Colistin Resistance and ESBL Production in Salmonella and Escherichia coli from Pigs and Pork in the Thailand, Cambodia, Lao PDR, and Myanmar Border Area

The study aimed to examine the prevalence and genetic characteristics of ESBL-production and colistin resistance in Salmonella and Escherichia coli from pigs and pork in the border area among Thailand, Cambodia, Lao PDR, and Myanmar. Salmonella (n = 463) and E. coli (n = 767) isolates were collected from pig rectal swab from slaughterhouses (n = 441) and pork from retail markets (n = 368) during October 2017 and March 2018. All were determined for susceptibility to colistin and cephalosporins, ESBL production and mcr and ESBL genes. Salmonella was predominantly found in Cambodia (65.8%). Serovars Rissen (35.6%) and Anatum (15.3%) were the most common. The E. coli prevalence in pork was above 91% in all countries. Colistin-resistance rate in E. coli (10.4%) was significantly higher than Salmonella (2.6%). ESBL-producing Salmonella (1.9%) and E. coli (6.3%) were detected. The blaCTX-M-55 and blaCTX-M-14 were identified. The mcr-1 gene was detected in Salmonella (n = 12) and E. coli (n = 68). The mcr-1/blaCTX-M-55 and mcr-3/blaCTX-M-55 co-concurrence was observed in one Salmonella and three E. coli isolates, respectively. In conclusion, pigs and pork serve as carriers of colistin and new generation cephalosporins resistance. Testing for resistance to last line antibiotics should be included in national AMR surveillance program using One Health approach.


Introduction
Multidrug resistance (MDR) in bacterial pathogens has escalated and become a significant cause of morbidity and mortality worldwide [1]. Resistance to last resort antibiotics (e.g., colistin and newer generation cephalosporins) is of particular concern because it may limit antibiotic therapy for bacterial infections in the near future [2,3].
Cephalosporins (3rd, 4th, and 5th generations) are used for disease treatment in humans and animals [4] and categorized as critically important antimicrobials in both human [5] and veterinary medicine [6]. Over recent years, resistance to cephalosporins have been increasingly reported in humans and animals [7,8] and role of livestock as important sources of extended-spectrum β-lactamase (ESBL) producers potentially spread to humans through the food chain was highlighted [9]. Resistance to cephalosporins is mainly due to the production of ESBLs that confer resistance to most β-lactams, including new generation cephalosporins and monobactams (e.g., aztreonam), but not cephamycin (e.g., cefoxitin) and carbapenems (e.g., imipenem) [10]. ESBL-encoding genes are primarily located on plasmid that can be horizontally transferred to inter and intra bacterial species. The genes frequently co-exist with other resistance genes on the same large plasmids, resulting in MDR phenotype of ESBL producers [9,11]. Colistin or polymyxin E has a narrow antibacterial spectrum mainly against Gram-negative bacteria [12]. Colistin is widely used in pig production for treatment and control of Enterobacteriaceae infections. Colistin-resistant E. coli carrying plasmid-borne mcr-1 were isolated from food animals in China in 2016 [13]. Since then, several mcr variants have been identified in livestock in many parts of the world, including countries with zero to marginal use of colistin (e.g., the United States) [14]. The mcr variants can occur due to transference of multiple mobile genetic elements. The co-occurrence of variants of mcr genes and other antimicrobial resistance (AMR) genes can increase the widespread of colistin resistance and pose a global public health risk [15].The border-crossing areas between Thailand and its land neighboring countries including Cambodia, Lao PDR and Myanmar has flourished and become a major route for cross-border trade and tourism owing to their geographical advantages situated along the Mekong river [16]. Purchasing and selling live animals and retail meat, especially pigs and pig products, is one of the high-trade volume and value. The cross-border trade has been emphasized on improvement of their commerce, investment, and economic cooperation [16]. This raises a particular concern over the role of food animals and products as reservoirs of diseases and AMR bacteria that possibly spread to other regions and countries.
Currently, knowledge on resistance to new generation cephalosporins and colistin in bacteria from livestock and products remains largely undiscovered in most cross-border regions globally. This study aimed to examine the prevalence and genetic characteristics of ESBL-production and colistin resistance in Salmonella and E. coli isolated from pigs and pork in the border area between Thailand, Cambodia, Lao PDR, and Myanmar.

