Genomic epidemiology and antimicrobial resistance of non-typhoidal Salmonella from retail meats in Beijing, China: implications for foodborne transmission and phage susceptibility

Abstract

Objectives

To investigate the prevalence, antimicrobial resistance (AMR) profiles, genomic characteristics and phage susceptibility of non-typhoidal Salmonella (NTS) isolated from retail meats in Beijing, China, and to assess the potential for foodborne transmission and risks associated with phage application.

Methods

A total of 583 retail meat samples (pork, chicken, beef, mutton) were collected across Beijing (2021–2022). NTS isolates were identified by serotyping and MALDI-TOF. Antimicrobial susceptibility was tested by broth microdilution. Whole-genome sequencing was performed on 153 isolates for resistome, virulome, plasmid typing and core-genome MLST (cgMLST). Phage lysis assays used 46 phages against representative isolates, and prophages were predicted in silico.

Results

NTS prevalence was highest in pork (52.4%) and chicken (43.8%). Dominant serotypes were S. London and S. Enteritidis. Resistance rates were high for sulfisoxazole (74.6%), ampicillin (68.6%) and tetracycline (68.6%); 64.4% of isolates were multidrug-resistant (MDR). Genomic analysis revealed widespread resistance genes (e.g. bla_CTX-M-55, sul, mph(A)) associated with mobile genetic elements (ISEcp1, IS91). cgMLST showed high genetic similarity between foodborne isolates and contemporaneous human clinical isolates from multiple provinces. Phage lysis was highly effective against S. Enteritidis (83.3%) but limited against S. Kentucky; 34.0% of isolates carried prophages containing transposons.

Conclusions

Retail meats in Beijing, particularly pork and chicken, represent an important reservoir of multidrug-resistant NTS. Comparative genomic analysis revealed close genomic relatedness between retail meat isolates and human clinical isolates, suggesting the potential for foodborne transmission along the food chain. In vitro phage lysis assays demonstrated serotype-dependent susceptibility patterns, highlighting the potential of phage-based control strategies for specific Salmonella serotypes. However, prophages containing transposon-associated sequences were identified in a subset of isolates, and their contribution to horizontal gene transfer warrants further investigation. Enhanced surveillance integrating food, animal, environmental and human clinical Salmonella monitoring, together with serotype-targeted intervention strategies, is needed to support a One Health approach to foodborne disease control.

Introduction

Non-typhoidal Salmonella (NTS) is a major cause of foodborne gastroenteritis and invasive disease worldwide.¹ The burden is greatest among young children, older adults and immunocompromised individuals, in whom infection may cause severe dehydration, bacteraemia and other complications. Global estimates attribute approximately 4.07 million disability-adjusted life years to diarrhoeal and invasive NTS infections.² Unlike typhoidal Salmonella, which primarily causes systemic enteric fever, NTS is commonly acquired through contaminated food and encompasses numerous serotypes with distinct epidemiological characteristics. This substantial and heterogeneous disease burden highlights the need for surveillance of NTS reservoirs across the food chain.

Retail meat is an important vehicle for NTS because contamination can occur during slaughter, processing, transport and retail handling. European surveillance has identified poultry and pork as important food-chain sources.³ Studies in China have likewise detected Salmonella in retail pork, chicken and duck, with reported prevalence varying across commodities and study settings.^4,5 For example, one Guangdong survey reported detection rates of 73.1% in pork and 63.6% in chicken.⁶ Taken together, these findings identify retail meat as a recurrent reservoir of NTS across diverse food-supply settings.

The public health significance of foodborne NTS is further amplified by increasing antimicrobial resistance (AMR). Fluoroquinolone-resistant NTS is classified as a high-priority pathogen by the World Health Organization.⁷ Resistance to β-lactams, tetracyclines, sulfonamides and other agents has also been documented across animal and food sectors.^5,8,9 Antimicrobial use in food animals can select resistant populations, which may subsequently expand clonally, while resistance determinants disseminate through horizontal gene transfer.¹⁰ Plasmids, insertion sequences and transposons facilitate this process; ISEcp1 and IS91 have been associated with the mobilization of bla_CTX-M and qnr genes.^11–13 Accordingly, the proposed ‘selection–enrichment–diffusion’ cycle can be described more explicitly as antimicrobial selection, amplification of resistant populations and dissemination of resistant clones or mobile resistance genes. Integrating whole-genome sequencing with mobile genetic element (MGE) analysis is therefore important for identifying high-risk lineages and resistance-gene contexts within a One Health surveillance framework.^14,15

