Introduction

Beta-lactam antibiotics are the most important antimicrobial agents in veterinary medicine. For instance, they are the treatment of choice in established practices such as dry cow therapy and weaning pig therapy1. The global emergence of antimicrobial resistance is a crisis of increasing importance and urgency. Bacteria commonly counteract beta-lactams by producing inactivating enzymes, such as β-lactamases2. Resistance to third-generation and fourth-generation cephalosporins, which are considered critically essential antimicrobials by the World Health Organization3, is often conferred by extended-spectrum β-lactamases (ESBLs). ESBLs are a group of hydrolytic enzymes that confer resistance to most β-lactam antibiotics, including extended-spectrum cephalosporins and monobactams, with the exception of carbapenems and cephamycins. However, they are susceptible to clavulanic acid, sulbactam, and tazobactam4. TEM, SHV, and CTX-M types are commonly observed ESBLs. The extensive use of antibiotics in humans, animals, and agriculture has contributed to the selection and dissemination of drug-resistant strains within the Enterobacteriaceae family. Food-producing animals are recognized as a significant reservoir for antimicrobial-resistant bacteria. Several studies suggest that ESBL-producing E. coli may spread between food-producing animals and humans through the food chain and environmental pathways5. Since similar ESBLs and plasmid profiles have been identified in food-producing animals, animal-derived foods, and humans6, food may act as a source of drug-resistant E. coli7. However, other studies have shown that nonhuman reservoirs contribute minimally to invasive ESBL-producing E. coli infections in humans8. The prevalence of ESBL-producing E. coli in diseased swine and cattle has been sporadically reported in Taiwan9,10. The purpose of this study was to identify ESBL-producing E. coli isolates collected from diseased livestock and poultry in Taiwan between 2011 and 2020. In this study, livestock refers to pigs, cattle, goats, and other domestic animals, while poultry includes turkeys and domesticated fowl. Because these samples were collected across Taiwan for over a long time, we hope that this retrospective survey will offer a more comprehensive understanding of the characteristics of ESBL-producing E. coli in diseased food-producing animals in Taiwan. Our study aligns with the Global Action Plan objective of the World Organisation for Animal Health, which aims to strengthen knowledge through surveillance and research.

Result

Occurrence of ESBL-producing E. coli

From 2013 to 2019, a total of 297 E. coli isolates were collected from pigs (n = 239), cattle (n = 28), and goats (n = 30). Among these, 36 isolates (12.1%) were identified as ESBL-producing E. coli. The prevalence of ESBL-producing E. coli was 7.9% (19/239) in pigs, 39.2% (11/28) in cattle, and 20% (6/30) in goats. For all the ESBL producers, the most frequently found was from pigs (52.8%, 19/36), followed by cattle (30.5%, 11/36) and goats (16.7%, 6/36). Between 2011 and 2020, a total of 269 E. coli isolates were obtained from poultry, including chicken (n = 153), duck (n = 43), goose (n = 58), and other birds (such as turkey and show chickens; n = 15). We identified 22 ESBL-producing E. coli isolates from poultry, representing 8.2% (22/269) of the total. The prevalence of ESBL-producing E. coli in chickens, ducks, geese, and other birds is 5.2% (8/153), 11.6% (5/43), 8.6% (5/58), and 26.7% (4/15), respectively. For the poultry-derived ESBL producers, chickens were the most common source (34.8%, 8/23), followed by ducks and geese (21.8, 5/23), and others (17.3%, 4/23). The annual detection rates of ESBL-producing E. coli in livestock from 2013 to 2019 were as follows: 7.3% (3/41), 7.3% (3/41), 13.3% (6/45), 7.9% (5/63), 17.0% (9/53), 10.3% (3/29), and 28.0% (7/25). In poultry, the detection rates from 2011 to 2020 were: 0% (0/8), 0% (0/42), 6.8% (3/44), 3.3% (1/30), 9.7% (3/31), 0% (0/17), 6.9% (2/29), 11.1% (1/9), 14.3% (1/7), and 21.2% (11/52).

