A total of 83 S. Enteritidis strains were collected. Clinical isolates originated from samples collected by the clinical laboratory of the First Medical Center, Chinese People’s Liberation Army (PLA) General Hospital, Beijing, China. All strains were isolated from different patients. Thirty-seven strains were isolated from aseptic sites in the human body that were devoid of local inflammation and damage, and were unconnected to the respiratory system, digestive system, urogenital system, or other external environments. These included locations such as venous blood, joint cavities, and cerebrospinal fluid. A further 11 strains were isolated from the feces of diarrheal patients. The other 35 isolates were of chicken-origin obtained from six provinces in China and cultured by the Chinese Institute for Food and Drug Control. The strains were isolated and cultured according to the National Clinical Laboratory Operation Procedures. The genus was identified by a VITEK MS RUO system v3.4 (BioMérieux, Lyon, France), and the serotype by the Salmonella serum agglutination test (Lot: 20210101, Statens Serum Institut, Copenhagen, Denmark).

Genomic DNA was extracted using the QiAamp DNA Mini Kit (Lot: 172030683, Qiagen, Dusseldorf, North Rhine-Westphalia, Germany) and then sequenced using an Illumina X-ten platform to generate 150-bp paired-end reads from a library with an average insert size of 500 bp. Raw reads were first filtered by fastQC v0.11.9 (Babraham Institute, Babraham, Cambridgeshire, England) to remove low-quality reads, and then assembled using SPAdes v3.13 (Saint Petersburg Academic University, Saint Petersburg, Leningrad Oblast, Russia) using the default parameters [17, 18]. The assembled genomes were evaluated by Quast v5.0.2 (Saint Petersburg State University, Saint Petersburg, Leningrad Oblast, Russia). Gene prediction was performed with Prokka v1.14.6 (University of Melbourne, Melbourne, Victoria, Australia), and BLAST v2.2.18 (National Center for Biotechnology Information, Bethesda, MD, USA) was used to compare the virulence genes carried by the strains with the virulence gene database VFDB. The parameters were set to a minimum coverage of 60% and a minimum similarity of 80%. The ResFinder v4.1.11 (Technical University of Denmark, Copenhagen, Denmark) was used to assemble strain sequences related to antibiotic resistance gene screening. The parameters were set to a minimum coverage of 60% and minimum similarity of 80%. The evolutionary tree was constructed with WGS single nucleotide polymorphism (wgSNP) data. MUMmer v3.23 (University of Maryland, College Park, MD, USA) was used to find SNPs in all strains and to merge these. To improve the overall quality of data, a random strain was selected as the reference sequence (FC15971). SNPs less than 5 bp away and those carrying unspecified nucleotides (“N”) were removed, and an evolutionary tree for wgSNP sequences was constructed using fasttree [19, 20, 21, 22, 23]. Antibiotic resistance genes (ARGs) were identified based on the best alignment with the ResFinder database, with thresholds of 90% DNA sequence identity and minimum coverage of 80%. A multi-locus sequence typing (MLST) scheme was used to subtype the isolates using BLASTn and 7 housekeeping genes (aroC, dnaN, hemD, hisD, purE, sucA, and thrA) [24].

Antimicrobial susceptibility testing (AST) of the S. Enteritidis isolates was performed with the broth microdilution method (Lot: D-012XS, Xingbai Biotechnology, Shanghai, China) and interpreted according to guidelines from the 2017 Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST; https://eucast.org/). A bacterial suspension with a precisely adjusted concentration of 0.5 McFarland units was first prepared. The recommended protocol for introducing the control sample and thoroughly amalgamating it with the bacterial solution was meticulously adhered to. Subsequently, the resultant mixture was kept in an incubation chamber at a constant temperature of 35 °C for 18 to 24 hours. This process was carefully conducted with the ultimate aim of determining the minimum inhibitory concentration (MIC). Susceptibility to the following antimicrobials was assessed: $\beta$-lactam antimicrobials, which includes ampicillin (AMP), ceftazidime (CAZ), cefoxitin (CFX), cefotaxime (CTX), cefazolin (CFZ), ampicillin-sulbactam (AMS) and imipenem (IPM); the tetracycline antibiotic, which includes tetracycline (TET); the quinolone antibiotic, which includes nalidixic acid (NAL) and ciprofloxacin (CIP); the macrolide azithromycin (AZM); the amphenicol antibiotic chloramphenicol (CHL); aminoglycosides, which included gentamicin (GEN); and the sulfonamides, which included the pediatric compound sulfamethoxazole (SXT). We defined an MDR strain as being resistant to three or more different classes of antibiotics at the same time. Strains were otherwise defined as non-MDR. Escherichia coli ATCC 25922 was used as the quality control strain. ResFinder software was used to detect antibiotic resistance-related genes based on the WGS data, and their association with MIC phenotypes.