Since the 1970s, bacteriophages have been investigated as non-antibiotic approaches for controlling Salmonella.¹⁶ Commercial phage products have also been applied for decontamination in slaughterhouses and farm environments.¹⁷ However, many available phages have narrow, strain- or serotype-dependent host ranges and reported candidates have primarily been evaluated against common serotypes such as S. Enteritidis and S. Typhimurium.¹⁸ Regional phage collections covering a broader range of Salmonella serotypes may therefore improve susceptibility screening. Nevertheless, candidate phages require evaluation of lytic activity, genomic safety and potential gene-transfer risks before practical application.^17,18

In this study, we focus on Beijing, China—a populous city with high demand for animal-derived foods and a complex supply chain connected to most major livestock production regions across the country—to provide updated insights into the epidemiology of NTS along the retail food supply chain. Previous studies have primarily investigated the prevalence or AMR of Salmonella in retail meats, whereas integrated analyses combining antimicrobial susceptibility testing, whole-genome sequencing, comparative genomic analysis with contemporary human clinical isolates, and phage susceptibility assessment remain limited in Beijing. Therefore, this study combined these complementary approaches to comprehensively characterize the genomic epidemiology, AMR, genetic relatedness to contemporary human clinical isolates and phage susceptibility of NTS circulating in retail meats, thereby providing evidence to support molecular surveillance and targeted intervention strategies.

Materials and methods

Sample collection, bacterial isolation and susceptibility testing

A total of 583 retail raw meat samples (231 pork, 242 chicken, 71 beef and 39 mutton) were collected from 16 districts of Beijing between July 2021 and July 2022. Samples were homogenized in buffered peptone water and pre-enriched at 37°C for 16 h, followed by selective enrichment in tetrathionate broth (TTB) at 42°C for 20 h. Enriched cultures were plated on chromogenic Salmonella agar base and incubated at 37°C for 24 h. Presumptive colonies were purified on brain heart infusion agar, identified by MALDI-TOF MS and serotyped by slide agglutination using commercial antisera (SSI Diagnostica, Denmark). The pre-enrichment step was used to resuscitate stressed bacterial cells, whereas selective enrichment at 42°C suppressed competing microbiota and enhanced the recovery of Salmonella.

Antimicrobial susceptibility testing was performed on 118 of the 255 NTS isolates selected by stratified sampling to represent geographic districts, meat types, retail settings and serotypes, while avoiding disproportionate representation of isolates with identical epidemiological characteristics. Minimum inhibitory concentrations were determined using a commercial broth microdilution panel (Shanghai Xingmo Biotechnology Co., Ltd., Shanghai, China) and interpreted according to the Clinical and Laboratory Standards Institute (CLSI) M100 guidelines, 31st edition (2021). Escherichia coli ATCC 25922 was used as the quality-control strain, as recommended by CLSI.¹⁹

Whole-genome sequencing and molecular analysis

Whole-genome sequencing was performed on 153 representative NTS isolates selected according to their spatiotemporal distribution (Table S1, available as Supplementary data at JAC-AMR Online). Genomic DNA was extracted using a commercial kit and sequenced on the Illumina NovaSeq 6000 platform with 150 bp paired-end reads. Raw reads were quality-checked using FastQC v0.11.9, trimmed with Trimmomatic v0.39 and assembled de novo using SPAdes v3.15.4.^20,21 Virulence genes were identified against the Virulence Factor Database using a 90% identity threshold, AMR determinants were detected with ResFinder v4.3, and plasmid replicons were identified using PlasmidFinder v2.1 at 80% identity.^22–24 Serotypes and sequence types were assigned using the Salmonella In Silico Typing Resource, and core-genome multilocus sequence typing (cgMLST) was performed with chewBBACA v3.8.5.^25,26 A minimum-spanning tree was constructed using GrapeTree with the MSTreeV2 algorithm, based on the seven-locus MLST allelic profiles (aroC, dnaN, hemD, hisD, purE, sucA and thrA). Edge lengths were proportional to the number of allelic differences between sequence types.²⁷

Human clinical Salmonella genomes used for comparative analysis were retrieved from the National Center for Biotechnology Information database. Genomes were included when they originated from hospitalized salmonellosis cases in China during the same sampling period and contained metadata describing collection year, geographic origin and host source. Assemblies with incomplete metadata or poor assembly quality were excluded. A total of 473 human clinical Salmonella genomes meeting these criteria were included for comparative genomic analyses.

