Whole-Genome Sequence Analysis of Colistin-Resistant, mcr-Harboring Escherichia coli Isolated from a Swine Slaughterhouse in Thailand

Abstract

The emergence of colistin-resistant Escherichia coli (E. coli) in livestock poses a major public health concern due to its zoonotic potential and multidrug resistance (MDR). The study aimed to detect mobilized colistin resistance genes (mcr-1 to mcr-9) in E. coli isolates, along with characterizing their antimicrobial susceptibility, resistance genes, virulence genes, and whole genome sequencing. We investigated E. coli contamination in a swine slaughterhouse in Nakhon Si Thammarat Province, Thailand. A total of 200 fecal samples were collected and screened for E. coli using selective media supplemented with colistin. A total of 200 fecal samples were collected from a swine slaughterhouse and screened for E. coli using selective media supplemented with colistin. Presumptive E. coli isolates were confirmed by PCR, followed by molecular detection of mcr-1 to mcr-9 genes. Antimicrobial susceptibility testing was performed using the disk diffusion method. Selected isolates were further analyzed for additional antimicrobial resistance genes and virulence associated genes by PCR. Whole genome sequencing was conducted on representative isolates exhibiting high levels of antimicrobial resistance. Our results showed that out of 200 fecal samples, 124 presumptive E. coli isolates were recovered from a swine slaughterhouse using selective media containing colistin. PCR confirmation identified 112 isolates (90.32%) as E. coli. Molecular detection of mobilized colistin resistance (mcr) genes (82 isolates, 73.21%) demonstrated that mcr-1 (50.89%) was the most prevalent, followed by mcr-9 (25.89%) and mcr-3 (24.11%). Overall, the 82 mcr E. coli isolates showed the highest level of resistance to ampicillin (97.56%), followed by tetracycline (95.12%), piperacillin (73.17%), and chloramphenicol (65.85%). For non-mcr E. coli isolates, the highest resistance percentage was observed for ampicillin (96.67%), followed by piperacillin (80%) and tetracycline (73.33%). Among the isolates, 75% exhibited MDR phenotypes, showing 22 distinct resistance profiles. The most common MDR pattern was AMP-PIP-TE-C-S (12.5%). Additional antimicrobial resistance genes, including aadA, ampC, and bla_TEM, were detected in over 60% of a subset of 30 tested isolates. The virulence gene analysis revealed that eae (74.10%), associated with enteropathogenic E. coli (EPEC), was the predominant pathotype. Whole genome sequencing of five selected isolates confirmed the presence of multiple antimicrobial resistance and virulence determinants. In conclusion, this study reveals a high prevalence of MDR E. coli harboring colistin resistance genes (mcr-1 to mcr-9) in a swine slaughterhouse in southern Thailand. The findings highlight the potential risk of zoonotic transmission of antimicrobial resistant E. coli through the food production chain and emphasize the importance of continuous genomic surveillance and prudent antimicrobial use in livestock production systems.

1. Introduction

Escherichia coli is recognized as a major foodborne infection, affecting millions of individuals annually [1,2]. The food involved in primary transmission is livestock products, including raw beef, pork, and chicken meat [3,4,5]. Existing research highlights that microbial contamination predominantly occurs throughout the production chain, from farm to fork, with a significant prevalence during the slaughtering processes [6,7]. Several pathogenic E. coli have been identified, based on bacterial genetics and associated pathologies, including Enteropathogenic E. coli (EPEC), Shiga toxin-producing E. coli (STEC), Enterohemorrhagic E. coli (EHEC), Enterotoxigenic E. coli (ETEC), Enteroinvasive E. coli (EIEC), Enteroaggregative E. coli (EAEC), Diffuse adhering E. coli (DAEC), and Adherent-invasive E. coli (AIEC). The pathogens cause various intestinal diseases, such as watery or bloody diarrhea, hemolytic uremic syndrome (HUS), and colitis [8]. In swine production, colistin, an antimicrobial medicine, has been excessively used in the treatment of E. coli infection or colibacillosis, as well as a growth promoter, contributing to the emergence of antimicrobial resistance (AMR) [9,10] and the spread of zoonotic colistin-resistant strains [11].

Swine farms have been reported as major reservoirs for mobile colistin resistance (mcr) genes, as mcr-positive E. coli have been identified from swine farms in Central China [12]. Moreover, colistin-resistant E. coli has isolated from swine and pork across Southeast Asian countries, including Thailand, the Lao PDR, and Cambodia [4,11]. This highlights the emerging concern of this issue since the detection of colistin-resistant E. coli in swine and pork reflects the intensive use of colistin in swine production [13]. The occurrence of mcr-encoded colistin resistance in E. coli has not only been documented in swine but also in farm workers, suggesting potential zoonotic transmission between animals and humans [14]. The persistence of colistin-resistant bacteria harboring plasmid-mediated mcr genes is concerning, as they may cause serious nosocomial infections in human hospitals [10,15]. Moreover, colistin, a priority antimicrobial (HPCIA) drug, should be reserved as a last-resort treatment for severe infections caused by Gram-negative multidrug-resistant (MDR) bacteria in humans [16]. This means that bacterial infections have become more difficult to treat effectively.

Transportation immediately prior to slaughter is a major stressor in livestock and can trigger neuroendocrine activation, including increased cortisol release, which may impair immune competence and compromise gastrointestinal barrier function [17,18]. These stress-related physiological changes can disrupt the gut microbial ecosystem, leading to microbial imbalance and facilitating the expansion of enteric pathogens [19]. Such alterations have been linked to elevated intestinal pathogen burdens and increased fecal shedding, notably Salmonella, thereby raising the risk of environmental dissemination and carcass contamination [7,18,20]. In parallel, antimicrobial resistance (AMR) poses a growing public health and economic burden in Thailand, including increased hospitalization, treatment costs, and mortality, underscoring the importance of AMR surveillance and integrated prevention and control strategies under the One Health concept [21,22,23]. Given the global rise in AMR-associated morbidity and mortality [24,25], this study aimed to characterize antimicrobial resistance phenotypes and determine the prevalence of mobile colistin resistance (mcr) among E. coli isolates recovered from swine feces at a swine slaughterhouse in Southern Thailand. We further screened for resistance genes relevant to antimicrobials commonly monitored in pork production chains, including phenicols, aminoglycosides, β-lactams, and tetracyclines [26], and performed whole-genome sequencing of five selected isolates for in-depth molecular characterization.

