In our previous study (Hu et al., 2018a), 2,629 Salmonella were recovered from 1,680 retail whole chicken carcasses (including freshly slaughtered, chilled and frozen samples) between August 2010 and March 2012 at seven sites located in six provinces of China referring to North, Northeast, Northwest, South and East part of China. All strains were confirmed as belonging to the Salmonella genus and 463 (17.6%, 463/2629) of them were then identified as S. Indiana by serotyping, 224 of which were further identified as CIP-CTX co-resistant S. Indiana by antimicrobial susceptibility testing (AST), and these isolates were distributed as below: Beijing (n = 99); Jilin (n = 60); Jiangsu (n = 42); Shaanxi (n = 16); and Guangdong (n = 7).

A panel of fourteen antimicrobial agents, representing ten classes, were selected for these 224 selected CIP-CTX co-resistant S. Indiana isolates for AST by broth micro-dilution, the drugs included ampicillin (AMP), cefotaxime (CTX), ceftazidime (CAZ), chloramphenicol (CHL), ciprofloxacin (CIP), gentamicin (GEN), imipenem (IPM), nalidixic acid (NAL), ampicillin-sulbactam (SAM), trimethoprim-sulfamethoxazole (SXT), tetracycline (TET), meropenem (MEM), florfenicol (FFC), and colistin (CT). Minimal inhibitory concentration (MICs) value data for all compounds were recorded manually, and these data obtained were interpreted following the recommendations of CLSI (M100-S28 and M31-A3) and EUCAST (version 2018); extended-spectrum beta-lactamase (ESBL) production was then screened according to the protocol and breakpoints described in CLSI (M100-S28) as previously described (Hu et al., 2017). Escherichia coli ATCC 25922 and Klebsiella pneumoniae ATCC™700603 were included as quality control microorganisms in AST assays and ESBL confirmation, respectively.

Molecular methods were used to determine AMR genes and to show the relevant resistance mechanism. Genomic DNA was extracted with a TIANamp bacterial DNA kit (DP302, Tiangen Biotech) from all isolates. PCR detections for both quinolone resistance determinant regions (QRDRs) including gyrA, gyrB, parC, and parE genes and PMQR-encoding genes including qnrABCDS, qepA, oqxAB, and aac-(6′)-Ib-cr were carried out according to the study reported previously (Ling et al., 2003; Park et al., 2006; Jacoby et al., 2008; Liu et al., 2008; Yamane et al., 2008; Cavaco et al., 2009; Kim et al., 2009; Bai et al., 2016). Additionally, two multiplex PCR reactions were used for the detection of ESBL genes including bla_CTX-M groups and subtyping was also performed based on the methods reported previously (Eckert et al., 2004; Xu et al., 2005).

All 224 CIP-CTX-resistant S. Indiana isolates were screened to investigate the presence of mcr-1–mcr-10 genes by multi-target PCR methods as previously reported (Rebelo et al., 2018; Borowiak et al., 2020; Lei et al., 2020). To achieve the aim of addressing the full resistance phenotypic pattern for any mcr gene positive S. Indiana isolate screened in this study, an extended AST was carried out, using additional selected antimicrobial agents relevant to Enterobacteriaceae, including 13 classes (composing 27 compounds) in total and all procedures of AST were processed according to the AST processing workflow described above.

A single colony of each CIP-CAZ-CT co-resistant isolate was cultured overnight in brain heart infusion broth (Beijing Landbridge, China) at 37°C. Genomic DNA was extracted and purified using a TIANamp Bacterial DNA extraction kit (DP302, TIANGEN BIOTECH, China) and then sequenced with the SMRT^® Pacific Biosciences (PacBio) RS II platform (Tianjin Biochip Corporation, Tianjin, China) with a 10-kbp template library preparation step with PacBio^® Template Prep Kit. SMRT Analysis v2.3.0 was used for de novo assembly according to RS Hierarchical Genome Assembly Process (HGAP) workflow v3.0. Subsequently, Consed version 28.0 was used to manually inspect and trim duplicate ends to generate single, complete, and closed sequences for each chromosome and plasmid. For data error correction, Pilon v1.23 was used with Illumina MiSeq sequencing read data, for which a library was prepared with a NEBNext^® Ultra DNA Library Prep Kit for Illumina (NEB#E7370) followed by sonication fragmentation (350 bp insert), and then loaded on the Illumina HiSeq platform with PE 150 sequencing strategy (Novogene, Beijing, China) using a HiSeq X Ten Reagent Kit v2.5 (Illumina, San Diego, CA). The closed genomes were deposited in National Center for Biotechnology Information (NCBI) and automatically annotated using the NCBI Prokaryotic Genomes Automatic Annotation Pipeline (PGAP, version 4.8).