Evaluation of phage lysis efficiency and prophage detection

Forty-six Salmonella phages were isolated from sewage and animal viscera by the Shandong Academy of Agricultural Sciences, using Salmonella isolates as hosts. For phage susceptibility testing, 133 representative Salmonella isolates were selected using a stratified approach to maximize diversity in serotypes, sequence types, AMR phenotypes, multidrug resistance profiles and sample origins.

For phage lysis assays, Salmonella cultures were grown in Lysogeny broth (LB) at 37°C to mid-exponential phase (OD₆₀₀ ≈ 0.6; ∼10⁸ cfu/mL). Bacterial suspension (100 μL) was mixed with 3 mL of molten soft LB agar (0.7%) and overlaid onto LB agar plates (1.5%). Each phage lysate (10 μL; titre ≥10⁹ PFU/mL) was spotted onto the bacterial lawn, and plates were incubated at 37°C for 12 h. Lysis was considered positive when the plaque diameter was ≥1 mm. Prophages were predicted using Phigaro v2.3.1 with default parameters.²⁸

Statistical analysis and visualization

Statistical analyses were performed using SPSS v28 and R v3.6.3. AMR rates, serotype distributions and prophage prevalence were assessed using Pearson’s χ² test, with Fisher’s exact test used when expected frequencies were <5. AMR heatmaps, cgMLST tree annotations and genetic-context diagrams were generated using ggplot2 v3.3.5, iTOL v6.7 and Adobe Illustrator, respectively.²⁹

Results

Prevalence of Salmonella in retail meat

Among 583 retail meat samples collected across 16 districts of Beijing, 255 (43.7%) were NTS-positive. Prevalence was highest in pork (52.4%, 121/231), followed by chicken (43.8%, 106/242), beef (26.8%, 19/71) and mutton (23.1%, 9/39). NTS prevalence was higher in wet markets (49.4%, 158/320) than in supermarkets (36.9%, 97/263; χ² = 9.16, P = 0.0025). Serotyping of 153 representative isolates identified 33 serotypes (Figure 1). S. London (41.7%), S. Rissen (18.1%) and S. Derby (11.1%) predominated among pork isolates (n = 72), whereas S. Enteritidis (30.4%) and S. Kentucky (15.9%) predominated among chicken isolates (n = 69). The seven beef isolates represented five serotypes, whereas each of the five mutton isolates belonged to a different serotype.

Figure 1.

Non-typhoidal Salmonella (n = 153) isolation from Beijing. The Sankey diagram illustrates the distribution and associations of non-typhoidal Salmonella among 16 districts in Beijing, different market types, various meat types and Salmonella serotypes. All bands are labelled with corresponding items to visually represent the flow and associations between different categories.

AMR phenotypes

Antimicrobial susceptibility testing was conducted on 118 NTS isolates from pork and chicken. Resistance was highest to sulfisoxazole (SOX) (74.6%, 88/118), ampicillin (68.6%, 81/118) and tetracycline (68.6%, 81/118). Resistance to amoxicillin/clavulanic acid was low (2.5%, 3/118), and all isolates were susceptible to colistin and meropenem. Trimethoprim–sulfamethoxazole resistance was 26.3 percentage points higher in pork-derived isolates than in chicken-derived isolates (71.0% versus 44.6%; χ² = 8.40, P = 0.004), whereas ceftiofur resistance was 23.7 percentage points higher in chicken-derived isolates (28.6% versus 4.8%; χ² = 12.27, P < 0.001) (Figure 2a).

Figure 2.

Antimicrobial Susceptibility of Non-typhoidal Salmonella from Different Sources and Predominant Serotypes. (a) Antimicrobial susceptibility of Non-typhoidal Salmonella in chicken and pork among all meats. (b) Antimicrobial susceptibility of the top five serotypes with the highest isolation rates of non-typhoidal Salmonella.