Antimicrobial susceptibility test

Figures 1 and 2 illustrate the drug resistance rates of ESBL-producing E. coli compared to non-ESBL-producing E. coli in livestock and poultry, respectively. ESBL-producing E. coli exhibited higher or equivalent drug resistance rates compared to non-ESBL-producing strains, except for amoxicillin/clavulanic acid in all animals. Both ESBL and non-ESBL-producing E. coli were susceptible to imipenem. ESBL-producing E. coli from livestock were all resistant to ampicillin, cefalexin, ceftiofur, and tetracycline. Additionally, they demonstrated over 50% resistance to gentamicin, enrofloxacin, and trimethoprim/sulfamethoxazole. However, these ESBL producers exhibited more than 50% susceptibility to amoxicillin/clavulanic acid. All ESBL-producing E. coli from poultry were uniformly resistant to ampicillin, cephalexin, ceftiofur, and tetracycline. Additionally, they exhibited over 50% resistance to gentamicin, enrofloxacin, and trimethoprim/sulfamethoxazole, while showing more than 50% susceptibility to amoxicillin/clavulanic acid.

Fig. 1
figure 1

Antimicrobial susceptibility testing of ESBL-producing E. coli (n = 36) and non-ESBL-producing E. coli (n = 261) isolated from livestock. The numbers on the Y-axis represent the percent resistant, and the antimicrobials tested are listed on the X-axis.

Full size image
Fig. 2
figure 2

Antimicrobial susceptibility testing of ESBL-producing E. coli (n = 22) and non-ESBL-producing E. coli (n = 247) isolated from poultry. The numbers on the Y-axis represent the percent resistant, and the antimicrobials tested are listed on the X-axis.

Full size image

The drug-resistant patterns of ESBL-producing E. coli compared to non-ESBL-producing E. coli isolated from livestock and poultry are highly similar. For both livestock and poultry, the drug resistance rate was higher in ESBL producers than in non-ESBL producers, particularly for ampicillin, cephalexin, ceftiofur, gentamicin, enrofloxacin, and tetracycline. Interestingly, the drug resistance rate of non-ESBL-producing E. coli from livestock was still higher than that of non-ESBL-producing E. coli from poultry. Amoxicillin/clavulanic acid was relatively effective against ESBL producers from livestock and poultry. For trimethoprim/sulfamethoxazole, ESBL-producing E. coli from livestock exhibited a slightly higher drug resistance rate than non-ESBL-producing E. coli, though the difference was small. However, in poultry, non-ESBL-producing E. coli demonstrated a slightly higher drug resistance rate than ESBL-producing E. coli.

β-lactamase gene detection

The ESBL-encoding genes in both livestock and poultry included blaCTX−M−1 group, blaCTX−M−9 group, and blaTEM. However, no blaCTX−M−2 group, blaCTX−M−8 group, blaCTX−M−25 group, and blaSHV detected in livestock and poultry. Table 1 lists the distribution of these bla genes in ESBL-producing E. coli from livestock and poultry. In livestock, the most commonly detected gene was blaCTX−M−1 group, within which blaCTX−M−55 being the major type of bla gene. blaCTX−M−15 and blaCTX−M−216 were also found in the blaCTX−M−1 group. Although various blaTEM genes were detected, only blaTEM−116 was classified as an ESBL gene. In ESBL-producing E. coli isolated from poultry, blaCTX−M−55 was the only bla gene type in the blaCTX−M−1 group. The blaCTX−M−9 group contained blaCTX−M−65 and blaCTX−M−130. None of the blaTEM types detected from poultry were classified as ESBL.

Table 1 Distribution of the bla genes in ESBL-producing E. coli from livestock and poultry.
Full size table

The bla genes and sequence types (ST) of the ESBL-producing E. coli are detailed in Table 2. E. coli carrying blaCTX−M−55 was the most common in both livestock and poultry. While ESBL-producing E. coli from livestock and poultry had distinct STs, ST162 and ST1196 were the only STs found in both sources. The most common ST was ST48 (n = 4), followed by ST162 (n = 3), ST457 (n = 3), ST468 (n = 3), and ST11476 (n = 3). The following STs were observed twice: ST38, ST69, ST349, ST617, ST744, ST1196, ST15697, and ST15699. All other STs were detected once. One ST10 and two ST38 isolates were detected from livestock, while one ST617 and two ST69 isolates were identified from poultry. These STs are recognized as common sequence types of extraintestinal pathogenic E. coli (ExPEC). The STs can be grouped into clonal complex 10 (CC10), CC29, and CC38. CC10 included 11 isolates from livestock and 7 from poultry, while both CC29 and CC38 had 4 isolates from livestock each. Figure 3 provides a visual representation of the MLST results for the 58 ESBL-producing E. coli isolates, showing the relatedness of each ST based on the degree of allele sharing between strains. The MLST analysis data for all ESBL-producing E. coli isolates are provided in Supplementary Table S1. The sequences of each bla gene, along with the seven housekeeping genes (adK, fumC, gyrB, icd, mdh, purA, and recA) used for MLST analysis, were individually registered in NCBI, with the accession numbers for each gene listed in Supplementary Table S2.