E. coli Prevalence in Pork
The overall percentage of E. coli in pork from the markets was 44.7% (343/767). The high prevalence (>88%) was observed in pork in all countries. The high recovery rate (>94%) for commensal E. coli was observed for rectal swab sample, indicating the high competence of E. coli isolation method used ( Table 1). Subtotal of E. coli was not identified in Table 1 to avoid the comparison among the sampling sources since commensal E. coli regularly present in rectal swab. Seven hundred and sixty-seven isolates were collected from Thailand (n = 368), Cambodia (n = 136), Laos (n = 133), and Myanmar (n = 130) for further experiments (Table 1).
The bla CTX-M55 gene in one ESBL-producing Salmonella from Thai pig and seven ESBLproducing E. coli from Cambodia (one from pig and one from pork) and Myanmar (three from pigs and two from pork) was conjugally transferred.

Discussion
A major finding in this study was the high prevalence of E. coli contamination in pork in retail markets, in agreement with a previous study [17]. Concurrently, the Salmonella contamination rate was approximately 1.9 times lower than that of E. coli. These observations emphasize that legal framework of microbiological quality of raw meat must be effectively implemented for food safety in retail market facilities among border provinces. In general, retail meat is delivered in open buckets and sold in open-air under ambient temperature. Unsold meat is stored overnight in an icebox for re-selling in the following day. These unhygienic practices contribute to the high bacterial contamination observed. Importantly, Salmonella contamination increased at retail markets, suggesting cross-contamination during slaughtering process or transportation or from environment at retail markets. However, the samples were not collected immediately after slaughtering and before transporting to retail markets, therefore, the hypothesis cannot be scrutinized. Together with the high E. coli contamination, it indicates that fresh meat at retail markets serve as a major source of foodborne pathogens and highlights the need for the effective-hygienic practice along food production chain.
Thailand has launched law and regulations to contain Salmonella in poultry since 2010 [18]. The law includes the enforcement of sample collection for Salmonella detection in broiler, laying hens and breeders. Such law has not been specifically issued for pig farm. However, the microbiological quality of meat is regulated under the Microbiological Quality Criteria for Food and Food Contact Containers No. 3 (2017), indicating that Salmonella must not be present in 25 g of fresh and frozen raw meat (either pork or chicken meat) intended for human consumption [19]. Therefore, good manufacturing practice is required at both farm and slaughtering process levels to produce Salmonella free raw meat.
In this study, the most common Salmonella serovars were Rissen and Anatum, in agreement with a previous study [20]. Salmonella Enteritidis was not observed in this study. This maybe because the major reservoir of Salmonella Enteritidis is not in swine production in agreement of previous study [17]. ESBL-producing Salmonella and E. coli were present at lower rate when compared to previous studies in Thailand, [21,22] and other countries e.g., Denmark [23], Czech [24], Nigeria [25], and Germany [26]. The differences could be associated with many factors (e.g., sampling period, geographical variations and different antimicrobials usage). This study covered the border provinces of Thailand and 3 neighboring countries (i.e., Cambodia, Lao PDR and Myanmar). However, when compared to our previous study conducted in Thailand-Cambodia border province, the prevalence of ESBL-positive Salmonella (2.2%) in this study was higher than that observed (1.7%) in the same area [27]. Due to high cost of cephalosporins, use of these antimicrobials in livestock production will increase investment cost and reduce profits. Therefore, cephalosporins may not be commonly used in food-animals in the study areas. In addition, distribution of ESBL producers is not always a direct effect of cephalosporin usage [28] and could be due to coselection of ESBL genes by other antibiotics, e.g., ampicillins, aminoglycosides, tetracyclines that are commonly used in swine production in Thailand [29].
Most ESBL-producing Salmonella (8%) and E. coli (54%) carried bla TEM-1 and almost all carried CTX-M group 1 and 9, in agreement with a previous study showing that bla CTX-M-positive E. coli isolates usually non ESBL bla TEM-1 [30]. This was also supported by the statistical significance of association between ESBL production and bla TEM-1 (p < 0.05) observed. In addition, not all ESBL producers in this study carried the ESBL genes tested, indicating the existence of other resistance mechanisms and/or ESBL genes that were not included in this study.
A previous study in the same region demonstrated that CTX-M β-lactamases were common among ESBL-producing isolates in this study [31]. This was in agreement with the current study where bla CTX-M-55 of CTX-M-1 and bla CTX-M-14 of CTX-M-9 were predominant. Several studies previously reported the common presence of bla CTX-M-1 in food animals in Switzerland, Tunisia, the United Kingdom, and Germany [9,10,32,33]. The bla CTX-M55 genes were located on conjugative plasmids, in agreement with our previous studies [27] and supporting its distribution in this area. Horizontal transfer of bla CTX-M-55 was under selective pressure of ampicillin in conjugation experiment, highlighting the role of the old generation antibiotic that has been widely used for a long time as a selective pressure for the new generation of clinically important antibiotics. In addition, the bla CTX-M-14 gene was most frequently found in ESBL-producing E. coli, in agreement with food animals in China [34] and Denmark [23].
In Thailand, the use of medicated feed containing polymyxins, cephalosporins, fluoroquinolones and others is regulated by the law on "Characteristics and conditions of animal feed containing drugs prohibited from producing, importing, selling and using" that was issued in 2018 [35]. The regulation of antimicrobials drugs (including polymyxins B, colistin and other drugs in penicillin and fluroquinolone groups) that must not be mixed in animal feed for prophylactic purposes was later announced [36]. The use of colistin for short-term treatment can be performed in swine production [37]. Colistin-resistant E. coli was found in pigs and pork, reflecting the use of colistin in pig production in the region. Interestingly, colistin-resistant Salmonella was limited to pork. This is rather a result of cross contamination during slaughtering, transportation, and in retail markets.
Most colistin-resistant Salmonella and E. coli harbored either mcr-1 or mcr-3, supporting the significance of mcr genes on spread of colistin resistance worldwide. However, some colistin-resistant isolates lacked both genes, indicating the existence of other colistinresistance mechanisms not characterized in this study. Colistin-resistant Salmonella and E. coli harboring both mcr-1 and mcr-3 were isolated, in agreement with previous studies [38]. Their colistin MICs were 4 and 8 µg/mL (data not shown) and not different from those with mcr-1 and mcr-3 alone, suggesting the variable effect of mcr genes on colistin-resistance level. The common presence of mcr-3 (either alone or with mcr-1) has warned about the possibility of its global spread as observed for mcr-1. None of mcr-2 was observed, in agreement with a previous study [39].
The findings of Salmonella with mcr-1/bla CTX-M-55 (n = 1) and E. coli with mcr-3/bla CTX-M55 (n = 3) are of particular concern. This is consistent with previous studies [40,41]. Despite the low percentage, the co-occurrence generates impact on human and veterinary medicine due to the potential contribution to the spread of bacteria with clinically important antimicrobial resistance, resulting in limited choice for bacterial infection therapy. Resistance genes in these groups are plasmid borne and potentially distribute to bacteria in humans, other animals and environment, so implementation of One Health approach is needed to alleviate the AMR issue.

