The clinical Salmonella isolates were collected from different hospitals in Sichuan province, China between 2014 and 2017. The Salmonella isolates were recovered from patients following the previous methods (Xia et al., 2009). The mcr-positive isolates were screened by the multiplex PCR method and confirmed by Sanger sequencing (Rebelo et al., 2018). All mcr-positive isolates were stored at -80°C in LB broth containing 20% glycerol for further investigation.

The MICs (minimum inhibitory concentrations) of mcr-positive isolates against different antimicrobials, including kanamycin, ciprofloxacin, meropenem, ampicillin, colistin, enrofloxacin, tetracycline and ceftiofur, were tested using broth microdilution as per Clinical and Laboratory Standards Institute (CLSI) standards (CLSI, 2020). E. coli ATCC25922 was used for quality control. In order to test the transferability of mcr genes, conjugation experiments were performed using E. coli C600 (Rif^R) as recipients. Briefly, the donor and recipient strains were grown in LB broth until they reached the logarithmic growth phase, then mixed at the ratio of 1:1 and cultured overnight on LB agar plates. The transconjugants were screened on LB agar plates with rifampin (300 mg/L) and colistin (2 mg/L), and validated using PCR targeting the mcr genes and 16S rDNA gene sequencing.

In order to decipher the genetic structure features of mcr genes in these Salmonella isolates. We first extracted their genomic DNA using FastPure Bacteria DNA Isolation Mini Kit (Vazyme™, China) according to the instructions of the manufacturer. The purity and concentration of genomic DNA were evaluated using the Titertek-Berthold Colibri (Berthold™, Germany) and the Qubit^R Fluorometer (Thermo Fisher™, US) respectively. Then, extracted genomic DNA was sequenced using both short-read and long-read sequencing methods. Short-read DNA sequencing was conducted on the BGISEQ-50 platform with the SE50 strategy. The long-read sequencing was performed on ONT (Oxford Nanopore Technologies) MinION platform using the SQK-LSK109 library preparation kit in R9.4 flow cells.

The complete genomic sequences of mcr-positive Salmonella isolates were obtained using Unicycler (Wick et al., 2017) with hybrid assemble strategy based on short- and long-read data. An online fully-automated service, RAST was used to annotate the complete genomes (https://rast.nmpdr.org/). Antimicrobial resistance genes (ARGs), insertion sequences (ISs) and plasmid replicon genes were detected using abricate tool (https://github.com/tseemann/abricate) based on NCBI AMRFinderPlus, ISFinder and PlasmidFinder databases. The plasmid data were downloaded from NCBI RefSeq database (https://ftp.ncbi.nlm.nih.gov/genomes/refseq/plasmid/). The Salmonella serovars were identified using SISTR (Yoshida et al., 2016). The multi-locus sequence type (MLST) was analyzed using an mlst tool (https://github.com/tseemann/mlst). BRIG v.0.95 and Easyfig v.2.2.3 were used to generate plasmid comparison maps (Alikhan et al., 2011; Sullivan et al., 2011).

In total, 12 mcr-positive isolates (ten mcr-1 and two mcr-3) were detected from 1046 clinical Salmonella isolates using PCR. The prevalence and basic genomic features have been identified in our previous study (Li et al., 2022). For the 12 mcr-positive S. enterica, five isolates each year were detected in 2016 and 2017. The remaining two isolates were found in 2014. No mcr-positive isolates were detected in 2015. Antimicrobial susceptibility testing showed that the 12 mcr-positive S. enterica isolates were all resistant to colistin but sensitive to meropenem. In addition, the majority of them were resistant to ampicillin, tetracycline, aztreonam and kanamycin, whereas they were almost all sensitive to ciprofloxacin and enrofloxacin ( Table 1 ). The majority of 12 mcr-positive S. enterica isolates showed multidrug resistance against three or more types of antimicrobials, designated as MDR isolates. According to the conjugation assay, the colistin resistance phenotype of the 12 S. enterica isolates could be transferred into E. coli C600. Antimicrobial susceptibility testing showed that all transconjugants were resistant to colistin. In addition, majority of transconjugants displayed resistance to other antimicrobials ( Table 1 ). This indicated that the resistance genes in donor S. enterica were probable carried by conjugative MDR plasmids.

