Plasmid-mediated colistin resistance and ESBL production in Escherichia coli from clinically healthy and sick pigs

This study aimed to determine the percentage of colistin resistant and ESBL-producing Escherichia coli from clinically sick and healthy pigs and understand the molecular mechanisms underlying colistin resistance and ESBL production. A total of 454 E. coli isolates from healthy pigs (n = 354; piglets, n = 83; fattening pigs, n = 142 and sows, n = 100) and sick pigs (n = 100) were examined for antimicrobial susceptibility, chromosomal and plasmid-mediated colistin resistance mechanisms and ESBL genes. The healthy (41%) and sick pig (73%) isolates were commonly resistant to colistin. Three mcr genes including mcr-1 (10.4%), mcr-2 (1.1%) and mcr-3 (45%) were detected, of which mcr-3 was most frequently detected in the healthy (33%) and sick pig (57%) isolates. Coexistence of mcr-1/mcr-3 and mcr-2/mcr-3 was observed in piglets (23%), fattening pig (3.5%) and sick pig (13%) isolates. Three amino acid substitutions including E106A and G144S in PmrA and V161G in PmrB were observed only in colistin-resistant isolates carrying mcr-3. The percentage of ESBL-producing E. coli was significantly higher in the sick pigs (44%) than the healthy pigs (19.2%) (P = 0.00). The blaCTX-M group was most prevalent (98.5%), of which blaCTX-M-14 (54.5%) and blaCTX-M-55 (42.9%) were predominant. The blaTEM-1 (68.8%) and blaCMY-2 (6.3%) genes were identified in ESBL-producers. All ESBL producers were multidrug resistant and the majority from piglets (97%), fattening pigs (77.3%) and sick pigs (82%) carried mcr gene (s). ESBL producers from piglets (n = 5) and sick pig (n = 1) simultaneously transferred blaTEM-1 (or blaCTX-M-55) and mcr-3 to Salmonella. In conclusion, pigs are important reservoirs of colistin-resistant E. coli that also produced ESBLs, highlighting the need for prudent and effective use of antimicrobials in pigs and other food-producing animals.

In recent times antimicrobial resistance (AMR) has rapidly increased and become one of the greatest threats to public health globally. The highest-increasing rates of AMR have been reported in low and middle-income countries, especially those in Southeast Asia 1 . Extensive use of antimicrobials in either human medicine or animal farming is considered a major contributor to emergence and spread of AMR 2 . In livestock production, the purposes of antimicrobials are either to treat infections, control or promote growth 3 . Different countries have different policies and regulations with respect to antibiotic growth promoter (AGP). For example, Thailand phased in AGP ban in 2011 and implemented total ban in 2015 4 . The US FDA prohibited the use of medically important antibiotics for AGP in 2017 but not for non-medically important ones 5 . Consumer's demand for livestock products has risen globally and is effectively driving antimicrobial consumption in food animals to maintain animal health and increase productivity. Some of these actions are consequently resulting in increasing levels of AMR 1 . The emergence of multi-drug resistant E. coli has been frequently reported not only in clinical medicine but also in livestock production. Particular concern has been raised to the dissemination of E. coli resistant to clinically important antibiotics (i.e. colistin, new generation cephalosporins and carbapenems) that may diminish antibiotics of choice for infection treatment in the near future.

Materials and methods
Bacterial isolates. A total of 454 E. coli isolates were obtained from two bacterial culture stocks isolated between 2007 and 2018 as described below. All E. coli strains were isolated by using standard method as previously described 20 . One E. coli colony from each positive sample was collected and stored in 20% glycerol at -80 °C.
Isolates from healthy pigs. Isolates were obtained from the bacterial stock of Department of Veterinary Public Health, Faculty of Veterinary Science, Chulalongkorn University (n = 354). These isolates originated from fecal samples collected from clinically healthy pigs, confirmed by farm veterinarians, piglets at 4-8 weeks of age (n = 83), fattening pigs at 9-18 weeks of age (n = 142) and sow at 37-45 weeks of age (n = 129) between 2007 and 2018 as part of our AMR studies. A yearly distribution of the isolates is shown in Fig. 1. The samples originated from farms located in Central and Northeast Thailand including Aungthong, Chachoengsao, Chonburi, Kanchanaburi, Ratchaburi, Suphanburi, Nakhonratchasima, Burirum, and Udonthani regions. These provinces have high densities of pig population, with farm sizes varying from small scale (51-500 pigs) to large scale (> 5000 pigs). Faecal samples were randomly collected from pigs of different age (one sample from one pig) by farm veterinarians. One isolate from each group of pigs at different age in each farm was used for antimicrobial susceptibility testing.