Table 2 bla genes and sequence type (ST) of the ESBL-producing E. coli.
Full size table
Fig. 3
figure 3

Minimal spanning tree (MSTree) of ESBL-producing E. coli. Each circle represents a ST divided into a sector for each isolate and delimited by the ST number or unknown. The ESBL-producing E. coli obtained from livestock were shown in green, while those E. coli obtained from poultry were shown in red. The numbers on the connecting line between STs within the MSTree indicate the number of different alleles. Solid lines represent an allele difference of three or less, whereas dotted lines and faint lines indicate an allele difference of four or more. The STs belonging to the same clonal complex (CC) are shaded gray.

Full size image

Conjugation test

All 36 ESBL-producing E. coli from livestock and 22 from poultry were tested for their ability to transmit bla genes. All the 58 ESBL-producing E. coli successfully transferred their bla genes to the recipient strain E. coli J53. The bla gene(s) of each ESBL-producing E. coli and its corresponding transconjugant, along with the antimicrobial resistance pattern of ESBL-producing E. coli, are depicted in Supplementary Table S3.

Discussion

All E. coli strains in this study were isolated from diseased food-producing animals, although the specific characteristics of these E. coli strains were not described. Avian pathogenic E. coli (APEC) causes colibacillosis, and most APEC strains are classified as ExPEC. However, we did not classify the E. coli isolates obtained from diarrheic livestock into specific pathotypes, which represents a limitation of our study. A study in Spain examining E. coli from pig farms revealed the presence of ESBL producers among both commensal and pathogenic E. coli strains11. We assume that both commensal and pathogenic strains of E. coli are likely present among the ESBL producers in our study.

Our study found that the prevalence of ESBL-producing E. coli was 7.9% in pigs, 39.2% in cattle, and 8.6% in poultry. In comparison, the GERM-Vet monitoring program conducted in Germany, as referenced in our investigation, showed the presence of ESBL-producing E. coli in 4.8% of pigs, 11.2% of cattle, and 0.8% of poultry12. A European study conducted across 11 countries found that the prevalence of third-generation cephalosporin (3GC)-resistant Escherichia coli, primarily mediated by ESBL and plasmid-mediated AmpC β-lactamases (pAmpC), was 3.6% in pigs, 8.1% in cattle, and 6.1% in broilers13. In contrast to similar studies conducted in Western countries, the present study reports a higher incidence rate of ESBL-producing E. coli. A study from another Asian country Thailand indicated that the occurrence of ESBL-producing E. coli was higher in the sick pigs (44%) than the healthy pigs (19.2%)14. The variation in the prevalence of ESBL-producing E. coli collected from different countries may be attributed to several factors. A scoping review on turkeys suggested that antimicrobial usage, biosecurity measures, and management practices are associated with antimicrobial resistance15.

This study exclusively investigated ESBL. ESBL-producing E. coli exhibited higher or comparable rates of drug resistance compared to non-ESBL-producing E. coli, with the exception of amoxicillin/clavulanic acid in all food-producing animals. Amoxicillin/clavulanic acid is anticipated to demonstrate notable efficacy in treating ESBL-producing E. coli due to the susceptibility of ESBL to clavulanic acid, which is a defining characteristic of ESBL. In E. coli strains harboring AmpC β-lactamases, whether plasmid-mediated or chromosomally encoded, resistance to third- and fourth-generation cephalosporins is anticipated. Notably, AmpC β-lactamase is resistant to clavulanic acid. The elevated resistance rate of non-ESBL producers to amoxicillin/clavulanic acid may be due to the presence of AmpC β-lactamase in these strains. Although both ESBL-producing and non-ESBL-producing E. coli were susceptible to imipenem, a carbapenem, this drug is not commonly used in veterinary medicine due to its critical role as a last-resort antimicrobial for humans.