2. Results

2.1. Prevalence of Mcr Escherichia coli from a Swine Slaughterhouse

In a total of 200 feces samples, 124 isolates were selected, based on lactose fermentation on MacConkey agar containing 4 mg/L of colistin and the presence of colonies with a metallic sheen color on EMB agar, which were predicted to be colistin-resistant E. coli. Among the isolates, 112 (90.32%) were confirmed as E. coli by the detection of the uidA gene using PCR analysis. For colistin resistance genes mcr-1 to 9, there were 82 isolates of mcr E. coli from the 112 isolates (73.21%). The prevalence of each mcr gene (mcr-1 to mcr-9) is shown in

Table 1. Prevalence of mcr E. coli isolated from a swine slaughterhouse (N = 112).

mcr Gene	No. of E. coli Isolates
mcr-1	57 (50.89%)
mcr-2	0 (0%)
mcr-3	27 (24.11%)
mcr-4	1 (0.89%)
mcr-5	0 (0%)
mcr-7	1 (0.89%)
mcr-8	2 (1.79%)
mcr-9	29 (25.89%)
mcr-1 and mcr-3	14 (12.50%)
mcr-1 and mcr-9	10 (8.93%)
mcr-3 and mcr-7	1 (0.89%)
mcr-3 and mcr-8	1 (0.89%)
mcr-3 and mcr-9	4 (3.57%)
mcr-7 and mcr-8	1 (0.89%)
mcr-1, mcr-3, and mcr-9	3 (2.68%)
mcr-3, mcr-7, and mcr-8	1 (0.89%)

The prevalence of mcr-1 E. coli was 50.89% (57/112), followed by 25.89% (29/112) of mcr-9, and 24.10% (27/112) of mcr-3, respectively. The prevalences of mcr-8 E. coli, mcr-4 E. coli, and mcr-7 E. coli were 1.78% (2/112), 0.89% (1/112), and 0.89% (1/112), respectively. There were no mcr-2 E. coli or mcr-5 E. coli. Among the 82 mcr-positive isolates, the most common combination detected was mcr-1 and mcr-3 (12.50%), followed by mcr-1 and mcr-9 (8.93%). We found less frequency of E. coli mcr genes, such as mcr-3 with mcr-7, mcr-8, or mcr-9 (0.89–3.57%), as well as the rare coexistences of three genes: mcr-1, mcr-3, and mcr-9 (2.68%); and mcr-3, mcr-7, and mcr-8 (0.89%) (Table 1).

2.2. Antibiotic Susceptibility of E. coli

The antimicrobial susceptibility of the 112 E. coli isolates was tested against ten antimicrobial drugs. The percentages of antimicrobial resistance in the mcr E. coli isolates (n = 82) and non-mcr E. coli isolates (n = 30) are demonstrated in Figure 1A and 1B, respectively. Both the mcr E. coli and non-mcr E. coli groups showed no resistance (0%) to amoxicillin-clavulanic (AMC) acid. For the mcr E. coli group, 97.56 percent of the isolates showed a resistance to ampicillin (AMP), followed by tetracycline (TE, 95.12%), piperacillin (PIP, 73.17%), chloramphenicol (C, 65.85%), streptomycin (S, 40.24%), gentamicin (GEN, 26.83%), enrofloxacin (ENR, 10.98%), ceftriaxone (CRO, 3.66%), and amikacin (AMK, 2.44%) (Figure 1A).

For non-mcr E. coli isolates, the highest resistance percentage was observed for AMP (96.67%), followed by PIP (80.00%), TE (73.33%), S (56.67%), C (56.67%), GEN (33.33%), ENR (10%), and there was no resistance (0%) to CRO and AMK (Figure 1B). Overall, mcr-1-positive isolates exhibited the broadest resistance spectrum, showing high resistance rates to AMP (68%), PIP (50%), TE (70%), C (51%), and S (28%). Moderate resistances to CRO (4%) and ENR (9%) were also observed. In contrast, mcr-3 and mcr-9 isolates demonstrated intermediate resistance profiles, particularly to AMP (32–35%), PIP (22–28%), TE (29–34%), and C (20–23%). Other mcr variants (mcr-2, mcr-4, mcr-5, mcr-7, and mcr-8) showed minimal or no resistance to the tested antibiotics. These findings indicated that mcr-1 is predominantly associated with multidrug resistance, whereas other mcr types tended to confer narrower resistance patterns (Figure 2). Fisher’s Exact test indicated a significant association between the antimicrobial resistance percentages and the percentages of positive mcr gene variants (p = 0.002).

2.3. MDR Patterns

The AMR patterns in mcr-positive E. coli and non-mcr-positive E. coli were presented in

Table 2. Antimicrobial resistance patterns of E. coli isolates (n = 112) based on disk diffusion results.

Pattern	Profile	No. of Drug Classes	No. mcr Isolates		No. of Non-mcr Isolates		Total
1	AMP-ENR-PIP-TE-C-CN-S-AK	5	1	0.89%	0	0%	0.89%
2	AMP-ENR-PIP-TE-C-CN-S	5	2	1.78	0	0	1.78
3	AMP-ENR-PIP-TE-C-CN	5	1	0.89	0	0	0.89
4	AMP-CRO-PIP-TE-C-CN-S	5	1	0.89	0	0	0.89
5	AMP-CRO-PIP-TE-C-CN	5	2	1.78	0	0	1.78
6	AMP-ENR-PIP-TE-C	4	1	0.89	2	1.78	2.67
7	AMP-PIP-TE-C-CN-S	4	5	4.46	2	1.78	6.25
8	AMP-PIP-TE-C-CN	4	1	0.89	2	1.78	2.67
9	AMP-PIP-TE-C-S	4	10	8.92	4	3.57	12.5
10	AMP-TE-C-CN	4	4	3.57	0	0	3.57
11	AMP-TE-C-S	4	1	0.89	2	1.78	2.67
12	AMP-ENR-PIP-TE	3	1	0.89	0	0	0.89
13	AMP-ENR-PIP-C	3	0	0	1	0.89	0.89
14	AMP-ENR-TE-C	3	1	0.89	0	0	0.89
15	AMP-PIP-TE-C	3	11	9.82	1	0.89	10.71
16	AMP-PIP-TE-CN	3	2	1.78	2	1.78	3.57
17	AMP-PIP-TE-S	3	5	4.46	2	1.78	6.25
18	AMP-PIP-C-S	3	1	0.89	0	0	0.89
19	AMP-PIP-C-CN-S	3	0	0	2	1.78	1.78
20	AMP-TE-C	3	9	8.03	1	0.89	8.921
21	AMP-TE-CN	3	1	0.89	0	0	0.89
22	AMP-TE-S	3	0	0	3	2.67	2.67
	Total		60	53.57%	24	21.42	75%

AK—amikacin; AMP—ampicillin; C—chloramphenicol; ENR—enrofloxacin, CN—gentamicin, PIP—piperacillin, S—streptomycin, CRO—ceftriaxone.

. The results showed that 60 of the mcr isolates (53.57%) and 24 of the non-mcr isolates (21.42%) were resistant to at least three different antimicrobial classes, indicating multidrug resistance (MDR). The most prevalent AMR pattern (12.5%) was AMP-PIP-TE-C-S, which was found in both mcr-positive isolates (8.92%) and non-mcr E. coli (3.57%). The second-most prevalent AMR pattern was AMP-PIP-TE-C (10.71%), a vast majority of which (9.82%) was found in mcr-positive E. coli. Similarly, a large majority of the third-most prevalent AMR pattern, which was AMP-TE-C, was found in mcr-positive E. coli (8.03%). Interestingly, there were seven mcr-positive E. coli isolates (6.25%) that showed resistant to five antimicrobial drug classes, while there was no isolate from non-mcr E. coli in this category. Moreover, the AMR patterns that were only found in the mcr-positive E. coli group were AMP-TE-C-CN, AMP-ENR-PIP-TE, AMP-ENR-TE-C, AMP-PIP-C-S, and AMP-TE-CN, which accounted for 3.57%, 0.89%, 0.89%, 0.89%, and 0.89% prevalence, respectively.