Isolates from sick pigs. The isolates were obtained from the strain collection of Department of Veterinary Medicine, Faculty of Veterinary Science, Chulalongkorn University (n = 100). All were isolated from fecal swab samples routinely collected from sick pigs at 2-21 weeks old displaying clinical signs of diarrhea during 2011-2018 ( Fig. 1). Farm veterinarians collected and submitted samples for clinical diagnosis at Veterinary Diagnostic Laboratory (VDL), Livestock Animal Hospital, the Nakornpathom campus. The farms from which these samples were obtained were located in Central (i.e. Nakornpathom, Saraburi and Suphanburi), Eastern (i.e. Chachoengsao and Chonburi), Western (i.e. Kanchanaburi and Ratchaburi) and Southern (i.e. Trang) regions of Thailand. Antibiotic use history was not available. Detection of ESBL production was performed by the disk diffusion method using antibiotics (quantity of antibiotic, zone diameter breakpoint) as follows: cefotaxime (30 µg, ≤ 27 mm), cefpodoxime (10 µg, ≤ 17 mm) and ceftazidime (30 µg, ≤ 22 mm) 21 . The antibiotic disks were obtained from Oxoid (Oxoid™, Hamshire, England). E. coli ATCC® 25922 and K. pneumoniae ATCC® 700603 served as quality control strains.
The E. coli isolates exhibiting resistance to at least one cephalosporin tested were phenotypically confirmed for ESBL production using the combination disk method including cefotaxime and cefotaxime (30 mg)/clavulanic acid (10 mg), and ceftazidime and ceftazidime (30 mg)/clavulanic acid (10 mg) (Oxoid™, Hamshire, England). The inhibition zone difference of ≥ 5 mm in the combination with clavulanic acid versus the inhibition zone in the cephalosporin alone was interpreted as positive for ESBL production 21 .
DNA isolation, PCR and DNA sequencing analysis. DNA template for PCR was prepared by whole cell boiled lysates as previously described 23 . All PCR amplifications were performed using TopTaq™ Master Mix Kit (QIAGEN, Germantown, MD, USA) according to the manufacturer's instruction. Primers used in this study are listed in Supplementary Table 1. PCR products were separated on 1.5% agarose gel electrophoresis (Sigma-Aldrish®) in 1XTris-acetate/EDTA (TAE) buffer. The gels were stained in RedSafe™ Nucleic Acid Staining Solution (iNtRON Biotechnology, NJ, USA) and visualized using the Omega Fluor™ Gel Documentation System (APLEGEN™ Gel Company, CA, USA). The PCR products were purified using Nucleospin® Gel and PCR clean up (Macherey-Nagel, Düren, Germany) and submitted for DNA sequencing at First Base Laboratories (Selangor Darul Ehsan, Malaysia). The DNA sequences obtained were compared with the reference sequence available at GenBank Database using the Blast algorithm (http:// www. ncbi. nlm. nih. gov).
Detection of mutations in the pmrAB system. Twenty colistin-resistant E. coli isolates from healthy pigs (n = 10) and sick pigs (n = 10) were arbitrarily selected (n = 20) for PCR amplification of pmrA and pmrB genes 9 . Five colistin-susceptible isolates from healthy pigs were included as control. The PCR amplicons were gel purified and submitted for DNA sequencing using PCR primers. DNA sequences were compared to those of E. coli K12 (U00096.2) available at GenBank database.
Conjugation experiments. Biparental filter mating method was performed to test transferability of mcr and ESBL genes. All the E. coli isolates carrying mcr and/or ESBL genes served as donors and the spontaneous rifampicin-resistant Salmonella Enteritidis (SE12Rif R , rifampicin MIC = 256 µg/ml), was used as recipient 30 . The Salmonella transconjugants were confirmed on Xylose Lysine Deoxycholate agar (Difco, MD, USA) containing 32 µg/mL rifampicin and an appropriate antibiotic (i.e. 100 μg/mL ampicillin, or 2 μg/mL colistin). Transfer of mcr and ESBL genes was confirmed by PCR as described above.
Statistical analysis. Comparisons of the association between antimicrobial resistance phenotype and resistance encoding gene were performed by using Pearson's chi-squared test (χ 2 ) (SPSS, version 22.0). A p-value of < 0.05 was considered statistically significant. Odds ratios with 95% confidence intervals (CIs) were calculated.