Among the 118 pork- and chicken-derived isolates, 76 (64.4%) were multidrug-resistant (MDR). Resistance to six antimicrobial agents was the most frequent MDR level, occurring in 18 isolates (23.7%, 18/76), with profiles including AMP-GM-TE-FLC-SOX-SXT and AMP-SPT-TE-FLC-SOX-SXT (Table S2). Chicken-derived isolates included strains resistant to 12 or 13 antimicrobial agents; one S. Indiana ST17 isolate was resistant to 13 agents, including AMP-GM-SPT-TE-FLC-SOX-SXT-TIO-CAZ-ENRO-OFX-APR-MEQ. Eight ESBL-producing isolates were also resistant to quinolones and azithromycin. Four were S. Kentucky ST198, while the remaining isolates were S. Indiana ST17, S. Saintpaul ST27, S. Mbandaka ST413 and S. Schwarzengrund ST96. S. Kentucky ST198 showed broad resistance to clinically important agents, including enrofloxacin, ofloxacin, ceftiofur and ceftazidime, whereas the S. Indiana ST17 isolate showed the broadest individual resistance profile, with resistance to 13 agents (Figure 2b).

Whole genome sequencing analysis

A total of 153 NTS isolates were analysed (Figure 3a), revealing a resistome of 71 distinct acquired resistance genes that conferred resistance to nine antimicrobial classes: β-lactams, aminoglycosides, phenicols, tetracyclines, trimethoprim, sulfonamides, quinolones, macrolides and fosfomycin. The aminoglycoside resistance gene aac(6’)-Iaa_1 was universally present (153/153, 100%). Resistance genes associated with β-lactams (66.7%, 102/153), sulfonamides (65.4%, 100/153), tetracyclines (60.8%, 93/153), phenicols (52.9%, 81/153), quinolones (49.0%, 75/153) and trimethoprim (45.8%, 70/153) were frequently detected. For β-lactams, bla_TEM variants were the most common (45.1%, 69/153), followed by extended-spectrum β-lactamase (ESBL) genes, predominantly bla_CTX-M-55 and bla_CTX-M-65 (21.6%, 33/153). Notably, several high-prevalence resistance genes of particular concern were identified, including the plasmid-mediated quinolone resistance gene qnrS1 (19.0%, 29/153), the macrolide resistance gene mph(A) (18.3%, 28/153) and the macrolide-streptogramin resistance gene pair (msr(E)-mph(E)) (17.0%, 26/153). The high carriage rates and known association of these genes with MGEs warranted further investigation into their genetic contexts.

Figure 3.

Distribution and functional annotation of antimicrobial resistance and virulence genes in Salmonella (n = 153). (a) Distribution of antimicrobial resistance genes in Salmonella strains. (b) Distribution of virulence factors across Salmonella strains. (c) Distribution of plasmid types across Salmonella strains.

Among the analysed isolates, 98.7% (151/153) carried Type III Secretion System (T3SS) genes from Salmonella pathogenicity islands-1 (SPI-1), including sipA, sipB, sipC, sopD, sopE and related virulence genes (Figure 3b). SPI-2 virulence-associated genes were present in 99.35% (152/153) of isolates, while SPI-5 genes (pipB, sigD, sopD and sopB) were found in all isolates. Virulence plasmid genes (spvB, spvC, spvD, spvR) were less common, with carriage rates ranging from 6.5% (10/153) to 13.7% (21/153). Plasmid typing revealed Col44I_1 as the most common plasmid (26.1%, 40/153), followed by IncFIB(S)_1 and IncFII(S)_1 (13.7%, 21/153) (Figure 3c). Most isolates carried a single plasmid (n = 51), while only a few carried four or more plasmids (n = 8). The isolates carrying plasmids accounted for 85.6% (131/153) of the total, with an average of 2.24 plasmids per strain.

