Prevalence and genomic insights into Yersinia enterocolitica in Southeastern China (2008–2022)

Abstract

Abstract

Yersinia enterocolitica is a significant foodborne pathogen causing gastrointestinal illnesses worldwide. This study investigates the prevalence and genomic characteristics of Y. enterocolitica to assess potential health risks in southeastern China, a region lacking mandatory yersiniosis monitoring. From 2939 samples collected between 2008 and 2022, 105 isolates were recovered. The highest prevalence was found in rodents (8.1%), followed by retail meats (7.1%), other foods (3.7%), and human clinical cases (0.8%). In addition to meats and rodents, ready-to-eat salads, seafood, and frozen food products were identified as potential transmission vehicles. Various bioserotypes and sequence types (STs) was identified, including twelve previously unreported STs. Biotype 1A, exhibiting greater genetic diversity than more pathogenic biotypes (3 and 4), was frequently found in human clinical cases. Phylogenetic analysis revealed two main lineages, with isolates primarily clustered by biotype and pathogenic traits. Antimicrobial susceptibility testing revealed 46.7% (49/105) of isolates were multidrug resistant (MDR), with frequent resistance to polymyxin B (100%), azithromycin (50.5%), and sulfanilamide isoxazole (31.4%). These findings highlight the ecological complexity and diversity of Y. enterocolitica, especially non-pathogenic biotype 1A strains, and underscore the need for enhanced food safety and antimicrobial stewardship to mitigate the public health impact of Y. enterocolitica infections.

Key points

• Biotype 1 A strains exhibited greater genetic diversity than pathogenic biotypes.

• Pathogenic strains were mainly associated with lineage HC1490_2, not HC1490_10.

• Higher MDR levels were observed in biotype 3 and 4 strains.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00253-025-13500-7.

Introduction

Yersinia enterocolitica is a zoonotic foodborne pathogen commonly found in both wild and domestic animals (Shoaib et al. 2019). It is typically transmitted to humans through contaminated food or water, leading to cause acute yersiniosis, which usually manifests as self-limiting bloody diarrhea (Galindo et al. 2011). In severe cases, the infection can lead to extra-intestinal complications, such as mesenteric lymphadenitis, and post-infectious sequelae (Bottone 1997; Huovinen et al. 2010). Although yersiniosis can affect individuals of all ages, children under 5 years old are particularly vulnerable and may develop serious complications (Quion and Torga 2021). In immunocompromised individuals, antimicrobial treatment may be necessary due to the risk of sepsis or invasive infections (Fàbrega and Vila 2012). The emergence of antimicrobial resistance in Y. enterocolitica is of growing concern, with increasing resistance reported to fluoroquinolones, ampicillin, amoxicillin/clavulanic acid, cefazolin, and cefalotin (Frazão et al. 2017).

The pathogen is classified into six biotypes and over 60 serotypes, with varying levels of clinical and epidemiological significance (Bancerz-Kisiel et al. 2018) Biotypes 1B and 2–5 are the most commonly implicated in human and animal infections, while biotype 1 A is generally regarded as non-pathogenic due to the absence of key virulence factors such as the pYV plasmid (Bottone 1997). Similar to Salmonella, Y. enterocolitica exhibits a broad host range and is frequently isolated from animals and environments in biodiverse landscape; however, colonization in swine presents heightened risk for human infections (Bissett et al. 1990; Huang et al. 2023; Nesbakken et al. 2007). The widespread occurrence of Y. enterocolitica in the natural environments, combined with intensive farming practices in the meat, produce, and seafood industries, as well as global food trade, has reinforced its prominence in the global food chain (Gupta et al. 2015; Palau et al. 2024). After campylobacteriosis and salmonellosis, yersiniosis ranks as the third most prevalent zoonotic infection in the European Union and remains a leading cause of foodborne illness worldwide (European Food Safety Authority 2022).

Although Y. enterocolitica infections are closely monitored in developed countries, diagnostic efforts are often insufficient in many developing regions, including China (Chlebicz and Śliżewska 2018; Wang et al. 2021). Only two large outbreaks of yersinosis have been officially documented in China, dating back to the 1980 s, with subsequent cases reported sporadically (Duan et al. 2017; Wang et al. 2008, 2009, 2019). As the world’s largest producer, importer, and consumer of pork, China faces an elevated risk of yersiniosis due to potential contamination of pork and pork-derived products (Qi et al. 2021). Evidence also suggests that other food sources, including beef, poultry, fresh produce, seafood, frozen foods, and dairy products, may contribute to Y. enterocolitica infections (Bursová et al. 2017; Espenhain et al. 2019; Verbikova et al. 2018; Ye et al. 2015). Nevertheless, the distribution of Y. enterocolitica in retail meats and other food products in China, as well as its connection to human yersinosis, remain poorly understood.