The predicted serotype and multi-locus sequence typing (MLST) types were identified using the Salmonella In Silico Typing Resource (SISTR). Plasmid replicon types (Incompatibility groups or Inc. groups) and antimicrobial resistance genes were identified through the Center for Genomic Epidemiology (CGE) website with PlasmidFinder (v2.0) and ResFinder (v3.0), respectively. Virulence factors and related virulence genes were predicted with VFanalyzer of the VFDB (virulence factor database, current status of August 10, 2022). All genes, plasmids, and chromosome sequences used in this study were managed, aligned, and analyzed by Geneious prime (v2019.2.3) software. Plasmids referring to several publicly published S. Indiana sequences were selected for genomic structure comparison against the three plasmids of S. Indiana CFSA664 in this study, and map generation was performed by BRIG (v0.95). Genetic environments were analyzed and displayed using Easyfig (v2.2.2).

In line with a previous study (Zhang et al., 2021), the transferability and frequency of colistin resistance were investigated by broth mating conjugation experiments with plasmid-free and sodium azide-resistant E. coli J53 as the recipient strain. The transconjugants were selected on MacConkey agar plates (Beijing Landbridge, China) supplemented with 100 mg/L sodium azide (Sigma-Aldrich) and 2 mg/L colistin (Sigma-Aldrich). Two different conjugation temperatures (30°C and 37°C) were used for the transfer of mcr-1-carrying plasmid in this study. Transfer frequencies were calculated as the number of transconjugants after confirmation by PCR. The colistin MIC values of the E. coli J53 and transconjugants were tested according to the description above.

The percentages related to AMR for all 244 CIP-CTX co-resistant S. Indiana recovered from poultry are shown in Table 1 and Supplementary Table 1. All expressed resistance to at least four categories of antimicrobial compounds. The resistance rates were divided into three classes: (1) higher than 78%: CIP, CTX, NAL, AMP, CHL, SAM, FFC, GEN, SXT, and TET; (2) 31.3%: CAZ; (3) lower than 2%: CT, IMP, and MEM. No isolate was found to be resistant to carbapenem-type compounds (IMP and MEM). Minor resistance differences were observed between S. Indiana isolated from different provinces, with the exception of CAZ and CT. Isolates from different regions showed a CAZ-resistant rate ranging from 11.1% (Beijing) to 54.8% (Jiangsu), and one colistin-resistant strain recovered from Jiangsu province, CFSA664, was identified and designated as a CIP-CTX-CT co-resistant isolate. Klebsiella pneumoniae ATCC™700,603 showed a 3 twofold concentration MIC value decrease for CTX tested in combination with clavulanate vs. when tested alone, the ESBL confirmation for all tested isolates was acceptable. All except seven isolates (3.1%, 7/224) from Beijing were confirmed as ESBL producers.

The antimicrobial resistance patterns are shown in Table 2 and Supplementary Table 1 for all 224 CIP-CTX co-resistant S. Indiana isolates tested. There were 2(0.89%), 1(0.45%), 12 (5.36%), 81 (36.16%), 127 (56.70%), and 1(0.45%) isolates resistant to 4, 5, 6, 7, 8, and 9 classes of antimicrobial agent tested, respectively. All 224 isolates were defined as MDR (resistant to three or more classes of antimicrobial drugs) and 209 (93.3%) of them were identified as high-level MDR (resistant to seven or more classes of antimicrobials). Among which, only one isolate was identified to be co-resistant to nine in ten classes of all drugs tested, with the resistant pattern of GEN-CHL-CIP-NAL-AMP-SAM-TET-CTX-SXT-FFC-CT. In total, 26 different AMR profiles were recorded among 224 S. Indiana isolates. The most prevalent MDR profiles identified were GEN-CHL-CIP-NAL-AMP-SAM-TET-CTX-SXT-FFC (86/224, 38.39%), followed by GEN-CHL-CIP-NAL-AMP-SAM-TET-CAZ-CTX-SXT-FFC (38/224, 16.96%). The isolates from different regions showed some MDR difference, for instance, the isolates showed co-resistance to 4, 5, and 6 classes of drugs only recovered from Beijing, Jilin, and Jiangsu provinces, while the isolates recovered from Shaanxi (n = 16) and Guangdong (n = 7) provinces showed co-resistance to 7, 8, and 9 classes of drugs.

