Fecal samples were collected from cattle and a farmer on a single cattle farm in the southern part of Kyushu Island, a major cattle farming area in Japan, in 2013. The fecal samples were directly inoculated onto deoxycholate-hydrogen sulfide-lactose agar plates and incubated at 37 °C for 24 h. Three colonies were randomly selected, and the species were identified using matrix-assisted laser desorption ionization-time-of-flight mass spectrometry (Vitek MS system; bioMérieux, Co., Ltd.). The isolates identified as E. coli were used for analyses. The Ethical Review Committee of the Teikyo University School of Medicine (No. 13–118) approved the study protocol, and the farmer provided written informed consent to participate in the study.

Antimicrobial susceptibility of the isolates was determined through agar microdilution according to the Clinical & Laboratory Standards Institute guidelines, and quality control was performed using the reference strain E. coli ATCC 25922 (CLSI, 2025). The presence of ESBL genes in the third-generation cephalosporin-resistant E. coli isolates was identified using multiplex PCR (Dallenne et al., 2010). Gene-specific PCR was performed to identify the genotype, and the amplified products were confirmed using DNA sequencing (Ohnishi et al., 2013). Sequence alignment and analysis were performed on the National Center for Biotechnology Information website using the Basic Local Alignment Search tool (Altschul et al., 1990)¹.

Genomic DNA was extracted using the QIAGEN Genomic-tip 500/G kit (Qiagen, Germany). WGS was performed on both the Illumina MiSeq platform (Illumina Inc., USA) and Oxford Nanopore MinION platform (Oxford Nanopore Technologies, UK). Hybrid de novo assemblies were generated using Unicycler v0.5.0 (Wick et al., 2017), and genome annotation was conducted via DFAST v1.6.0 on default parameters.

Plasmid replicon types were determined with PlasmidFinder, and plasmid/genome structures were compared and visualized using Easyfig v2.2.2. Antimicrobial resistance genes were identified using ResFinder² with thresholds ≥90% identity and ≥60% minimum length. Multilocus sequence typing (MLST) of E. coli isolates was performed, and STs were assigned using the PubMLST database³ (Wirth et al., 2006). Pairwise nucleotide identity among bla_CTX-M-14-encoding plasmids was evaluated using the JSpeciesWS online platform⁴ (Richter and Rosselló-Móra, 2009). Average Nucleotide Identity based on MUMmer (ANIm) was calculated for all plasmid pairs under default settings.

cgMLST was performed using cgMLSTFinder 1.2 (Center for Genomic Epidemiology).⁵ Assembled genomes of all isolates were submitted to the pipeline, and allele calling was performed against the E. coli cgMLST scheme (2,513 loci). The output data included the total number of called alleles, ST assignments [core-genome Sequence Type (cgST)], and allelic distances between isolates. A minimum-spanning tree was constructed from the cgMLST allelic profiles using GrapeTree v2.2 (Zhou et al., 2018). The resulting phylogenetic relationships were further visualized and annotated with FigTree v1.4.4.⁶ Isolates with identical cgSTs and zero allelic distance were considered clonally indistinguishable, whereas those with small allelic differences (≤10 alleles) were regarded as highly related, according to previously described criteria (Zhou et al., 2021).

Chromosomal core-SNP analysis was performed to assess the genetic relatedness. Therefore, plasmid sequences were excluded, and only chromosomal sequences were used. Sequence reads were mapped to the E. coli TK3888 genome (GenBank accession no. AP044770), which was used as the reference, using the CLC Genomics Workbench v24.0 (Qiagen) with default mapping parameters. To ensure accuracy, SNPs were called under the following thresholds: minimum coverage of 10×, minimum variant frequency of 90%, and minimum base quality score of 20. Putative SNPs located in repetitive or low-complexity regions were excluded, and only high-confidence SNPs in conserved chromosomal regions were retained for downstream analyses.