To address this gap, we conducted a comprehensive study to analyze the prevalence and diversity of Y. enterocolitica in southeastern China. Over a 15-year period, we collected 2939 samples and sequenced 105 genomes of Y. enterocolitica isolates. This study aims to characterize the prevalence, antimicrobial resistance profiles, and genetic diversity of Y. enterocolitica strains from various sources, including food products, humans, and rodents. We also examined the transmission dynamics between environmental reservoirs and human clinical cases, to better understand the public health implications of this emerging pathogen.

Materials and methods

Sample collection

Between 2008 and 2022, a total of 2939 samples were collected in southeastern China as part of the national foodborne pathogen surveillance program, including human clinical cases (n = 977), rodents (n = 542), retail meats such as poultry (n = 399) and livestock (n = 530), as well as various food products (n = 491).

Retail meat and food samples, excluding frozen products, were stored at 4 °C for up to 24 h before processing to maintain sample integrity. Rodents were trapped in both rural and urban areas, including various farm types, to ensure a broad representation of domestic and wild environments. Upon capture, rodents were transported alive to the laboratory in mobile cages, where they were euthanized, and gastrointestinal samples were immediately collected for further analysis.

Human specimens were collected from patients with diarrhea at sentinel hospitals in the greater Wenzhou area. A total of 977 specimens were obtained through anal swabs or fecal samples from sporadic cases of diarrheal illness. All human specimens were stored in Cary-Blair transport medium (Oxoid, Basingstoke, UK) with ice packs and assessed within 4 h of collection.

Isolation and species identification

Rodent intestinal contents, as well as patient feces and anal swabs, were inoculated into 10 mL of modified phosphate-buffered saline (PBS) (Hopebio, Qingdao, China) for enrichment. Retail meats and food samples (25 g each) were suspended in 225 mL of PBS, homogenized for 1 min, and processed according to the National Food Safety Standard of China (GB4789.8–2016). Samples were incubated at 4 °C for 7, 14, and 21 days, considering the slow-growing nature of Yersinia at this temperature. After incubation, 0.5 mL of each enrichment was subjected to alkaline treatment (4.5 mL of 0.5% potassium hydroxide) for 15 s and streaked onto Cefsulodin Irgasan Novobiocin agar (CIN) with Yersinia supplement (Hopebio, Qingdao, China), followed by incubation at 26 °C for 48 h. For biochemical tests, presumptive colonies were streaked onto modified Kligler Iron Agar (KIA) slopes (Hopebio, Qingdao, China) and cultivated at 26 °C for 24 h. Positive colonies, characterized by no gas production and yellowish slant tubes (acid positive), were screened for urea hydrolysis (Thermo Fisher Scientific, Waltham, MA, USA). Urease-positive colonies were further evaluated for motility at 26 °C and 37 °C. Colonies exhibiting motility at 26 °C but not 37 °C were confirmed as Y. enterocolitica using VITEK2 compact system (BioMérieux, Lyon, France) and matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS, Bremen, Germany) according to the manufacturer’s instructions. All isolates were stored at – 80 °C in Microbank vials (ProLab Diagnostics Inc., Richmond Hill, Ontario, Canada) containing tryptic soy broth supplemented with 20% glycerol.

Bioserotyping

The biotype of each Y. enterocolitica isolate was determined through metabolic profiling, which included tests for lipase, esculin hydrolysis, salicin, trehalose, xylose fermentation, pyrazinamidase activity, nitrate reduction, and indole production, following the methodology described previously (Yuchun et al. 2010). Serogroup identification was performed using slide agglutination assays with commercial antisera (Denka Seiken, Tokyo, Japan) for O-serogroups O:1, O:2, O:3, O:5, O:8, and O:9. Positive reactions, indicated by visible agglutination, were interpreted as a match with the respective serogroup. Strains that did not agglutinate with any of the antisera were classified as “unknown” (UN). Y. enterocolitica CICC21669 was used as a positive control.

