Genetic Characteristics of Multidrug-Resistant Salmonella Isolated from Poultry Meat in South Korea

Given the lack of genetic characterization data for multidrug-resistant (MDR) Salmonella in South Korean poultry, we analyzed 53 MDR Salmonella strains from 1232 poultry meat samples (723 chicken, 509 duck) using whole-genome sequencing. Five serotypes were identified: S. Infantis (30/53, 56.6%), S. Enteritidis (11/53, 20.8%), S. Virchow (9/53, 17.0%), S. Agona (2/53, 3.8%), and S. Indiana (1/53, 1.9%). Sequence types (STs) included ST32, ST11, ST16, ST13, and ST17, with three major clusters, each having two subclusters. Eight core genome sequence types (cgSTs) were identified: 225993, 2268, 58360, 150996, 232041, 96964, 117577, and 267045. Salmonella Infantis and S. Enteritidis had two (117577, 267045) and three (225993, 2268, 58360) cgSTs, respectively, whereas S. Virchow showed allelic differences in identical cgSTs. The S. Enteritidis subcluster was classified as chicken or duck. Twenty-eight antimicrobial resistance genes (ARGs), 10 plasmid replicons, 11 Salmonella pathogenicity islands (SPIs), and 230 virulence genes were identified, showing distinct profiles by cluster and subcluster. Salmonella Infantis, the primary MDR Salmonella, carried the IncFIB (pN55391) plasmid, 10–11 ARGs, nine SPIs, and approximately 163 virulence genes. Three major MDR Salmonella serotypes (S. Infantis, S. Enteritidis, and S. Virchow) had specific genetic profiles that can inform epidemiological surveillance.


Introduction
Salmonella is the major etiological agent of diarrheal diseases, presenting an important worldwide health concern with 1.9 billion cases annually [1].Poultry and egg products are the primary sources of Salmonella infection [2].This phenomenon is also observed in South Korea.According to the Korean Ministry of Food and Drug Safety (MFDS), Salmonella is the third most common foodborne pathogen associated with foodborne illnesses in South Korea over the past two decades [3].
Various antimicrobials are widely used for disease prevention and treatment in the global livestock industry [4].However, continuous exposure to antimicrobials decreases microbial diversity and increases the number of antimicrobial-resistant bacteria via selective pressure [5].Therefore, livestock products continuously exposed to antimicrobials are exposed to the threat of antimicrobial-resistant bacteria.Livestock contaminated with antimicrobial-resistant bacteria serve as reservoirs and ensure transmission to the community through food [6].Several studies have shown a strong causal link between antimicrobial usage in livestock and the emergence of antimicrobial resistance (AR) in pathogenic bacteria that cause human diseases [7,8].

Salmonella Isolation and Identification
Salmonella spp.were isolated using an analytical method certified by the MFDS Food Code [12].Briefly, approximately 25 g of the sample was dispensed with 225 mL of buffered peptone water (BPW, Merck, Darmstadt, Germany) into a sterilized blender bag, homogenized for 30 s, and incubated at 37 • C for approximately 24 h.Then, 0.1 mL and 1 mL of incubated BPW were transferred into 10 mL of Rappaport-Vassiliadis broth (BD, Franklin Lakes, NJ, USA) and 10 mL of tetrathionate broth (MBcell, Seoul, Republic of Korea), and incubated at 42 • C and 37 • C for 24 h, respectively.After incubation, each culture solution was spread on xylose lysine deoxycholate agar (XLD; Oxoid, Basingstoke, UK) as well as Brilliant Green Sulfa Agar (Remel, Lenexa, UK) and incubated at 37 • C for 24 h.The presumed Salmonella colonies were selected, spread on Tryptone Soya Agar (Oxoid), and incubated at 37 • C for 24 h.After incubation, the bacteria were identified at the species level using matrix-assisted laser desorption ionization-time of flight.At least one Salmonella isolate per sample was selected for further analysis.Salmonella isolates were stored at −80 • C in Tryptic Soy Broth (Oxoid) with 10% glycerol.

Antimicrobial Susceptibility
All identified Salmonella strains were subjected to an antimicrobial minimal inhibitory concentration (MIC) assay.The MIC test was performed using the KRNV5F (TREK Diagnostic Systems, Cleveland, OH, USA) panel for this assay according to the manufacturer's instructions.E. coli ATCC 25922 was used as the reference strain.Each panel contained a total of 16 agents of 13 antimicrobial subclasses (Table 2).The MIC results were interpreted according to the breakpoint guidelines of the Clinical and Laboratory Standards Institute [13].
As the CLSI has no breakpoint guidelines for ceftiofur and streptomycin, these data were interpreted according to the National Antimicrobial Resistance Monitoring System [14].