An SNP alignment was generated from all isolates, and a maximum-likelihood phylogenetic tree was constructed in CLC Genomics Workbench using the Jukes–Cantor model with 1,000 bootstrap replicates. Pairwise SNP distances were also calculated to determine the degree of genetic relatedness between isolates. Based on prior studies on Enterobacteriaceae, isolates differing by 0–5 core-SNPs were considered clonally identical, whereas larger distances (e.g., >15–25 core-SNPs) were interpreted cautiously as indicative of more distant relationships (Dallman et al., 2015; Ludden et al., 2021).

Conjugation experiments were performed using the broth mating method with E. coli J53 (sodium azide–resistant) as the recipient, as previously described (Masui et al., 2022). Briefly, donor and recipient strains were grown to the exponential phase in Luria–Bertani (LB) broth, mixed at a 1:1 ratio (vol/vol), and incubated overnight at 37 °C without shaking. After which, aliquots of the mating mixtures were plated onto LB agar supplemented with cefpodoxime (8 μg/mL) and sodium azide (100 μg/mL) to select for transconjugants. The presence of the resistance gene bla_CTX-M-14 in transconjugants was confirmed using PCR. Conjugation frequency was calculated as the number of transconjugant colonies (cfu/mL) divided by the total number of recipient cells (cfu/mL).

Overall, 15 isolates (three each from one parent cattle, three calves, and the farmer) were obtained and identified as 13 E. coli and 2 Klebsiella pneumoniae isolates. Among them, seven E. coli isolates exhibited resistance to third-generation cephalosporins with an ESBL phenotype (Table 1). One isolate each was recovered from the parent cattle, two calves, and the farmer, whereas three isolates were obtained from the remaining calf. These seven isolates displayed diverse susceptibility profiles to other antimicrobials. Five isolates were resistant to levofloxacin and carried mutations in the quinolone resistance-determining regions of gyrA and parC (Figure 1). Four isolates carried tet(A) and exhibited resistance to tetracycline, two of which also carried aph(3′)-Ia or aph(3′)-IIa and were resistant to kanamycin. All seven isolates harbored the bla_CTX-M-14 gene. Additionally, two isolates obtained from calf 3 carried further resistance determinants: TK3887 harbored aac(3)-IIa and was resistant to gentamicin, whereas TK3888 carried mcr-3.1 and was resistant to colistin.

MLST analysis revealed five distinct STs among the seven isolates (Table 1). Notably, the three isolates recovered from calf 3 belonged to different STs. Shared STs were identified between the parent cattle and calf 1 (ST533), as well as between calf 3 and the farmer (ST448).

To investigate the genetic relatedness of the isolates, both cgSNP analysis and cgMLST were performed (Table 1; Figure 1). The two ST533 isolates from the parent cattle (TK3893) and calf 1 (TK3946) differed by only one SNP in the core genome, strongly suggesting clonal identity. Consistently, both isolates were assigned to the same cgST (138832), further confirming their close genetic relationship. This represents a direct example of clonal dissemination between livestock individuals within the same farm.

Contrastingly, ST448 isolates from calf 3 (TK3889) and the farmer (TK3896) differed by 3,891 core SNPs and had distinct cgSTs (137,535 and 153,627), showing a distant relation, despite their sharing of the same ST (Supplementary Figure 1). Other isolates (ST1261, ST1148, and ST1431) differed by over 15,000 SNPs with distinct cgSTs, consistent with independent lineages. The three isolates from calf 3 belonged to different STs and cgSTs, indicating coexistence of multiple unrelated bla_CTX-M-14-positive E. coli lineages within a single animal.

These findings highlight clonal dissemination among livestock (ST533) and demonstrate that shared STs alone (e.g., ST448) do not necessarily indicate recent transmission, underscoring the value of integrating SNP and cgMLST analyses.

Hybrid assembly and comparative analysis revealed that the bla_CTX-M-14 gene in all positive isolates was consistently located on highly similar IncI1-type plasmids of approximately 114 kb in size (Figure 2). These plasmids exhibited a conserved backbone structure with only minor variations in accessory regions. Pairwise ANIm analysis using JSpeciesWS demonstrated extremely high sequence similarity among the seven IncI1 plasmids, with ANIm values ranging from 99.98 to 100%, and alignment coverage between 99.1 and 100% (Supplementary Table 1). These results clearly indicate that nearly identical bla_CTX-M-14-encoding IncI1 plasmids were shared among the cattle and farmer within the same farm.