The genetic contexts of mcr-1 in the ten mcr-1 positive Salmonella isolates were classified as five types ( Figure 1 ). The genetic structures of mcr-1 in IncX4 plasmids were identical. No mobile elements were detected around the mcr-1 gene in IncX4 plasmids. In IncHI2 plasmids, four different mcr-1 genetic structures were found. Apart from plasmid pSC2017167-mcr-256k, we found an ISApl1 in the upstream of mcr-1 in the other IncHI2 plasmids. In addition, we detected a reversed ISApl1 in the downstream of mcr-1 in plasmid pSC2017297-mcr-249k. ISApl1 was an important element associated with the rapid dissemination of mcr-1, which is usually located in both upstream and downstream of mcr-1 forming an ISApl1-mcr-1-pap2-ISApl1 (Tn6330) transposon (Snesrud et al., 2018). Another study demonstrated that mcr-1 genetic structures with only one ISApl1 or with ISApl1 absent were formed by deletion of ISApl1 from the ancestral Tn6330 (Snesrud et al., 2018). We detected truncated Tn6330 in many mcr-1-bearing plasmids, suggesting that the genetic structures of mcr-1 in these plasmids have evolved for a long time.

The genetic structures of mcr-3 in plasmids pSC2016091-mcr-116k and pSC2016290-mcr-147k were similar and shared a common core structure tnpA-mcr-3-dgkA-ISKpn40 ( Figure 1 ). In plasmid pSC2016091-mcr-116k, we found a truncated IS26 present upstream and an intact IS26 present downstream of mcr-3. Only an intact IS26 present upstream of mcr-3 was detected in plasmid pSC2016290-mcr-147k. Similar to the mobilization of mcr-1 medicated by Tn6330, the mobility of mcr-3 is usually associated with a composite transposon structure ISKpn40-mcr-3.1-dgkA-ISKpn40 (He et al., 2020). Apart from ISKpn40, the highly mobilized IS26 around mcr-3 deserves our attention, as it may facilitate the horizontal translocation of mcr-3.

Long-read nanopore sequencing and short-read Illumina sequencing were performed on the 12 S. enterica isolates. The complete chromosomes and plasmids of these isolates were obtained using a hybrid assembly strategy and single-read analysis, as described in our previous study (Li et al., 2018). According to the assembly results, we found two chromosomally integrated IncHI2 plasmids from isolates SC2016025 and SC2016090 as a similar phenomenon previously reported ( Figures 2A, D ) (Chang et al., 2022). Similar mcr-1-bearing IncHI2/HI2A plasmids were found in other mcr-1 positive isolates from our investigation including SC2017030, SC2017100, SC2017297 and SC2017167. Moreover, these mcr-1-bearing IncHI2/HI2A plasmids were also found in NCBI nr database by BLASTn analysis ( Figure 2A ). Almost identical mcr-1 bearing plasmids were simultaneously detected in different genus, different geographical locations, different sources of isolates, implying that such plasmids play a crucial role in the horizontal transmission of mcr-1.

Comparative analysis found that there are some structure differences in these mcr-1-bearing IncHI2/HI2A plasmids from different isolates. A replicon gene of IncN plasmids was absent in plasmids pSC2017100-mcr-218k, pSC2017297-mcr-249k and pSC2017030-mcr-251k. In plasmid pSC2017100-mcr-218k, a region encoding multiple resistance genes containing cmlA1, floR and sul2 was lost ( Figure 2A ). The evolution of this type of plasmids in different isolates may result in their structural changing. IncHI2/HI2A plasmids with mcr-1 are commonly detected in E. coli (Lu et al., 2020). This type of mcr-1 positive plasmid has recently been found frequently in Salmonella (Vazquez et al., 2021), indicating that the mcr-1 gene has the ability to transfer to other species of bacteria with their assistance. We should pay more attention to monitor the spread of such mcr-1-bearing plasmids.

We identified three mcr-1-bearing IncX4 plasmids in isolates SC2014238, SC2016042 and SC2017057. The three IncX4 plasmids shared 100% identity and 100% coverage with each other. In addition, there are many IncX4 plasmids from other genus of bacteria that are identical to the three plasmids ( Figure 2B ). Like IncHI2/HI2A plasmids, IncX4 plasmids were also a common plasmid type carrying mcr-1, because they are self-transferable with high conjugation frequencies (Sun et al., 2017). Apart from IncHI2/HI2A and IncX4 plasmids, an IncI2 plasmid harboring mcr-1 with 62kb in length was identified in isolate SC2014107 ( Figure 2C ). The IncI2 plasmids have recently been noticed due to they are usually associated with the transmission of ESBLs genes and mcr-1 (Wong et al., 2015). A previous investigation showed that IncI2 plasmids were the most common type of mcr-1-carrying plasmids (Ricker et al., 2022). In addition, we found many mcr-1-bearing IncI2 plasmids from different bacteria in nr database ( Figure 2C ), indicating that the IncI2 plasmids were broad-host-range plasmids.