In ESBL-producing E. coli from livestock and poultry, blaCTX−M−55, belonging to the blaCTX−M−1 group, was the most commonly identified bla type. blaCTX−M−55 is a derivative of blaCTX−M−15, distinguished only by the Ala77Val substitution16. Currently, the blaCTX−M−55 gene is increasingly reported, particularly in China, where blaCTX−M−55 is the second most common CTX-M variant in food-producing animals17. Our previous investigation of pigs with diarrhea revealed a high prevalence of blaCTX−M−55 in ESBL-producing E. coli isolated from these animals9. In a study conducted in Thailand, the blaCTX−M group was found to be the most prevalent ESBL-encoding gene, with blaCTX−M−14 (54.5%) and blaCTX−M−55 (42.9%) being the predominant variants14. The presence of blaCTX−M−55 has also occurred outside Asian countries. In the United Kingdom, the prevalence of blaCTX−M−15 producers has declined over several years in favor of new variants, particularly blaCTX−M−5518. In France, blaCTX−M−55 was found in E. coli of various animal species and localized to different plasmid types, suggesting that a wide variety of E. coli clones and plasmid types supported the spread of blaCTX−M−5519. blaCTX−M−14, which belongs to the blaCTX−M−9 group, was the second most frequently detected bla gene in ESBL-producing E. coli from livestock. This bla gene was common in humans and several animal species in Asia20.

We identified four STs (ST10, ST38, ST69, ST617) that are classified as important ExPEC lineages and cause extraintestinal infection in humans21. Additionally, ST457, which occurred three times in livestock, is part of ExPEC, although rarely reported22. From a food safety perspective, ExPEC isolated from diseased food-producing animals is unlikely to cause infection in humans as these animals would ultimately not enter the market. However, there may be a potential risk to people who are in close contact with these animals. In addition, studies have shown that bacteria excreted from the feces of diseased pigs can contaminate the environment and serve as a reservoir for the exchange of drug-resistant genes, some of which could eventually enter the food chain23. The most frequently found ST was ST48 (n = 4) from poultry. This particular ST is often associated with various lactamases, including ESBL and New Delhi metallo-β-lactamase (NDM), a major type of carbapenemases24. E. coli ST48 has been associated with extraintestinal pathogenic strains for avian species25. Significantly, the ST48 strains were found to be the dominant hosts for the mcr-1–IncX4 plasmid, which has a highly conserved sequence and plasmid structure26. In summary, the percentage of ExPEC in ESBL-producing E. coli isolated from diseased livestock and poultry was 15.5% (9/58), a statistic that should be kept in mind.

The findings of this study have important implications for both public health and agricultural practices. The identification of ESBL-producing E. coli in livestock and poultry highlights the role of food-producing animals as significant reservoirs of antimicrobial resistance. The widespread presence of the blaCTX−M−55 gene in these isolates, along with the ability of ESBL-producing strains to transfer their resistance genes through conjugation, raises concerns about the potential for these resistant bacteria to spread within animal populations and possibly to humans.

Research into alternative farming practices, such as reducing the use of antibiotics or using probiotic interventions, could help mitigate the development and spread of resistance. Expanding the study to include healthy food-producing animals and other animal species could lead to a more comprehensive understanding of the global impact of antimicrobial resistance in agriculture.

Materials and methods

Sample collection

In Taiwan, the examination of livestock and poultry diseases occurring at the farm level is conducted by registered veterinarians working at the state-operated local livestock disease control center. When necessary, samples from affected animals are sent to the Veterinary Diagnostic Laboratory at the Veterinary Research Institute for accurate diagnosis. E. coli was isolated by following this process: after a 24-hour incubation at 37 °C, suspected pink colonies on MacConkey agar (Becton Dickinson, Franklin Lakes, USA) were carefully selected, purified, and then identified using the VITEK® 2 Compact automated system (Biomérieux, Marcy-l’Étoile, France) with GN ID cards. In this study, 297 E. coli isolates were obtained from fecal swabs of diseased farm animals showing symptoms of diarrhea, collected between 2013 and 2019. Additionally, 269 E. coli isolates were recovered from the organs of diseased poultry affected by colibacillosis, gathered between 2011 and 2020. These bacterial isolates were preserved in Microbank System cryovials (Pro-Lab Diagnostics, Richmond Hill, ON, Canada) and stored at -80 °C for future research.