Illumina reads were mapped to the reference genome using Bowtie 2 v2.2.8 (Johns Hopkins University, Baltimore, MD, USA), SNPs were identified using Samtools v1.9 (Shanghai Institute of Life Sciences, Shanghai, China), and the data was combined as described previously [25]. All high-quality SNPs (HQ snps) supported by $>$5 reads with a mapping quality of 20 were investigated further (https://github.com/generality/iSNV-calling). Phylogenetic analysis was performed based on the remaining core genome sequences. The best-fitting substitution model (K3P+ASC+R2) was identified using ModelFinder and selected to build a maximum likelihood phylogenetic tree using IQ-TREE v2.0.6 (Center for Computational Biology, Stockholm, Sweden). The phylogenetic relationships and distribution of resistance genes were displayed using iTOL v6.8.1 (Max Planck Institute for Developmental Biology, Tübingen, Germany) [26, 27, 28].

Statistical analysis was performed using SPSS software (Version 16.0, IBM, Armonk, NY, USA). Comparisons of antibiotic resistance between clinical isolates and chicken-origin isolates were performed by Chi-square/Fisher exact test for categorical variables. p 0.05 was considered to indicate a statistically significant difference.

All 83 isolates were identified as S. Enteritidis ST11 by MLST after WGS analysis (see Supplementary Table 1).

A total of 124 virulence genes were detected in the 83 Salmonella genomes. All 83 isolates were positive for 111 virulence genes. Seven known virulence factors showed different distributions in the VFDB matrix (rck, pefB, pefD, pefC, spvR, spvC, spvB), with a low carriage rate in chicken isolates (see Supplementary Table 1). These virulence genes are located on the virulence plasmid of S. Enteritidis.

Analysis of the antibiotic resistance phenotypes of S. Enteritidis isolated from different sources revealed a minor level of resistance in the clinical isolates. The multiple drug resistance rate of chicken-origin isolates was higher than that of clinical isolates (p $>$ 0.05). For the penicillin antibiotics AMP, CAZ, CFX, CTX, and CFZ, the resistance rates of chicken isolates were slightly higher (except for CFZ) than those of the clinical isolates, but the difference was not statistically significant (p $>$ 0.05). All strains were sensitive to AZM and resistant to the first-generation quinolone drug NAL. The clinical isolates were sensitive to CIP, but there were some resistant strains in chicken-origin isolates. In addition, clinical isolates were sensitive to the aminoglycoside drug GEN and the chloramphenicol antibiotic CHL, but chicken isolates were significantly more resistant (p 0.05) (see Table 1).

A maximum-likelihood tree of all 83 S. Enteritidis strains was constructed using SNPs (see Fig. 1). The phylogenic tree was broadly divided into four subclades. Of note, the genomes of MDR and non-MDR strains were distributed into different subclades. Subclade A strains were resistant to two groups of antibiotics, except strain FC9132 which was resistant to 6 groups of antibiotics. All strains of subclade B were resistant to two or three groups of antibiotics. Some strains of subclade C were resistant to 5 or 6 groups of antibiotics, and were resistant to $>$3 $\beta$-lactam antibiotics. Almost half of subclade D strains were resistant only to NAL, although strain FC15955 was resistant to 5 groups of antibiotics. Additionally, subclade C were all chicken-origin, except for strain 15. After mapping the SNP phylogenic tree with antibiotic resistance-related genes, subclade D strains were indeed found to have a lower carry rate for some of the resistance genes (e.g., tetA, tetR, bla${}_{\mathrm{\text{TEM-1}}}$, and sul2) compared with other subclades.

Fig. 1.

Phylogenetic analysis showing antibiotic resistance patterns and the profiles of specific genetic loci in S. Enteritidis strains. Neighbor-joining tree based on whole-genome level single-copy genes in the 83 S. Enteritidis strains. The type indicates the different isolation source of strains, and the year indicates the year of isolation. The presence or absence of specific loci are denoted by blue and grey rectangles, respectively. The phylogenetic tree was divided into (A), (B), (C), and (D) subclades, with the adjacent values representing the minimum and maximum single nucleotide polymorphism (SNP) divergence between any two pairs of clades.