Whole genome sequencing of Y. enterocolitica

Positive isolates were subjected to Illumina sequencing (Suzhou IOTAbiome Biotechnology Co., Ltd., China). Briefly, a single colony picked from Trypticase Soy Agar (TSA) and cultivated in 1 mL of Trypticase Soy Broth (TSB; Hopebio, Qingdao, China) for 24 h in a shaking incubator at 26 °C. Genomic DNA was extracted from overnight cultures using the Wizard® Genomic DNA Purification Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Paired-end libraries (~ 300 bp inserts) were prepared using the VAHTS Universal DNA Library Prep kit (Vazyme Biotech, Nanjing, China) for Illumina V3 and sequenced on an Illumina NovaSeq 6000 using the S4 reagent kits (v1.5) according to the manufacturer’s instructions. Library construction and sequencing were conducted by the staff at Suzhou IOTAbiome Biotechnology Inc.

Bioinformatics analysis

Raw reads were quality-checked using FastQC v0.11.5 (Andrews 2010), trimmed with Trimmomatic v0.36, and de novo assembled using EToKi “assemble.” Genomes were aligned using EToKi v1.3 (Zhou et al. 2020) align onto a reference genome (GCF_001304755) to obtain a multiple sequence alignment of core genomic regions that were shared by ≥ 95% of the genomes. A maximum-likelihood phylogenetic tree was constructed using IQ-TREE, with recombinant regions eliminated by EToKi RecHMM (Zhou et al. 2020). Visualizations were created using GrapeTree v1.5 (Zhou et al. 2018) and iTOL v6.9 (https://itol.embl.de/).

Gene annotation was performed with PROKKA v1.14.6 (Seemann 2014). Antibiotic resistance genes were identified using AMRFinderPlus (Feldgarden et al. 2021), and the virulence determinants were predicted by comparing against the Virulence Factor Database (VFDB; http://www.mgc.ac.cn/VFs/) using BLAST v2.9. Multi-locus sequence types (STs) and core genome multi-locus sequence typing (cgMLST) for Yersinia were assigned via EnteroBase (https://enterobase.warwick.ac.uk/species/index/yersinia). The Achtman MLST scheme, which includes seven housekeeping genes (aarF, dfp, galR, glnS, hemA, rfaE, and speA), was used to determine STs of Yersinia (Laukkanen-Ninios et al. 2011). Hierarchical clustering (HierCC) was performed based on cgMLST profiles of 1553 coding loci (Zhou et al. 2020). To detect plasmids, all 6982 complete plasmids sequences from the NCBI RefSeq database (October 2022) were downloaded. The assembly of each Y. enterocolitica isolate was aligned against the complete plasmid sequences using BLAST, and the alignment scores for each plasmid were summarized. To ensure the reliability and representativeness of alignment, only hits with query coverage > 60% and sequence identity > 85% were considered reliable. All sequence data were deposited in EnteroBase (https://enterobase.warwick.ac.uk/species/index/yersinia) and their accession numbers were provided in Supplemental Table S1.

Antimicrobial susceptibility testing

Antimicrobial susceptibility of 105 Y. enterocolitica isolates was assessed using the microbroth dilution method across 27 antimicrobials from 12 classes, with resistance interpreted according to Clinical and Laboratory Standards Institute (CLSI) guidelines (Malvern 2022). The 27 antimicrobials and lowest minimum inhibitory concentrations (MICs, μg/mL) of the non-susceptible population were as follows: ampicillin (AMP, ≥ 32), ampicillin-sulbactam (AMS, ≥ 32/16), cefazolin (CFZ, ≥ 8), cefepime (FEP, ≥ 16), cefotaxime (CTX, ≥ 4), ceftazidime (CAZ, ≥ 16), cefoxitin (CFX, ≥ 32), aztreonam (AZM, ≥ 16), azithromycin (AZI, ≥ 32), ciprofloxacin (CIP, ≥ 1), gentamicin (GEN, ≥ 16), tetracycline (TET, ≥ 16), chloramphenicol (CHL, ≥ 32), nalidixic acid (NAL, ≥ 32), imipenem (IMI, ≥ 4), meropenem (MEM, ≥ 4), amikacin (AMI, ≥ 64), kanamycin (KAN, ≥ 64), doxycycline (DOX, ≥ 16), minocycline (MIN, ≥ 16), levofloxacin (LEV, ≥ 2), gemifloxacin (GEM, ≥ 1), trimethoprim-sulfamethoxazole (SXT, ≥ 4/76), sulfanilamide isoxazole (SIZ, ≥ 512), amoxicillin-clavulanic acid (AMC, ≥ 32/16), polymyxin B (PMB, ≥ 4) and colistin (CS, ≥ 4). Antibiotic classes consisted of penicillins (AMP), β-lactamides (AMS, AMC and AZM), cephalosporins (CFZ, FEP, CTX and CAZ), cephamicins (CFX), carbapenems (IMI and MEM), aminoglycosides (AMI, KAN, and GEN), tetracyclines (TET, DOX, and MIN), quinolones (NAL, LEV, GEM, and CIP), sulfonamides (SXT and SIZ), phenicols (CHL), peptides (PMB and CS), and macrolides (AZI). Escherichia coli ATCC 25922 was employed as a control organism. Strains resistant to three or more classes (excluding ampicillin, amoxicillin-clavulanic acid and cefazolin) were defined as multidrug-resistant (MDR).

