Prevalence and Genomic Characteristics of mcr-Positive Escherichia coli Strains Isolated from Humans, Pigs, and Foods in China

ABSTRACT Colistin is one of the last-resort antibiotics for treating infections caused by multidrug-resistant (MDR) Gram-negative bacteria. However, mcr genes conferring resistance to colistin have been widely identified, which is considered a global threat to public health. Here, we investigated the prevalence and characteristics of mcr-harboring Escherichia coli strains isolated from humans, animals, and foods in China by PCR, antimicrobial susceptibility testing, conjugation experiments, molecular typing, genome sequencing, and bioinformatics analysis. In total, 135 mcr-1-harboring E. coli isolates were acquired from 847 samples, and 6 isolates carried mcr-3. Among them, 131 isolates were MDR bacteria. Sixty-five resistance genes conferring resistance to multiple antimicrobials were identified in 135 isolates. The diverse pulsed-field gel electrophoresis (PFGE) patterns and sequence types (STs) of mcr-1-carrying isolates demonstrated that clonal dissemination was not the dominant mode of mcr-1 transmission. Seven types of plasmids were able to carry mcr-1 in this study, including IncI2, IncX4, IncHI2, p0111, IncY, and two hybrid plasmids. The genetic structures carrying mcr-1 of 60 isolates were successfully transferred into the recipient, including 25 IncI2 plasmids, 23 IncX4 plasmids, and an IncHI2 plasmid. mcr-1–pap2 was the dominant mcr-1-bearing structure, followed by ISApl1–mcr-1–pap2–ISApl1 (Tn6330) and ISApl1–mcr-1–pap2, among 7 mcr-1-bearing structures of 135 isolates. In conclusion, IncI2, IncX4, and IncHI2 plasmids were the major vectors spreading mcr-1 from different geographical locations and sources. The prevalence of Tn6330 may accelerate the transmission of mcr-1. Continuous surveillance of mcr-1 and variants in bacteria is vital for evaluating the public health risk posed by mcr genes. IMPORTANCE The spread of polymyxin-resistant Enterobacteriaceae poses a significant threat to public health and challenges the therapeutic options for treating infections on a global level. In this study, mcr-1-bearing ST10 E. coli was isolated from pigs, pork, and humans simultaneously, which demonstrated that ST10 E. coli was an important vehicle for the spread of mcr-1 among animals, foods, and humans. The high prevalence of mcr-1-positive E. coli strains in pigs and pork and the horizontal transmission of mcr-1-bearing plasmids in diverse E. coli strains suggest that pigs and pork are important sources of mcr-1-positive strains in humans and pose a potential threat to public health. Additional research on the prevalence and characteristics of mcr-1-positive E. coli is still required to facilitate early warning to improve polymyxin management in hospitals.

Molecular typing. The vast majority of isolates were distributed on different genetic branches according to the pulsed-field gel electrophoresis (PFGE) patterns, and the genetic relationships were distant. Therefore, there was no absolute dominant PFGE spectrum (see Fig. S1 in the supplemental material). In addition, some mcr-1-positive E. coli isolates exhibited genetically similar PFGE types with 100% homology, such as CP9 and CP10, CP18 and CP20, CP103 and CP113, as well as CP111 and CP112 (Fig. S1), all of which were obtained from pig feces, implying that the clonal spread of mcr-1-positive E. coli strains occurred among farmed pigs. The E. coli isolates of the present study were classified using Clermont typing, and the majority belonged to group A (101/135; 74.8%), followed by group B1 (23/ 135; 17.0%). Groups C, D, E, F, and G were also observed in the present study (Fig. 1). The vast majority of isolates from Sichuan clustered into two distinct subclades in the phylogenetic tree based on single-nucleotide polymorphisms (SNPs) of core genomes (Fig. 1). The remaining isolates were scattered across various branches of the phylogenetic tree (Fig. 1). The phylogenetic tree revealed pronounced genotypic diversity among mcr-1-bearing isolates, and clonal spread occurred in samples from the same region and source but was not the dominant mode of mcr-1 transmission (Fig. 1).