WGS Analysis
Based on the MIC assay, strains resistant to five or more antimicrobial classes (n = 53) were selected for the WGS analysis.The selected strains were subjected to WGS at Senigen, Inc. (Seoul, Republic of Korea).Briefly, a MagListo 5M Genomic DNA Extraction Kit (Bioneer, Daejeon, Republic of Korea) was used for DNA extraction according to the manufacturer's instructions.WGS was analyzed using an Illumina MiSeq desktop sequencer (Illumina Inc., San Diego, CA, USA) with paired-end reads of approximately 300 bp in length.Trimmomatics (version 0.38) was used for the trimming process.SPAdes (version 3.13.0)was used to assemble raw reads.The assembled sequence data were filtered out with a length of 1000 bp and a depth of at least 5.The assembled contig number ranged between 21 and 66 and from 101 to 236 with average depths of 39 and 153, respectively.

Serotyping and Homology Analysis
Salmonella serotypes were determined using SeqSero (version 1.2) [15].Additionally, bacteria whose serotypes were not confirmed using SeqSero were tested using a slide agglutination test according to the Kauffman-White scheme using commercially available antisera (S&A Reagents Lab, Bangkok, Thailand).Homologies were compared using MLST and cgMLST.Seven housekeeping genes (aroC, dnaN, hemD, hisD, purE, sucA, and thrA) were obtained from the MLST database [16].MLST (version 2.0) was used in silico on the Center for Genomic Epidemiology (CGE) website to determine the sequence type.For cgMLST, we used the cgMLSTfinder (version 1.2) on the CGE website to predict allelic profiles.A cgMLST-based minimum-spanning tree was constructed using GrapeTree (version 1.5.0).

In Silico Characterization of WGS
The genetic characterization of Salmonella was performed using WGS.ARGs were identified using ResFinder (version 4.1), with minimum identity and coverage thresholds set at 90% and 60%, respectively.Plasmid types and Salmonella pathogenicity islands (SPIs) were predicted using Plasmidfinder (version 2.1) and SPIFinder (version 2.0), with minimum identity and coverage thresholds of 95% and 60%, respectively.Mobile genetic elements (MGEs) were identified using mobile element finder (software version 1.0.3 and database version 1.0.2) [17], with minimum identity and coverage thresholds of 90% each.Virulence factors were predicted using the Virulence Factor Database [18], with minimum identity and coverage thresholds set at 90% and 50%, respectively.

Prevalence of Salmonella and MDR Salmonella in Poultry Meat Samples
The prevalence rates of Salmonella were 27.4% (n = 198) and 41.3% (n = 210) in 723 and 509 chicken and duck meats, respectively; 46.0% (n = 91) and 28.1% (n = 59) of Salmonella isolated from chicken and duck meats were multidrug-resistant (MDR; resistant to three or more antimicrobial classes) [19].Of the 113 Salmonella resistant to at least five antimicrobial classes, 53 strains were selected and subjected to WGS analysis.

Salmonella Infantis was classified based on cgSTs (cgST117577 and cgST267045). Salmonella
Enteritidis clusters were classified based on cgST and origin (i.e., subcluster A was identified as cgST58360 and was isolated from chicken meat, whereas subcluster B was identified as cgST225993 and cgST2268 and was isolated from duck meat).The S. Virchow cluster was separated based on allelic differences.