Antimicrobial resistance genes detected for the 224 CIP-CTX co-resistant S. Indiana were listed in Table 3 and Supplementary Table 1. Five types of amino acid substitution were identified in QRDRs. Among the isolates studied, 164 (73.2%, 164/224) possessed the dominant amino acid substitution in GyrA (S83F and D87N) and ParC (T57S and S80R), along with other substitutions in GyrA (S83F and D87G) and ParC (T57S and S80R) being observed in 56 isolates (25.0%, 56/224). These two allelic types accounted for 98.2% (220/224) of all CIP-CTX co-resistant S. Indiana isolates. Two isolates were identified with amino acid substitutions in GyrA (D87G) and ParC (T57S), while one isolate showed a substitution type of GyrA (S83F) and ParC (T57S). One isolate containing an amino acid substitution in GyrA (S83L and D87N) together with several nucleotide insertions and deletions (InDels) in the parC gene were also found. No amino acid change was observed for GyrB and ParE.

PMQR-encoding genes, especially oqxA, oqxB, and aac (6′)-Ib-cr, were identified in all 224 S. Indiana tested, while qnrS and qepA genes were also detected in 25 (11.2%) and 5 (2.2%) isolates, respectively (Table 3 and Supplementary Table 1). One hundred and ninety-five isolates were harboring only the oqxAB and aac (6′)-Ib-cr genes without any qnr or qepA genes, which represented 87.1% (195/224) of this collection. In addition to the oqxAB and aac (6′)-Ib-cr genes, there were 24 and 4 isolates carrying qnrS gene and qepA gene separately, and one isolate was observed to carry both qnrS and qepA genes, which also possessed two point mutations in QRDRs including amino acid substitutions in GyrA (S83F) and ParC (T57S). The PMQR genes qnrA, qnrB, qnrC, and qnrD were not detected.

Among 217 ESBL positive CIP-CTX co-resistant S. Indiana, we identified five subtypes of bla_CTX-M gene, including bla_CTX-M-65 (n = 165), bla_CTX-M-14 (n = 19), bla_CTX-M-27 (n = 11), bla_CTX-M-28 (n = 9), and bla_CTX-M-79 (n = 8), while five isolates were bla_CTX-M group negative. Of these, isolates with the bla_CTX-M-65 gene ranked the top with a rate of 76.0% (165/217) overall. The QRDRs, PMQR, and ESBLs genotypes of 224 CIP-CTX co-resistant S. Indiana are shown in Table 3 and Supplementary Table 1.

The complete collection was tested for the presence of the mcr-1–mcr-10 genes and only one S. Indiana (0.45%, 1/224) denoted as CFSA664 was found to be positive for mcr-1 without any other mcr genes being detected, and this isolate was also detected as the only isolate exhibiting CTX-CIP-CT co-resistance. This isolate was recovered from a chilled, packaged chicken sample collected from a supermarket in Yangzhou city in Jiangsu province in 2011.

The MIC values and resistance phenotypes against an extended panel of 27 antimicrobial compounds for this isolate were listed in Table 4. The isolate showed an MIC value of 4 mg/L which was a relevant low level of colistin resistance (2–8 mg/L) but was susceptible to ceftriaxone, cefepime, three carbapenem drugs, aztreonam, and tigecycline, while expressing an intermediate resistant phenotype to cefoxitin, nitrofurantoin, and polymyxin B. Further, this isolate was confirmed as an ESBL positive strain and demonstrated an MDR phenotype against ten classes (16 kinds) of antimicrobial compounds.

Based on the genome data, CFSA664 was confirmed as S. Indiana and identified as the MLST type of ST17 by the SISTR platform. Sequencing provided 88,121 reads in total with a mean read length of 8,319 bp and coverage of 85.69x. There were 4,760 coding genes and 176 pseudogenes along with 120 RNA genes (22 rRNAs, 86 tRNAs and 12 ncRNAs) within this genome from the prediction of NCBI PGAP annotation, and it consisted of a single circular chromosome (4,733,813 bp, 52.1% GC content) and three circular plasmids: plasmid pCFSA664-1 (255,327 bp, 47.9% GC content), plasmid pCFSA664-2(41,696 bp, 45.4% GC content), and plasmid pCFSA664-3(61,841 bp 42.4% GC content). The replicon type of plasmid pCFSA664-1 included IncHI2, IncHI2A, IncN, and RepA-pKPC-CAV1321, presenting a multiple Inc. type in the same plasmid; plasmids pCFSA664-2 and pCFSA664-3 were denoted as IncP1 and IncI2 replicon types, respectively. The virulence factors (VFs) of CFSA664 predicted from VFDB database were shown in Supplementary Table 2.