Characteristics of mcr-1-bearing plasmids. Conjugation experiments with 135 mcr-1-carrying isolates were performed to determine the transmissibility of mcr-1. The genetic structures carrying the mcr-1 genes of 60 E. coli isolates with polymyxin B resistance phenotypes were successfully transferred into the recipient E. coli strain 26R 793 (Fig. 1). Using draft sequences, we performed a comprehensive analysis of the location of mcr-1. Among mcr-1carrying isolates from Hebei, mcr-1 in 17 isolates was transferred to the recipients, and the mcr-1 gene existed mainly on IncI2 plasmids (n = 12), followed by IncX4 plasmids (n = 2) ( Fig. 1 and Table S1). Among mcr-1-carrying isolates from Henan, mcr-1 in 10 isolates was transferred, and the mcr-1 gene was located on the IncI2 (n = 5) and IncX4 (n = 5) plasmids ( Fig. 1 and Table S1). Among mcr-1-carrying isolates from Sichuan, mcr-1 in 33 isolates was transferred, and the mcr-1 gene was mediated mainly by IncX4 (n = 16) and IncI2 (n = 8) plasmids, followed by IncHI2 plasmids (n = 1) ( Fig. 1 and Table S1). These results showed that IncI2 and IncX4 plasmids were the dominant vehicles responsible for disseminating the mcr-1 gene in different geographical locations and sources through horizontal transfer. However, whether the most dominant vector was IncI2 or IncX4 plasmids varied by region.

DISCUSSION
Numerous studies have reported the prevalence and characteristics of mcr-1-positive Enterobacteriaceae in China, and E. coli was found to be a critical host of mcr-1 in both medical and veterinary settings (4). The positivity rate for mcr-1-bearing E. coli isolates and September 2017 showed that the rate of carriage of mcr-1 was extremely high (457/ 600) (76.2% on average, ranging from 45.0% to 100% in different provinces) (23), which was considerably higher than that observed previously. In the current study, we determined that the positivity rate for mcr-1-carrying E. coli isolates in pig feces in Sichuan was 54.6% (54/99) in 2016. These investigations showed that the prevalence of mcr-1carrying E. coli varied by sampling region, sampling time, and research method. In general, E. coli carrying mcr-1 was highly prevalent in China before polymyxin was banned as an animal feed additive. A significant decrease in the prevalence of mcr-1 was observed after the polymyxin withdrawal policy for animal feed was implemented (16,17). However, continuous monitoring of polymyxin resistance is essential, particularly with regard to early warning for polymyxin management in hospitals.
In the present study, the prevalence of mcr-1-positive E. coli was the highest in pig feces (54.6%), while among retail foods samples, the prevalence of mcr-1-positive E. coli was the highest in pork (32.9%). As one of the most important food-producing animals, pigs are reared in proximity to humans around the world. Therefore, mcr-1-positive E. coli may be transmitted to humans through the food production chain. The high prevalence of mcr-1-positive E. coli in pigs and pork poses a potential threat to public health. Importantly, mcr-1-bearing ST10 E. coli was found in pigs, pork, and humans simultaneously, which demonstrates that ST10 E. coli is an important vehicle for the Plasmids are extrachromosomal DNA elements that can confer new determinants to bacteria for adaptation to new environments (27). mcr-1-bearing plasmids markedly contribute to the prevalence of polymyxin resistance. Diverse types of plasmids can mediate the transmission of mcr-1 in field and clinical isolates, including IncHI2, IncX4, IncI2, IncP, IncY, IncFII, IncHI1, IncFI, IncK2, IncX1, IncFIA, IncFIB, IncR, IncQ1, and hybrid plasmids (IncX1-IncX2, IncI2-IncFIB, and IncX3-IncX4) (20,(28)(29)(30)(31)(32)(33). The global distribution of mcr-1 is mediated mainly by IncHI2, IncX4, and IncI2 plasmids (10,12). An mcr-1-carrying IncHI2 plasmid could impose fitness costs on the host cell, but this cost was largely compensated for after long-term culturing (34). In addition, fitness advantages and high efficiencies of transfer of mcr-1-bearing IncI2 and IncX4 plasmids in E. coli were observed (35). This may explain the prevalence of mcr-1-harboring IncI2, IncX4, and IncHI2 plasmids globally. In addition, mcr-1-bearing IncI2, IncX4, and IncHI2 plasmids are typically conjugative plasmids (12). However, some mcr-1-bearing IncI2, IncX4, and IncHI2 plasmids were not transferred to recipients, which warrants further investigation.