Antimicrobial Resistance Patterns of MDR Salmonella
The five serotypes showed different AR profiles, with S. Enteritidis and S. Virchow subclusters showing notable differences (Figure 2).Both A and B clusters of S. Infantis were resistant to seven antimicrobial classes (aminoglycoside, aminopenicillin, cephalosporin III, folate pathway inhibitor, phenicol, quinolone, and tetracycline).No specific differences in AR were observed between the S. Infantis subclusters.Both A and B clusters of S. Enteritidis were resistant to three antimicrobial classes (aminopenicillin, quinolone, and tetracycline).In addition, S. Enteritidis subclusters showed clear differences in resistance to four antimicrobial classes (aminoglycosides, cephalosporin III, cephalosporin IV, and folate pathway inhibitors).Both A and B clusters of S. Virchow were resistant to three antimicrobial classes (aminopenicillin, cephalosporin III, and quinolone).In addition, S. Virchow subclusters showed clear differences in resistance to two antimicrobial classes (β-lactam/β-lactamase inhibitor combination and tetracycline).Both A and B clusters of S. Virchow were resistant to three antimicrobial classes (aminopenicillin, cephalosporin III, and quinolone).In addition, S. Virchow subclusters showed clear differences in resistance to two antimicrobial classes (β-lactam/β-lactamase inhibitor combination and tetracycline).The S. Agona cluster was resistant to five antimicrobial classes (aminoglycoside, aminopenicillin, folate pathway inhibitor, phenicol, and tetracycline).The S. Indiana singleton was resistant to seven antimicrobial classes (aminoglycoside, aminopenicillin, fluoroquinolone, folate pathway inhibitor, phenicol, quinolone, and tetracycline).

Antimicrobial Resistance Patterns of MDR Salmonella
The five serotypes showed different AR profiles, with S. Enteritidis and S. Virchow subclusters showing notable differences (Figure 2).Both A and B clusters of S. Infantis were resistant to seven antimicrobial classes (aminoglycoside, aminopenicillin, cephalosporin III, folate pathway inhibitor, phenicol, quinolone, and tetracycline).No specific differences in AR were observed between the S. Infantis subclusters.Both A and B clusters of S. Enteritidis were resistant to three antimicrobial classes (aminopenicillin, quinolone, and tetracycline).In addition, S. Enteritidis subclusters showed clear differences in resistance to four antimicrobial classes (aminoglycosides, cephalosporin III, cephalosporin IV, and folate pathway inhibitors).Both A and B clusters of S. Virchow were resistant to three antimicrobial classes (aminopenicillin, cephalosporin III, and quinolone).In addition, S. Virchow subclusters showed clear differences in resistance to two antimicrobial classes (β-lactam/β-lactamase inhibitor combination and tetracycline).Both A and B clusters of S. Virchow were resistant to three antimicrobial classes (aminopenicillin, cephalosporin III, and quinolone).In addition, S. Virchow subclusters showed clear differences in resistance to two antimicrobial classes (β-lactam/β-lactamase inhibitor combination and tetracycline).The S. Agona cluster was resistant to five antimicrobial classes (aminoglycoside, aminopenicillin, folate pathway inhibitor, phenicol, and tetracycline).The S. Indiana singleton was resistant to seven antimicrobial classes (aminoglycoside, aminopenicillin, fluoroquinolone, folate pathway inhibitor, phenicol, quinolone, and tetracycline).

Detection of Virulence Factor Genes
In total, 230 pertinent genes belonging to 14 virulence factor classes were detected, namely fimbrial adherence determinants, macrophage-inducible genes, magnesium uptake, non-fimbrial adherence determinants, regulation, secretion system, serum resistance, stress adaptation, toxin, adherence, iron uptake, autotransporter, immune evasion, and invasion.The majority of genes belonged to fimbrial adherence determinants (94/230) and secretion systems (98/230).In total, 120 genes were identified (Table 3), of which 110 had different virulence gene profiles (Figure 4).Several genes were dominant in each serotype.For example, in the fimbrial adherence determinant virulence factor class, eight (pefABCD and pegABCD) and seven (staABCDEFG) genes were predominant in S. Enteritidis and S. Agona, respectively.Iron uptake genes (irp2, psn/fyuA, and ybtAPQSTUX) were exclusively found in S. Infantis.There were clear differences in several VFs, even between subclusters of the same serotype, such as the secretion system (spvD) and immune evasion (gtrA) in S. Enteritidis; the secretion system (spiC/ssaB) and invasion (ibeB) in S. Virchow; and fimbrial adherence determinants (fimW and safD) in S. Infantis.