Genetic environment analysis

The geographic origins of the NTS with the same cgMLST type were widely distributed (Figure 4). Genetic analysis was conducted on Salmonella isolates exhibiting ESBL resistance and simultaneous fluoroquinolone resistance. Ten isolates carrying bla_CTX-M-55 were identified, including five S. Kentucky isolates with ST198. Among these, four isolates sourced from fresh chicken carried bla_TEM-1B, bla_CTX-M-55 and bla_LAP genes (Figure 5b). These isolates also harboured the Tn3 family transposase Tn2, IS1380 family transposase ISEcp1, and IS903 gene transfer elements near their β-lactam resistance genes. The quinolone resistance gene qnrS1 was surrounded by ISKra4 family transposase ISKpn19 and IS2 transposase (Figure 5c). Two isolates, 106 (from fresh mutton) and 110 (from fresh chicken) also carried IS3 family transposase ISEc36. Additionally, isolate 134 carrying qnrS1 was found to have the tetracycline-resistance regulatory gene tetR nearby.

Figure 4.

Systematic phylogenetic tree illustrating the distribution of host, cgMLST, specific antimicrobial resistance genes and plasmid carriage among 153 non-typhoidal Salmonella isolates from Beijing.

Figure 5.

Genetic environments of Salmonella genes. (a) Genetic environments of mph(A) resistance in non-typhoidal Salmonella. (b) Genetic environments of β-lactam resistance genes in non-typhoidal Salmonella. (c) Genetic environments of fluoroquinolone genes in non-typhoidal Salmonella. (d) Complex multidrug-resistance region in an S. Indiana ST17 isolate. Arrows indicate the direction of transcription, colours represent different genes or functional categories and grey shading indicates nucleotide identity >99%.

Twenty-eight isolates carrying mph(A) were detected (Figure 5a). Of these isolates, 10 carried both IS6100 and IS91; nine carried IS6100 and intI-1; and eight carried IS6100, IS91 and intI-1. Adjacent genetic regions contained several resistance elements, including multidrug efflux pumps (Tap), TetR/AcrR family transcriptional regulators, N-acetyltransferase family enzymes, disinfectant tolerance gene qacE, quinolone resistance gene qnrB, aminoglycoside resistance gene ant(3'')-Ia, rifampicin resistance aar-3 and quinolone gene aac(6')-Ib-cr. Additionally, genetic analysis of the ERM efflux pump gene emrE revealed multiple resistance genes in an S. Indiana isolate from fresh chicken. These genes included aac(6')-Ib-cr, bla_OXA-1, cat, aar-3, rmtB, msr(E)-mph(E), along with gene transfer elements IS91, ISEc35 and ISEc29 interspersed among them (Figure 5d).

This study identified two Salmonella ST17 (S. Indiana) isolates from diverse geographic regions and distinct sample sources. Despite their varied origins, these isolates exhibited a close genetic relationship and carried resistance genes for extended-spectrum β-lactams, fluoroquinolones and azithromycin. To examine the geographic distribution and allelic relationships of selected sequence types among retail-meat and human clinical isolates, a minimum-spanning tree was constructed from the seven-locus MLST allelic profiles (Figure 6). Several sequence types, including ST34, ST11, ST17, ST198, ST2040 and ST469, were represented among isolates from multiple geographic regions. Further cgMLST analysis revealed close allelic similarity between Beijing isolates and clinical isolates from Zhejiang, Guangdong, Shanghai, Jiangsu and Shandong (Figure 7).

Figure 6.

Minimum-spanning tree based on the seven-locus MLST allelic profiles of selected retail-meat and human clinical Salmonella sequence types from different regions of China. Each node represents a sequence type, node size is proportional to the number of isolates, and coloured sectors indicate the geographic composition of isolates within each sequence type. Edge lengths are proportional to the number of allelic differences between sequence types.

Figure 7.

cgMLST phylogenetic tree of 153 retail-meat isolates and 473 human clinical Salmonella isolates (n = 626). Different coloured branches represent different regions, and the colour of the outermost circle of the phylogenetic tree indicates the origin of the strain. The enlarged portion within the red box represents strains of non-typhoidal Salmonella with high homology from both human and animal sources.