2.4. Detection of Antimicrobial Resistance Gene

The distribution of antimicrobial resistance (AMR) genes among the studied isolates showed a significant deviation from a uniform expectation (χ² = 162.04, df = 12, p < 0.05), indicating a substantial heterogeneity in resistance gene prevalence. Among β-lactamase genes, ampC and bla_TEM were detected at the highest frequencies, each present in 93.3% of isolates (95% CI: 0.787–0.982), whereas bla_SHV and bla_OXA were not detected in any isolate. For tetracycline resistance, tetA and tetB were observed in 26.7% (95% CI: 0.142–0.444) and 20.0% (95% CI: 0.095–0.373) of isolates, respectively, while tetC was absent. Within the phenicol class, cmlA was highly prevalent (73.3%; 95% CI: 0.556–0.858), whereas cat1 was not detected. Aminoglycoside resistance determinants were also common, with aadA detected in 86.7% (95% CI: 0.703–0.947) and aphA1 in 56.7% (95% CI: 0.392–0.726) of isolates. In contrast, aad3I was absent, and aad3IV was observed at a low frequency (3.3%; 95% CI: 0.006–0.167) (

Table 3. Antimicrobial resistance gene profiles of E. coli isolates (n = 30).

Class	Gene	Positive (n)	Proportion	95% CI
Beta-lactam	ampC	28	0.933	0.787–0.982
	blaTEM	28	0.933	0.787–0.982
	blaSHV	0	0.000	0.000–0.114
	blaOXA	0	0.000	0.000–0.114
Tetracycline	tetA	8	0.267	0.142–0.444
	tetB	6	0.200	0.095–0.373
	tetC	0	0.000	0.000–0.114
Phenicols	cat1	0	0.000	0.000–0.114
	cmlA	22	0.733	0.556–0.858
Aminoglycosides	aadA	26	0.867	0.703–0.947
	aphA1	17	0.567	0.392–0.726
	aad3I	0	0.000	0.000–0.114
	aad3IV	1	0.033	0.006–0.167

).

An assessment of the AMR gene distribution across mcr genotypes revealed distinct co-occurrence patterns (Figure 3). Isolates harboring mcr-1 exhibited the highest level of multidrug resistance gene co-selection, particularly for ampC and bla_TEM (60.0% each), tetA (23.3%), cmlA (46.7%), aadA (20.0%), aphA1 (36.7%), and aad3-IV (3.3%). The mcr-3 genotype also showed moderate AMR gene associations, with ampC and bla_TEM detected in 30.0% of isolates, and several additional determinants, including tetA (6.7%), tetB (10.0%), cmlA (16.7%), aadA (23.3%), aphA1 (23.3%), and aad3-IV (3.3%). In contrast, the isolates carrying mcr-4, mcr-5, mcr-6, and mcr-7 lacked all of the screened β-lactam, tetracycline, phenicol, and aminoglycoside resistance determinants. The mcr-8 genotype displayed a limited co-occurrence, with only low-frequency detection of ampC, tetA, cmlA, and aadA (all 3.3%). Meanwhile, mcr-9 was associated with moderate prevalence of several genes, including ampC (26.7%), bla_TEM (23.3%), cmlA (20.0%), and aphA1 (16.7%). Together, these findings demonstrated that isolates carrying mcr-1 and mcr-3 harbored a wider range of co-occurring AMR genes, whereas other mcr variants showed minimal coselection of resistance determinants. However, this difference was not statistically significant with Fisher’s Exact test (p = 0.955).

2.5. Detection of Virulence Gene in E. coli

We found the differences in the proportion of E. coli pathotypes. A high detection rate was found in Enteropathogenic E. coli (EPEC), which is encoded by the eae and bfp genes. The highest prevalence was in the eae gene, with 83/112 isolates (74.10%). On the other hand, the bfp gene showed the lowest prevalence, with 1/112 isolates (0.89%). Enterotoxigenic E. coli (ETEC) showed low prevalence (less than 2% each). The pathogenic lt (heat-labile toxin) gene and stII (heat-stable toxin) gene were detected in 2/112 isolates (1.78%). Moreover, Enteroinvasive E. coli (EIEC) was encoded by virF and ipaH. The virF gene was detected in 2/112 isolates (1.78%), while the ipaH gene was not detected (0%). Similarly, Shiga Toxin-Producing E. coli (STEC: stx-1 and stx-2) and Enteroaggregative E. coli (EAEC: aafII) were not detected in our study. We also detected the eaeAO157:H7 with 10/112 isolates (8.92%). A Chi-square goodness-of-fit test demonstrated a significant difference in the distribution of virulence genes (χ² = 802.18, p < 0.001), with uidA and eae being the predominant markers (

Table 4. Prevalence of pathotype E. coli from swine slaughterhouse.

Detection	Target Genes	No. of the Isolates	p-Value
E. coli	uidA	112/124 (90.32%)	p < 0.001
STEC	stx-1	0/112 (0%)
	stx-2	0/112 (0%)
EPEC	bfp	1/112 (0.89%)
	eae	83/112 (74.10%)
ETEC	lt	2/112 (1.78%)
	stII	2/112 (1.78%)
EIEC	virF	2/112 (1.78%)
	ipaH	0/112 (0%)
EAEC	aafII	0/112 (0%)
O157:H7	eaeAO157:H7	10/112 (8.92%)

Note: Percentages are calculated from 112 E. coli isolates. Chi-square goodness-of-fit test demonstrated a significant difference in the distribution of virulence genes (χ² = 802.18, p < 0.001), with uidA and eae being the predominant markers.

).

The mcr E. coli isolates were found to harbor virulence genes. The EPEC-type markers dominated the virulence profile of mcr carriers (Figure 4). The eae gene was the most detected gene found in mcr-1 (46.34%), mcr-9 (29.27%), and mcr-3 (23.17%). Also, we sporadically detected mcr-4/7/8 (each 1.22%). However, there was no mcr-2/5. In addition, the bfp gene was rare (≤1.22%, detected only in mcr-3). For STEC markers, the detection rate was low, with stx-2 at 2.44% in mcr-1, at 1.22% in mcr-3, and stx-1 at 0% in mcr1-9.

For ETEC markers, the detection rate was also low, with stII at 1.22% and lt at 0% in mcr-1. The EAEC aafII was not detected. Furthermore, the detection of the O157:H7-specific eaeAO157:H7 gene was low, with 4.88% in mcr-1, 3.66% in mcr-3, and 6.10% in mcr-9. Yet, the EIEC targets (virF, ipaH) were not detected. The relationship between pathotype and mcr E. coli was not significantly different (p = 0.864).

2.6. Genome Characteristics of E. coli Isolates Based on Whole Genome Sequencing

Whole genome sequencing was performed on five representative isolates with the most pronounced antimicrobial resistance phenotypes (EMCR1–EMCR5), generating high-quality assemblies with a deep sequencing coverage (455–584×), ensuring reliable genomic characterization. The total number of raw reads obtained for each isolate ranged from 6.9 to 9.2 million, resulting in assembled genome sizes between 4.49 and 5.18 Mb, which is consistent with the typical genome size of E. coli. Assembly quality metrics also indicated generally good contiguity, with N50 values ranging from 47,716 to 108,063 bp and N75 values from 19,939 to 44,622 bp. The number of contigs varied among isolates (115–362), reflecting differences in assembly fragmentation, likely due to a variation in sequencing depth, read quality, or genome complexity. All genomes had GC contents around 50.5%, which aligned with the expected GC% for E. coli (

Table 5. Genome characteristics and accession numbers of E. coli strains.

ID	Depth of Coverage	Totals Raw Reads (M)	Genome Size (bp)	No. of Contigs	GC (%)	N50 (bp)	N75 (bp)
EMCR1	455	6.895	4,675,856	115	50.59	108,063	44,622
EMCR2	534	9.228	5,181,179	362	50.51	47,716	19,939
EMCR3	584	8.546	4,495,920	149	50.81	72,472	41,866
EMCR4	581	9.057	4,764,291	193	50.61	66,249	35,854
EMCR5	572	8.827	4,744,383	149	50.77	76,012	43,569

).