Apart from mcr-1, another mcr variant mcr-3 was detected in isolates SC2016090 and SC2016091. Whole genome analysis of the two isolates found that the mcr-3 gene was located on plasmids. In isolate SC2016090, mcr-3 was carried by an IncC plasmid pSC2016290-mcr-147k with a size 147kb. Besides, the plasmid also carried bla _CTX-M-55, bla _TEM-1B, catA2, aph(6)-Id, tet(A) and floR. We noticed that a few of such mcr-3-bearing plasmids have been detected in other Salmonella isolates ( Figure 3A ). Of note, the plasmid pSC2016290-mcr-147k showed a partial backbone similar to an E. coli plasmid pECAZ155_KPC carrying the carbapenemase gene bla _KPC-2. This phenomenon implied that the plasmid pSC2016290-mcr-147k could integrate into other plasmids, causing the mcr-3 gene to spread to other bacteria. Another mcr-3-bearing plasmid found in isolate SC2016091 was an IncFIB type plasmid pSC2019091-mcr3-116k with 116kb. Coincidently, the plasmid also carried ESBL gene bla _CTX-M-55 ( Figure 3B ). Furthermore, the IncFIB plasmids belonged to a type of broad host plasmids, and there are numerous plasmids from various bacteria sharing a backbone comparable to the IncFIB plasmid.

In order to further investigate the prevalence and plasmid backbone structures of mcr-positive plasmids, all plasmid data in the RefSeq database were downloaded and utilized for comparative analysis of plasmid backbone structures. Through BLASTn analysis, a total of 136 complete plasmids encoding mcr were identified among over twenty-seven thousand plasmids in database. The plasmid replicon types of plasmids were further determined using the plasmidfinder database. The major of mcr-1 variants identified in both the database and this study was found in plasmids of replicon types IncX4 (34/122), IncI2 (49/122), and IncHI2/HI2A (31/122), and all plasmids of those types carried only mcr-1, suggesting those plasmids has a strong correlation with mcr-1 horizontal transfer. Additionally, eleven plasmids encoding mcr-3 identified in the database and this study, belonging to four different replicon types, including IncFII (5/11), IncC (4/11), IncI1 (1/11), IncFII/FIB (1/11). Furthermore, the plasmids carrying mcr-8, with replicon types IncFII (1/4), IncFII/FIA (2/4) and IncFII/FIB (1/4), and eight additional plasmids carrying mcr-10 was determined to be of replicon type IncFII/FIB (5/8), IncFII (1/8), IncFII/FIA (1/8) and IncFII (1/8).

The main hosts of mcr-positive plasmids in the RefSeq database were identified as Salmonella enterica (18/136) and Escherichia coli (96/136).

The plasmids from both the database and this study were merged for further investigations. Among these plasmids, those belonging to replicon types IncX4 and IncI2 were most abundant, often in the size range of approximately 30kb and 60kb, respectively, followed by IncHI2A/HI2 type plasmids mainly distributed between 220kb to 270kb in length ( Figure 4 ). Alignment of plasmid backbone structures revealed that, aside from IncFII/FIB replicon type mcr-positive plasmids, those with the same plasmid replicon types displayed similar backbone structures ( Supplement Figure 1 ). These plasmids with similar structures were found in different host bacteria especially Escherichia coli, indicating the horizontal transmission of mcr-positive plasmids carrying different mcr variants among different hosts.

To investigate the formation mechanism of the integrated plasmid, we employed third-generation DNA sequencing to obtain accurate genome structures of strain SC2016090. The chromosome of SC2016090 was 5,247,576-bp in length, exhibited a GC content of 51.9% and comprised 4,927 predicted coding sequences (CDSs). Comparative analysis using BLASTn revealed that the chromosome of SC2016090 showed remarkable similarity (93% coverage, 99.97% identity) to the chromosome of Salmonella enterica strain ST101 (CP050731.1), isolated from a diarrheal patient in Shanghai. However, the SC2016090 chromosome contained an extra 261,335 bp DNA fragment with 46.8% GC content ( Figure 5 ) that was not present in the ST101 chromosome.

Further investigation revealed that this extra fragment encompassed 249 CDSs and displayed 99% identity with plasmids pMCR_NDM (CP049111.1) and pSI_16E242 (ON960347.1), with coverage of 95% and 99%, respectively. Notably, within the extra DNA fragment of strain SC2016090, several others resistance genes, including bla _CTX-M , sul3, aadA2, and fosA were detected, IncHI2 replicon was also coding in the extra DNA fragment. Furthermore, a comprehensive analysis of the extra DNA fragment unveiled the presence of numerous mosaic genetic elements, including the clustered Tn26. Interestingly, we found the CDS for a D-serine transporter (DsdX, QIU08742.1) was truncated by insertion of the extra DNA fragment. Additionally, the presence of the ISApl1 in both areas of the fragment’s ends and the reverse repeating base sequence located on both sides of the extra plasmid segment, suggested the involvement of ISApl1 in the insertion of the integrated plasmid ( Figure 5 ).