Inferring transmission frequency

Transmission or host transfer events were inferred by comparing host and country information between ancestral and descending nodes of phylogenetic branches. Transmission frequency between different hosts was evaluated using the method as previously described (Li et al. 2024).

Statistical analysis

Statistical analysis was performed using SPSS (Statistics 20, IBM, Armonk, NY, USA). Differences in the occurrence of positive samples and Y. enterocolitica isolates from various sources were analyzed using non-parametric tests. Statistically significant associations between resistance phenotypes and variables such as source, isolation year, and bioserotyping were determined using Chi-square tests. A P value < 0.05 was considered as statistically significant.

Results

Prevalence and distribution of Y. enterocolitica in southeastern China

A total of 105 Y. enterocolitica isolates were recovered from 2939 collected between 2008 and 2022 across various sources in Zhejiang Province, China (Table 1). The distribution of Y. enterocolitica varied significantly across source types (P < 0.001). The highest recovery rate was found in rodents (8.1%, 44/542), followed by retail meats (7.1%, 66/929) and miscellaneous food products (3.7%, 18/491). Among retail meats, Y. enterocolitica was more prevalent in livestock products (9.1%, 46/508) compared to poultry (20/421, 4.8%; P < 0.01). Despite nearly 1000 specimens collected from patients with diarrhea, only eight cases of yersiniosis were confirmed (0.8%, 8/977). All human samples positive for Y. enterocolitica were carefully screened to exclude co-infections with other enteropathogens. Detailed information regarding all positive isolates is provided in Supplemental Tables S2 and S3.

Table 1.

Prevalence of Yersinia enterocolitica by sample type during 2008–2022

Source	Sample type	Sample size	Positive samples	Positive rate (%)
Retail meat	Livestocka	508	46	9.1%
	Poultry	421	20	4.8%
	Total	929	66	7.1%
Miscellaneous food	Salad, seafood, frozen food products	491	18	3.7
Human	Anal swabs, feces	977	8	0.8
Rodents	Intestinal contents	542	44b	8.1

^aThe meat types in the category of livestock consist of frozen and fresh beef, frozen and fresh pork, frozen and fresh mutton, imported frozen pork and beef

^b31 strains died during recovery and the remaining 13 rodent strains were used for later characterization

The 105 isolates were divided into nine bioserotypes, with 57 strains being unassignable due to a lack of reactivity with available diagnostic sera (Table 2). Among the assignable isolates, the most common bioserotype was 1 A/O:5 (n = 25), followed by 1 A/O:8 (n = 17) and 1 A/O:1,2 (n = 3). In rodents, the dominant bioserotype was 1 A/unknown (n = 10), with smaller numbers of 1 A/O:8 (n = 2) and 4/unknown (n = 1). Retail meats exhibited a broader diversity of bioserotypes, although some serotypes were exclusive to specific sample types. For instance, O:3 and 4/unknown were found only in human and rodent samples, respectively, while bioserotypes 1 A/O:1,2, 1 A/O:9, and 4/O:3 were found only in retail meats. Importantly, no biotype 1B strains were detected among the isolates.

Table 2.

Biotype and serotype distribution of Yersinia enterocolitica isolates

l		Retail meat (n = 66)	Miscellaneous food (n = 18)	Human (n = 8)	Rodents (n = 13)	Total (n = 105)
Bioserotype	1 A/O:1,2	3	0	0	0	3
	1 A/O:5	18	6	1	0	25
	1 A/O:8	10	3	2	2	17
	1 A/O:9	1	0	0	0	1
	1 A/UNa	31	8	4	10	53
	2/UN	2	1	0	0	3
	3/O:3	0	0	1	0	1
	4/O:3	1	0	0	0	1
	4/UN	0	0	0	1	1

^aUN unknown

Genetic characterization and sequence type diversity

In addition to bioserotyping, genomic analysis identified 54 distinct sequence types (STs), including 12 novel STs: four from meat, four from rodents, three from food products, and one shared between human and meat isolates. The most common STs were ST3 (n = 13), ST157 (n = 6), ST304 (n = 5), ST656 (n = 5), ST278 (n = 4) and ST359 (n = 4), collectively accounting for 35.2% (37/105) of the isolates. Notably, ST157 and ST656 were detected across human, meat, and rodent sources, indicating potential cross-sector transmission. Four STs (ST359, ST726, ST8, and ST125) found in human isolates were also frequently identified in both meat and food samples, while ST429 was unique to clinical isolates.