2.7. Presence of MLST by Whole-Genome Sequencing

Whole genome sequencing based MLST analysis showed that the five E. coli isolates belonged to distinct sequence types, reflecting broad genetic diversity. EMCR1 was classified as ST877, which is associated with the ST86 clonal complex. EMCR2 belonged to ST206 and corresponded to the ST206 clonal complex. EMCR3 was identified as ST542, a sequence type that was not assigned to any currently defined clonal complex. Similarly, EMCR4 was classified as ST2935, which also lacked an established clonal-complex designation. In contrast, EMCR5 belonged to ST48, a member of the widely distributed ST10 clonal complex (Figure 5).

2.8. Virulence Gene Detection by Whole-Genome Sequencing

The WGS based assessment of the virulence genes demonstrated a marked variation in virulence gene profiles among the five E. coli isolates. EMCR1 carried a moderate set of genes, including adhesion-related genes (csgA, fdeC, and lpfA), stress- and survival-associated markers (hlyE, nlpI, ompT, and terC), and the yehA–yehD gene cluster. EMCR2 displayed the most extensive virulence profile, harboring multiple adhesins (AslA, csgA, fdeC, and fimH), enterotoxin and effector genes (astA, cif, espA, espF, espJ, tir, nleA, and nleB), siderophore-associated genes (fyuA and irp2), and several secretion- and plasmid-associated genes (colE2-like, etpD, traJ, and traT). EMCR3 possessed a smaller but diverse set of virulence genes, including AslA, csgA, fdeC, fimH, hlyE, nlpI, and the yeh gene cluster. EMCR4 contained AslA, colE6, csgA, fimH, hlyE, nlpI, terC, and several yeh genes, indicating a limited but functionally relevant virulence repertoire. We revised this sentence as: EMCR5 was found adhesins genes (AslA, csgA, fdeC, and fimH), a toxin-associated gene (hlyE), a regulatory gene (anr), stress-response marker genes (nlpI and terC), plasmid-related genes (traJ and traT), and the yehA–yehD cluster. Overall, the isolates exhibited a substantial heterogeneity in the virulence content, with EMCR2 showing the broadest array of virulence determinants and the remaining isolates carrying more restricted but distinct sets of the virulence genes (Figure 5).

2.9. Presence of AMR Genes by Whole-Genome Sequence

Analysis of antimicrobial resistance (AMR) determinants among the five EMCR isolates revealed substantial variation in both genetic profiles and corresponding phenotypic resistance patterns. EMCR1 carried aminoglycoside-modifying enzyme genes aph6-Id and aph3-Id, the β-lactamase bla_TEM-1B, the plasmid-mediated quinolone resistance gene qnrS1, and the tetracycline resistance determinant tetM. These genotypes were consistent with its phenotypic resistance to ampicillin, piperacillin, streptomycin, tetracycline, and chloramphenicol (AMP-PIP-S-TE-C). EMCR2 exhibited the most extensive AMR repertoire, including multiple aminoglycoside resistance genes (aadA1, aadA2, aac3-IId, aph6-Id, aph3″-Ib, and aph3″-Ia), the β-lactamase bla_TEM-1B, the colistin resistance gene mcr-1.1, macrolide–lincosamide–streptogramin genes (mphA and ermB), phenicol resistance genes (cmlA1, flor), sulfonamide resistance genes (sul1 and sul3), and multiple tetracycline resistance genes (tetM, tetX4, and tetA), along with trimethoprim resistance genes (dfrA12 and dfrA5). This broad AMR gene profile corresponded with resistance to ampicillin, pirlimycin, streptomycin, tetracycline, chloramphenicol, and gentamicin (AMP-PIP-S-TE-C-CN). EMCR3 possessed genes associated with aminoglycoside resistance (aadA1), phenicol resistance (floR), plasmid-mediated quinolone resistance (qnrS13, OqxA, and OqxB), tetracycline resistance (tetA), and folate-pathway resistance (dfrA12 and dfrA14). The observed genotype aligned with resistance to AMP-PIP-S-TE-C. EMCR4 harbored multiple aminoglycoside resistance genes, including aph6-Id, aph3″-Id, aac3-IId, and aph3′-Ia, as well as a diverse set of β-lactamases (bla_TEM-220, bla_TEM-135, bla_TEM-126, bla_TEM-106, and bla_TEM-1B). Additionally, it encoded mcr-1.1, qnrS1, sul1, and tetracycline resistance genes (tetM and tetA), along with dfrA5. These determinants were reflected in its phenotype, showing resistance to AMP-PIP-S-TE-C-CN. Finally, EMCR5 carried aadA1, aadA2, bla_TEM-1B, cmlA1, and tetracycline and trimethoprim resistance genes (tetM, tetA, and dfrA12), consistent with resistance to AMP-PIP-S-TE-C. Overall, the concordance between AMR genotypes and phenotypes across isolates supports the functional expression of the detected resistance determinants (Figure 6).

2.10. Plasmid Replicon Detection

Whole-genome sequencing identified a diverse range of plasmid replicon types among the EMCR isolates, reflecting substantial plasmid-mediated genomic plasticity and a high potential for horizontal gene transfer. EMCR1 carried three replicon groups—IncFIB(K), ColE10, and IncX1—all of which are frequently associated with Enterobacterales plasmids, involved in antimicrobial resistance dissemination. EMCR2 exhibited the most extensive plasmid repertoire, comprising IncHI2, IncHI2A, IncFIA(HI1), IncFIB(AP001918), IncFIB(K), IncFII, IncN, and IncX1, together with multiple Col-type elements (Col156, ColRNAI, and Col440II). This combination of large conjugative plasmids and small mobilizable plasmids is characteristic of highly transmissible multidrug-resistant lineages. EMCR3 harbored IncFIB(K), IncFIA(HI1), and IncX1, indicating the presence of well-recognized conjugative backbones that are commonly found in clinically relevant Enterobacterales. EMCR4 contained replicons IncHI2, IncHI2A, Col156, IncQ1, and p0111, a profile consistent with plasmid groups known to accommodate a wide variety of acquired-resistance determinants. EMCR5 carried multiple replicon types, including IncFIC(FII), IncX1, p0111, IncFIB(AP001918), Col440I, IncFIA(HI1), and IncFIB(K). This reflected the co-existence of several large conjugative plasmids together with smaller Col-type plasmids within a single genome. Collectively, the identification of plasmid families, such as IncF, IncHI2, IncX1, and IncN, across these isolates underscores the prominent role of plasmid-edited mechanisms in shaping the genomic architecture of the EMCR strains and supports their capacity for efficient transmission of antimicrobial resistance determinants. Collectively, the WGS data indicate that the EMCR isolates showed highly plastic genomes shaped by extensive plasmid-mediated gene acquisition. Isolates with the broadest AMR and virulence profiles, particularly EMCR2 and EMCR4, also harbored multiple large conjugative plasmids (e.g., IncHI2, IncF, and IncN), suggesting that resistance and virulence determinants are embedded within mobile genomic backbones. The diversity of sequence types, including lineages previously reported in both animal and human settings, further supports the role of slaughterhouse environments as convergence points for heterogeneous E. coli populations. Together, these genomic features highlight the potential for horizontal dissemination and persistence of high-risk E. coli genotypes along the food chain (Figure 7).