As shown in Table 4, two point mutations were identified in the gyrA and parC genes on the chromosome of S. Indiana CFSA664, and these gave rise to amino acid substitutions of GyrA (S83F, D87N) and ParC (T57S, S80R), mediating the quinolone resistance phenotype detected. A total of 26 antimicrobial resistance-encoding genes were identified in the genome of S. Indiana CFSA664, including two copies of the sul2 gene on the same plasmid pCFSA664-1. No AMR gene was found on the IncP1-type plasmid pCFSA664-2. The aac (6′)-Iaa gene and the mcr-1 gene were the only resistance genes localized to the chromosome and the IncI2 plasmid pCFSA664-3, respectively. All of the other 24 plasmid-mediated antimicrobial resistance-coding genes, including three ESBL genes (bla_CTX-M-65, bla_OXA-1, bla_TEM-1B), were located on the same plasmid pCFSA664-1. Twenty-one of these 24 genes were mapped within an ~56-kbp locus (230,792–31,620 bp, Figure 1) that contained more than 22 mobile genetic elements (MGE) including insertion sequence (IS) elements and transposon CDSs, such as IS3, IS4, IS5, IS6, IS91, and Tn3 family transposase, and also IS6-, IS91-and Tn3-like element (IS26, IS1006, ISVsa3, and TnAs3) family transposase. Additionally, 16 ORFs and 11 hypothetical proteins were also included in this region. BLASTn comparisons indicated that a high similarity (100% coverage value and 99.99% nucleotide identity) of this region was observed mapping to a plasmid pD90-1 (GenBank no. CP022451.1) from an S. Indiana D90, which was isolated from a whole chicken carcass collected in a poultry slaughterhouse in 2012 in Henan, China and carrying four different Inc. type plasmids (HI2/HI2A/N/Q1-, I2-, N/X1-and unknown-Inc types). In terms of the remaining three of the 24 resistance genes [mph (A), aph (3′)-IIa, and tet (A)], they were separated and positioned at ~84-, 88- and 203-kbp, respectively on plasmid pCFSA664-1, with the number of transposase CDSs varying from 1 to 4 (Figure 1). Other than these AMR genes observed in plasmid pCFSA664-1, a ter gene operon (terABCDFWXZ), required for mediating resistance to potassium tellurite, was situated in front of tet (A) gene, positioning at 180–200-kbp (Figure 1).

A total of thirteen plasmids, which were collected from Enterobacteriaceae (E. coli, n = 8; Salmonella, n = 4; Klebsiella, n = 1) isolates from GenBank and had an almost identical backbone structure, were used for structural comparative analysis with plasmid pCFSA664-3 (Figure 2). Plasmid pCFSA664-3 could be differentiated from these other plasmids by an insertion sequence element IS3 transposase, a phage tail protein, and hicA-coding protein. In terms of the mcr-1 gene region, all plasmids demonstrated a high degree of sequence conservation. In plasmid pCFSA664-3, the sequences relevant to shufflon C appeared to be disrupted by IS3 transposase, a feature which was missing in all other plasmids. Plasmid pHNSHP45 (Accession number: KP347127), belonging to the first isolate reported to contain an mcr-1 gene (Escherichia coli strain SHP45), was devoid of three more sequences for encoding DNA primase, DnaJ, and XRE transcriptional regulator. Several conjugal transfer proteins were observed in these plasmids, and the mcr-1 gene was the only acquired gene conferring resistance to a known antimicrobial agent. A zinc transporter gene was also located near the mcr-1 and relaxase locus.

To better understand the genetic environment of the mcr-1 locus of plasmid pCFSA664-3, sequences in different plasmids, belonging to three replicon types, were extracted and compared (Figure 3). This analysis revealed that the mcr-1 genes in three IncI2 type plasmids (pCFSA664-3, pHNSHP45, and pCFSA244-2) were found to be located between a PAP2 family protein-encoding gene (arrowed in yellow) and a relaxase-encoding gene (arrowed in green). In the case of plasmid pHNSHP45, an IS30 family element ISApl1 was followed by its relaxase-encoding gene downstream. In comparison with these three plasmids, three other plasmids used in this study did not have a relaxase-encoding gene upstream of the mcr-1 gene but processed some hypothetical proteins and some ORFs. In terms of plasmid pCFSA664-3 in this study, together with plasmids pCFSA231 (IncX4), pCFSA1096 (IncHI2A/HI2), and pCFSA244-2 (IncI2), the mcr-1 genes were identified locating with a pap2-encoding gene distal to this site, but without any insertion sequences (ISs) coding genes, and this was different when compared with plasmids pHNSHP45 and pCFSA122-1.

In this study, plasmid pCFSA664-3, the mcr-1-carrying plasmid in S. Indiana CFSA664, could transfer into E. coli J53 at relatively high frequencies (1.6 × 10⁻⁴) per recipient (30°C) and 2.2 × 10⁻⁴ per recipient (37°C). This was confirmed following the selection of 5 mcr-1-positive transconjugants (E. coli J53 + pCFSA664-3) by PCR, with MIC values of 4 mg/L to colistin, similar to that described previously for CFSA664.