of 495 pathogenic E. coli strains were isolated from 1,901 clinical specimens obtained from diarrhea patients of foodborne outbreaks in Gyeonggi-do, South Korea from 2014 to 2018. The presence of virulence genes associated with pathogenesis of E. coli was determined by PCR amplification and they could be classified into four pathotypes based on the presence of virulence genes; 254 (51.3%) isolates were ETEC having st/lt, 126 (25.5%) isolates were EAEC having aggR, 111 (22.4%) isolates were EPEC having eaeA, and 4 (0.8%) isolates were EHEC having vt1/vt2. However, EIEC having an ipaH was not detected (Table 1). ETEC and EAEC were identified as the two most frequently isolated E. coli pathotypes, consistent with the previous reports from Iran (Alizade et al., 2019) and Bangladesh (Fahim et al., 2018). The hybrid strains of pathogenic E. coli, which contain multiple virulence genes that may confer higher virulence, have recently been reported worldwide (Santos et al., 2020). However, no hybrid strains were detected in the present study. Antibiotics resistance of the 495 pathogenic E. coli isolates were tested by CHROMagar ESBL selective medium and 80 isolates (16.2%) showed resistance to β-lactam antibiotics. The pathotypes of these 80 β-lactam-resistant isolates were 44 ETEC, 26 EAEC, and 10 EPEC. But none of them were EHEC or EIEC (Table 1). The prevalence of ESBL-producing pathogenic E. coli in this study was similar to the previously published data from South Korea and China (Song et al., 2009; Xu et al., 2018). These results highlight the diverse distribution of pathogenic E. coli pathotypes in diarrhea patients in Gyeonggi-do, South Korea. Moreover, the emergence of antibiotic resistance, particularly the 16.2% resistance to ESBLs, emphasizes the need for continuous monitoring and surveillance.

Table 1 Distribution of pathotypes among pathogenic E. coli and ESBL-producing pathogenic E. coli isolated from foodborne disease patients
Full size table

Antimicrobial susceptibility of the ESBL-producing E. coli

The antimicrobial susceptibilities of 80 ESBL-producing pathogenic E. coli isolates were tested with 17 antibiotics as described in the Materials and Methods and analyzed the results according to CLSI criteria. All tested isolates showed resistance or intermediate resistance to penicillins (e.g., ampicillin), first-generation cephalosporins (e.g., cefalotin and cefazolin), and third-generation cephalosporins (e.g., cefotaxime and ceftriaxone), whereas only four isolates showed resistance or intermediate resistance to second-generation cephalosporins (e.g., cefotetan and cefoxitin). All ESBL-producing pathogenic E. coli isolates were susceptible to carbapenems (e.g., imipenem) (Table S2). While bacteria with resistance to three or more categories of antibiotics were defined as multidrug resistance (MDR) (Nath et al., 2020; Magiorakos et al., 2012), all ESBL-producing pathogenic E. coli isolates examined in this study showed multidrug resistance to five or more antibiotics. Interestingly, 32 out of 44 of the ETEC strains (72.7%, 32/44) showed resistance to five antibiotics with an AMP-CEF-CFZ-CTX-AXO pattern and 12 EAEC strains (46.2%, 12/26) showed resistance to eight antibiotics with an AMP-SAM-CEF-CFZ-CTX-TET-CHL-SXT pattern (Table 2). Remarkably, all isolates displayed multidrug resistance to five or more antibiotics, and EAEC/ETEC pathotype-specific resistance was observed.

Table 2 Multidrug resistance patterns of 80 ESBL-producing pathogenic E. coli
Full size table