2.11. Analysis of Secondary Metabolite Biosynthesis Gene Clusters

Whole-genome analysis using antiSMASH version 7.0 (https://antismash.secondarymetabolites.org, accessed on 1 November 2024) identified multiple secondary metabolite biosynthesis gene clusters across all five isolates (Figure 6). EMCR1 contained two clusters, comprising a complete NRP-metallophore/NRPS cluster with 100% similarity to the enterobactin biosynthetic pathway, together with a low-similarity thiopeptide cluster (14%). EMCR2 exhibited the greatest diversity, harboring four clusters, including an enterobactin-like NRPS cluster (90%); an NRPS–T1PKS hybrid cluster, showing partial similarity to the yersiniabactin system (17%); and two additional clusters of the thiopeptide and RiPP showed low similarity (~14%). EMCR3 carried one thiopeptide cluster (14%) and one NRPS cluster, both lacking close matches to known references, indicating potentially divergent or uncharacterized secondary metabolite pathways. EMCR4 contained a low-similarity thiopeptide cluster (14%) together with a fully conserved enterobactin-associated NRPS cluster (100%). EMCR5 similarly encoded one thiopeptide cluster (14%) and one enterobactin-like NRPS cluster with 100% similarity.

2.12. Compare Genome Mapping

Comparative circular genome mapping of the five E. coli EMCR isolates (EMCR1–EMCR5) showed a largely conserved chromosomal backbone (~5 Mb scale) with similar CDS density and generally stable GC content/GC-skew profiles, consistent with shared core genome organization. All genomes displayed multiple annotations linked to reduced antimicrobial accumulation, including Acr-family efflux components (e.g., acrB/acrF) and Mdt-family multidrug transporters, suggesting a common intrinsic capacity for tolerance to diverse toxic compounds. However, genome plasticity differed among strains: EMCR1 contained a clearly annotated Tn3-family transposase (Tn2), and EMCR4 also showed a Tn3-family/IS-associated transposase feature, supporting strain-specific horizontal acquisition or rearrangement potential. CRISPR–Cas content varied markedly across isolates, with EMCR1 showing a single Cas cluster, EMCR3 carrying a CAS-Type IIE cluster, EMCR2 and EMCR4 containing both a Cas cluster and a CAS-Type IIE annotation, and EMCR5 encoding the most extensive defense repertoire, including CAS-Type IIE plus additional Type I-associated cas annotations (e.g., cas3). Together, these data indicate that while EMCR isolates share conserved efflux or transport features relevant to antimicrobial tolerance, they differ in mobile-element signatures and CRISPR–Cas defenses that may influence their propensity to acquire or restrict additional resistance determinants (Figure 8).

3. Discussion

The high prevalence of colistin-resistant E. coli observed in this study is consistent with previous reports from swine production systems in Southeast Asia, where colistin has been frequently used as a growth promoter and therapeutic agent [27,28]. A study by Nguyen et al. (2022) in Vietnam found a similar prevalence (88%) of colistin-resistant E. coli in swine farms, highlighting the widespread occurrence of resistance in food-producing animals [29]. The dominance of mcr-1 among our isolates aligns with global patterns, as mcr-1 remains the most frequently detected plasmid-mediated colistin resistance gene in swine, pork products, and farm environments worldwide [30,31]. The additional detection of mcr-3 and mcr-9 further supports other emerging reports on the multiple mcr variants co-circulating in swine systems. The patterns are more likely driven by long-term antimicrobial exposure and complex plasmid-mediated gene-transfer dynamics [32,33]. The coexistence of multiple mcr genes within the same isolate, such as mcr-1/mcr-3 and mcr-1/mcr-9 combinations, suggests active horizontal gene-transfer events and the accumulation of diverse resistance determinants on multireplicon plasmids. The phenomenon has also been described in swine-associated E. coli in Thailand, China, and Spain [34,35,36]. These multi-mcr plasmids have been shown to persist in bacterial populations even after a reduction or a withdrawal of colistin use due to co-selection from other antimicrobials and heavy metals commonly used in swine feed formulations [13,37]. The presence of less common variants mcr-4, mcr-7, and mcr-8, although at low frequencies, indicates an ongoing diversification of mcr reservoirs within the swine gut microbiome. Since slaughterhouses serve as critical junction points where fecal contamination, carcass surfaces, and the environment intersect, the detection of diverse mcr-positive isolates at this stage raises substantial concern regarding potential spillover into the food chain, farm workers, and downstream environments [38,39]. Overall, the high prevalence and diversity of mcr-mediated colistin resistance in this setting highlight an urgent need for targeting antimicrobial stewardship in swine production systems. Strengthened regulations of colistin use, enhanced hygiene measures in slaughterhouses, and routine genomic monitoring of mcr-carrying plasmids are essential steps toward mitigating the transmission of colistin-resistant E. coli. Given the established evidence of animal-to-human transmission of mcr-positive strains through direct contact, contaminated meat, and environmental pathways, these findings underscore the importance of integrated One Health strategies across veterinary and public-health sectors [36,40]. Although broth microdilution is the reference method recommended for colistin susceptibility testing, it was not included in the present study. Consequently, the absence of MIC data limits the ability to assess the level of phenotypic resistance. Nevertheless, the combination of selective phenotypic screening and molecular detection was appropriate for the primary objective of this study, which focused on the genomic characterization of mcr-harboring E. coli. Future studies incorporating standardized MIC determination would be beneficial to further strengthen genotype–phenotype correlations.

Among the aminoglycosides, the isolates remained highly susceptible to amikacin (resistance in 2.44% of mcr-positive isolates and 0% in non-mcr isolates), while gentamicin resistance showed moderate frequency (26.83% vs. 33.33%). Streptomycin resistance was observed (40.24% vs. 56.67%). The results align with the previous study, which shows the prevalence of streptomycin resistance in livestock farms [41]. Enrofloxacin resistance was relatively low (~10–11% across groups), whereas tetracycline resistance was high (~89% overall), reflecting widely reported resistance trends in livestock production systems [42]. These findings emphasized the need for antibiotic stewardship and surveillance in animal husbandry, especially in regions where antibiotics are frequently used without strict regulatory control. The high frequency of MDR patterns in E. coli from swine aligned with global trends of increasing resistance in livestock-associated bacteria. A study by Wu et al. (2018) found that 70% of E. coli isolates from swine in China showed MDR [43], while in Thailand, MDR rates exceeding 60% have been reported [44,45]. The extensive use of antibiotics in animal farming, particularly for growth promotion and disease prevention, has been implicated as a major driver of this resistance [46]. These results underscore an urgent need for enhanced antimicrobial stewardship in livestock sectors, including stricter regulations on antibiotic use, routine surveillance, and the promotion of alternative disease management strategies to curb the emergence and dissemination of MDR pathogens. Colistin resistance mediated by plasmid-borne mcr genes represents a major public health concern due to the drug’s role as a last-resort antibiotic for multidrug-resistant Gram-negative infections [28,47].

The detected ARGs span several major antibiotic classes, including aminoglycosides, β-lactams (including extended-spectrum β-lactamase [ESBL] and ampC β-lactamase), tetracyclines, and phenicols, which are commonly used in veterinary practice. The most prevalent resistance genes identified were aadA, associated with aminoglycoside resistance, and ampC, a β-lactamase that confers resistance to cephalosporins and penicillin. The widespread presence of these genes is consistent with prior studies that report the high detection rates of aadA and ampC genes in E. coli isolated from livestock, particularly in intensively farmed swine [47,48]. These findings support concerns over the dissemination of plasmid-mediated ampC β-lactamases, which can reduce the efficacy of β-lactam antibiotics in both veterinary and human medicine [49]. The results of the bla_TEM gene aligned with the report on E. coli in swine, especially in association with mobile genetic elements that facilitate its horizontal transfer [43]. Its prevalence highlights the ongoing selection pressure exerted by β-lactam usage in animal agriculture.