Discussion
Recently, S. Infantis carrying the pESI-like megaplasmid has been reported worldwide [20][21][22].Several studies have reported that the presence of pESI or pESI-like megaplasmids increases antibiotic resistance and toxin levels in S. Infantis [23].Salmonella Infantis carrying the pESI plasmid was reportedly predominant in feces and dust from commercial broiler farms in Korea [24].Salmonella Infantis was the most frequently identified serovar in eggs, and pESI-like megaplasmids have been identified in the broiler industry [25,26].
The pESI plasmid has the potential to spread S. Infantis carrying the pESI plasmid to the community in a short period of time [20,21,27].Despite these reports, the genetic analysis and timing of the spread of S. Infantis remain unclear.Therefore, we performed a WGS analysis of MDR Salmonella isolated from Korean poultry meat in 2020-2021, earlier than the previous report.The analysis results showed that S. Infantis carrying the pESI plasmid was isolated.In addition to S. Infantis, S. Enteritidis and S. Virchow were shown as the major MDR Salmonella in Korean poultry.
In this study, 53 MDR Salmonella spp.isolates were identified from 1232 poultry samples (723 chicken and 509 duck meat samples) collected in five areas of South Korea from 2020 to 2021.Of the 53 strains, 94% (50/53) were serotypes (S.Infantis, S. Enteritidis, and S. Virchow) typically found in poultry in South Korea [28][29][30].Salmonella Infantis, S. Enteritidis, and S. Virchow, belonging to ST32, ST11, and ST16, respectively, showed the same results as those reported previously [31,32].Most Salmonella showed clearly different homology depending on the source of isolation, even though they were the same serotype and sequence type.However, in S. Infantis subcluster A strains, 21 strains were sourced from chicken meat and one from duck meat, suggesting that cgST267045 may have been transmitted from chickens to ducks or as a result of contamination in meat processing.
In this study, various antibiotic resistance, plasmid, Salmonella pathogenicity island, mobile genetic element, and virulence factor genes were detected and clustered into similar types according to serotype and cgST.S. Infantis was divided into two subclusters based on cgST; however, the genetic difference was not clear.The IncFIB (pN55391) plasmid replicon was detected in all S. Infantis.Extended-spectrum beta-lactamase (ESBL)-producing S. Infantis carrying an IncFIB(pN55391)-like plasmid was first isolated in Israel and quickly disseminated worldwide [27,33,34].Recently, S. Infantis carrying the pESI-like megaplasmid was reported for the first time in eggs in Korea in 2022 [24].IncFIB(pN55391) was one of the replicons typical of the "parasitic" pESI-like megaplasmid found [24].In this study, S. Infantis carrying the IncFIB(pN55391) showed resistance to various antimicrobial subclasses (tetracycline, beta-lactam, aminoglycoside, sulfonamide, phenicol, trimethoprim, and disinfectant), consistent with other reports [9,24,25,33].This study used samples from 2020 to 2021, earlier than the previously reported 2022 [24][25][26].S. Infantis carrying the IncFIB(pN55391) has not been reported among Salmonella isolates from poultry samples prior to 2020 in South Korea [10,11,35].To the best of our knowledge, this study is the earliest time to isolate S. Infantis carrying the IncFIB (pN55391) in Korea.Therefore, we suspect the time when the new S. Infantis carrying the IncFIB(pN55391) began spreading was in 2020.
Contrastively, S. Enteritidis and S. Virchow subclusters showed different genetic profiles.The ABR gene profile of the S. Enteritidis subcluster A was common among MDR S. Enteritidis isolated from the chicken industry in South Korea [11,35].Salmonella with a genetic profile similar to that of S. Enteritidis subcluster B was also isolated from Chinese ducks and is a strong candidate to be the major MDR Salmonella in ducks [36].Both subclusters carried the beta-lactam resistance gene (bla CTX-M-15 and bla TEM-1B ), but in the 3rd and 4th clusters, cephalosporin (ceftiofur, ceftazidime, and cefepime) resistance was clearly different.Additionally, a marked disparity in gentamicin resistance was observed, which was S. Enteritidis subclass A carrying the aac(3)-IId gene [37].Both subclusters carried IncFIB(S) and IncFII(S), but only subcluster B contained the IncX1 plasmid.IncX1 is reported to carry beta-lactam, aminoglycoside, and sulfonamide resistance genes [38,39], which is consistent with our study.In addition, Tn2 in all cluster B strains carried bla TEM-1B , and one strain was detected with the composite transposon cn_35009_IS26 that also carried aph(6)-ld, aph(3 ′′ )-lb, sul2, and bla TEM-1B with the IncX1 plasmid.Therefore, the composite transposon cn_35009_IS26 was a strong candidate for