Although mcr-1-bearing plasmids are diverse, the mcr-1 gene is frequently accompanied by the ISApl1 element, which belongs to the IS30 family, surrounded by imperfect terminal inverted repeat sequences (36). It has been reported that the mcr-1 gene can be mobilized by the ISApl1-mediated composite transposon (Tn6330) (37). However, mcr-1-pap2 cassettes lacking ISApl1 or possessing one ISApl1 element upstream have been commonly identified and were demonstrated to be formed by the deletion of ISApl1 from the ancestral Tn6330 element (38). In the current study, mcr-1-pap2 (58/135) was the dominant mcr-1-bearing structure, followed by ISApl1-mcr-1-pap2-ISApl1 (39/135) and ISApl1-mcr-1-pap2 (25/135). The mcr-1-bearing structures on IncX4 plasmids were only mcr-1-pap2 Characteristics of mcr-Bearing Strains in China Microbiology Spectrum cassettes, which was consistent with the results of a previous study suggesting that ISApl1 was presumably involved in the transposition of the mcr-1-pap2 cassette and then was lost (39), and mcr-1 could be further propagated by conjugative IncX4 plasmids. The remaining four structures were considered derivatives of Tn6330. These derivatives may further evolve into mcr-1-pap2 cassettes and spread mcr-1 by plasmids. The high prevalence of Tn6330 may accelerate the transmission of mcr-1 and represents a significant threat to global public health.
Conclusion. In conclusion, IncI2, IncX4, and IncHI2 plasmids were the major vectors for the spread of mcr-1 from different geographical locations and sources, and the prevalence of Tn6330 may accelerate the transmission of mcr-1. The high prevalence of mcr-1-positive E. coli strains in pigs and pork as well as the horizontal transmission of mcr-1-bearing plasmids in diverse E. coli strains suggest that pigs and pork are vital sources of mcr-1-positive strains in humans and pose a potential threat to public health. Continuous surveillance of mcr-1 and variants in bacteria after polymyxin withdrawal is essential for evaluating the public health risk posed by mcr genes. Typical lactose-fermenting colonies growing as red or pink colonies were purified and subcultured onto tryptic soy agar plates for species identification, which was carried out using a Vitek2 compact automatic microbial identification system, a matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) system, and 16S rRNA gene sequencing. These isolates were subsequently screened for the presence of mcr-1 by PCR with specific primers (mcr-1-F [ATCAGCCAAACCTATCCTATCG] and mcr-1-R [ATAGATGTTG CTGTGCGTCTGC]) and Sanger sequencing.
Conjugation experiments. Isolates carrying mcr-1 were subjected to conjugation assays to study the transferability of mcr-1. Briefly, mcr-1-positive isolates were used as donors, and E. coli 26R 793 (resistant to rifampin) was used as the recipient. Cultures of donors and the recipient with a culture density of a 0.5 McFarland standard were mixed at a ratio of 1:1, respectively, and the mixtures were then incubated at 37°C statically. After incubation for 12 to 14 h, we subsequently 10-fold serially diluted the mixtures in sterile saline and aliquoted 100 mL of the diluted culture onto selective LB agar plates containing polymyxin B (4 mg/mL) and rifampin (100 mg/mL). The mcr-1-positive transconjugants were screened by PCR and polymyxin B resistance phenotypes.
Genomic DNA sequencing and bioinformatic analysis. The genomic DNAs of mcr-1-harboring isolates were extracted using the FastPure bacterial DNA isolation minikit (Vazyme, Nanjing, China) according to the manufacturer's recommendations. Whole-genome sequencing was performed on the Illumina HiSeq X ten platform to acquire short-read data. The short-read Illumina raw sequences of mcr-1-bearing isolates were separately assembled using SPAdes (41). The plasmid replicons and AMR genes were analyzed using PlasmidFinder and ResFinder (https://www.genomicepidemiology.org/services/). We determined the location of mcr-1 by analyzing the replicon types on the contigs where mcr-1 resides. In addition, the sequences coharboring mcr-1 and replicons could be used as the reference plasmid sequences to understand the localization of mcr-1 in other bacteria. The core-genome MLST allelic profiles of E. coli were built using phyloviz (42). The phylotyping of E. coli was performed using clermont.py software (https://github.com/A-BN/ClermonTyping). The draft genomes were annotated using Prokka (43). A pangenome analysis was conducted on the mcr-1-bearing E. coli isolates using Roary (44). The phylogenetic tree of all mcr-1-positive isolates was constructed using FastTree based on SNPs of core genomes (45).
The phylogenetic tree was visualized by using iTOL (https://itol.embl.de/). TBtools was used to visualize the distributions of AMR genes (46). BRIG was used to generate a plasmid comparison map (47).
Data availability. The whole-genome sequencing data for 129 mcr-1-positive isolates can be found in the NCBI database under BioProject accession number PRJNA560609.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.3 MB. We declare no conflicts of interest.