Biotype 1 A strains exhibited the greatest genetic diversity, representing 50 distinct STs. In contrast, biotype 2 strains were restricted to only two STs (ST26 and ST296). Biotypes 3 and 4 were associated with unique STs, with biotype 3 linked to ST429 and biotype 4 to ST18.

Phylogenomic insights and global comparison

To explore the genetic relationship of the isolates, we compared them to a global collection of 2,513 Y. enterocolitica genomes available on EnteroBase. Using a minimum spanning tree (MST) constructed from the cgMLST scheme, we observed clear clustering of isolates into divergent microclades (Fig. 1A). Hierarchical clustering at the HC1490 level (also known as cgMLST eBurstGroup), a recommended method for clustering Y. enterocolitica, grouped the isolates into four microclades (Achtman et al. 2022). The majority of isolates (97.1%, 102/105) clustered within HC1490_10, with three isolates assigned to HC1490_2. These two microclades, HC1490_2 and HC1490_10, likely represent the dominant lineages of Y. enterocolitica circulating in southeastern China.

Fig. 1.

Population structure of Yersinia enterocolitica in southeastern China during 2008 to 2022. (A) A maximum-likelihood phylogeny tree was based on 182,063 core SNPs from 549 representative genomes of Y. enterocolitica. Isolate relatedness was assessed using core genome multilocus sequence typing (cgMLST). Isolates recovered in this study (n=105) are red outlined. Two clades associated with the isolates in our collection are shaded in blue and carneose, respectively. (B) The maximum-likelihood phylogenetic structure of Y. enterocolitica with China southeastern isolates plus 443 global genomes from NCBI database. Different shades of branches in the tree represent different hierarchical clustering group. Data on the biotype, continent, source, sequence type and isolation year are mapped on the tree from the inner to the outer circle. Shades of red indicate the isolates obtained from our collection. UN = unknown

Further phylogenomic analysis, comparing southeastern isolates to 443 global isolates from the HC1490_2 and HC1490_10 clades, revealed a broad distribution of these lineages across multiple countries (Fig. 1B). This suggested that globally circulating lineages were well-established in southeastern China. Additionally, within specific clades, isolates from diverse sources—including human, rodent, and meat samples—were closely related, implying frequent transmission across environmental and clinical settings (Supplemental Fig. S1). Biotype 1 A isolates were predominantly found in HC1490_10, while biotypes 1B and 2–5 were more commonly associated with HC1490_2, further reinforcing the relationship between biotypes and their respective genomic lineages.

Virulence determinants and plasmid characterization

A total of 181 virulence determinants spanning 12 functional categories were identified across the 105 strains (Supplemental Table S4). Of these, 32% (58/181) were highly conserved across nearly all isolates (≥ 98%), including genes involved in regulation, type III secretion system (TTSS), and flagella production. However, other virulence factors exhibited substantial variability between lineages.

The presence of key virulence markers did not differ significantly between clinical and nonclinical isolates but showed some correlation with specific clonal groups and biotypes. For example, the attachment invasion locus gene (ail) was exclusively present in HC1490_2 strains and absent from all HC1490_10 strains (Fig. 2). Similarly, genes such as invA (invasin) and yst (thermostable enterotoxins) were also confined to HC1490_2 strains, suggesting a potential link between these markers and the pathogenicity of this lineage.

Fig. 2.

Hierarchically clustered heatmap illustrating the distribution of virulence genes in 105 Y. enterocolitica isolates. The maximum-likelihood phylogenetic tree was based on 221,919 core SNPs from 105 genomes. The tree metadata showed the source of isolates, biotype and virulence factors (VFs). The virulence genes with more than 98% carriage were not displayed in this figure. For more details, see Supplemental Table S3

Plasmid pYV and associated TTSS genes (including Icr, syc, ysc, tyeA, and virG/yscW)—critical for Y. enterocolitica virulence—were found exclusively in biotype 3 and 4 strains from HC1490_2, with no presence in biotype 1 A and 2 strains from HC1490_10. Despite the absence of pYV and other key chromosomal virulence genes, biotype 1 A and 2 strains in HC1490_10 still exhibited a wide range of virulence traits. These included auto-transportation (98.0%), iron acquisition (92.1–97.1%), regulatory functions (99.0–100%), adherence (2.9–59.8%), toxin production (47.1–98.0%), flagella formation (63.7–100%), and capsule formation (1.0–99.0%).