Sampling Location and Sample Collection
The sampling area are boundary provinces along the border between Thailand and Cambodia (Sa Kaeo and Banteay Meanchey), Thailand and Lao PDR (Nong Khai and Vientiane), and Thailand and Myanmar (Chiang Rai and Tarchileik) (Figure 1). The samples were collected during 2017-2018. All pig samples were obtained from one municipal pig slaughterhouse and one municipal fresh market in each province.

Isolation and Identification of Salmonella and E. coli
Salmonella was isolated following ISO 6579:2002(E) [42] and subjected to serotyping by slide agglutination based on the Kaufman-White scheme [43]. A single colony of each serotype was collected from each positive sample. E. coli was isolated and biochemically confirmed [44,45]. One isolate was collected from each positive sample and all were stored as 20% glycerol stock at −80 °C.

Determination of Minimum Inhibitory Concentration (MICs) of Colistin
All the Salmonella (n = 463) and E. coli (n = 767) isolates were tested for susceptibility to colistin by MIC determination using two-fold agar dilution method [46]. The colistin concentrations ranged from 0.0625 to 64 µ g/mL. The clinical breakpoint for defining col-

Isolation and Identification of Salmonella and E. coli
Salmonella was isolated following ISO 6579:2002(E) [42] and subjected to serotyping by slide agglutination based on the Kaufman-White scheme [43]. A single colony of each serotype was collected from each positive sample. E. coli was isolated and biochemically confirmed [44,45]. One isolate was collected from each positive sample and all were stored as 20% glycerol stock at −80 °C.

Determination of Minimum Inhibitory Concentration (MICs) of Colistin
All the Salmonella (n = 463) and E. coli (n = 767) isolates were tested for susceptibility to colistin by MIC determination using two-fold agar dilution method [46]. The colistin concentrations ranged from 0.0625 to 64 µ g/mL. The clinical breakpoint for defining col-

Isolation and Identification of Salmonella and E. coli
Salmonella was isolated following ISO 6579:2002(E) [42] and subjected to serotyping by slide agglutination based on the Kaufman-White scheme [43]. A single colony of each serotype was collected from each positive sample. E. coli was isolated and biochemically confirmed [44,45]. One isolate was collected from each positive sample and all were stored as 20% glycerol stock at −80 °C.