increasing the antibiotic resistance threat of duck-isolated Salmonella by carrying four AMR (bla TEM-1B , aph(3 ′′ )-Ib, aph(6)-Id, and sul2) and IncX1 plasmid genes.No mobile colistin resistance genes were detected in S. Enteritidis despite the fact that most S. Enteritidis have colistin resistance.Colistinresistant mcr-negative S. Enteritidis may be associated with chromosomal mutations, such as those in components of lipopolysaccharide and outer membrane synthesis and modification (RfbN, LolB, ZraR) and the multidrug efflux pump (MdsC) [40].The S. Virchow cluster showed extremely different ABR genetic profiles between subclusters and had the lowest ARG presence.The S. Virchow cluster A was resistant to ampicillin, ceftiofur, cefoxitin, nalidixic acid, streptomycin, and tetracycline, consistent with several reports [29,35].Subcluster B carried only two ABR genes, with bla CMY-2 being unique among the 53 strains, conferring resistance to amoxicillin/clavulanic acid, including penicillin and cephalosporin.Although bla CMY-2 is a plasmid-mediated gene [41,42], no promising plasmid candidates were detected in the current study.
Salmonella chromosomes and plasmid regions encoding virulence-related genes, such as those involved in invasion, survival, and extraintestinal spread, are named SPIs [43].In this study, 11 Salmonella SPIs (SPI1, SPI2, SPI3, SPI4, SPI5, SPI9, SPI10, SPI13, SPI14, C63PI, and CS54) were identified, and the SPI profiles were found to be identical at the serotype cluster level.SPI1, SPI2, SPI3, SPI4, and SPI5, which are more critical for Salmonella pathogenesis than other SPIs [9], were commonly identified among the Salmonella strains tested.SPI9 was also one of the most common SPIs in the current study, encoding a type 1 system similar to SPI4 [43].In this study, the three common MDR serotypes (S.Infantis, S. Enteritidis, and S. Virchow) may have been influenced by SPI-13 and SPI-14, which promoted the colonization of chicken spleen [44].In total, 230 virulence factor genes were identified, with 120 genes related to five virulence factor classes (fimbrial adherence determinants, macrophage-inducible genes, magnesium uptake, non-fimbrial adherence determinants, regulation, and secretion system) being common in this study.In contrast, 47.8% of the virulence factor-related genes showed different virulence gene profiles among the serotype clusters.In particular, diverse genes from two virulence factor classes (attachment and iron uptake) have been identified in S. Infantis.The diverse attachment factor genes may play distinct roles in chick infection [45], and a robust iron uptake system may contribute to Salmonella fitness and pathogenicity in vivo, potentially allowing rapid dissemination [46].
This study has several limitations.Due to the nature of the poultry industry in Korea, we were unable to identify which farm the meat samples originated from.If we could identify which farm a particular sample originated from, we would be able to have a more in-depth discussion.The study design was limited to bacterial selection for WGS analysis.Of the 408 Salmonella isolates (198 from chicken and 210 from duck meat), 53 strains with high levels of MDR were selected and analyzed using WGS.Salmonella isolates sensitive to antimicrobials were excluded from the study.Therefore, the distribution of Salmonella serotypes in South Korean poultry may differ from those observed in this study.To address these limitations, increasing the number of WGS analyses to cover all Salmonella isolates would provide greater insight into the traceback investigation of Salmonella from Korean poultry.
Our study suggests that S. Infantis carrying the pESI plasmid first emerged in Korea in 2020 and quickly became the major serotype in the poultry industry in 2022.This new major serotype in the Korean poultry industry spreads rapidly and carries a large number of lethal virulence genes.Therefore, we analyzed the genetic characteristics of MDR Salmonella from the Korean poultry industry, including S. Infantis carrying the pESI plasmid.Ultimately, this study may help to increase the understanding of MDR Salmonella in poultry meat in Korea and to help control its spread.

Figure 3 .
Figure 3. Identified antimicrobial resistance, plasmid, Salmonella pathogenicity island, and mobile genetic element gene profiles of multidrug-resistant Salmonella spp.

Figure 4 .
Figure 4. Identified virulence factor genes of multidrug-resistant Salmonella spp.except common virulence factor genes.

Figure 4 .
Figure 4. Identified virulence factor genes of multidrug-resistant Salmonella spp.except common virulence factor genes.

Table 1 .
Sample tested in this study.

Table 2 .
Antimicrobials tested in this study.

Table 3 .
Identified common virulence factor genes of multidrug-resistant Salmonella spp.