Genetic diversity of ESBL-producing pathogenic E. coli

To analyze genetic diversity of ESBL-producing pathogenic E. coli, whole genome sequencing (WGS) and comparative phylogenetic analysis were performed. For this study, multilocus sequencing typing (MLST), phylogroup analysis, and average nucleotide identity (ANI) analysis based on WGS data were conducted. Bioinformatic analysis of the genomes of 80 ESBL-producing pathogenic E. coli strains revealed that they have an average genome size of 5,346,277 bp, with 1,450 coding sequences (CDSs), seven rRNA genes, and 83 tRNA genes. MLST analysis of 80 ESBL-producing pathogenic E. coli strains using seven house-keeping genes showed that they could be classified into 12 MLST sequence types, including sequence type (ST)4, ST34, ST189, ST301, ST382, ST414, ST517, ST1491, ST2040, ST2178, ST4069, and ST6272. However, four strains could not be classified into any sequence type. Most E. coli pathotypes were classified into specific MLST sequence types with exception of 4 strains: most EAEC isolates (n = 26) were ST414 (84.6%, 22/26), most ETEC isolates (n = 44) were either ST2040 (54.5%, 24/44) or ST1491 (40.9%, 18/44), but only 40% (4/10) of EPEC isolates (n = 10) were ST517 (Table 3). The Clermont phylo-typing analysis revealed that phylogroup A (61%, 49/80) and D (28%, 22/80) were the predominant phylogroup in 80 ESBL-producing pathogenic E. coli strains. Phylogroup A and D of E. coli have been reported to be the most prevalent in patients with diarrhea in India, Sweden, and Spain (Ljungquist et al., 2020; Modgil et al., 2020; Valverde et al., 2009). Phylogroups also showed an association with E. coli pathotypes. All ETEC isolates belong to phylogroup A (100%, 44/44), most EAEC isolates belong to phylogroup D (85%, 22/26), and EPEC isolates belong to phylogroups A/B1 (40%, 4/10) respectively (Table 3). The ANI phylogenetic trees of 80 ESBL-producing pathogenic E. coli strains showed that two phylogenetic groups were distinctly separated below the 95% ANI threshold: A and B groups (A, 25/80 and B, 55/80). The A group strains were divided into A1 (2 EPEC strains) and A2 (23 EAEC strains), and B group strains were divided into B1 (1 EAEC, 4 EPEC, and 20 ETEC strains) and B2 (2 EAEC, 4 EPEC, and 24 ETEC strains) (Fig. 1). The analysis of genetic diversity in 80 ESBL-producing pathogenic E. coli strains revealed a strong phylogenetic association among different E. coli pathotypes.

Table 3 Distribution of E. coli pathotypes in the MLSTs and phylogroups of 80 ESBL-producing pathogenic E. coli strains
Full size table
Fig. 1
figure 1

ANI phylogenetic tree of 80 ESBL-producing pathogenic E. coli clinical isolates from foodborne disease in Gyeonggi-do. ESBL-producing pathogenic E. coli strains are grouped into two major groups (A and B). The information about pathotypes, MLST, and phylogroups for 80 E. coli isolates is indicated on the left side

Full size image

Analysis of antimicrobial resistance genes of ESBL-producing pathogenic E. coli

Analysis of the bla genes of 80 ESBL-producing pathogenic E. coli isolates revealed following families of β-lactamases: blaCTX-M (− 14, 15, 27, 55, and 65), blaCMY (− 2, 5, 55, 60, 61, 130 and 153), and blaTEM (− 1B, 30, 99, 122, 141, 163, 164, 206, and 207). All these isolates, except one, harbored the blaCTX-M gene alone or in combination with either blaTEM gene or blaCMY gene, suggesting that the blaCTX-M gene is the most prevalent among pathogenic E. coli from foodborne illnesses (98.8%, 79/80). Prevalence of the blaCTX-M variants were as follows; 47 isolates have blaCTX-M-15 (44 ETEC and 3 EPEC), 27 isolates have blaCTX-M-14 (25 EAEC and 2 EPEC), 4 isolates have blaCTX-M-55 (1 EAEC and 3 EPEC), 2 isolates have blaCTX-M-27 (1 EAEC and 1 EPEC), and 3 isolates have blaCTX-M-65 (1 EAEC and 2 EPEC). In addition, one EPEC strain was found to have both blaCTX-M-14 and blaCTX-M-15 genes while one EAEC strain have blaCTX-M-14, blaCTX-M-55 and blaCTX-M-65 genes (Fig. 2). It is interesting to note that E. coli strains in the ANI-A2 group showed a significantly higher proportion of ST414 and phylogroup D, with blaCTX-M-14 being predominant in this group. ST1491 and ST2040 were highly prevalent in the ANI-B group (B1/B2) and in phylogroup A, and blaCTX-M-15 was widespread in this group. Phylogenetic studies have demonstrated that ST1491 and ST2040 are closely related. EAEC strains were associated with ANI-A group and ETEC strains were related to the ANI-B group. These results suggest that each E. coli pathotype has distinctive phylogenetic background.