Phage lysis and prophage detection analysis

Forty-six phages demonstrated the highest average lytic rate against S. Enteritidis (ST11), achieving 83.3%. The second highest lytic rate was against S. Typhimurium variant (ST34) at 36.0%, followed by S. Agona (ST13, 34.1%), S. Derby (ST40, 25.2%), S. Rissen (ST469, 25.0%), S. Typhimurium (ST19, 19.0%), S. Schwarzengrund (ST241, ST96, 18.8%) and S. Kentucky (ST198, 16.30%) (Table S3). Phages isolated from S. Rissen (phages 12–16) showed a 65.00% lytic rate against S. Rissen. Four S. Kentucky isolates were resistant to all 46 phages, while one S. Kentucky strain was susceptible to all phages. Isolates with lytic rates exceeding 90% included 9 S. Enteritidis (ST11) isolates, 1 Salmonella strain of serotype I 9,46:g,m:1,6 (ST11), 1 S. Kentucky (ST198) and 1 S. Rissen (ST469). However, the phages were ineffective against 24 Salmonella isolates. Among the 8 MDR isolates (resistant to ampicillin, ceftiofur, ceftazidime, enrofloxacin, ofloxacin and azithromycin), 3 were not lysed by any of the phages.

Prophages were detected in 151 of the 153 Salmonella isolates (98.7%), with most belonging to the Caudovirales order (Table S4). Among these, 33.99% (52/153) of isolates contained prophages associated with transposons. Prophages from the Myoviridae family were present in 42 isolates, Siphoviridae in 36 isolates and Podoviridae in 2 isolates. Among the 52 isolates carrying transposon-associated prophages, the predominant serotypes were S. Enteritidis (n = 21), S. Rissen (n = 14), S. Corvallis (n = 4), S. Schwarzengrund (n = 3), S. Give (n = 2) and S. Derby (n = 2), with one isolate each of S. Mbandaka, S. Muenster, S. Ruanda, S. Worthington, S. Indiana and serotype I 9,46:g,m:1,6. All 21 S. Enteritidis ST11 isolates carried transposon-associated prophages. Overall, ST11 isolates accounted for 44.2% (23/52), comprising 21 S. Enteritidis, one S. Ruanda and one I 9,46:g,m:1,6 isolate.

Discussion

NTS is a critical zoonotic pathogen, and contaminated food is an important route of transmission to humans, particularly vulnerable populations. This study provides a molecular epidemiological assessment of NTS in retail meats across Beijing, a major consumption hub with a complex supply chain. By integrating antimicrobial susceptibility testing, whole-genome sequencing, comparison with contemporary human clinical isolates and phage susceptibility testing, it extends previous surveillance of foodborne Salmonella in Beijing. The high prevalence of NTS in pork (52.4%) and chicken (43.8%) was consistent with previous surveillance in China.^30,31 Notably, the lower isolation rate in supermarkets than in wet markets may reflect differences in hygiene practices, cold-chain management and cross-contamination risks between retail settings.

A central finding was the high prevalence of multidrug resistance (MDR), with 64.4% (76/118) of tested isolates classified as MDR. Resistance to ampicillin, tetracyclines and sulfonamides involved antimicrobial classes widely used in China’s livestock sector.³² The concordance between these retail meat-derived resistance profiles and those increasingly reported in human clinical Salmonella cases suggests that these similarities support retail meat as a potential reservoir of antimicrobial-resistant Salmonella.³³ This is particularly relevant to Beijing, which receives meat from multiple production regions and may therefore be exposed to diverse resistant lineages. Supporting this concern, the high resistance rates to tetracycline and ampicillin observed in Salmonella from diarrheal patients in Beijing closely mirror the resistance profiles documented in animal-derived strains, which, together with the genomic evidence generated in this study, supports the possibility of dissemination through the food production and distribution chain.

Whole-genome sequencing and cgMLST identified clusters containing both retail meat isolates and contemporary human clinical isolates. The close relatedness of Beijing isolates to human clinical isolates from Jiangsu, Zhejiang and Guangdong suggests circulation of genetically similar Salmonella lineages across regions and sectors. However, these genomic similarities do not establish direct transmission. Similar resistance-gene profiles in food and human isolates are nevertheless consistent with antimicrobial selection acting across interconnected One Health sectors.³⁴

At the molecular level, several resistance determinants were associated with mobile genetic elements (MGEs). The spread of the dominant extended-spectrum β-lactamase gene, bla_CTX-M-55, was strongly linked to the insertion sequences ISEcp1 and IS903, recapitulating a conserved genetic context observed in clinical isolates. Critically, ISEcp1 has been demonstrated to directly mediate the transposition of bla_CTX-M genes across plasmids and chromosomes, providing a key mechanism for cephalosporin resistance mobilization.¹¹ The quinolone-resistance gene qnrS1 was associated with ISKpn19, whereas multidrug-resistance regions containing bla_OXA-1 and aac(6′)-Ib-cr were flanked by IS91 and ISEc29.³⁵

Notably, IS91 is a potent facilitator known to mobilize entire resistance gene clusters across diverse Enterobacteriaceae.^13,36 The co-localization of this IS91/ISEc29 pair within a single extensively drug-resistant (XDR) S. Indiana (ST17) isolate suggests an emerging and high-risk mechanism for the rapid assembly of pan-resistant phenotypes, warranting focused surveillance. Collectively, these findings support an important role for MGEs in the assembly and dissemination of AMR determinants in foodborne Salmonella populations.