Determination of Minimum Inhibitory Concentration (MICs) of Colistin
All the Salmonella (n = 463) and E. coli (n = 767) isolates were tested for susceptibility to colistin by MIC determination using two-fold agar dilution method [46]. The colistin concentrations ranged from 0.0625 to 64 µ g/mL. The clinical breakpoint for defining col- All pigs were from commercial production farms that provided meat for domestic consumption. In Thailand, the pig slaughterhouse in Nong Khai was a large-scale facility with greater than 80 pigs were slaughtered each day, while slaughterhouses in Sa Kaeo and Chiang Rai provinces were small-scale modern facilities with no more than 50 pigs were processed daily. The pig slaughterhouses in Banteay Meanchey province, Cambodia and from Tachileik, Myanmar were traditionally small slaughterhouses with a throughput of 30 or fewer pigs per day. In Lao PDR, the pig slaughterhouse in Vientiane was large-scale modern facility with a throughput of greater than 200 pigs per day. A total of 809 samples were collected by rectal swab from pigs in slaughterhouses (n = 441) and carcass swab in retail markets (n = 368) from Thailand (n = 380), Cambodia (n = 149), Lao PDR (n = 140) and Myanmar (n = 140) ( Table 1).
At slaughterhouses, rectal swab samples were obtained by rectal evacuation after bleeding but prior to scalding. At fresh markets, the pig carcass swab was obtained by swabbing an area of at least 50 cm 2 on each fresh carcass. The sample collection was performed three times at each sampling site. All samples were stored in an icebox and transported within 24 h after collection for immediate processing or held in refrigerator at 4 • C until exanimation 48 h of collection.

Isolation and Identification of Salmonella and E. coli
Salmonella was isolated following ISO 6579:2002(E) [42] and subjected to serotyping by slide agglutination based on the Kaufman-White scheme [43]. A single colony of each serotype was collected from each positive sample. E. coli was isolated and biochemically confirmed [44,45]. One isolate was collected from each positive sample and all were stored as 20% glycerol stock at −80 • C.

Determination of Minimum Inhibitory Concentration (MICs) of Colistin
All the Salmonella (n = 463) and E. coli (n = 767) isolates were tested for susceptibility to colistin by MIC determination using two-fold agar dilution method [46]. The colistin concentrations ranged from 0.0625 to 64 µg/mL. The clinical breakpoint for defining colistin resistance was 2 µg/mL [47]. Multidrug resistance (MDR) was defined as being resistant to at least three different classes of antibiotics.

Determination of ESBL Production
ESBL production was initially screened and confirmed using disk diffusion method in all the Salmonella and E. coli isolates [46]. Initial screening was performed using ceftazidime (30 ug), cefotaxime (30 ug) and cefpodoxime (10 ug) (Oxoid, Hampshire, England). All the isolates resistant to at least one of the indicator cephalosporins were subjected to confirm ESBLs production by using combination disk diffusion method with ceftazidime (30 µg) and cefotaxime (30 µg) alone and in combination with clavulanic acid (10 µg) (Oxoid). A difference of ≥5 mm between the inhibition zone of the cephalosporin/clavulanic acid combination and corresponding cephalosporin disks alone was interpreted as positive ESBL phenotype E. coli ATCC 25922 was used as a quality control strain.

PCR and Nucleotide Sequencing
PCR-template DNA was prepared by whole cell boiled lysate procedure [48]. All the primers used are listed in Table 5. All PCR reactions were performed using GeNei™mastermix (Merck, Munich, Germany). PCR amplicons were gel purified using Nucleospin ® Gel (Macherey-Nagel, Düren, Germany) and submitted for sequencing using PCR primers at First Base Laboratories (Selangor Darul Ehsan, Malaysia). The obtained DNA sequence was BLAST compared with GenBank database (http://www.ncbi.nlm.nih.gov/ BLAST accessed on 24 May 2021).

Statistical Analysis
The significance differences between the prevalence of Salmonella and E. coli and the occurrence of AMR stratified by population, location, and sample type were determined using Pearson's Chi-square or Fisher's exact test. The statistical analyses were performed with IBM SPSS Statistics for Windows, Version 18.0 (SPSS Inc., Chicago, IL, USA). Logistic regression analysis was used to measure the strength of association between phenotype and genotype of resistance (STATA SE12, Stata Corp LLC.: College Station, TX, USA). A p-value of <0.05 was considered statistically significant. Odds ratio and 95% confidence interval (CI) were calculated.

Conclusions
Healthy pigs can serve as reservoirs for colistin-resistant and ESBL-producing Salmonella and E. coli that may constitute a public health threat. It highlights the need for surveillance program as well as control and prevention strategic plan for AMR in bacteria of food animal origin at national and regional levels. Standard hygiene and sanitation practices should be enhanced in retail markets to minimize the cross contamination. The epidemiological data will help to understand the root cause of AMR. Guidance of the interventions and evaluation of the success of the inventions from farm to fork is essential for the region.