Antimicrobial susceptibility and resistance patterns

Antimicrobial susceptibility testing was conducted across 24 antimicrobial agents, excluding ampicillin, amoxicillin-clavulanic acid, and cefazolin due to the presence of chromosomally encoded β-lactamases (BlaA and BlaB) in Y. enterocolitica (Bonke et al. 2011). Among the remaining antibiotics, Y. enterocolitica isolates exhibited moderate resistance to azithromycin (50.5%, 53/105), sulfanilamide isoxazole (31.4%, 33/105), and nalidixic acid (30.4%, 32/105). In contrast, 99% (104/105) of the strains were sensitive to carbapenems, aminoglycosides, and chloramphenicol (Fig. 3A). Notably, all isolates were resistant to polymyxin B, and 46.7% (49/105) were classified as multidrug resistance, including two biotype 1 A strains (from rodent and meat sources) that were resistant to five or more antibiotic classes (Fig. 3B-C).

Fig. 3.

Antibiotic resistance profiles among 105 Y. enterocolitica isolates from 2008-2022. (A) Antimicrobial resistant rate of 105 Y. enterocolitica isolates against twenty-seven antimicrobials. The number of drug resistance patterns classified by (B) sources and (C) biotypes. Resistance to ampicillin, amoxicillin-clavulanic acid and cefazolin were excluded in accounting MDR levels. (D) Heat plot (right) that shown phenotypic (blue and turquoise boxes) and genotypic (brown boxes) antimicrobial resistance identified among 105 Y. enterocolitica isolates, which are ordered along the phylogeny (left). AMP, ampicillin; AMS, ampicillin-sulbactam; CFZ, cefazolin; FEP, cefepime; CTX, cefotaxime; CAZ, ceftazidime; CFX, cefoxitin; AZM, aztreonam; AZI, azithromycin; CIP, ciprofloxacin; GEN, gentamicin; TET, tetracycline; CHL, chloramphenicol; NAL, nalidixic acid; IMI, imipenem; MEM, meropenem; AMI, amikacin; KAN, kanamycin; DOX, doxycycline; MIN, minocycline; LEV, levofloxacin; GEM, gemifloxacin; SXT, trimethoprim-sulfamethoxazole; SIZ, sulfanilamide isoxazole; AMC, amoxicillin-clavulanic acid; PMB, polymyxin B; CS, colistin

Resistance profiles varied by source. Rodent isolates showed universal susceptibility to cefoxitin and colistin, whereas human isolates displayed higher resistance to cefoxitin (50%, 4/8), ciprofloxacin (25%, 2/8), and colistin (12.5%, 1/8) compared to other sources. One meat strain exhibited resistance to nine different antibiotic classes. However, no significant differences in MDR prevalence were observed between human and non-human sources. Biotype 4 strains demonstrated significantly higher MDR rates (P = 0.03) compared to biotype 2 strains, with biotypes 3 and 4 generally showing higher average MDR levels than biotypes 1 A and 2 (Fig. 3C).

Genotypic analysis revealed the presence of 21 antimicrobial resistance (AMR) determinants, with 11 located on plasmids and 10 integrated into chromosomal AMR cassettes. Due to the low occurrence of resistance genes in some antibiotic categories (e.g., penicillins, peptides, and cephalosporins), these were excluded from further analysis. Comparison between genotypic and phenotypic resistance profiles revealed discrepancies, with genotypic predictions showing high specificity (over 96%) but low sensitivity (< 22%) for aminoglycosides, β-lactams, sulfonamides, quinolones, and tetracyclines. These findings highlight the challenges of relying solely on genomic data for precise antimicrobial resistance surveillance (Fig. 3D).

Discussion

Y. enterocolitica is a significant zoonotic pathogen with multiple transmission routes, making it an emerging concern for food safety globally. Understanding the genetic and phenotypic diversity of Y. enterocolitica across various environmental and host sources is essential for evaluating its pathogenic potential and public health implications. In this study, we present a comprehensive 15-year analysis of Y. enterocolitica strains collected from southeastern China, integrating genomic data and AMR profiles to enhance our understanding of this pathogen within a “One Health” framework. Our findings reveal a highly diverse population of Y. enterocolitica, particularly biotype 1 A, which exhibited a broad range of sequence types (STs), including 12 novel STs previously unreported in the literature. This underscores the genetic heterogeneity of Y. enterocolitica in this region, with implications for both transmission dynamics and virulence potential.