Serotype-specific resistance patterns warrant particular attention. Consistent with national trends, S. Indiana (ST17) and S. Kentucky (ST198) were among the most resistant lineages identified in our study.³⁷ MDR S. Kentucky ST198 has been increasingly reported in China and internationally and is recognized as an emerging high-risk clone associated with AMR dissemination.³⁸ The detection of isolates resistant to over 10 antibiotics, including third-generation cephalosporins and fluoroquinolones, underscores the difficulty of managing infections caused by these high-risk clones. The co-occurrence of mph(A) with quinolone-resistance determinants in several isolates may further compromise the utility of azithromycin as an alternative treatment.³⁹

Given the escalating threat of AMR, bacteriophages have attracted increasing attention as potential biocontrol agents.⁴⁰ Our in vitro phage lysis assays demonstrated high susceptibility of S. Enteritidis (ST11) isolates to the tested phages. However, the broad phage resistance observed in several S. Kentucky isolates highlights an important limitation and suggests that broader host-range phage collections may be required to improve coverage against diverse Salmonella serotypes.¹⁸

Furthermore, our genomic analysis revealed a high prevalence of predicted prophages (98.7%), with prophages containing transposon-associated sequences identified in 52 of 153 isolates (34.0%). Although the functional contribution of these prophages to AMR dissemination was not evaluated in the present study, their occurrence warrants further investigation regarding their potential role in horizontal gene transfer.⁴¹ Previous studies have reported prophages capable of carrying resistance-associated genes, including bla_CTX-M-27.⁴² Therefore, future development and safety assessments of phage-based interventions should include rigorous evaluation of the potential risks associated with phage-mediated gene transfer.

Several limitations should be acknowledged. First, beef- and mutton-derived isolates were underrepresented compared with pork and chicken isolates, which may limit the generalizability of serotype distribution and AMR patterns observed in these meat categories. Second, although comparative genomic analyses revealed close genomic relatedness between retail meat isolates and human clinical isolates, direct epidemiological linkage could not be established because exposure histories and traceback information were unavailable. Third, phage susceptibility was evaluated exclusively through in vitro lysis assays, and the efficacy, stability and safety of phage-based interventions require further validation under practical application conditions.

Conclusions

This study identifies retail meats in Beijing, particularly pork and chicken, as important reservoirs of MDR Salmonella. Genomic analysis revealed diverse resistance determinants and mobile genetic elements, including ISEcp1 and IS91, that may facilitate the organization and dissemination of β-lactam, quinolone and macrolide resistance determinants. Comparative genomic analysis identified closely related retail-meat and human clinical isolates from several regions of China, although direct epidemiological transmission was not established. In vitro phage-lysis assays demonstrated high susceptibility of S. Enteritidis ST11 but limited susceptibility among several S. Kentucky ST198 isolates. These findings support integrated One Health surveillance linking retail-food monitoring with routine human enteric surveillance, strengthened antimicrobial stewardship, targeted monitoring of high-risk lineages and cautious evaluation of phage-based biocontrol strategies.

Funding

This research was supported in part by the Beijing Natural Science Foundation of China (6222033).

Transparency declarations

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Author contributions

A.C.: Conceptualization, Methodology, Writing—original draft. H.L.: Investigation, Writing—original draft. Y.W., X.H., T.L. and Y.Z.: Data curation. L.S. and P.H.: Formal analysis. Y.X. and Y.W.: Writing—review & editing. Y.L. and Y.Z.: Conceptualization. S.W.: Writing—review & editing, Funding acquisition, Project administration, Conceptualization.

Supplementary data

Tables S1 to S4 are available as Supplementary data at JAC-AMR Online.