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was conducted using the minimum inhibitory concentration (MIC) method with the VITEK 2 AST-GN 96 card. The results were recorded for the following antibiotics: ampicillin, amoxicillin-clavulanic acid, ceftiofur, cefalexin, imipenem, enrofloxacin, gentamicin, tetracycline, and trimethoprim/sulfamethoxazole.

ESBL-producing E. coli phenotypic screening and confirmation

ESBL-producing E. coli were initially screened using the VITEK 2 system, where the GN ID card reported ESBL results. To confirm ESBL production, the combination disk test was performed using cefotaxime and ceftazidime (30 µg) with and without clavulanic acid (10 µg) (Becton Dickinson), following the standards set by the Clinical and Laboratory Standards Institute (CLSI). In brief, E. coli (0.5 × McFarland) was plated on Muller–Hinton agar (Becton Dickinson) and incubated at 35 °C for 16–18 h. A difference of 5 mm or more in the inhibition zones between cephalosporins alone and in combination with clavulanic acid was considered positive for ESBL production. Klebsiella pneumoniae ATCC 700,603 and E. coli ATCC 25,922 were used as positive and negative controls, respectively.

bla gene analysis and genotyping of the ESBL-producing E. coli

The bla genes of the ESBL-producing E. coli isolates were analyzed using polymerase chain reaction (PCR). The template DNA of the bacteria was extracted using the boiling method27. The most commonly encountered ESBL gene groups are blaSHV, blaTEM, and blaCTX−M. Therefore, the bla genes, including blaCTX−M−1−group, blaCTX−M−2−group, blaCTX−M−8−group, blaCTX−M−9−group, blaCTX−M−25−group, blaSHV, and blaTEM, were the target genes for screening in the present study. The PCR conditions were as follows: initial denaturation at 95 °C for 5 min, 35 cycles at 95 °C for 30 s, annealing at 52–55 °C for 30 s, and a 72 °C extension for 1 min. Next, 45 µL of each PCR sample was loaded onto a 1.5% agarose gel and electrophoresed at 100 V for 30 min. The gel was stained with ethidium bromide dye at a concentration of 10 mg/mL (Bionovas, Toronto, Canada) for 20 min and examined under ultraviolet illumination Quantum ST4-1100 (Vilber, Marne-la-vallée, France). The agarose gel containing the correct size of the bla gene was incised with a blade and placed into a microcentrifuge tube and then subjected to sequencing (Mission Biotech, Taipei, Taiwan). The DNA sequences were analyzed using the Beta-Lactamase DataBase (www.bldb.eu) and NCBI Beta-Lactamase Data Resources (https://www.ncbi.nlm.nih.gov/pathogens/beta-lactamase-data-resources/).

The ESBL-producing E. coli were genotyped using multilocus sequence typing (MLST)28. Internal fragments of adk, fumC, gyrB, icd, mdh, purA, and recA were amplified through PCR. The PCR products were purified and sequenced, and the sequence data were uploaded to PubMLST (https://pubmlst.org) for comparison. Phylogenetic analysis of the ESBL-producing E. coli strains was performed using BioNumerics version 7.6 (Applied Maths, Sint-Martens-Latem, Belgium). The primers used for bla gene detection and genotyping of E. coli are listed in Supplementary Table S4 and S5, respectively.

Conjugation test

To assess whether the ESBL-producing E. coli strain could horizontally transfer its bla genes, a conjugation test was performed. The ESBL-producing E. coli strain served as the donor, while sodium azide-resistant E. coli J53 (ATCC BAA-2730™) was used as the recipient. In brief, sodium azide-resistant E. coli J53 (recipient) and donor strains were grown overnight in 3 mL of LB broth at 35 °C with shaking at 120 rpm. The overnight cultures of the donor strains were then serially diluted 10-fold up to 10⁻³. A mixture of 100 µL of each donor strain and 100 µL of the undiluted recipient strain was incubated in 1.5 mL Eppendorf tubes for 24 h at 35 °C with shaking at 120 rpm. The conjugation mixtures were then serially diluted, and 100 µL of the appropriate dilutions was plated in duplicate onto MacConkey agar plates containing 2 µg/mL cefotaxime and 150 µg/mL sodium azide (Merck, Darmstadt, Germany). The plates were incubated at 37 °C for up to 48 h29. When transconjugant colonies were observed, a lysate was prepared from the colony to serve as a DNA template. PCR was then performed on this DNA template using primers specific to the bla genes of the donor E. coli strain to confirm the transmission of the bla genes.