Moderate frequencies were observed for cmlA and bla_SHV, representing resistance to phenicols and β-lactams, respectively. The cmlA gene encodes a chloramphenicol efflux pump and has previously been linked to resistance in both swine and poultry E. coli isolates [49,50]. In contrast, bla_OXA, tetA, and tetB were detected at lower frequencies. These genes are responsible for resistance to β-lactams (OXA-type β-lactamases) and tetracyclines, respectively. The relatively low prevalence of tetracycline resistance genes was surprising, considering the extensive historical use of tetracyclines in swine farming. However, regional differences in antibiotic usage policies or horizontal gene transfer dynamics may explain these variations [51,52]. Importantly, no isolates carried the cat1, aac(3)-I, aac(3)-IV, or tetC genes. The absence of these genes suggests either limited use of corresponding antibiotics or their replacement by more dominant resistance mechanisms in this population. The lack of detection aligns with previous reports, showing that the prevalence of specific resistance genes can fluctuate based on antimicrobial usage patterns, geographical location, and plasmid compatibility [53]. Collectively, these findings emphasized the genetic diversity of the antimicrobial resistance mechanisms in E. coli at the slaughterhouse level and highlight the necessity for continuous monitoring and responsible antimicrobial stewardship in animal husbandry.

A variety of virulence genes were found in E. coli in this study. Interestingly, the EPEC was the most prevalence, this result was agreed with previous studies [54,55], together with the extremely low detection of bfp, strongly suggests that most isolates belong to atypical EPEC (aEPEC; eae⁺/bfp⁻) rather than typical EPEC. This is consistent with recent work, indicating that aEPEC is currently the predominant EPEC subtype in both humans and animals and is recognized as a major cause of diarrhea, especially in low- and middle-income settings [56]. In contrast to the predominance of eae, ETEC markers (lt and stII) were detected in only 2/112 isolates (1.78% each), and STEC markers (stx1, stx2), as well as EAEC (aafII) and EIEC (ipaH). This is noteworthy because ETEC and STEC are classically considered the main enteric pathotypes in swine, particularly in neonatal and post-weaning diarrhea [57]. Many published swine studies target clinically diarrheic piglets or farm outbreaks, in which ETEC and STEC are enriched [58]. In contrast, the present work focuses on slaughterhouse isolates, which are more likely to originate from subclinical colonized animals and from environmental contamination along the processing line. Under such conditions, EPEC/aEPEC-like E. coli may outnumber classical enterotoxigenic or toxigenic strains. However, the positive detection of the eaeAO157:H7 gene showed a higher prevalence than a previous study [59]. Among the mcr-positive E. coli and virulence genes, this study found that EPEC-type markers dominated, which disagreed with the study by Göpel et al. [60]. The previous study detected the highest frequency of the ETEC/STEC hybrid. However, our results align with the highest frequency of the mcr-1 detection [60]. From One Health perspectives, the convergence of the critical last-resort resistance (mcr) and intestinal virulence (eae) in slaughterhouse isolates is particularly alarming. It indicates that slaughterhouse environments may act as hubs where resistant and virulent E. coli genotypes mix and disseminate along the food chain, especially if hygiene and antimicrobial stewardship are suboptimal [61]. The high prevalence of eae among slaughterhouse isolates highlights the potential role of subclinically colonized animals as reservoirs of intestinally adherent E. coli. Given that eae-positive strains are capable of efficient intestinal colonization in humans, their presence in slaughterhouse environments raises concerns regarding food-chain contamination and zoonotic transmission. From a One Health perspective, the coexistence of intestinal virulence determinants and antimicrobial resistance in these isolates underscores the importance of strengthened hygiene control and surveillance at the animal–food–human interface.

Whole-genome sequencing (WGS) of five E. coli isolates (EMCR1-EMCR5) was conducted using Illumina technology, yielding high-quality assemblies with genome sizes ranging from 4.49 to 5.18 Mb and GC content between 50.51% and 50.81%, consistent with typical E. coli genomes [62,63]. The depth of coverage (455*–584*) and total raw reads (6.9–9.2 million) ensured reliable sequencing performance, comparable to other high-throughput E. coli studies [64]. Assembly metrics, such as N50 (47,716–108,063 bp) and contig counts (115–362), indicated moderate variation in genome continuity, potentially influenced by repetitive elements or strain complexity [65]. Multilocus sequence typing (MLST) based on WGS revealed diverse sequence types (STs) and clonal complexes (CCs), including ST877/CC86 and ST48/CC10, both previously linked to antimicrobial resistance and pathogenicity [66,67]. EMCR3 displayed genetic heterogeneity (ST542). However, EMCR4 (ST2935) may represent an under characterized lineage. The genomic diversity identified highlights the ecological adaptability and potential public health relevance of these isolates. WGS analysis revealed a diverse set of virulence genes among the five E. coli isolates. EMCR1–EMCR5 harbored typical virulence-associated genes, such as astA, csgA, fdeC, fimH, hlyE, ompT, and the yeh operon, associated with adhesion and biofilm formation [68]. EMCR2 exhibited the broadest virulence profile, including eae, espA, tir, nleA/nleB, and cif, consistent with an EPEC-like pathogenic potential [69].

Additional genes, such as colicin genes, katP, and yersiniabactin system (irp2/fyuA), suggest enhanced oxidative stress tolerance and iron acquisition [65]. The presence of traJ/traT in EMCR2 and EMCR5 reflects the potential for horizontal gene transfer and serum resistance. The antimicrobial resistance gene (ARG) profiles showed broad resistance across isolates, particularly in EMCR2 and EMCR4, which harbored mcr-1.1 and multiple β-lactamase genes (bla_TEM variants). Also, a diverse resistance determinant against aminoglycosides, macrolides, tetracyclines, and sulfonamides aligns with reports of MDR E. coli from swine in Asia [30,47]. The concordance between genotype and phenotype across all isolates supports WGS as a reliable predictor of resistance patterns [70]. The detection of mcr-1.1 highlights a public health concern due to its plasmid-mediated colistin resistance [36].

All isolates carried plasmids with known associations to resistance and virulence. Common replicons included IncFIB(K), IncX1, and Col-type plasmids, consistent with global E. coli plasmid patterns [71]. EMCR2 and EMCR5 harbored the most diverse plasmid types, including IncHI2/HI2A, IncFIA, and p0111, which have been linked to ESBL and mcr gene dissemination in livestock and clinical settings [72,73]. The broad plasmid content observed, especially in MDR isolates, underscores their potential for horizontal transfer and persistence in diverse environments.

4. Materials and Methods

4.1. Sample Size and Sampling Techniques

The sample size was calculated using a formula for estimating true prevalence (https://epitools.ausvet.com.au (accessed on 2 January 2024)). Owing to reported variability in the prevalence of colistin-resistant Escherichia coli, an expected prevalence of 10.4% was used for the calculation. The parameters applied included a test sensitivity of 95%, specificity of 90%, a desired precision of 10%, and a 90% confidence level. Based on these assumptions, a total of 200 fecal samples were collected and analyzed.