Ecological reservoirs, foodborne sources, and demographic patterns of Y. enterocolitica in southeastern China

Rodents exhibited the highest recovery rate of Y. enterocolitica (8.1%), consistent with previous reports showing similar prevalence in rodents and wildlife globally (Arden et al. 2022; Liang et al. 2015; Shayegani et al. 1986). These results support the notion that rodents, due to their adaptability to both wild and anthropogenic environments, may play a pivotal role in pathogen transmission at the human-animal-environment interface (Young et al. 2014). While Y. enterocolitica-related foodborne outbreaks are infrequent and often without clear reservoirs, it is plausible that wild rodents, as a natural reservoir, contribute to such outbreaks indirectly via livestock or food products (Hayashidani et al. 1995). Among retail meat samples, livestock-derived products (pork, beef, and mutton) had a higher prevalence of Y. enterocolitica compared to poultry, although poultry contamination (4.8%) was also notable and exceeded the rates reported in a prior study from Shaanxi Province (3.3%) (Lü et al. 2022). Our findings are consistent with reports from Europe, including studies from Poland (Morka et al. 2021) and Switzerland (Stevens et al. 2025), that have identified similar trends and emphasize the need for targeted surveillance of livestock meats.

Y. enterocolitica was also detected in a variety of non-meat food commodities—including seafood, ready-to-eat (RTE) salads, and frozen products—at a combined rate of 3.7%. These results highlight the expanding spectrum of potential transmission vehicles beyond conventional meat sources. While meat has historically been considered the main source of yersiniosis, outbreaks involving non-meat items, such as an RTE salad outbreak in Norway (MacDonald et al. 2012) and a bean sprout-associated outbreak in Pennsylvania (USA) (Bissett et al. 1990), indicate the necessity for broader control strategies across the food supply.

Demographic analysis revealed a higher infection rate among adults (0.72%) compared to children, which contrasts with literature identifying young children as the primary risk group (Chakraborty et al. 2015; Jones et al. 2003). This contrast may reflect regional epidemiological variations or limitations in diagnostic capacity, especially given that adult infections often manifest as pseudo-appendicitis and are easily misdiagnosed in the absence of active surveillance (Bottone 1997). The clinical bioserotypes identified in this study included 1 A/O:8 (n = 2), 1 A/UN (n = 4), and 3/O:3 (n = 1). Notably, bioserotype 3/O:3 is dominant in other Chinese regions but is infrequently reported in Europe, where 4/O:3 accounts for approximately 80% of human infections (Duan et al. 2017; Yue et al. 2023). While Y. enterocolitica O:3 has also been reported in the USA, UK, and Japan, often associated with infant infections or handling of contaminated chitterlings (pig intestines), the sources of many outbreaks remain unidentified (Blumberg et al. 1991; Fukushima et al. 1997). Previous research showed that strains 3/O:3 and 4/O:3 have comparable susceptibility to O:3-specific phages, suggesting that differences in phage susceptibility do not fully explain the low prevalence of 4/O:3 in China (Liang et al. 2016; Shoaib et al. 2020). These findings underscore the need for improved diagnostic sensitivity and age-inclusive surveillance to fully capture the epidemiological landscape of Y. enterocolitica.

Genomic diversity and phylogenetic insights into Y. enterocolitica

Genomic analysis further confirmed significant heterogeneity in gene content, even among isolates from the same host. Shared STs identified from both human and other sources suggest potential links between them. Notably, we identified 12 novel STs in this study, which were not previously represented in the public database, thereby expanding the genomic diversity of Y. enterocolitica and supporting the concept of a complex genetic landscape. The variability observed among isolates is likely to contribute to the occurrence of mixed infections during outbreaks, as different Y. enterocolitica strains may circulate simultaneously within a given population.