The 20 resistance genes associated with resistance tests were the $\beta$-lactamase resistance-related genes (bla${}_{\mathrm{\text{TEM-1}}}$, bla${}_{\mathrm{\text{CTX-M-55}}}$, bla${}_{\mathrm{\text{TEM-206}}}$, bla${}_{\mathrm{\text{TEM-141}}}$, bla${}_{\mathrm{\text{TEM-135}}}$, bla${}_{\mathrm{\text{TEM-126}}}$, bla${}_{\mathrm{\text{TEM-106}}}$, bla${}_{\mathrm{\text{TEM-206}}}$, bla${}_{\mathrm{\text{TEM-141}}}$, bla${}_{\mathrm{\text{TEM-164}}}$, bla${}_{\mathrm{\text{TEM-163}}}$, bla${}_{\mathrm{\text{TEM-122}}}$), aminoglycoside resistance gene (rmtB), amidohydrin resistance gene (floR), sulfonamide-resistance genes (dfrA1, dfrA12, dfrA17, sul2, sul1) and tetracycline resistance gene (tet(A)). Numerically dominant resistance genes were observed for different antibiotics, with bla${}_{\mathrm{\text{TEM-1}}}$ present in 67 strains, rmtB in 4 strains, and floR in 3 strains. sul2 was present in 56 isolates and tet(A) in 17 isolates. WGS data was correlated with the phenotypic profiles to evaluate its ability to predict phenotypic resistance. Some resistance genes showed excellent correlation, for example bla${}_{\mathrm{\text{TEM-1}}}$ with AMP, bla${}_{\mathrm{\text{CTX-M-55}}}$ with CAZ and CTX, rmtB with GEN, gyrA with NA, and tet(A) with TET (see Supplementary Table 1). The low carry rate of other resistance genes meant they were difficult to analyze. It is worth noting that three ampicillin-resistant strains (strains 54, 15955, 15968) did not present known $\beta$-lactamase resistance genes, and one cefotaxime-resistant strains did not have known resistance genes (strain 9132). Two tetracycline-resistant S. Enteritidis did not have known tetracycline resistance genes (strains 19 and 49), and one phenotype that was resistant to cotrimoxazole had no known drug resistance genes (strain 15955). No known resistance genes were present in S. Enteritidis with the gentamicin resistance phenotype, and three chloramphenicol-resistant strains (FC15955, 9099, 9132) had no known drug resistance genes.

We next screened for mutations within the Quinolone resistance determination region (QRDRs). All strains carried the gyrA mutation, with the mutation types being D87Y (33/48 clinical isolates vs 28/35 chicken-origin isolates), D87N (10/48 vs 3/35), and D87G (5/48 vs 4/35). No significant differences in the mutation types were observed between the different isolates (p $>$ 0.05). No mutations were found in parC or parE.

The WGS data was used to detect the presence and absence of transposons, plasmid replicons, and virulence genes in the 83 Salmonella genomes. Six types of plasmid replicons were detected: IncI1, col, IncN, IncX1, IncFIB, and IncFII. Of these, IncFIB (90.4%), IncFII (89.2%), and IncX1 (79.5%) were the most common. Only one strain (FC15968) had no plasmids, but was resistant to ampicillin and nalidixic acid. Three strains (FC15966, 17, and 78) that showed multiple drug resistance carried four types of plasmids and were resistant to 7, 7, and 3 types of antibiotics, respectively. Strains were divided into three clusters in the Inc_type matrix graph (see Fig. 2). In cluster A, the vast majority of strains were resistant only to NAL, and carried plasmids IncFIB and IncFII. In cluster C, most strains were resistant to two groups of antibiotics (with some up to 6 groups), and carried plasmids IncX1, IncFIB, and IncFII. Although most isolates in cluster B carried only the IncX1 plasmid, they still showed MDR.

Fig. 2.

Distribution of Inc types in different strains. The abscissa shows the Inc type and the ordinate shows the strain number. The presence or absence of Inc are denoted by red and beige rectangles, respectively.