Results
Antimicrobial susceptibility. Healthy pigs. Overall, 78% of the E. coli isolates from healthy pigs (n = 384) were resistant to at least one antimicrobial agent tested (Table 1). Most isolates from fattening pigs (96.5%, 137/142) and all the isolates from piglets and sows were resistant to at least one antimicrobial agent tested. Concurrently, the majority of the isolates in this study including all piglet isolates, 99.2% of the sow isolates and 94.4% of the fattening pig isolates, were MDR (resistant to at least three antimicrobial agents in different classes). However, there was no significant difference of MDR proportion among the E. coli isolates from different groups of the healthy pigs. Overall, the percentage of colistin-resistant E. coli was 40.7%.
Colistin resistance was predominant among the piglet isolates (95.2%), followed by the isolates from fattening pigs (43.7%) and sows (2.3%). The colistin resistance rate in the piglet isolates was significantly higher than that in the sow and fattening pig isolates (p < 0.05).
Sick pigs. All the E. coli isolates from sick pigs were resistant to at least one antimicrobial agent and up to 99% were MDR. Resistance to colistin was found in 73% of the isolates. Forty-four E. coli isolates (44%) in this group were ESBL-producers, that exhibited resistance to ceftazidime (53%), cefotaxime (53%) and cefpodoxime (37%). The percentage of ESBL-producing E. coli isolates was significantly higher in sick pigs than healthy pigs (p < 0.05). The majority of the sick pig isolates were resistant to tetracycline (100%) and ampicillin (97%) (Fig. 2).

Amino acid alterations in pmrAB.
In comparison to E. coli K12, sequence variations in pmrAB were found in all E. coli tested (n = 25) ( Table 3). Four amino acid substitutions including S29G, E106A, G144S and E184D were identified in PmrA and five amino acid substitutions including H2R, V161G, D283G, Y358N, A360V were detected in PmrB. The amino acid changes S29G in PmrA and H2R, D283G and Y358N in PmrB were found in both colistin-resistant and susceptible E. coli isolates. Among the healthy pig isolates, two mcr-3 carrying isolates (i.e. E.453 and E.454) carried G144S amino acid substitution in PmrA and additionally harbored V161G in PmrB. The colistin MIC of both isolates was 16 µg/mL. One sick pig isolate (i.e. EC.P.45, colistin MIC = 8 µg/mL) carried both mcr-3 and E06A amino acid substitution in PmrA. A colistin-susceptible isolate (i.e. GCa13, colistin MIC = 0.25 µg/mL) harbored E184D amino acid substitution in PmrA that was not observed in any colistin-resistant isolates tested.
Horizontal transfer of β-lactamase genes was observed in 14 E. coli isolates. Five piglet isolates transferred bla TEM-1 and co-transferred mcr-3. One sick pig isolate was capable of transferring bla CTX-M55 and mcr-3 simultaneously. Seven E. coli isolates including 2 isolates from piglets (one isolates with bla CTX-M-14 and the others with bla CTX-M-55 ) and 4 isolates from sick pigs (2 isolates with bla CTX-M-14 and 2 isolates with bla CTX-M-55 ) were able to transfer bla CTX-M . One isolate from sick pig could transfer both bla CTX-M-55 and bla TEM-1 gene at the same time.