Fig. 2
figure 2

Distribution of antimicrobial resistance (AMR) genes in ESBL-producing pathogenic E. coli isolates. The orange color indicates the presence of β-lactamase genes and blue color indicates the presence of other antimicrobial resistance genes based on ResFinder

Full size image

The CTX-M-type ESBLs in E. coli are the most predominant and rapidly disseminated type in both humans and animals (Smet et al., 2010; Carvalho et al., 2021). Among several CTX-M variants, CTX-M-9 type ESBL was dominant at the end of the 1990s, but appeared to be rapidly displaced since 2011 by CTX-M-15 and CTX-M-14 types. (Cantón & Coque, 2006). The blaCTX-M-15 has been identified as the most common ESBL gene in environment, livestock, and human-associated E. coli (Cantón & Coque, 2006; Zurfluh et al., 2015). The CTX-M-15- and CTX-M-14-type ESBLs (92.5%) were the most predominant ESBLs in the present study, which is similar to the reports from Asia, Africa, Europe, America, Australia, and South Korea (Chen et al., 2014; Iroha et al., 2012; Livermore et al., 2007; Pietsch et al., 2017; Sidjabat et al., 2010; Song et al., 2009).

Co-harboring of extended-spectrum β-lactamase encoding genes and other antimicrobial resistance genes

It is well known that most of the ESBL-producing pathogenic E. coli carry additional antibiotics resistant genes (Park et al., 2022). Here, all ETEC and EPEC strains harboring the blaCTX-M gene also carried plasmid-mediated quinolone resistance (PMQR), namely qnrS1 (98.0%, 48/49) and qnrS2 (2.0%, 1/49) (Fig. 2). Several previous studies have reported co-existence of PMQR and ESBL genes in E. coli (Azargun et al., 2018; Nazik et al., 2011 Viana et al., 2013). The qnr genes were highly specific to ETEC strains in this study (Fig. 2). Previous studies have also reported the co-existence of qnr and blaCTX-M genes in ESBL-producing ETEC, which may be because both genes encoding enterotoxins (LT/ST) and PMQR are plasmid-borne (Gyles et al., 1974; Jiang et al., 2008). However, further research is needed to understand these results better. Likewise, a high prevalence of qnr genes among ESBL-producing E. coli has been previously reported in South Korea (Park et al., 2007). In addition, most of the EAEC strains harboring the blaCTX-M-14 gene also carried the chloramphenicol resistance gene catA1 (92%, 23/25), folate pathway antagonist gene sul1/dfrA5 (88%, 22/25), and tetracycline gene tet(A) (92%, 23/25) (Fig. 2). To confirm the antibiotic resistance attributable to the co-existence of antimicrobial resistance genes, the minimal inhibitory concentrations (MICs) of ciprofloxacin, tetracycline, chloramphenicol, and trimethoprim/sulfamethocxazole were determined (Table 4). The MICs of tetracycline, chloramphenicol, and trimethoprim/sulfamethoxazole were higher than the breakpoints of CLSI and EUCAST guidelines (Table S3) in most of the CTX-M-14 type ESBL-producing EAEC strains carrying catA1, sul1, dfrA5, and tet(A). This suggests that the phenotype was consistent with the genotype in these strains.

Table 4 Antimicrobial susceptibility of representative ESBL-producing pathogenic E. coli
Full size table

Although E. coli is a common commensal gut bacterium, our findings emphasize the high prevalence of ESBL and PMQR gene in pathogenic E. coli strains isolated from diarrhea patients of foodborne disease outbreaks. These results suggest that the prevalence of ESBL-producing pathogenic E. coli strains may become an important public health concern, highlighting the urgent necessity for monitoring the spread of these foodborne pathogens.