We next analyzed the insertion sequence data for the 83 S. Enteritidis strains. A total of 562 insertion sequences were found, including ISEcp1, ISCro1, IS1R, ISSso4, and IS186B unique to aseptic clinical sites, and IS5075, ISVsa3, and ISCR1 unique to chicken. Most strains contained ISSaen1, IS1230B, IS285, IS26, ISSwi1 and IS1351. Isolates that were resistant only to nalidixic acid showed the insertion pattern ISSaen1-IS1230B-IS285-IS1351-IS1351. Some strains that were resistant to two or three types of drugs carried IS26 and ISSwi1, while all strains resistant to four types of drugs carried IS26 and ISSwi1 (except strain 59). FC9130 and FC9131 were resistant to 6 groups of drugs and carried the additional insertion sequence of IS50R-IS1294-IS5075-ISVsa3-ISCR1.

The only subtype of the bla${}_{\mathrm{\text{CTX-M}}}$ resistance gene found in this study was CTX-M-55. A total of 9 S. Enteritidis strains carried bla${}_{\mathrm{\text{CTX-M-55}}}$, of which 5 were clinical isolates and 4 were chicken-origin isolates. These accounted for 10.42% (5/48) and 11.43% (4/35) of the respective groups, with no significant difference in the carry rate between them (p $>$ 0.05). Further analysis showed the bla${}_{\mathrm{\text{CTX-M-55}}}$ resistance gene was located on the plasmid in all clinical isolates carrying this gene, whereas in half of the chicken isolates it was located on the chromosome (see Fig. 3). After alignment, two clinical strains with bla${}_{\mathrm{\text{CTX-M-55}}}$ had the same genetic background and were located on the insertion sequence ISEcp1. The two strains (strains FC10061 and FC9131) with the bla${}_{\mathrm{\text{CTX-M-55}}}$ gene located on the chromosome had the same ISSwi1-CTX-M-55 transposable element.

Fig. 3.

Genetic environment of bla${}_{\mathrm{\text{CTX-M-55}}}$ in five S. Enteritidis isolates (FC10061, FC9131, 17, 78, FC9092, and FC9130). (a) bla${}_{\mathrm{\text{CTX-M-55}}}$ of strains located on the chromosome. (b) bla${}_{\mathrm{\text{CTX-M-55}}}$ of strains located on the plasmid.

The most common $\beta$-lactam antibiotic-related gene, bla${}_{\mathrm{\text{TEM-1}}}$, was carried by 37 strains of clinical isolates and 30 strains of chicken-origin isolates. The TEM-1 resistance gene was located on the plasmid in 55 strains, with no significant difference in frequency between clinical and chicken-origin isolates (33/37 vs 22/30 respectively, p $>$ 0.05). ISSwi1 was the most frequent accompanying insertion sequence in these strains.

A total of 8 isolates carried the drfA resistance gene, of which two strains were clinical isolates (all located on plasmids) and 6 strains were chicken-origin isolates (four strains located on plasmids and two strains located on chromosomes, all belonging to the dfrA17 subtype) (see Fig. 4). The two strains (FC10081 and FC9102) with dfrA17 located on the chromosome had similar genetic contexts to the one strain (strain 13) located on the plasmid. A total of 56 S. Enteritidis isolates had the sul2gene (29 clinical isolates and 27 chicken-origin isolates). The sul1 gene was found in one clinical isolate and in four chicken-origin isolates. All sul1 and sul2 genes were located on plasmids, with no significant difference in frequency between clinical and chicken-origin isolates (p $>$ 0.05), and no accompanying insertion sequence.

Fig. 4.

Genetic environment of drfA in eight S. Enteritidis isolates (FC10081, FC9102, 13, FC9092, FC9130, FC9131, FC10061, and 59). (a) Comparison between dfrA17-positive strains. The dfrA17 of strains FC10081 and FC9102 were located on the chromosome. The dfrA17 of strain 13 was located on the plasmid. (b) The dfrA12 of strains were located on the plasmid. (c) The dfrA1 of strain was located on the plasmid.

Three S. Enteritidis isolates, all of chicken-origin, carried the floR gene on the plasmid, thereby conferring a higher resistance rate to chloramphenicol in the chicken-origin isolates compared to clinical isolates.

Seventeen strains carried tet(A) resistance genes, including 7 clinical isolates and 10 chicken-origin isolates. In all cases the genes were located on plasmids, with no significant difference in the frequency between the clinical and chicken isolates (p $>$ 0.05).

Several highly resistant chicken isolates were examined (strains FC15955, FC9092, FC9130, FC9131, FC9132, and FC10061). Most were found to carry special resistance genes such as floR, tet(A), sul1, and rmtB, and to carry more insertion sequences (e.g., IS1230B, IS285, ISSaen1) than other isolates.