No additional resistance genes were encoded on these IncI1 plasmids, indicating that they primarily functioned as dedicated vectors for the dissemination of bla_CTX-M-14. Other resistance determinants—mcr-3.1, aac(3)-IIa, aph variants, tet(A), and bla_TEM-1B—were carried on separate plasmids belonging to different incompatibility groups (e.g., IncFIB/FII, IncFIB/FIC, IncX1; Figure 1), highlighting the diversity of plasmid backgrounds contributing to multidrug resistance within the same farm.

Notably, the two clonally identical ST533 isolates from the parent cattle (TK3893) and calf 1 (TK3946) carried indistinguishable IncI1 plasmids, as well as nearly identical IncFIB/FIC plasmids encoding tet(A) and aminoglycoside resistance genes. Moreover, the bla_CTX-M-14-encoding IncI1 plasmids exhibited high sequence similarity (>99.9%) to an IncI1 plasmid deposited in GenBank (accession no. LC567051), based on BLAST analysis (Figure 2). This reference plasmid was isolated from a clinical E. coli strain in Japan in 2015. This finding is consistent with the possibility that highly conserved plasmids can disseminate across human and livestock reservoirs.

Conjugation experiments were performed with all seven CTX-M-14–producing E. coli isolates. The transfer frequencies of the plasmids ranged from 10⁻² to 10⁻³ per recipient, indicating a relatively high conjugation potential (Table 1). All transconjugants consistently carried only the bla_CTX-M-14–encoding IncI1 plasmid, whereas plasmids harboring additional resistance determinants, such as aac(3)-IIa and mcr-3.1, were not co-transferred. The antimicrobial susceptibility profiles of the transconjugants were uniform, and their MIC values for third-generation cephalosporins were comparable across all donor strains (Supplementary Table 2).

To the best of our knowledge, this is the first report in Japan providing genome-level evidence that nearly identical IncI1 plasmids carrying bla_CTX-M-14 were shared between humans and animals in a farm. CTX-M-14–producing E. coli are frequently isolated from livestock (Hayashi et al., 2018; Nakano et al., 2023) and prevalent among individuals (Yano et al., 2013; Komatsu et al., 2018; Masui et al., 2022). This suggests widespread dissemination across human and animal reservoirs. Similar patterns have been reported in other Asian countries, especially China, highlighting the regional significance of CTX-M-14 in both clinical and agricultural contexts (Liu et al., 2018; Zheng et al., 2019; Chen et al., 2024).

A major factor in this predominance is the association of bla_CTX-M-14 with IncI1 plasmids, exhibiting high conjugation efficiency, broad host range within Enterobacterales, and stability during bacterial replication (Cottell et al., 2011; Di Pilato et al., 2019). These properties facilitate persistence and rapid dissemination across bacterial lineages, host species, and ecological niches (Beyrouthy et al., 2021). The strong linkage between bla_CTX-M-14 and IncI1 plasmids likely underlies the successful establishment and widespread distribution of bla_CTX-M-14-positive E. coli in both clinical and livestock environments.

Notably, among all isolates recovered from cattle and the farmer, bla_CTX-M-14 was the only ESBL gene detected. The absence of other ESBL types within the same farm suggests a limited diversity of ESBL-producing E. coli at the time of sampling, which may reflect a relatively restricted introduction of ESBL determinants into this farm environment.

cgSNP analysis revealed that E. coli isolates from the parent cattle and calf 1 belonging to ST533 differed by only a single SNP, indicating clonal identity. Consistent with this, cgMLST analysis noted that these two ST533 isolates shared the same cgST, further supporting their close genetic relationship. This represents an example in which a parent cattle and her calf on the same farm shared a clonally identical strain at the chromosomal level (0–1 core SNP difference), a finding that is biologically plausible given the close contact between animals. Such a precise one-to-one chromosomal match between a specific animal pair (parent cattle and calf) within the same farm has been infrequently reported; most previous studies have instead described low-SNP clusters involving multiple animals and/or humans within a shared farm or environmental setting (Pietsch et al., 2018; Peng et al., 2022; Bachmann et al., 2024).