Discussion
The present study was conducted in E. coli isolates from clinically healthy and clinically sick pigs collected during the time period 2007-2018. One significant finding was high MDR rates in the isolates from healthy (97.5%) and sick pigs (99%). It is expected that only healthy pigs are slaughtered for human consumption, but their health status does not guarantee the absence of resistant bacteria. This is because antibiotics may be previously   www.nature.com/scientificreports/ administered to the pigs that were the source of the isolates for disease prevention and growth promotion, which may have resulted in commensal bacteria developing antibiotic resistance. Similarly, carrying resistant bacteria does not infer having a disease. As the complete ban of AGP in all animal feed was implemented in 2015, use of AGP could influence the high AMR rates observed in earlier years in this study. In consideration of the dynamics of AMR, antibiotic administration and AMR development may not simultaneously occur. At the same time, it is still unclear to what extent antibiotic use must be reduced, and how long the interventions must be made to effectively to reverse the spread of AMR. The highest frequency of resistance among the isolates from healthy and sick pigs was to tetracycline and ampicillin, in agreement with a previous study in E. coli isolated from pig farms in Thailand 31 . However, it was not possible to obtain the antibiotic use history in each farm. The antibiotics are usually administered to piglets by oral route either in feed or in water for controlling gastrointestinal tract infection in piglets including polypeptides (e.g. colistin) and aminoglycosides (e.g. apramycin). Tylosin, tilmicosin and chlortetracycline were used in fattening pigs. Cephalosporins (e.g. ceftiofur and ceftriaxone) are occasionally used for treatment of respiratory diseases, lameness, and reproductive infections. It was estimated that approximately 39.7% of medicated feed was used in suckling and nursery pigs followed by fattening pigs (37.3%) and breeding pig (23%) in Thailand 7 . Some antibiotics mixed in medicated feed used in pig production in the country are included in WHO list of Critically Important Antimicrobials for Human Medicine e.g. amoxicillin, colistin and lincomycin 7,32 . Up to date, Thailand has launched law and regulations to contain AMR associated with food animals, for example, Notification of the Ministry of Agriculture and Cooperatives that specifically prohibits the use of all antibiotics in animal feed as growth promoters was released in 2015 4 . Law on "Characteristics and conditions of animal feed containing drugs prohibited from producing, importing, selling and using" was issued in 2018, of which medicated feed containing polymyxin B, cephalosporins, fluoroquinolones and others are covered by this law 33 .
A year later, regulation of antimicrobial drugs that must not be mixed in animal feed for prophylactic purposes was announced 34 . The latter included polymyxins B, colistin and other drugs in penicillin and fluroquinolone groups. Effective enforcement of these regulations is expected and the outcomes of implementation may be seen through national AMR surveillance data in coming years.
A note could be made for ciprofloxacin resistance (35.9% for the healthy pig isolates and 87% for the sick pig isolates) that was defined by clinical breakpoints used. Being ciprofloxacin susceptible does not always warrant being wildtype lacking alterations in target fluoroquinolone genes. The different contribution of a certain amino acid substitution to fluroquinolone resistance was previously suggested 35 . However, interactions with target site mutations and ciprofloxacin resistance level were not pursued in this study.
Isolates from healthy (40.7%) and sick pigs (73%) exhibited a high colistin resistance rate that was higher than in a previous study conducted in E. coli from healthy and diseased pigs in Japan between 2012-2013 14 . The highest colistin resistance rate was found in the isolates from piglets (95.2%), followed by sick pigs (77%). This is likely because colistin has been used for treatment of gastrointestinal tract infections caused by E. coli, especially in post-weaning diarrhea in piglets 7 . Colistin was commonly formulated into medicated feed for suckling and nursery pigs for the prevention of gastrointestinal tract infection in Thailand 7 . Approximately 40 tons of colistin were mixed in medicated feed and about 87.2% were intended for piglets in Thai pig production 7 . Such extensive use of colistin may contribute to high colistin resistance rate observed in the present study. Implementation to minimize use and encourage prudent use of colistin and other antimicrobials is mandatory. In addition, the colistin resistance rates in the piglets (95.2%) and sow (2.3%) isolates were quite different. Piglets usually acquire intestinal flora including E. coli from the mother at birth and therefore, the similar resistance rates are expected in the piglets and sow isolates. The discrepancy observed in this study could be attributed to different pattern of colistin administration to pigs at the different stages. Colistin is most often administered orally in medicated feed or individually by a feeding bottle to suckling piglets and nursery pigs to treat post-waning diarrhea (PWD) and colibacillosis 7 . However, this is not the case for sows. Another explanation could be involved in the sources of the isolates, of which the piglet and sow isolates were obtained from several studies in different years and from different pig farms with different pattern of antimicrobial usage.
Chromosomal mutations in the two-component regulatory system of PmrAB were previously shown to be significantly associated with colistin resistance in bacterial pathogens such as Klebsiella pneumoniae and Salmonella enterica, Acinetobacter baumannii and Pseudomonas aeruginosa 36 . However, mutations in PmrAB is rarely reported in E. coli. A previous study demonstrated amino acid substitutions S39I and R81S in PmrA and V161G in PmrB in colistin-resistant E.coli isolates from pigs in Spain 9 . However, the S39I and R81S amino acid substitutions in PmrA were not found in this study. In addition to mobile colistin resistance (mcr) genes, research studies focusing on the chromosomal-mediated colistin resistance and their regulatory mechanism have increased 36 . Some mutations (i.e. E106A and G144S in PmrA and V161G in PmrB) were observed only in colistin-resistant isolates carrying mcr-3 in this study. However, individual contribution and cumulative effects of the genes to colistin resistance was not determined and needs further investigations. At the same time, some amino acid changes (e.g. S29G in PmrA and D283G, Y358N and H2R in PmrB) were identified in both colistinresistant and colistin-susceptible isolates, suggesting the lack of impact on colistin resistance phenotype. Studies of other TCSs and their regulators such as PhoPQ, MgrB, and PmrD are suggested 36 .