4.2. Sample Collection

A total of 200 fecal samples were randomly collected via rectal swabbing from finishing and breeder swine across 20 batches at a slaughterhouse in Nakhon Si Thammarat Province, Southern Thailand, between January and July 2024. All laboratory procedures were conducted at the Research Building of Walailak University. The study was approved by the Walailak University Animal Care and Use Committee (WU-ACUC-66-047) and the Institutional Biosafety Committee (WU-IBC-66-024). Samples were transported in sterile plastic bags within ice boxes to the laboratory and processed for microbiological analysis within one hour of collection.

4.3. Escherichia coli Isolation and Confirmation

All samples were analyzed for the presence of colistin-resistant E. coli. Approximately 10 g of fecal material was placed into a sterile plastic bag and mixed with 90 mL of Peptone Saline Buffer, containing 0.1% peptone (Oxoid, Hampshire, UK), followed by homogenization using a stomacher. A 10 μL loop of the resulting suspension was inoculated onto MacConkey agar (Oxoid, Hampshire, UK) with 4 mg/L of colistin. The agar plates were then incubated at 37 °C for 24 h [15]. From plates exhibiting bacterial growth, we characterized bacterial colonies by red coloration, smooth spherical morphology, and a diameter >2.5 mm. We selected a colony, subcultured it onto Eosin Methylene Blue (EMB) agar (HiMedia, Mumbai, India), and incubated it at 37 °C for 24 h. Colonies with metallic green sheen on the EMB agar were considered indicative of E. coli and were subsequently subcultured onto tryptic soy agar (HiMedia, India) and incubated at 37 °C for an additional 24 h. Polymerase Chain Reaction (PCR) was performed to confirm the identity of each presumptive isolate by targeting the uidA gene. Genomic DNA was extracted from all E. coli-positive samples using a bacterial genomic DNA extraction kit (Geneaid, New Taipei City, Taiwan), following a previously described protocol [37]. PCR amplification was performed, and the products were confirmed using electrophoresis on a 1.5% agarose gel (Major Science, Irvine, CA, USA). PCR bands were visualized under UV illumination using a G-BOX F3 Gel Documentation System (Syngene, Cambridge, UK). The primer sequences, expected amplicon sizes, and annealing temperatures are detailed in

Table 6. Primers are used for the amplification of E. coli genes, mcr-encode genes, pathotype E. coli genes, and antibiotic-resistant genes of E. coli.

Target Genes	Primer Sequences	Product Size (bp)	Annealing Temperature	References
uidA	F: 5′-GTCACGCCGTATGTTATTG-3′	530	58	[74]
	R: 5′-CCAAAGCCAGTAAAGTAGAAC-3′
mcr-1	F: 5′-AGTCCGTTTGTTCTTGTGGC-3′	320	58	[75]
	R: 5′-AGATCCTTGGTCTCGGCTTG-3′
mcr-2	F: 5′-CAAGTGTGTTGGTCGCAGTT-3′	715	58	[75]
	R: 5′-TCTAGCCCGACAAGCATACC-3′
mcr-3	F: 5′-AAATAAAAATTGTTCCGCTTATG-3′	929	58	[75]
	R: 5′-AATGGAGATCCCCGTTTTT-3′
mcr-4	F: 5′-TCACTTTCATCACTGCGTTG-3′	1116	58	[75]
	R: 5′-TTGGTCCATGACTACCAATG-3′
mcr-5	F: 5′-ATGCGGTTGTCTGCATTTATC-3′	1644	58	[75]
	R: 5′-TCATTGTGGTTGTCCTTTTCTG-3′
mcr-7	F: 5′-GTCAGTTACGCCATGCTCAA-3′	791	59	[76]
	R: 5′-TTCTTGTCGCAGAACTGTGG-3′
mcr-8	F: 5′-AAACTGAACCCGGTACAACG-3′	943	59	[76]
	R: 5′-GCCATAGCACCTCAACACCT-3′
mcr-9	F: 5′-GCGGTTGTAAAGGCGTATGT-3′	635	59	[76]
	R: 5′-CAAATCGCGGTCAGGATTAT-3′
stx-1	F: 5′-CAGTTAATGTGGTGGCGAAGG-3′	348	58	[77]
	R 5′-CACCAGACAATGTAACCGCTG-3′
stx-2	F: 5′-ATCCTATTCCCGGGAGTTTACG-3′	584	58	[77]
	R: 5′-GCGTCATCGTATACACAGGAGC-3′
bfp	F: 5′-GGAAGTCAAATTCATGGGGGTAT-3′	300	58	[77]
	R: 5′-GGAATCAGACGCAGACTGGTAGT-3′
eae	F: 5′-TCAATGCAGTTCCGTTATCAGTT-3′	482	58	[77]
	R: 5′-GTAAAGTCCGTTACCCCAACCTG-3′
lt	F: 5′-GCACACGGAGCTCCTCAGTC-3′	218	58	[77]
	R: 5′-TCCTTCATCCTTTCAATGGCTTT-3′
stII	F: 5′-AAAGGAGAGCTTCGTCACATTTT-3′	129	58	[77]
	R: 5′-AATGTCCGTCTTGCGTTAGGAC-3′
virF	F: 5′-AGCTCAGGCAATGAAACTTTGAC-3′	618	58	[77]
	R: 5′-TGGGCTTGATATTCCGATAAGTC-3′
ipaH	F: 5′-CTCGGCACGTTTTAATAGTCTGG-3′	933	58	[77]
	R: 5′-GTGGAGAGCTGAAGTTTCTCTGC-3′
aafII	F: 5′-CACAGGCAACTGAAATAAGTCTGG-3′	378	58	[77]
	R: 5′-ATTCCCATGATGTCAAGCACTTC-3′
eaeAO157:H7	F: 5′-AAGCGACTGAGGTCACT-3′	473	60	[77]
	R: 5′-ACGCTGCTCACTAGATGT-3′
cat1	F: 5′-AGTTGCTCAATGTACCTATAACC-3′	547	58	[50]
	R: 5′-TTGTAATTCATTAAGCATTCTGCC-3′
cmlA	F: 5′-CCGCCACGGTGTTGTTGTTATC-3′	698	58	[50]
	R: 5′-CACCTTGCCTGCCCATCATTAG-3′
aadA	F: 5′-TGATTTGCTGGTTACGGTGAC-3′	284	58	[50]
	R: 5′-CGCTATGTTCTCTTGCTTTTG-3′
aac(3)-I	F: 5’-ACCTACTCCCAACATCAGCC-3′	157	58	[50]
	R: 5′-ATATAGATCTCACTACGCGC-3′
aphA-I	F: 5′-ATGGGCTCGCGATAATGTC-3′	600	58	[50]
	R: 5′-CTCACCGAGGCAGTTCCAT-3′
aac(3)-IV	F: 5′-CTTCAGGATGGCAAGTTGGT-3′	286	58	[50]
	R: 5′-TCATCTCGTTCTCCGCTCAT-3′
blaTEM	F: 5′-TCGCCGCATACACTATTCTCAGAATGA-3′	445	60	[78]
	R: 5′-ACGCTCACCGGCTCCAGATTTAT-3′
blaOXA	ACA CAA TAC ATA TCA ACT TCG C	813	60	[79]
	AGT GTG TTT AGA ATG GTG ATC
blaSHV	F: 5′-ATGCGTTATATTCGCCTGTG-3′	747	51	[78]
	R: 5′-TGCTTTGTTATTCGGGCCAA-3′
tetA	F: 5′-GTAATTCTGAGCACTGTCGC-3′	965	57	[50]
	R: 5′-CTGCCTGGACAACATTGCTT-3′
tetB	F: 5′-CTCAGTATTCCAAGCCTTTG-3′	414	52	[50]
	R: 5′-ACTCCCCTGAGCTTGAGGGG-3′
tetC	F: 5′-CCTCCTGCGGGATATCGTCC-3′	505	65	[50]
	R: 5′-GGTTGAAGGCTCTCAAGGGC-3′
ampC	F: 5′-ATCAAAACTGGCAGCCG-3′	510	65	[80]
	R: 5′-GAGCCCGTTTTATGCACCCA-3′

.