The seven CTX-M-14–producing isolates encompassed five distinct STs (ST1431, ST533, ST1148, ST448, and ST1261), reflecting genetic diversity within the farm. The three isolates from calf 3 exhibited different STs, indicating multiple circulating lineages. Among these, ST533 has been frequently detected in livestock from both Asia and Europe, suggesting adaptation to animal hosts (Huber et al., 2013; Dahms et al., 2015). Conversely, ST448 has been detected in both humans and animals, highlighting its potential to cross host boundaries and disseminate between humans and animals (Blaak et al., 2014; Quiñones et al., 2020; Sivarajan et al., 2025).

The identification of clonally identical ST533 isolates between the parent cattle and calf 1, supported by both SNP and cgMLST data, strongly indicates recent transmission and local clonal expansion within the livestock population. However, the ST448 isolates from calf 3 and the farmer differed by more than 3,800 core SNPs. These findings underscore the utility of high-resolution genomic analysis in detecting direct animal-to-human transmission, which may be difficult to infer from conventional typing methods. Thus, chromosomal SNP analysis demonstrated that the clonal spread of bla_CTX-M-14-positive E. coli occurs independently of plasmid transfer, emphasizing the dual mechanisms whereby resistance persists and disseminates on farms. These relationships are schematically summarized in Figure 3.

The transfer frequencies of IncI1 plasmids have been reported at levels of 10⁻²–10⁻³ per donor cell, which is considerably higher than those of many other plasmid types. Additionally, they stably persist without imposing a significant fitness cost on the host bacterium (Carattoli, 2009; Rozwandowicz et al., 2018). These properties facilitate the long-term maintenance and interspecies transfer of bla_CTX-M-14.

Notably, we found that the 99.9% similarity between the plasmids detected in this study and those isolated in Japan further underscores the close genetic relationship between plasmids circulating in human and livestock populations. Therefore, IncI1 plasmids act as a “mobile resistome,” linking bacterial communities across different reservoirs. Similar observations have been reported in other countries, where IncI1 plasmids carrying bla_CTX-M-14 were found to disseminate across diverse Enterobacterales lineages and between human and animal hosts (Liao et al., 2015; Yu et al., 2024). Our results reinforce the fact that plasmid-mediated dissemination is a major driver of ESBL spread in both clinical and agricultural settings.

In addition, some isolates carried other clinically important resistance genes. One calf-derived isolate harbored aac(3)-IIa (gentamicin resistance), and another carried mcr-3.1 (colistin resistance). The presence of mcr genes in livestock is concerning, as colistin is a last-resort antimicrobial for multidrug-resistant Gram-negative infections (Liu et al., 2016; Yin et al., 2017; Wang et al., 2018). Tet(A) and aph(3′) genes, conferring resistance to tetracyclines and kanamycin, respectively, were also detected, reflecting selective pressure from long-standing antimicrobial use in livestock (Van Boeckel et al., 2019). The coexistence of bla_CTX-M-14 with additional resistance determinants on separate mobile elements may facilitate co-selection and persistence of multidrug resistance. These findings underscore the public health risk of accumulating multiple resistance genes in livestock-associated E. coli and highlight the need for ongoing genomic surveillance within a One Health framework.

This study has some limitations. First, the sampling was restricted to a single farm with a limited number of cattle and a single participant, which may not represent broader epidemiological trends. Second, although clonal and plasmid sharing were observed, the precise transmission direction (human-to-animal, animal-to-human, or environment-mediated) could not be determined. Third, longitudinal and environmental sampling was not performed, preventing assessment of temporal dynamics and external sources. Fourth, detailed epidemiological metadata such as the animal age, origin, housing conditions, or duration of cohabitation are lacking from data collected in this study, limiting our ability to fully reconstruct transmission pathways or exclude environmental or external sources. Finally, we did not investigate the detailed structural plasticity of the plasmids, such as transposable elements or recombination events.