Extended-spectrum β-lactamase-producing Escherichia coli with chromosomally integrated resistance genes isolated from a wild Asian black bear Authors: Michiyo SUGIYAMA, Tomoya MORIMOTO, Tetsuo ASAI Published: 2026 ABSTRACT Carriage of antimicrobial-resistant bacteria in wildlife is a growing public health concern. We isolated an extended-spectrum β-lactamase-producing Escherichia coli from the feces of a wild bear collected in a persimmon orchard in 2025. The bear isolate exhibited resistance to ampicillin, cefazolin, cefotaxime, tetracycline, and chloramphenicol. Whole-genome sequencing revealed that the strain belonged to ST5044-OUT:H29 and harbored the blaCTX-M-55 and other resistance genes integrated into the bacterial genome. Comparative genomic analysis showed close relatedness to isolates from a Vietnamese pig farmer and a Japanese slaughterhouse wastewater, with pairwise distances of 0–80 single-nucleotide polymorphisms. DNA metabarcoding revealed a diet dominated by wild vegetation, with a minor anthropogenic plant component. Thus, wild bears may contribute to the dissemination of antimicrobial-resistant E. coli at the human–wildlife interface. The presence of antimicrobial-resistant bacteria (ARB) in wild animals has become a growing public health concern, because ARB shed in wildlife excreta may pose risks to humans [10]. Of particular concern is the potential spillover of resistant Escherichia coli clones associated with human activities into wild animal populations. Wildlife may subsequently act as secondary reservoirs and contribute to the environmental circulation of clinically important resistant bacteria, potentially leading to human exposure [13]. In Japan, the prevalence of antimicrobial-resistant E. coli in wild mammals and birds has remained low [3, 5, 6, 14]. For example, one study reported a resistant E. coli carriage rate of approximately 2.1% in wild Japanese deer between 2016 and 2019 [14]. Among 366 wild mammals between 2018 and 2021, 26 (7.1%) carried E. coli resistant to at least one antimicrobial [5]. Overall, the reported prevalence of antimicrobial-resistant E. coli in wild animals has been far lower than that in humans, livestock, and companion animals [10]. Variation in ARB prevalence among wildlife species appears to depend on their proximity to human activities, feeding ecology, behavioral traits, and habitat characteristics [10]; however, genomic evidence linking ARB in wildlife to anthropogenic exposure remains limited in Japan. In 2025, the intrusion of bears into residential areas became a serious problem in Japan [15]. We isolated extended-spectrum β-lactamase (ESBL)-producing E. coli from bear feces collected in a persimmon orchard (Fig. 1Fig. 1.Bear feces collected in a persimmon orchard.). To investigate the potential route of ARB acquisition in the bear, we analyzed both the characteristics of the isolate and the feeding habits of bears in the area. Bear feces collected in a persimmon orchard. A fecal sample, presumed to be from a wild Asian black bear (Ursus thibetanus), was collected using a new plastic bag from a persimmon orchard in Ono-cho, Gifu Perfecture by a clinical veterinarian on October 28, 2025 (Fig. 1). E. coli resistant to third-generation cephalosporins and quinolones was examined using deoxycholate hydrogen sulfide lactose agar (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) supplemented with 1 mg/L cefotaxime or 50 mg/L nalidixic acid, as previously described [3]. The minimal inhibitory concentrations (MICs) of the antimicrobials were determined using the broth microdilution method on a custom-made plate (GDB7; Eiken Chemical Co., Ltd., Tokyo, Japan). E. coli ATCC 25922 was used as a quality control strain following the guidelines of the Clinical and Laboratory Standards Institute (CLSI) [9]. The MICs of all antimicrobials were interpreted according to CLSI standards [9]. Genomic DNA was extracted from bacterial cell suspensions inactivated with phenol–chloroform and subjected to whole-genome sequencing, as previously described [4]. Sequencing libraries were prepared using the Illumina DNA Prep Kit (Illumina, San Diego, CA, USA), and paired-end sequencing (2 × 150 bp) was performed using the iSeq system (Illumina), yielding an average genome coverage of more than 30-fold. After de novo assembly of short-read sequences using CLC Genomics Workbench (Qiagen, Hilden, Germany), the draft genome had a total length of 4,588,813 bp, consisting of 60 contigs, with an N50 value of 195,834 bp. The largest contig was 616,990 bp, and the GC content was 50.7%. The assembled contigs were analyzed using MLST 2.0, SeroTypeFinder 2.0, and ResFinder 3.1 to determine the sequence type (ST), serotype, and antimicrobial resistance gene profiles, respectively. Core genome single nucleotide polymorphism (cgSNP) analysis was performed using CSI Phylogeny. Strain FLAG-89322 (accession no. SRR25629957) was used as the reference genome for SNP calling. All web-based tools were provided by the Center for Genomic Epidemiology (http://www.genomicepidemiology.org). All analyses were performed using default parameters. The genomic locations of contigs harboring antimicrobial resistance genes were assessed using the Basic Local Alignment Search Tool (BLAST) against the National Center for Biotechnology Information nucleotide database to determine whether they were likely derived from chromosomal or plasmid sequences. For complete genome sequencing, genomic DNA was extracted using the Genomic DNA Buffer Set and QIAGEN Genomic-tip 100/G (QIAGEN). Libraries were prepared using the Rapid Barcoding Sequencing Kit (SQK-RBK114.24; Oxford Nanopore Technologies, Oxford, UK) and sequenced on the MinION platform (Oxford Nanopore Technologies) using an R10 flow cell. Basecalling was performed using Dorado basecall server v7.11.2 (Oxford Nanopore Technologies). Hybrid genome assembly was performed using Unicycler version 0.5.0 with default parameters [19], which includes internal polishing steps based on short-read data. The long-read sequencing data yielded an N50 of 4,647,345 bp with an average coverage of approximately 40×. To identify the genomic locations of antimicrobial resistance genes, contig sequences were annotated using DFAST (https://dfast.ddbj.nig.ac.jp/). The dataset generated and analyzed in this study has been deposited in the GenBank database under accession number AP045833. For diet analysis, DNA metabarcoding was performed as described previously [2, 4]. The interior portion of each fecal sample was collected using a sterile plastic spoon and used for subsequent analysis. Fecal DNA was extracted using the GeneCheck DNA Extraction Kit (GC002-1N; Fasmac Co., Ltd., Atsugi, Japan) following the manufacturer’s instructions. DNA was amplified using universal primers for mammals (MiMammal) [17], birds (MiBird) [18], fish (MiFish) [11], and plants [7], with slight modifications to the adapter sequences to incorporate Illumina indices (Supplementary Table 1). Polymerase chain reaction products were purified and used to prepare sequencing libraries with Illumina DNA/RNA unique dual indexes (Illumina), followed by paired-end sequencing on the Illumina iSeq platform. Reads were processed using QIIME2 (Quantitative Insights Into Microbial Ecology 2; QIIME 2 Development Team) [8]. Forward and reverse reads were merged, and low-quality tails (<98%) were trimmed. Primer sequences were removed, and reads with an expected error rate ≥1% or length ≤100 bp were discarded. Amplicon sequence variants were generated using the Unsupervised Noise Reduction algorithm and identified using BLAST with a 98.5% identity threshold. Plant species were further confirmed using YList (http://www.ylist.info/index.html). Candidate colonies were observed on plates containing cefotaxime but not on plates containing nalidixic acid. Three isolates were identified as E. coli. The bear isolates exhibited resistance to ampicillin, cefazolin, cefotaxime, tetracycline, and chloramphenicol (Table 1Table 1.Antimicrobial susceptibility profiles of bear-derived Escherichia coli isolates recovered using cefotaxime-supplemented deoxycholate hydrogen sulfide lactose agarAntimicrobials (breakpoint: mg/L)Strain IDwlxctx60wlxctx61wlxctx62Ampicillin (32)>128>128>128Cefazolin (8)>128>128>128Cefotaxime (4)>64>64>64Meropenem (4)≤0.25≤0.25≤0.25Gentamicin (16)888Kanamycin (64)321632Tetracycline (16)323232Nalidixic acid (32)888Ciprofloxacin (1)0.50.50.5Colistin (4)0.50.50.5Chloramphenicol (32)6412864Sulfamethoxazole/trimethoprim (76/4)19/119/119/1). Although an elevated MIC was observed for ciprofloxacin (0.5 mg/L), the MIC for nalidixic acid was low (8 mg/L). Among these three isolates, one (strain ID: wlxctx60) was selected for short-read sequencing. This strain belonged to ST5044-OUT:H29 and harbored the resistance genes blaCTX-M-55, aph(6)-Id, aph(3”)-Ib, tet(A), floR, sul2, and qnrS4. The presence of qnrS4 in the absence of quinolone-resistance–determining region mutations accounted for the low-level fluoroquinolone resistance phenotype observed in this isolate. Previous studies have reported blaCTX-M-55–producing E. coli in foxes, badgers, and great cormorants, tet(A) in deer, foxes, raccoons, masked palm civets, and great cormorants, and floR in deer, foxes, and great cormorants [3, 5]. Additionally, qnrS1 and qnrS2, rather than qnrS4, have been detected in great cormorants [3]. Table 1.Antimicrobial susceptibility profiles of bear-derived Escherichia coli isolates recovered using cefotaxime-supplemented deoxycholate hydrogen sulfide lactose agarAntimicrobials (breakpoint: mg/L)Strain IDwlxctx60wlxctx61wlxctx62Ampicillin (32)>128>128>128Cefazolin (8)>128>128>128Cefotaxime (4)>64>64>64Meropenem (4)≤0.25≤0.25≤0.25Gentamicin (16)888Kanamycin (64)321632Tetracycline (16)323232Nalidixic acid (32)888Ciprofloxacin (1)0.50.50.5Colistin (4)0.50.50.5Chloramphenicol (32)6412864Sulfamethoxazole/trimethoprim (76/4)19/119/119/1 Antimicrobial susceptibility profiles of bear-derived Escherichia coli isolates recovered using cefotaxime-supplemented deoxycholate hydrogen sulfide lactose agar All detected antimicrobial resistance genes were located on two contigs: one harbored aph(6)-Id, aph(3”)-Ib, tet(A), floR, and sul2, whereas the other contained blaCTX-M-55 and qnrS4. BLAST analysis suggested that the former was located on the chromosome, whereas the location of the latter could not be determined. Long-read sequencing revealed that blaCTX-M-55 and qnrS4 were located in adjacent positions, whereas aph(6)-Id, aph(3”)-Ib, tet(A), floR, and sul2 were located in other adjacent positions in the chromosome (Fig. 2aFig. 2.Genomic features and resistance gene context of Escherichia coli strain wlxctx60 from a wild bear. (a) Complete circular genome map of E. coli wlxctx60 (4,647,345 bp). From the outer to inner rings: annotated coding sequences (blue), GC skew (green, positive; purple, negative), and GC content (black). Two genomic regions are indicated as region A, containing aph(6)-Id, aph(3″)-Ib, tet(A), floR, and sul2, and region B, containing blaCTX-M-55 and qnrS4. The scale is shown in kilobase pairs (kbp). (b) Comparison of the genomic region (~10 kb) containing aph(6)-Id, aph(3″)-Ib, tet(A), floR, and sul2 among isolates from a bear (wlxctx60), a Vietnamese pig farmer (EC179), and slaughterhouse wastewater (T-R01940, T-R01950, T-R01955, T-R01957, and T-R03382). (c) Comparison of the genomic region containing blaCTX-M-55 and qnrS4 among the same isolates. Short-read sequences of the isolate wlxctx60 obtained in this study were de novo assembled using CLC Genomics Workbench (Qiagen, Hilden, Germany), whereas publicly available short-read datasets retrieved from the database were assembled using SPAdes v3.15.5 (https://github.com/ablab/spades) for comparative genomic analysis. Green and yellow arrows represent antimicrobial resistance genes and insertion sequences (IS elements), respectively. Grey shading between sequences denotes nucleotide identity (70–100%). The overall synteny and gene content are highly conserved, suggesting that this multidrug resistance region may be associated with mobile genetic elements and disseminated across different hosts and environments. Scale bar indicates 10 kb.), demonstrating integration of these genes into the bacterial genome. Antimicrobial resistance is generally associated with a biological cost, owing to the metabolic burden of plasmid replication, maintenance, and the expression of resistance genes [1]. In environments with little or no antimicrobial selective pressure, such as natural ecosystems and wildlife, bacteria may lose plasmid-encoded resistance genes over time as a consequence of this fitness cost. In contrast, when resistance genes are integrated into the chromosome, they cannot be readily lost, thereby enabling more stable persistence of resistance genes even in the absence of antimicrobial selective pressure [12]. Therefore, bacteria harboring chromosomally encoded resistance genes may contribute to the long-term persistence of ARB in wildlife populations. Collectively, these findings suggest that wild animals, including bears, harboring ESBL-producing E. coli, may contribute to the environmental dissemination of antimicrobial resistance within their habitats. Genomic features and resistance gene context of Escherichia coli strain wlxctx60 from a wild bear. (a) Complete circular genome map of E. coli wlxctx60 (4,647,345 bp). From the outer to inner rings: annotated coding sequences (blue), GC skew (green, positive; purple, negative), and GC content (black). Two genomic regions are indicated as region A, containing aph(6)-Id, aph(3″)-Ib, tet(A), floR, and sul2, and region B, containing blaCTX-M-55 and qnrS4. The scale is shown in kilobase pairs (kbp). (b) Comparison of the genomic region (~10 kb) containing aph(6)-Id, aph(3″)-Ib, tet(A), floR, and sul2 among isolates from a bear (wlxctx60), a Vietnamese pig farmer (EC179), and slaughterhouse wastewater (T-R01940, T-R01950, T-R01955, T-R01957, and T-R03382). (c) Comparison of the genomic region containing blaCTX-M-55 and qnrS4 among the same isolates. Short-read sequences of the isolate wlxctx60 obtained in this study were de novo assembled using CLC Genomics Workbench (Qiagen, Hilden, Germany), whereas publicly available short-read datasets retrieved from the database were assembled using SPAdes v3.15.5 (https://github.com/ablab/spades) for comparative genomic analysis. Green and yellow arrows represent antimicrobial resistance genes and insertion sequences (IS elements), respectively. Grey shading between sequences denotes nucleotide identity (70–100%). The overall synteny and gene content