Phylogenetic analysis identified two major lineages, HC1490_2 and HC1490_10, which partially corresponded to biotypes and key virulence markers. This suggests that the genomic diversity of Y. enterocolitica may be closely tied to its virulence potential, with certain genetic profiles correlating with varying levels of virulence. Previous studies have noted correlations between pathogenicity and clonal groups (Thomson et al. 2006; Wren 2003), which our data reaffirmed by showing that hallmark virulence determinants were primarily associated with lineage HC1490_2 rather than HC1490_10. In particular, the HC1490_2 lineage was associated with key virulence factors, including the chromosomal ail (attachment and invasion locus) and invA (invasin) genes, as well as the pYV plasmid, which is crucial for pathogenicity in Y. enterocolitica (Pierson and Falkow 1993; Reuter et al. 2014). In contrast, biotype 1 A strains were predominantly associated with HC1490_10, a lineage typically considered low pathogenic, yet these strains continue to be increasingly identified in human clinical cases, signaling a potential shift in their pathogenicity (Platt-Samoraj 2022).

A particularly noteworthy finding was the identification of two biotype 4 nonclinical strains within the HC1490_2 lineage, which suggests the existence of multiple transmission routes for Y. enterocolitica and highlights the potential for reservoirs beyond clinical settings. Furthermore, several isolates from indigenous rodents and meats harbored plasmids encoding virulence factors, including pYV, indicating the presence of highly virulent strains in nonclinical sources. These strains may represent an evolving pathogen capable of more efficient transmission and persistence in diverse environmental niches, though further investigation is needed.

Antimicrobial resistance patterns and therapeutic implications

Yersiniosis is typically self-limited, with drug therapy reserved for severely cases involving sepsis or focal infections (Guinet et al. 2011). Early treatment with trimethoprim-sulfamethoxazole remains effective in reducing Y. enterocolitica loads in southeastern China, though it may not shorten the duration of symptoms associated with uncomplicated enteritis or mesenteric adenitis (Quion and Torga 2021). Y. enterocolitica demonstrated relatively low resistance to critical antibiotics such as aminoglycosides, tetracyclines, phenicols, and carbapenems, consistent with prior reports on the susceptibility of Y. enterocolitica strains from humans, animals, and food sources (Koskinen et al. 2022; Piras et al. 2021, 2023). Five nonclinical strains exhibited reduced susceptibility due to the carriage of resistance-encoding plasmids, likely acquired from other enteric bacteria (Supplemental Fig. S2). However, human isolates demonstrated increasing resistance to therapeutically relevant agents, such as quinolones, macrolides, and third-generation cephalosporins, which could complicate treatment, especially in severe cases of yersiniosis or coinfections (Fàbrega and Vila 2012). The elevated MDR rates observed among biotype 3 and 4 strains are concerning, as these strains often carry critical virulence factors that may lead to severe outcomes (Leclercq et al. 2005).

A key observation in our study was the discrepancy between genotypic and phenotypic resistance profiles. Unlike the successful prediction of drug-resistant Salmonella (Fang et al. 2022), the ability of whole genome sequencing (WGS) to accurately identify antimicrobial resistance in Y. enterocolitica was limited. As expected, isolates lacking known resistance genes still exhibited resistance to penicillins, β-lactamides, and first-generation cephalosporins (cefazolin) due to the intrinsic production of β-lactamases (Schriefer et al. 2013). However, significant inconsistencies across other antibiotic classes highlight the need to further unravel the unknown resistome and intrinsic resistance mechanisms that could influence susceptibility in this species.

Conclusions

Y. enterocolitica may be more prevalent than previously recognized, with significant implications for both food safety and public health. Our study emphasizes the long-term persistence and genetic diversity of biotype 1 A strains in southeastern China, revealing a highly heterogeneous population compared to more pathogenic biotypes (3 and 4). Despite being considered less virulent, the increasing prevalence of biotype 1 A in clinical cases, coupled with rising antibiotic resistance, highlights an urgent need for improved diagnostic and treatment strategies. In addition to retail meats, our findings suggest that ready-to-eat salads, seafood, and frozen food products may also serve as important vehicles for human infections. In the context of globalized food systems and the rising burden of zoonotic diseases, this study underscores the complexity of Y. enterocolitica transmission and reinforces the need for a One Health approach. While pathogenic strains remain relatively scarce in southeastern China, continued surveillance and research are crucial for understanding the molecular mechanisms driving virulence and resistance, as well as the transmission dynamics of Y. enterocolitica across diverse environments and host settings.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

LF, AX, and YL conceived and designed the research. SL conducted the experiments with assistance from HL, ZX, YH, AX, and YW. Data analysis was carried out with contributions from SL, JR, and YZ. The manuscript was drafted by LF and YL and finalized and approved by all authors.

Funding

No funding was available for this work.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval

The studies using human specimens were reviewed and approved by the ethics committee of Zhejiang University. This study was conducted in accordance with the Declaration of Helsinki. All participants had completed informed consent.

Competing of interest

The authors declare no conflict interests.