In this study, mcr-3 was most predominant among the E. coli isolates from both healthy pigs (32.5%) and sick pigs (57%), while the lower percentage of mcr-1 was observed in healthy pigs (7.6%) and in sick pigs (20%). These results are inconsistent to a previous study reporting that mcr-1 was commonly detected in E. coli from healthy and diseased pigs in Japan (45%) and mcr-3 was found at lower rate (8.3%) in diseased pigs 14 . The discrepancies may be due to difference in antimicrobial usage patterns or in the prevalence of different clones and/or plasmids.
The mcr-1 gene is globally distributed and has been found in many bacterial species (e.g. E. coli, Salmonella spp., Klebsiella spp. and Pseudomonas spp.) from food animals, food stuff and human 8  However, mcr-3 appear to be common among healthy and sick pigs in this study. Further studies in different animal sources and other countries should be conducted to determine the role of this gene in the dissemination of colistin resistance. Moreover, a previous study showed that mcr-3 was commonly located on broad-host range plasmids (i.e. IncP) and several transposases and IS elements (i.e. IS4321, ΔTnAs2 and ISKpn40) were identified in the flanking regions of mcr-3. This might cause wider spread and stronger transmission capabilities of mcr-3 than mcr-1 37 . Further genetic characterization of mcr-3 carrying plasmid are needed to elucidate molecular mechanisms underlying dissemination of this gene.
The mcr-2 positive isolates were detected (n = 5) in fattening pigs. The mcr-2 gene was previously reported in colistin-resistant E. coli from pigs in Belgium (20.8%) 10 and China (56.3%) 13 . None of the isolates in this study carried mcr-4. Up to date, the report of mcr-4 has been limited to EU countries including Salmonella from pigs in Italy and E. coli from pigs in Spain and Belgium 11 . These variations suggest that spread and evolution of mcr genes should be monitored.
Coexistence of different mcr variants was observed, including mcr-1/mcr-3 (23% of piglets and 13% of sick pigs) and mcr-2/mcr-3 (3.5% of fattening pigs). The E. coli carrying mcr-1/mcr-3 were previously isolated from cattle in Spain, pig and poultry in China and humans in New Zealand 13,15 . The isolates carrying both mcr-1 and mcr-2 were previously identified in pigs in Canada 38 . By considering the colistin MIC, all mcr-1 harboring isolates exhibited resistance to colistin (colistin MIC 4-64 µg/mL). However, mcr-3 can be found in colistin susceptible strains (colistin MIC 0.5-2 µg/mL), in agreement with a previous study 14 . In addition, all the E. coli isolates harboring more than one mcr genes had colistin MIC of 4 or 8 µg/mL. Taken together, the observations indicate that the number of mcr derivatives is not always related to colistin resistance level. As the contribution of individual mcr genes, especially mcr-3, to colistin resistance level remains to be elucidated, monitoring mcr variants should be conducted in colistin-susceptible and resistant strains.
The ESBL E. coli of healthy pig origin (19.2%) in this study was less common than that in a previous report in the isolates obtained during 2012-2013 in the same country 39 . The presence of ESBL producers in sick pigs (44%) was significantly higher than that in healthy pigs (p < 0.05). Among the healthy pigs, the highest percentage of ESBL producers was observed in piglets (45.8%) (p < 0.05). This is presumably associated with the common use of β-lactam antibiotics (e.g. amoxicillin and third-generation cephalosporins) in the suckling period for treatment of respiratory disease as suggested by a study of antimicrobial use in pigs in Germany 40 . The percentage of ESBL-producing E. coli in sick pigs (44%) was significantly higher than that in healthy pigs (19.2%) (p < 0.05). This may be a result of antibiotics previously administered to treat sick pigs. Cephalosporins are generally more expensive than other antimicrobial agents and may not be commonly used in pig production in Thailand and other countries in South East Asia. Currently, cephalosporins are increasingly used in pig production due to its long-lasting potency and lower doses. However, the presence of ESBLs may be also a result of other antimicrobial usage. This is because ESBL genes commonly colocalize on the same plasmid as other resistance genes.