are highly conserved, suggesting that this multidrug resistance region may be associated with mobile genetic elements and disseminated across different hosts and environments. Scale bar indicates 10 kb. A strain of E. coli ST5044-OUT:H29 carrying blaCTX-M-55, aph(6)-Id, aph(3”)-Ib, tet(A), floR, mdf(A), and sul2 was previously reported in pig farmers in Vietnam [16]. The present isolate lacked mdf(A), but possessed qnrS4, suggesting distinct resistance profiles depending on the region and host population within the ST5044-OUT:H29 lineage. To clarify the genetic relatedness, 31 strains of E. coli ST5044 were selected from EnteroBase (https://enterobase.warwick.ac.uk/), because of the availability of detailed metadata, including source and country of origin, which allowed better contextual interpretation (Supplementary Table 2). cgSNP analysis revealed that isolates from a bear (wlxctx60), a Vietnamese pig farmer (EC179), and Japanese slaughterhouse wastewater (T-R01950, T-R01955, T-R01957, and T-R03382) formed a closely related cluster (Supplementary Fig. 1). Pairwise SNP distances within this cluster ranged from 0 to approximately 80 SNPs, indicating a high degree of genetic similarity. In contrast, other isolates included in the analysis were separated by substantially larger SNP distances and formed distinct lineages. These results indicate that wildlife-, human-, and wastewater-derived isolates are genetically closely related compared with other strains. The genetic context surrounding tet(A), consisting of IS26-ISVsa3-floR-ISVsa3-tet(A)-tetR(A)-aph(6)-Id-aph(3″)-Ib-sul2, was highly conserved among isolates from a bear (wlxctx60), a Vietnamese pig farmer (EC179), and wastewater samples from Japanese slaughterhouses (Fig. 2b). In addition, the genetic environments of blaCTX-M-55, with ISEcp1 located upstream and qnrS4 located downstream of blaCTX-M-55, were also highly similar among isolates from wildlife, a Vietnamese pig farmer, and slaughterhouse wastewater (Fig. 2c). These findings suggest that these resistance genes are embedded within mobile genetic elements, particularly insertion sequence-mediated structures. Although direct epidemiological links cannot be established among these isolates, the findings raise the possibility that multi-antimicrobial-resistant ST5044 strains circulate or disseminate globally across human, animal, and environmental compartments. To investigate potential associations among human activities, domestic animals such as pets and livestock, and wildlife, DNA metabarcoding was applied to characterize the feeding habits of the bear. More than 99% of the sequence reads were identified as black bear–derived using the MiMammal, MiBird, and MiFish primer sets. The ESBL-producing ST5044 strain was isolated from Japanese slaughterhouse wastewater, implying its distribution among food animals in Japan; however, no dietary DNA from other animal species was detected in the fecal samples. Consequently, the source of the ARB could not be inferred from the dietary DNA profile. In contrast, plant-derived sequences, including those of Aphananthe aspera (Muku tree), Diospyros morrisiana (Tokiwa persimmon), and Diospyros kaki (Kaki persimmon), were detected. Kaki persimmon includes both cultivated populations (orchards and abandoned farmland) and naturalized individuals, whereas Tokiwa persimmon represents a wild species occurring in mountainous regions. To improve clarity and transparency, we have summarized the read abundance of each plant taxon (Supplementary Table 3). The vast majority of the relative read abundance originated from the Muku tree (96.0%), with only minor proportions from Tokiwa persimmon (3.2%) and Kaki persimmon (0.8%), representing wild and anthropogenically influenced plant sources, respectively. These findings suggest that the bear subsisted primarily on wild vegetation, although the exact foraging location and timing of the bear remain uncertain. This study had several limitations. First, the findings were based on a single fecal sample obtained from a single wild Asian black bear, which limits the generalizability of the results to a broader bear population. Additionally, temporal information on bacterial acquisition was unavailable, making it impossible to determine whether colonization occurred before or after the intrusion of the bear into human-modified environments. In this study, although the route of ARB acquisition remains unclear, wild bears may still contribute to the dissemination of ARB between wildlife and human environments. Future studies with larger sample sizes, longitudinal sampling, and integrated surveillance across wildlife, humans, livestock, and the environment are necessary to clarify the ecology and transmission dynamics of antimicrobial-resistant E. coli strains in wildlife. CONFLICTS OF INTEREST The authors have nothing to disclose. Supplementary Material