4.4. Antibiotic Susceptibility Testing

Antimicrobial susceptibility testing of all isolates was performed using the disk diffusion method, following the guidelines of the Clinical and Laboratory Standards Institute (CLSI, 2019) [81]. Isolates were initially cultured on Tryptic Soy Agar (TSA) (HiMedia Laboratories, Mumbai, India). Each bacterial suspension was adjusted to match the 0.5 McFarland turbidity standard, and the standardized E. coli inoculum was evenly spread onto Mueller–Hinton Agar (MHA) (HiMedia Laboratories, Mumbai, India) using sterile cotton swabs. Antimicrobial disks were then placed on the agar surface, and plates were incubated at 37 °C for 24 h. Susceptibility was assessed against ten antimicrobial agents representing five major classes: beta-lactams (ampicillin, amoxicillin/clavulanic acid, and piperacillin), aminoglycosides (amikacin, gentamicin, and streptomycin), phenicols (chloramphenicol), cephalosporins (ceftriaxone), quinolones (enrofloxacin), and tetracyclines (tetracycline) (Oxoid Ltd., Basingstoke, UK). Multidrug resistance (MDR) was defined as resistance to at least one antimicrobial agent in three or more different antibiotic classes.

4.5. Molecular Detection of Colistin Resistance Genes, Pathotype Genes, and Antimicrobial Resistance Genes

A total of 112 E. coli isolates were identified as carrying colistin resistance genes, including mcr-1, mcr-2, mcr-3, mcr-4, mcr-5, mcr-7, mcr-8, and mcr-9. Multiplex PCR targeting mcr-1 to mcr-5 was conducted under the following conditions: initial denaturation at 94 °C for 15 min, followed by 35 cycles of denaturation at 94 °C for 30 s, annealing at 58 °C for 90 s, and extension at 72 °C for 60 s, with a final extension step at 72 °C for 10 min. The detection of mcr-7, mcr-8, and mcr-9 was carried out using a separate PCR protocol consisting of initial denaturation at 94 °C for 4 min, followed by 20 cycles of denaturation at 94 °C for 5 s, annealing at 59 °C for 15 s, and a final extension at 72 °C for 5 min. Additionally, the isolates were screened for pathotype-specific genes using PCR amplification. Ten virulence-associated genes were targeted: eae, bfp, stx-1, stx-2, lt, stII, virF, ipaH, aafII, and eaeAO157:H7. PCR conditions for pathotype gene detection were conducted as previously described (Table 6). Thirty randomly selected E. coli isolates were analyzed for the presence of antibiotic resistance genes using specific primers (Table 6) through PCR, following a previous study. The PCR conditions for amplifying resistance genes included an initial denaturation at 95 °C for 5 min, followed by 30 cycles consisting of denaturation at 95 °C for 35 s, annealing at the appropriate temperature for 45 s, and extension at 72 °C for 1 min. A final extension step was performed at 72 °C for 5 min [20].

4.6. Whole-Genome Sequencing

Five E. coli isolates exhibiting multidrug resistance (MDR) and demonstrating the highest multiple antibiotic resistance (MAR) indices were selected for whole-genome sequencing. Genomic DNA was extracted from these representative isolates using a commercial kit (Geneaid, New Taipei City, Taiwan). The quality and concentration of the extracted DNA were evaluated using a microvolume UV-Vis spectrophotometer (ThermoScientific™ NanoDrop™ One, Madison, WI, USA) and further verified by agarose gel electrophoresis. High-quality DNA samples underwent de novo sequencing, and sequence reads were assembled using Velvet (version 1.2.10). Contig extension was performed with GapFiller (version 1.10), followed by scaffold construction using SSPACE (version 3.0). Protein-coding genes were predicted from the assembled scaffolds using Prodigal (version 2.6.3). Functional annotation of the predicted genes was performed using BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi (accessed on 1 November 2024)), with significant matches defined by an E-value threshold of less than 0.00001.

4.7. Application of Bioinformatic Tools

Whole-genome sequencing (WGS) was performed using a paired end short read sequencing approach. Raw sequencing reads were quality checked and filtered using AfterQC v0.9.6, including adapter trimming and error correction. De novo genome assembly was conducted using Unicycler v0.5.0 and QUAST v5.0.2. Genome completeness and contamination were evaluated using CheckM v1.1.2, and genome annotation was performed with Prokka v1.14.6. Genome assemblies were further analyzed using web-based bioinformatics tools to confirm E. coli identification, determine sequence types, and characterize antimicrobial resistance genes (ARGs), mobile genetic elements (MGEs), and virulence factors (VFs). Species identification was performed using KmerFinder v3.2 (https://cge.food.dtu.dk/services/KmerFinder/, (accessed on 1 November 2024)) Multilocus sequence typing (MLST) and clonal-complex assignment were conducted using the PubMLST database (https://pubmlst.org/ (accessed on 1 November 2024), based on allelic profiles of seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA, and recA). Antimicrobial resistance genes were identified using ResFinder 4.0 (https://cge.food.dtu.dk/services/ResFinder/ (accessed on 1 November 2024)) with default thresholds of 90% sequence identity and a minimum alignment length of 60%, including the detection of chromosomal point mutations, following Bortolaia et al. (2020) [82]. The presence of MGEs and their associations with ARGs and VFs were analyzed using MobileElementFinder v1.0.3, as described by Johansson et al. (2021) [83]. Virulence-associated genes were identified using the Virulence Factor Database (VFDB) by applying a conservative BLASTp threshold with an E-value of 1 × 10⁻¹⁰ and a minimum query coverage of 40%, following the approach described by Liu et al. (2022) [84].

4.8. Statistical Analysis

Descriptive statistics were used to summarize the prevalence of E. coli, mcr-encoding isolates, antimicrobial resistance profiles, resistance genes, and virulence genes. Differences in proportions between groups were analyzed using Fisher’s exact test or the Chi-square test, as appropriate. Statistical analyses were performed using R (version 4.3.0), and a p-value < 0.05 was considered statistically significant. Binomial 95% confidence intervals (CIs) were calculated using the Wilson score method [85].

5. Conclusions

This study investigated colistin-resistant Escherichia coli from swine farms, revealing a high prevalence (90.32%) of confirmed E. coli isolates. Antimicrobial susceptibility testing showed widespread multidrug resistance (MDR), with the highest resistance to ampicillin and tetracycline. The mcr-1 gene was the most common colistin resistance gene, detected in over 50% of isolates. Whole-genome sequencing of selected isolates revealed diverse sequence types, virulence factors, and AMR genes, including mcr-1.1. Plasmid analysis indicated a high potential for horizontal gene transfer, particularly in isolates carrying IncF, IncHI2, and IncX1 plasmids. These results highlight a significant public health concern regarding the spread of MDR and colistin-resistant E. coli in the swine industry.