The bla CTX-M gene was the most prevalent ESBL gene in this study, in agreement with previous reports in Thailand 41 17 . Previous studies reported that bla CTX-M-55 was the major CTX-M subtype in ESBL-E. coli isolates from clinical isolates, food animals, farm waste and canals in Thailand 41 . The gene was predominant in E. coli from livestock and pets in other Asian countries e.g. China and Hong Kong 42 . The bla CTX-M55 gene was also detected in countries outside Asia but to less extent.
The β-lactamase gene, bla TEM-1 (72.3%) was commonly identified in this study. The gene has been frequently detected in the E. coli isolates from animals and is commonly co-harbored with ESBL genes 26 . This is in agreement with the current study where most ESBL producers (67.9%) carried TEM-1 and ESBL genes. The bla CMY-2 gene was detected at low frequency (5.4%). The gene was firstly identified in K. pneumoniae from human isolates and is increasingly reported in different bacteria from livestock e.g. E. coli from ground chicken and pig feces in Taiwan 43 , and E. coli from healthy chicken and sick animals in Spain 44 , in agreement with this study. In addition, the isolates carrying bla CMY-2 coharbored bla CTX-M-55 and bla CTX-M-14 , in agreement with previous studies 45 .
Most ESBL producers from piglets (97%), fattening pigs (77.3%) and sick pigs (82%) additionally carried mcr genes, of which the most common mcr gene among ESBL producers was mcr-3. However, a previous study in China showed that mcr-1 was more commonly found in ESBL E. coli than non ESBL producers 17 . β-lactams and colistin are bactericidal antibiotics that disrupt the outer membrane of bacterial cells. Recruiting mcr genes in the cell is a survival mechanism to maintain the cell wall integrity and may contribute to the increasing prevalence of ESBL producers coharboring mcr genes 17 . In addition, all ESBL-mcr carrying isolates were MDR, in agreement with a previous study 17 . These results highlight the continued need to encourage the prudent and effective use of antimicrobials in food animal production.
By using ampicillin as selectable marker, co-transfer of β-lactamase genes (bla TEM-1 and bla CTX-M55 ) and mcr gene (mcr-3) was detected, suggesting co-resistance of the gene on the same plasmid. This also suggest that distribution of mcr and ESBLs genes can be a result of co-selection by antibiotics in other classes.
In this study, the strength of the association between AMR phenotype and genotype was quantified. Strong positive correlation suggests possible genetic linkage of AMR genes, e.g., co-localization on the same plasmid. However, a wide confidence interval (CIs) was observed and likely due to a small sample size or variability of the study group. The significant association between AMR phenotype and genotype was observed. Positive Scientific Reports | (2022) 12:2466 | https://doi.org/10.1038/s41598-022-06415-0 www.nature.com/scientificreports/ associations were identified between phenotypic resistance to CIP-CTX-M-14, STR-mcr-1/ CTX-M-55, SULmcr-2/TEM-1/CTX-M-55, TET-mcr-2/TEM-1/CTX-M-14/CTX-M-55 and TMP-mcr-1/mcr-2/CTX-M14. This could be possible due to co-localization of multiple resistance genes on the same plasmid. The strongest association was observed between tetracycline resistance and bla CTX-M-55 (OR = 31) or mcr-2 (OR = 9.95), in agreement with previous studies 19 . The results emphasize that emergence and spread of AMR is a dynamic issue and selective pressure of resistance to various antimicrobials are linked. Therefore, regulation of antimicrobial use should be conducted using a whole-system approach, not at individual drug level.
In conclusion, the findings emphasize the role of commensal and pathogenic E. coli as an important reservoir of ESBL and mcr genes encoding resistance to the highest priority critically important antimicrobials (HP-CIAs). Horizontal transfer of the genes indicates their significance as a global health risk. The use of ampicillin could select for colistin resistance, confirming that the pandemic spread of mcr genes can be a result of co-selection by other antimicrobial classes. Coexistence of genes encoding resistance to multiple clinically important antimicrobials raises a particular concern of future challenges for infection treatment options in either veterinary or human medicine. Therefore, prudent and responsible use of antibiotics in food animal production should be encouraged and whole-system approach to optimize antimicrobial uses is required. Detection of ESBL production and colistin resistance at phenotypic and genotypic level should be included in national AMR surveillance program to allow epidemiological tracing of resistance trend. Further studies to characterize E. coli carrying different mcr genes and plasmid backbones of ESBL and mcr genes are warranted.